(Oxford Political Philosophy) Sobel, David - Vallentyne, Peter - Wall, Steven-Oxford Studies in Political Philosophy. Volume 2-Oxford University Press (2016)

OUP CORRECTED PROOF – FINAL, 03/02/16, SPi
OXFORD STUDIES IN POLITICAL PHILOSOPHY
Dictionary: NOSD
Dictionary: NOSD
Oxford Studies in
Political
Philosophy
Volume 2
Edited by
D AV I D S O B E L, PE T E R VA L L E N T Y N E,
A N D S T EV E N WA L L
Dictionary: NOSD
3
Great Clarendon Street, Oxford, OX2 6DP,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© the several contributors 2016
The moral rights of the authors have been asserted
First Edition published in 2016
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University Press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number: 2015946693
ISBN 978–0–19–875962–1 (Hbk.)
ISBN 978–0–19–875963–8 (Pbk.)
Printed in Great Britain by
Clays Ltd, St Ives plc
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
Dictionary: NOSD
Acknowledgments
This is the second volume of the Oxford Studies in Political Philosophy. The
chapters assembled here were first presented as papers at a workshop at the
University of Missouri, Columbia in September 2014. We would like to
thank all those who attended this event, with special thanks to Alex Howe,
who oversaw most of the organization. All of the chapters in this volume
were reviewed by referees, most of whom serve on the editorial board of
Oxford Studies in Political Philosophy (see <http://oxfordstudies.arizona.edu/
oxford-studies-political-philosophy>). We very much thank these referees
for their efforts in helping to make this second volume a success. Travel by
editor, Steven Wall, to the workshop at which the papers were presented was
made possible through the support of a grant from the John Templeton
Foundation. The opinions expressed in this publication are those of the
authors and do not necessarily reflect the views of the John Templeton
Foundation. Finally, we would like to express our gratitude to Peter
Momtchiloff for supporting this series and for his expert guidance.
Dictionary: NOSD
Dictionary: NOSD
Contents
List of Contributors ix
Introduction 1
Peter Vallentyne
Part I: Ideal Theory

1 Just and Juster 9
David Estlund
Part II: The Moral Assessment of States

2 Political Rule and Its Discontents 35
Niko Kolodny
3 Consent and Political Legitimacy 71
Amanda Greene
4 The Value of Self-Determination 98
Anna Stilz
5 Domination and the Rule of Law 128
Assaf Sharon
6 Elitism 156
Richard Arneson
7 Is the Gendered Division of Labor a Problem of Distribution? 185
Gina Schouten
Part III: Issues in Social Relations

8 Terms of Trust 209
Daniel Attas
Index 235
Dictionary: NOSD
Dictionary: NOSD
List of Contributors
Richard Arneson is Professor of Philosophy at the University of

California, San Diego.
Daniel Attas is Professor of Philosophy at the Hebrew University of
Jerusalem.
David Estlund is Lombardo Family Professor of the Humanities at Brown
University.
Amanda Greene is Lecturer in Philosophy at University College London.
Niko Kolodny is Professor of Philosophy at the University of California,
Berkeley.
Gina Schouten is Assistant Professor of Philosophy at Illinois State
University.
Assaf Sharon is Assistant Professor of Philosophy at Tel Aviv University.
Anna Stilz is Associate Professor of Politics at Princeton University.
Dictionary: NOSD
Dictionary: NOSD
Introduction
Peter Vallentyne
Since its revival in the 1970s, political philosophy has been a vibrant field
in philosophy, one that intersects with political theory (in political
science), jurisprudence, normative economics, and just war theory. The
Oxford Studies in Political Philosophy series aims to publish some of the
best contemporary work in political philosophy and these closely related
subfields. The chapters in this volume have been grouped into three
topical areas: ideal theory, the moral assessment of states, and issues in
social relations.
Part I examines ideal theory, which is roughly normative (e.g. moral)
theory that addresses normative assessment under idealized conditions. For
simplicity, we focus on justice. Rawls (1999, 2001), for example, holds that
full justice is based on the idealizing assumptions that (1) individuals fully
comply with the normative demands (strict compliance), and (2) external
circumstances (e.g. material wealth, social relations) are favorable to securing/
realizing justice.1 An important question is how ideal justice is related to
practical justice, which makes assessments relative to people’s actual choice
dispositions and their actual social and material circumstances.
A related issue is that justice can be understood as deontic assessment
(what is permitted by justice) or as axiological assessment (e.g. what states
of affairs are at least as just as others). Practical deontic assessment evaluates
what is permitted by justice, relative to what is feasible in the actual choice
situation, whereas practical axiological assessment ranks states of affairs
relative to that choice situation. Ideal deontic assessment evaluates what is
permitted by justice relative to idealized choice situations, whereas ideal
axiological assessments rank states of affairs relative to the idealized choice
situations.
1
For insightful discussion of ideal theory, see Stemplowska and Swift (2012).
Dictionary: NOSD
2 Peter Vallentyne
Amartya Sen (2006, 2011) has argued that ideal justice, which he
implicitly understands as ideal deontic justice, is practically irrelevant. If
ideal deontic justice is feasible, then it will be the most just feasible option.
If it is not feasible, then it is not relevant for choice. Those seeking justice
need to know what the most just feasible options are, but they don’t need to
know what is ideally just. Moreover, we can add that ideal axiological justice
is also practically irrelevant, since it too does not take the current
circumstances (and history) into account.
In his chapter within Part I, entitled “Just and Juster,” David Estlund
argues against Sen that there is an important role for threshold (i.e.
deontic) concepts of justice which cannot be filled by the comparative
axiological relation of at-least-as-just. Even if comparative assessments of
justice are sufficient for making choices, the generation of adequate
comparative assessments, he argues, seems to require threshold (deontic)
assessments.
Part II of this volume addresses the moral assessment of states. This can
be understood in many ways. First, there is the object of assessment: the
basic constitution, particular laws, state practices, or particular state
actions. For simplicity, let us just refer to the state. Second, there are
different kinds of moral assessment (and here I’ll focus on deontic
assessment). One is whether a given state is morally permissible (ideally or
practically). A second is whether the state is just. Unfortunately, “justice”
is understood in different ways by different authors: moral permissibility,
interpersonal permissibility (which leaves out impersonal wrongs),
enforceable duties, fairness (a particular moral concern, typically desert-
based), etc. So the content of such claims always needs to be clarified. A
third kind of moral assessment is whether the state is legitimate in the
sense that others (member or outsiders) are not permitted to forcibly
interfere with the state’s actions (etc.). This may be because its actions are
permissible or just, or because they are not sufficiently impermissible or
unjust to justify forcible interference. Unfortunately, “legitimacy” is also
understood in different ways by different authors (e.g. as permissible or
just use of force to enforce directives). So, here too, clarification is always
needed when the term is used.
A fourth kind of moral assessment is whether the state has political
authority in the sense that it has a moral power, by issuing dictates, to create
at least pro tanto moral duties on the part of its members to comply with
those dictates. One can hold that a state is permissible and perfectly just
without holding that it has any political authority.
Related to all of the above is a state’s right to rule, which can be understood
as consisting of some combination of the above four assessments. The
Dictionary: NOSD
Introduction 3
strongest form of a right to rule is that the state is permissible, just,

legitimate, and has political authority. Obviously, weaker forms are possible.
In the chapter entitled “Political Rule and Its Discontents,” Niko
Kolodny briefly considers various potentially problematic features of the
state and then focuses on the imposition of deterrents for violations of state
directives. The most serious problem, he argues, is that such imposition
violates a deontological constraint on using force even to achieve a greater
good. He argues that there is no relevant moral difference, at least for
sufficiently democratic states, between imposing deterrents for the violation
of natural prohibitions (e.g. natural moral rights) and imposing deterrents
for the violation of state directives. Thus, if, as many believe, the former are
just, then so too are the latter. The crucial question, of course, is whether the
two are morally equivalent.
In her chapter, “Consent and Political Legitimacy,” Amanda Greene
addresses the topic of the legitimacy of a state in the sense of having
“the appropriate standing to exercise power over its subjects.” She argues
that both the contractualist view (based on hypothetical consent) and the
voluntarist view (based on actual consent) involve unacceptable idealizations.
She then develops and defends the sovereignty conception, according to
which a regime is legitimate insofar as it achieves actual quality consent
to rule. Quality consent obtains when a subject consents to her state on
the basis of a judgment of governance success, provided that the judgment
does not conflict with the government’s minimal aim, i.e. basic security
for all subjects. She argues that a state comes to be legitimate by governing
in such a way as to be widely recognized as doing so successfully by its
subjects.
Anna Stilz, in her chapter “The Value of Self-Determination,” develops
and defends an account of self-determination as necessary for full legitimacy
and for being a morally ideal state. She argues that the value of collective
self-determination is rooted in a fundamental interest in appropriately
seeing oneself as a coauthor of the institutions that govern one’s own life.
When citizens affirm their participation in a self-determining political
group, they can relate to demands imposed by their coercive institutions as
self-imposed, and not a problematic restriction of their political freedom.
She further argues that, for a subgroup to have a right to self-determination,
it must show (a) that their aims are consistent with basic justice, (b) that
they do not involve the unjustifiable coercion of others, and (c) that a
feasible institutional alternative is available that would secure greater
collective self-determination for them at reasonable cost.
In the chapter “Domination and the Rule of Law,” Assaf Sharon argues
that contemporary republicanism is mistaken in its claim that the rule of
Dictionary: NOSD
4 Peter Vallentyne
law is compatible with individual liberty (and non-domination), because it

is non-arbitrary. He considers three definitions of non-arbitrariness—in
terms of consent, interests, and control—and argues that the rule of law
does not satisfy any of them. Finally, he argues that the republican idea
of eliminating discretionary power is not always desirable or essential to
political freedom.
In Richard Arneson’s chapter, “Elitism,” he discusses political elitism,
understood as the claim that those who know the political truths needed for
correct (e.g. just) public policy choice, and they alone, are entitled to a share
of political rule. He also discusses a weaker form that allows that all have a
right to a share of political rule, but those with greater relevant knowledge
have a right to a greater share. He argues against several purported
justifications for elitism and argues that elitism might, under the right
conditions, be justified by appeal to a (perhaps limited) moral duty to
cooperate with others to promote the fulfillment of justice.
In the final chapter of Part II, “Is the Gendered Division of Labor a
Problem of Distribution?,” Gina Schouten addresses the legitimacy (roughly,
the permissibility of using collective political power or force) of political
interventions to change the gendered division of labor (e.g. of unpaid
housework and childcare). Interventions such as work time regulation,
subsidized dependent care provisions, and paid family leave initiatives are
promising approaches, but they appear to violate a basic liberal requirement
that exercises of political power be publicly defensible within the justificatory
community of reasonable citizens. This in turn requires that the intervention
must be neutral among the conceptions of the good that citizens may
reasonably embrace. Although Schouten believes that gender egalitarian
interventions can be so neutral and be legitimate, she argues against the
view that they can be neutrally justified as necessary means to remedy unjust
distributions.
Part III of this volume addresses issues with respect to social relations
that, on some views, are relevant to the justice, legitimacy, authority, or
moral attractiveness of a state. There is one chapter in this Part, and it
addresses the development of trust (which is necessary for social cooperation).
In “Terms of Trust,” Daniel Attas analyzes the concept of trust and ways
of promoting it or of undermining it. He argues that the promotion of
empathy, fairness, and reciprocity can be an effective way of promoting
trust. More generally, he argues that creating non-prudential (e.g. moral)
reasons to reciprocate trust is typically more effective than merely creating
prudential reasons (e.g. guarantees, incentives, and sanctions). Indeed, he
argues that the latter can undermine trust.
Dictionary: NOSD
Introduction 5
Bibliography
Rawls, John (1999). A Theory of Justice, rev. ed. (Cambridge, MA: Harvard University
Press).
Rawls, John (2001). Justice as Fairness: A Restatement, Erin Kelly (ed.) (Cambridge,
MA: Harvard University Press).
Sen, Amartya (2006). “What Do We Want From A Theory of Justice?” Journal of
Philosophy 103: 215–38.
Sen, Amartya (2011). The Idea of Justice (Cambridge, MA: Harvard University Press).
Stemplowska, Zofia and Swift, Adam (2012). “Ideal and Non-Ideal Theory, in D.
Estlund (ed.), The Oxford Handbook of Political Philosophy (Oxford: Oxford
University Press), 373–89.
Dictionary: NOSD
Dictionary: NOSD
PA RT I
I D E A L T H E O RY
Dictionary: NOSD
Dictionary: NOSD
1
Just and Juster
David Estlund
Does it matter whether anything is just or unjust? Or does it matter only

which things are juster than other things?1 Amartya Sen asks, in the title of
an important paper, “What Do We Want from a Theory of Justice?”
Certainly, one thing we want is the ability to assess real available alternatives
for their comparative degrees of justice in order to make good social choices.2
For that practical purpose, we would not need a theory that gives any
meaning to the distinction between just and unjust. As long as a theory
allowed us to order the alternatives as if they lay on a line from less just to
more, we could pursue not only greater justice but also greater “expected
justice” (as I will explain). Then there is, arguably, no further practical value
in a threshold—a “partition”—between just and unjust. For that particular
purpose it does not matter whether slavery is unjust.3
While such a comparative theory would suffice for purposes of choice, it
seems like a disadvantage that it would not find any legitimate meaning in
1
In addition to those mentioned in footnotes, I am grateful to Gustaf Arrhenius, Jerry
Gaus, and Amartya Sen for conversations about a forerunner paper, “The Best and the
Rest.” For helpful comments on earlier drafts of the present incarnation, thanks to Ian
Carter, Peter Vallentyne, and an anonymous referee. I also benefitted from discussions at:
a conference on Amartya Sen’s book, The Idea of Justice, at Rutgers University April 2011;
“Workshop on Feasibility,” Jesus College, Oxford, March 2011; “Workshop on Political
Philosophy,” Vejlefjord, Denmark, 2011; Conference for Oxford Studies in Political
Philosophy, September 2014, University of Missouri.
2
Some readers may prefer to substitute degrees of severity of injustice for degrees of
injustice, leaving the latter as binary. I don’t believe that will affect any of the issues
discussed later.
3
The term “partition” is from Sen (2006). It denotes the boundary between just and
unjust, and is not to be confused with different meanings for that term in mathematics
and set theory. “Threshold” wouldn’t risk that confusion, but lacks a good adjectival form
in place of “partitioned.” “Binary” doesn’t refer to the boundary. So I stick with Sen’s
term. The central questions of this chapter were raised for me by his challenging article.
This is not a full response to Sen’s own approach to justice, but an attempt to grapple with
some central challenges for and from a comparativist approach to the theory of justice.
Dictionary: NOSD
10 David Estlund
the statement that slavery is unjust, since that judgment entails a partition.
If rendering such judgments meaningful, at the very least, is also something
we want from a theory of justice, then we want more than a theory that
would guide practical choice. My question is whether there are, nevertheless,
good reasons for biting this comparativist bullet—for resisting any role for
the partition and accepting that “slavery is unjust” is loose talk with no
legitimate meaning. I will argue that there are not sufficiently good reasons
for paying that price. Intuitive convictions that entail a justice partition
would be an important resource in developing a rich account of comparative
judgments of injustice, and there is no strong reason not to develop a
systematic theory of the partitioned standard—the standard of full justice.
For the sake of argument I will here accept the methodological reduction
of normative theorizing about justice to the theory of practical social choice
in the following way: the most just thing, from a set of alternatives, is defined
as whatever society should, morally speaking, choose.4 And for purposes of
rational choice there is no value in information about a partition so long as we
have a (rich) comparative measure. In a strong form, this would seem to
suggest that any alleged content of normative principles that has no role in
practical choice is pointless philosophical musing. I do not accept that view,
but my challenge to comparativism will accept this kind of practicalism for
the sake of argument.5
One argument against employing the just/unjust partition is, as I have
said, that it is of no use for purposes of choice between alternatives. I will
call this argument the sufficiency of comparisons. I will leave until the very
end a second reason for doubting the value of a partition between just and
unjust, namely the claim that full justice is not for this world, while practical
choices between unjust and less unjust conditions face us all the time.
Whatever full justice might consist in, the energy of political philosophers
is better spent, on this view, contributing to clear thinking about the
important comparative choices societies actually face. Let us call this
argument the practicality of comparisons. I will have to leave that line of
argument aside here, and I mention it only to mark clearly the difference
between this and the argument from the sufficiency of comparisons. In
both cases, the target is what I will call partitionism about justice: theorizing
about justice in a way that relies on a partition between just and unjust
social structures, and not (merely) on comparisons between the just and the
juster. A binary or bare partitionism would include only the partition (and
the comparisons that it supports) but no further comparisons with respect
4
Slightly more elaborately, there might be ties for “most just,” in which case society
should choose one of those.
5
I explore challenges to practicalism in Estlund (2011).
Dictionary: NOSD
Just and Juster 11
to relative justice or injustice. A rich partitionism would include both a

partition and further comparisons. The approach that eschews the partition
altogether is, as I have said, comparativism (weak or strong).6
Practicalism about justice has an advantage, namely that it is
straightforwardly an answer to the fair question, “what good would such
knowledge be?” If knowing the partition adds no practical value to the
comparative information in a theory of justice, it is fair to ask whether it adds
anything of value at all. More starkly, if justice should admit of only the
partition but no comparative information, we might wonder what good it is
to understand justice at all. We will see how, even if partition information is
of no use for purposes of choice over and above comparative information, this
does not establish that a wholly comparative theory of justice is adequate—
even for purposes of choice. The reason is that epistemology might favor the
richer measure, delivering a partitioned scale, or at least partition-entailing
judgments, in the first instance. In short, my argument will be that even if
practicalism were correct, and even though (as I will grant) comparisons are
sufficient for choice, the enterprise of a theory of justice does not get to take
such comparisons as given, but it must generate many of them from its own
resources. A wholly comparative conception of justice would limit itself to
impoverished resources, and for no good reason.
1. CATEGORICAL AND METHODOLOGICAL

COMPARATIVISM
Suppose that, while there is no appeal to any partition, we have good

warrant for more than merely an ordinal ranking, but also for an ordering
of degrees of justice on an interval scale. That is, we could not only rank
states of affairs as to which is juster, but we could compare pairs according
to which represented a greater difference in justice, and by what proportion.
That is, we could say not only that A is juster than B, but also that the
difference between A and B is greater than that between A and C. But
suppose we were not in any position to represent how distant any
arrangement was from full justice. This intermediate “interval” measure of
justice would still be highly limiting intuitively, since it would not allow any
6
The idea of a partition between just and unjust is only one kind of salient dividing
point that might be present in an ordering of relative justness. I will not explore the
possibilities, but will suppose, for simplicity, that if there is a partition, there is only one,
the division between just and unjust. Also, I allow and discuss the possibility of rankings
within the unjust category, although for simplicity I leave aside the question whether
there might be orderings above the threshold as well (a kind of supererogatory justice).
Dictionary: NOSD
12 David Estlund
meaning to be assigned to judgments such as that slavery is profoundly

unjust, since that judgment entails a partition. Moreover, the interval
ranking does not assign any meaning to the judgment that slavery is unjust
at all. That is meaningless if there is no partition between cases that are just
and cases that are unjust. I think this is such a big cost, intuitively, that I will
regard this as decisive against what I will call categorical comparativism, the
claim that justice does not (in reality or “metaphysically”) admit of richer
comparison than an interval scale allows: there is no partition between just
and unjust societies. On that view, which I will now put aside, there is only
“juster,” to varying degrees allowed by an interval scale. It denies that a
society organized around slavery (or any society at all, for that matter) is
either just or unjust.
There remains a more nuanced kind of philosophical reticence about the
just/unjust partition, and I will call it methodological comparativism. While
there is or may be a partition in fact, we ought to theorize wholly in terms
of comparisons.7
Here is an argument, which I will eventually reject, that methodological
comparativism would follow from the sufficiency of comparisons, combined
with methodological practicalism:
1. We ought to theorize only in whatever ways would best facilitate
rational social choice (methodological practicalism).
2. Comparisons are fully sufficient for rational choice (sufficiency of
comparisons).
3. We ought to theorize wholly in terms of comparisons, not in terms of
a partition between just and unjust (methodological comparativism).
As noted earlier, I am accepting methodological practicalism for the sake of
argument. The same goes for the second premise, the sufficiency of
comparisons, although I would like to sketch what I take to be a strong case
for its truth. Then, if all we want from a theory of justice is the structure and
information needed to make the most rational social choices it may seem
that there is no reason to build a partition into our working theory.8 I will
argue that this conclusion, methodological comparativism, does not follow
from the sufficiency of comparisons along with methodological practicalism.
7
Sen never denies that there might be a partition, and so I take his position to be that
of methodological comparativism.
8
Sen writes: “Perhaps the most important contribution of the social choice approach
to the theory of justice is its concern with comparative assessments. This relational, rather
than transcendental, framework concentrates on the practical reason behind what is to be
chosen and which decisions should be taken, rather than speculating on what a perfectly
just society (on which there may or may not be any agreement) would look like”
(Sen 2011, p. 106).
Dictionary: NOSD
Just and Juster 13
2. THE SCALE OF JUSTICE
To fully appreciate the case for the sufficiency of comparisons, let’s put the
question of the partition in a slightly broader context by reviewing quickly
and informally the several levels of richness that can be possessed by scales
for comparing things. Our concern is the comparison of alternative social
arrangements with respect to their justice or injustice, but these points
apply to comparisons generally. Three kinds of scales—ordinal, interval,
and ratio—can be distinguished by how much information they include
for purposes of comparing two or more cases. Ordinal contains the least,
and ratio the most, and they might be the most familiar to common sense.
An interval scale lies in between. An ordinal scale, the weakest, allows no
more than an ordering. A ratio scale tells us much more, since it incorporates
information about how much of something is present. Notice that no
amount of anything, such as an amount of preferredness, could be
recovered from a mere ordinal preference ranking. By contrast, an ordering
of, say, quantities of sugar by weight contains more than mere ranking.
There is a natural zero point (no sugar at all), and some quantities can be
expressed as fractions of other quantities (a half cup to a cup). The ordering
does not contain that information. Other cases that contain this richer
information including a natural or objective zero point and meaningful
ratios include: runners’ times in a race, length of anything, annual salary.
In many contexts, such as all of my examples so far, there are only positive
amounts but no negative amounts. In other cases, though, such as a
person’s net financial worth, the amount can be negative. In some cases, in
fact, there are only negative cases. We might think of purity as an upper
bound, with all levels of impurity on the negative side (note that there can
be a maximum level of impurity as well: 100 percent impurity). There is
nothing above the zero point in that case. As we’ll see, purity is one possible
way to think about justice.
Between these familiar ideas of rank (ordinal scale) and amount (ratio
scale) is the interval scale. The specific meaning of such a scale is more
complex, although there are plenty of familiar examples. The key idea in
an interval scale is distance between points on a line. For any pair of
points, each point on a line is at some distance from its mate, and we can
compare the distance of one pair to the distance of another pair as a ratio.
AB might be half as far apart as CD, for example. What is missing,
though, is any idea of the amount of distance represented by any point
and this is the way in which it is weaker than a ratio scale. That would
require a natural or objective zero point in addition to the distance or
ratio information, as we have seen in the examples. We might ask, for
example, how far Halifax is from here, or from Detroit, but there is no
Dictionary: NOSD
14 David Estlund
question of how far Halifax is—how much distance it has. That would
require a zero point or origin. But even without that, distance information
is obviously richer than mere ordering information. We might know that
Halifax is farther from Detroit than Toronto, but we might also want to
know how much farther. And whether we want to know it or not, there is
an answer—a fact of the matter. Points in space bear distance relations,
even though no point has any meaningful amount of distance itself. That
is because there is no privileged zero point or origin.9 Points in time can
easily be seen on an interval scale but not a ratio scale, assuming there is
no beginning or end of time.
The scale of justice, as I will call it, might take any of three forms:
ordinal, interval, or ratio. Since the ratio scale contains a partition or
zero point, a comparativist (that is, non-partitioned) scale of justice
must be on an ordinal or interval, but not a ratio scale. To say that a
ratio scale is richer than an interval scale, with both being richer than
an ordinal scale, is to say that they incorporate more comparative
information. That sounds like a good thing about them, but those are
features for which a theorist must pay extra, so to speak. If we don’t
know how much more Lori likes one film than another, but only her
ordinal preference ranking, then we are unable to upgrade our scale
from ordinal to interval. From a theorist’s point of view, the scales that
contain less information have the advantage of being, as it were,
informationally less expensive. Of course, a theory will use a scale for
certain purposes, and some purposes might require informationally
richer scales. A tool with fewer features might be less expensive, but that
is no advantage if it is inadequate for the job at hand.
A particularly severe version of the sufficiency objection to partitionism
would hold that all we need in a theory of justice are ordinal rankings of
certain available social structures as juster than others. On this ordinal
comparativism no use is made of the idea that one alternative is far juster
than another, only that it is juster. For that, a richer kind of comparison
would be required, namely one that supports comparisons of the intervals
between the injustice of one alternative and that of another—an interval
comparativism.
9
The measurement of temperature is a common but not perfectly clear example often
used to illustrate interval measure without a ratio scale: distance without amount. It can
be confusing because Fahrenheit and Celsius each posit a zero point. It is arbitrary rather
than objective, which is why the case can be useful, but the Kelvin scale seems to identify
a true objective zero point with nothing meaningful on the other side: zero kinetic energy
is the coldest possible temperature. So temperature is not a good case of interval scale
without ratio information after all.
Dictionary: NOSD
Just and Juster 15
3. A THEORY OF JUSTER: THE LIMITS OF ORDINAL

AND INTERVAL COMPARATIVISM
I want to emphasize the limits of comparativism, and I begin with the

weakest form, ordinal comparativism. Here are some familiar forms of
judgment about justice that are obviously not available on an ordinal
comparativist framework. To speak in a simple shorthand: If there is only
“juster,” then there is:
a. no “just,” no “unjust”
b. no “much more just,” no “much more unjust”
c. no “highly just,” no “gravely unjust”
d. no “nearly just,” no “far from just”
e. no “nearly as just,” no “nearly as unjust.”
There would be no meaning to the claim that slavery is unjust, or that it is
severely unjust. We could not meaningfully say that slavery is significantly
more unjust than gender differentials in wages, only that it is more unjust.
We could not argue that one policy would improve justice much more than
an alternative. And so on.
There is more. Suppose that you are devoted to the cause of eradicating a
certain source of corruption in the political process. Suppose, for example,
that Supreme Court Justices are exploiting a loophole in the law and
receiving highly lucrative favors and gifts from parties to cases they will
hear. (Many examples would serve as well.) As strongly as you believe in this
cause, there are other valuable causes too and your chances of success are
uncertain. You should consider working on something else if this loophole-
closing project is unlikely to succeed and other causes are not. For example,
if one condition were twice as much juster (so to speak) than the status quo
than another condition would be, then a 50 percent chance of attaining it
would be as good as a certainty of attaining the other. If you take its chances
to be less than that, you should work on the other cause. However, on a
simple ordinalist framework, that form of reasoning is incoherent. The
reason is that there is no measure by which the justice-difference of a certain
improvement is any multiple of the justice-difference (from the status quo)
of any alternative. Ordinal rankings contain no such information, not even
roughly. In that case, the information about probability is of no use. Simple
ordinalism, all things considered, is quite devastating to the purposes to
which we hope to put our thought about justice. For this reason, I will put
ordinal comparativism aside, and hereafter use the term “comparativism” to
refer to approaches that include richer than ordinal comparisons.
An interval scale, as we saw, is one that contains not only rank orderings,
but what we intuitively think of as distance information, on the model of
Dictionary: NOSD
16 David Estlund
points on a line. Some of the limitations in ordinal comparativism would be

removed if interval information were available. On an interval scale we can
now meaningfully say that the improvement from x to z is greater than the
improvement from x to y by some multiple. Whether or not a precise
multiplier could be ascertained, ordinal rankings do not allow that form of
comparison at all. That same point allows us to say that y is nearer in justice
to x than is z by some multiplier. Again, the point is not about precision.
The rough judgment that z is “many” times further from x than y is would
not be allowed at all by ordinal ranking. It implies an interval measure,
albeit maybe only a rough one. So, if enough richness were available to
support an interval comparison, this would support at least somewhat more
of our standard reasoning about justice.
However, even interval ordering would not allow any meaning for many
of the standard judgments listed above. To see this, think of a line with no
beginning or end, and each alternative state of justice lying at some point.
Since the line segment defined by any pair is some fraction of any other
line segment, the comparisons marked by (b) and (e) are meaningful, but
since there is no privileged place to put the line between just and unjust (no
“partition” as we are calling it), (a), (c), and (d) are meaningless.
4. JUSTICE IS NOT LIKE UTILITY
If comparativism leaves behind so much of our thought about degrees and

amounts of justice, then what countervailing advantages does it have over
partitionism? There are well-known reasons for comparativism about utility,
but a brief review will show that they have no application to the question of
the scale of justice. We turn, after that, to a more promising basis for
comparativism about justice, namely the tight relation between comparison
and choice.
Ordinalism has been extremely important in the theory of utility as
developed by economists and others, and this has been driven mainly by
arguments that richer scales require more information than we have any
way to obtain. In particular, many have argued that there is no scientifically
defensible way to measure amounts of psychological states of pleasure,
contentment, or satisfaction—as intuitively familiar as such states may be.
For one thing, it has often been argued that there is no scientifically
respectable epistemic access to how much pleasure or satisfaction an
individual experiences in any given case. This is a special difficulty in the
epistemology of psychology, the so-called “problem of other minds.”
Second, economics, along with utilitarian moral philosophy, has wanted to
aggregate utility across persons in pursuit, often, of the greatest sum or
Dictionary: NOSD
Just and Juster 17
average. Even if there were a way to know such things as a person’s relative
strength of desire across alternatives, there may not yet be any way to know
how it compares to any other person’s experience—the problem of
interpersonal comparisons of utility. As a proxy, theorists have substituted
the less rich idea of an individual’s preference between two available options.
(Arguably, there is no need even to suppose there is some richer mental state
that this is partly measuring.) The resulting scale is only ordinal, but it
comes at a low informational cost. This, put very briefly, is the epistemic
pressure in favor of a merely ordinalist account of individual utility.
These familiar epistemic arguments in the theory of utility, if sound,
militate against any measure of utility that has a partition between good and
bad, a zero point of any fixed kind. That is more than can be generated from
the merely ordinal information implicit in choices, or from the amplifications
of that ordinal information into interval information.
Our topic is not utility, of course, and that, in fact, is my point. The
theory of utility has its own purposes and faces its own epistemological
challenges. In the context of the theory of justice there is no similar
reason to limit ourselves to what can be built out of ordinal rankings.
First, since we are asking about a scale of the justice of a single society
(nothing rules out its comprising the whole globe), there is nothing akin
to the epistemic issue about interpersonal comparisons. (Whether some
similar epistemological problem would arise in comparing the justice of
one society to another, I leave aside.)
Second, apart from aggregation, recall the alleged epistemological
problem about how to know the psychological state of utility even in one
(other) person’s mind. Epistemological questions of the kind familiar from
the theory of utility do not apply to the question of the scale of justice, and
so they are no pressure toward a weaker, comparativist account.
Third, in the case of a justice ranking there is no clear basis for introducing
lotteries into the ranked elements in order to generate ratio information. In
the case of utility, it is evidently legitimate to suppose that a choosing agent
would have dispositions to choose even when at least one of the options was
a lottery—a probability distribution between two determinate options.
However, the justice of a state of affairs is not a measure of any agent’s
disposition to choose it.10 It would be possible to interpret justice as what
10
There are a number of proposed techniques in economics and decision theory for
generating interval orderings out of merely ordinal rankings. These tend to be still
informationally more demanding in the following respect. For any set of alternatives that
are merely ordinally ranked, there is no technique for placing any subset of them on an
interval scale without adding more ordinally ranked elements to the set. In the case of the
von Neumann-Morgenstern method, the additional elements are lotteries between the
initially ranked elements. See the perspicuous presentation in Resnick (1987, 88–91).
Dictionary: NOSD
18 David Estlund
would be chosen by some hypothetical chooser, such as an impartial spectator

(Gaus 2011). But since this is not a real chooser the difficulties familiar from
utility theory around ascertaining the chooser’s psychological attitudes would
not apply. There is no impartial spectator whose attitudes we might wish to
know. Whatever basis we might have for theoretically positing rankings of
lotteries into the utility function of a hypothesized impartial chooser would
serve just as well and more directly as a basis for introducing the stronger
interval information between determinate states.11
So, lotteries are not themselves juster or less just than states of affairs or other
lotteries, and so they are absent from rankings with respect to comparative
justice. This is a metaphysical point. In addition, there is the methodologial
point that there is no reason for a detour through ordinal rankings of lotteries,
even if such things made sense, since the aim would be to retrieve interval
information that could be directly posited with no mention of lotteries. This
point reinforces the severe limitations of ordinal comparativism in the theory
of justice. It cannot be ramped up into cardinal information in the way it
arguably can in the case of utility. So the only interesting comparativism is
interval comparativism, that is, an ordering of (as I assume) basic social
structures which contains interval information but no partition.
5.â•‡ IS “JUST” LIKE “TALL?”
Could comparativism accept that slavery is unjust after all? Suppose we

tried understanding “slavery is unjust” on the model of statements such as
“the John Hancock Building is tall.” Tallness statements might be contextual
and comparative, but that doesn’t mean that nothing is tall.12 Roughly, let’s
suppose (at first) that normally the statement, “the John Hancock Building
is tall,” sets the appropriate comparison class as the set of all buildings past
and present. Then, to say that this building is tall is to say that it is among
the tallest in that class, measured by comparison to the average (or by the
11
â•‡ The point holds even if the universe is not deterministic but essentially probabilistic.
The method of recovering (a) interval information from (b) a hypothetical agent’s
attitudes toward even objective lotteries (so to speak); (a) and (b) are equivalent, and
there is no independent epistemological step in first ascertaining (a) when the choosing
agent is merely hypothetical.
12
â•‡ The semantics of “tall” are not simple. In addition to several varieties of contextualism,
among the views that are now actively debated is the view that there is such a thing as
being “just plain tall,” in a way that is not relative to any speaker or context. On that view
tallness would be partitioned even without relation to a context. John MacFarlane discusses
several options in MacFarlane (2007); MacFarlane attributes the “just plain tall” position
to Cappelen and Lepore (2005). Thanks to Assaf Sharon and Sahar Akhtar for urging me
to consider the “tall” analogy, and to Sean Aas for helpful discussion.
Dictionary: NOSD
Just and Juster 19
fraction of the buildings that it is taller than, or combinations of such

things, any of which might naturally be partially vague). Call this kind of
threshold a context-statistical threshold.
I doubt that statements such as “slavery is unjust” can be understood in
the context-statistical manner. For one thing, this makes it hostage to the
arbitrary setting of the comparison class. Whether it is true to say that
chattel slavery in the American South was unjust should not depend on
which other conceivable arrangements land in the context-determined
comparison class in an instance of utterance. We could allow that its truth
might be relative to the class of available alternatives, but that is different.
Even Rawls says that slavery might be justified in unfortunate circumstances
where if war prisoners were not enslaved they would be killed.13 (He does
not say it would be “just” even in that case.) For our purposes, we can leave
these issues aside in the following way: Consider a case in which the available
alternatives are also specified along with the option in question, such as
American chattel slavery along with the alternatives that were available,
including freeing the slaves or not enslaving them in the first place. Call that
embedded case of slavery “S.” The context-statistical model still says that
whether S is unjust depends on what other (alternative-embedded) cases of
slavery land in the comparison class that is somehow set by a context of
utterance.14 So, the context-statistical approach would say that America’s
wholly optional enslavement of hundreds of thousands of Africans would
not truly be said to be unjust if the comparison class turned out to consist
mostly of even worse atrocities. And maybe it does turn out that way.
Consider two possible comparison classes: (a) all the ways people are or have
been actually treated, and (b) all the conceivable or possible ways people
might be treated.15 Even if slavery is quite low in the comparison class (a),
since people are relatively rarely treated as badly as slavery, this appears to
shift in the case of (b). Is there any reason to think that of all the conceivable
or possible ways of treating people, vastly more of them are better than
slavery? I see no reason to think so. But then whether it is true to say “slavery
is unjust” depends, implausibly, on which of those comparison classes is
triggered in the context of utterance. Unlike the case of someone being tall,
I doubt that the truth or meaning of statements about slavery’s injustice is
relative to the context of utterance in this way.
13
â•‡ “For example, suppose that city-states that previously have not taken prisoners of
war but have always put captives to death agree by treaty to hold prisoners as slaves
instead” (Rawls 1971, p. 248).
14
â•‡ That is not the same as the set of alternatives in the context of action.
15
â•‡ I’m allowing, for the sake of argument, that such “ways of treating” are countable in
some appropriate way. If not, the context-statistical approach cannot get off the ground.
Dictionary: NOSD
20 David Estlund
Second, even if the setting of the comparison class is not arbitrary and
variable, the context-statistical interpretation remains implausible for
another reason. Suppose that the comparison class triggered by the statement
“slavery is unjust” is always and automatically the class of all the ways people
are or have been treated. That would avoid my first objection. But in this
case, the context-statistical claim that slavery is unjust is still nothing but a
claim about the fraction of all the forms of treatment (considered in the
context of the available alternatives at the time of action) that are better
than slavery. In that case, “slavery is unjust” depends on—because it is
nothing but a claim about—whether people have, in fact, usually been
treated better than that. I doubt that slavery being counting as unjust (the
statement being true) is even partly a fact about how people have normally
been treated. Slavery would be unjust even if most people were slaves. A
person’s counting as tall is different, and it can be captured plausibly in the
context-statistical way. “Just,” then, does not appear to be like “tall.” That
analogy will not save comparativism from the objection that it cannot
accommodate such robust judgments as “slavery is unjust.”
6. THE VARIETY OF PARTITIONS
I am mostly concentrating on the idea of a partition between just and unjust

social arrangements, but the very idea of a partition is more complex than I
have been acknowledging. If someone denies that there is any value or truth
in partitioned measures of justice, it is important to know which kind of
partition is at issue. As I will explain, we can distinguish between the
following kinds of partitions: floors, ceilings, thresholds, and toggles (or
“bare” partitions).
First, there is a floor partition when there are comparisons, and also a
zero point, but no comparisons below that point. Height is a simple
example. Strength seems to be another. This is a partition because it marks
a line between things that have some strength and things that do not. The
ideal of a floor partition will not concern us in the context of justice, though
all the other three kinds will. Second, a threshold partition is present when
there is a significant origin or zero-point, but also comparisons above the
line and comparisons below the line. Wealth is like that, since one can have
positive wealth of different degrees, but also, in the case of debt, negative
wealth of various degrees. Similarly, for our purposes, cases below will be
unjust, cases at or above just.
Third, a ceiling threshold is where there are comparisons and a point of
fullness or perfection. Purity is like that; also verticality, scores on many
tests, and much else. Finally, the simplest kind of partition is what I will call
Dictionary: NOSD
Just and Juster 21
a toggle or binary partition, the case in which the standard or property is

either met fully or not at all, and there are no further comparative degrees.
An appliance might be on or off, but not partly on or partly off (at least for
many appliances). Famously, one is pregnant or not pregnant; never a little
bit pregnant. (When I speak of an account of justice being “comparative,”
I will mean that there are more comparisons than simply the binary division
given by a toggle partition.)
When Sen considers the “grand partition” between just and unjust, he is
explicit that he means justice as a perfection point, and so this would be a
ceiling partition (Sen 2006, pp. 216–17). He argues or at least assumes that
justice is also not a toggle or binary partition, so that there are further
comparisons among the unjust cases.16
16
Sen writes: “. . . Rawls explores in depth the nature of an entirely just society . . .”,
(2006) p. 216. However, contra Sen, Rawls understands justice as a threshold partition
rather than as a ceiling. Is Rawls’s conception of justice partitioned at all? One reason for
doubting it lies in Rawls’s second principle, the Difference Principle. It is often said that
the Difference Principle requires, in effect, maximizing the benefits (measured as an
index of “primary goods”) of those in the worst-off class. It is important to remember that
Rawls’s own view is that distributions are not directly evaluable by his principles of
justice. The basic social structure is just insofar as it tends to promote the satisfaction of
Rawls’s two distributive principles. Officially, distributions themselves are just or unjust
only in a purely procedural sense: just if they were produced by a just basic structure,
otherwise not. Nevertheless, the principles themselves might have a partitioned or a
comparative structure, and our present question is which structure they have. In speaking
of the justice of distributions here, I simply mean the conformity of the distributions to
the principles. Any distribution of primary goods that violates the equal basic liberties
principle (which is not, on Rawls’s mature view, a maximizing principle. See Rawls (1993)
Lecture VIII, p. 291: “. . . the words ‘a fully adequate scheme’ replace the words ‘the most
extensive total system’ which were used in Theory”), and where the inequality is not for
the sake of greater liberty for all, would be unjust. So that is a partition of a kind, a bright
line below which distributions are unjust. However, among the distributions above the
line, the second principle would favor adopting the one, from those available, that is best
for those who will be worst off. That would render the overall account of justice comparative,
involving no ceiling partition. Rawls explicitly interprets the first principle, the principle
of “equal basic liberties,” as being fully or perfectly satisfiable, and so having a ceiling
partition (see Rawls 1971, section 8). It is common, as I have said, to interpret the
Difference Principle itself as requiring the maximization of the goods held by the worst off.
In that case, the principle would be maximizing, with no ceiling or threshold partition. As
a result, the theory as a whole would not admit of full satisfaction. It would not be ceiling
or threshold partitioned either. Fortunately, Rawls addresses this question directly. He
explicitly says that an arrangement in which advantages to the better off could not be
reduced without harming the worst off is “just through and through, but not the best just
arrangement.” The best just arrangement, or the “perfectly just scheme” (earlier on the
page), would be where there are no possible changes, not even benefits to the better off, that
would further improve the position of the worst off. The resulting structure of the view,
then, is that there is a threshold partition of full justice—the complete absence of
injustice—but also the possibility of supererogation—further improvements to justice
within the not-unjust range, and so no ceiling partition of perfect justice. The view of just
Dictionary: NOSD
22 David Estlund
7. WHAT JUSTICE MIGHT BE LIKE (DOESN’T

FAVOR COMPARISONS)
One might be tempted toward comparativism by the following thought,

which, in the end, I think is mistaken: whether or not justice is partitioned,
surely it will at least be comparative, and that comparative information is all
we need for choice; thus we can dispense with the partition. This might
seem to put justice comparisons on firmer methodological ground than a
justice partition. I doubt that it does. After all, some perfectible standards
might not admit of any further rankings. They might involve no subpar
rankings and include only a toggle or binary partition. This point is similar
rhetorically to Sen’s use of the example of height (Sen 2006 and 2011). He
points out that mountains can be compared with respect to height even
without any identification of the tallest mountain in the world (Sen 2011,
p. 22). As stated, that is not quite pertinent, since his opponent, the partition
theorist, says nothing about the most just society in the world, but rather
makes appeal to a standard of full or complete justice. But Sen still has a
point, namely that mountains can be compared with respect to height even
without any identification of a standard of full or perfect height. (Indeed, it
is clear that there is no such thing as being fully or perfectly tall.) Sen’s point
is that, since some standards such as height are like that, justice might be
like that: comparisons of societies with respect to justice may not require
any reference to a standard of full or perfect justice.
However, it should not be thought that this is any advantage for
comparisons as compared with partitions. For one thing, some comparative
standards do depend on there being a partition. Purity is evidently like
this: there is, it seems to me, no way to understand comparative degrees of
impurity without positing such a thing as complete purity. It is not the
“purer than” relation as such that entails a ceiling partition, any more than
the “greater than” relation does. But greater purity is nothing but greater
proximity to perfect purity. Standards such as purity have a ceiling partition.
So, justice might be like that.
Second, just as there might be comparisons without a partition, for some
standards there can be a partition without comparisons—a toggle or binary
partition. Examples include truth and arithmetical equality. So, justice
might be like that. Consider, first, the standard of truth. It is a perfectible
property of a proposition (or belief, or statement, etc.). But it does not
versus unjust is evidently threshold partitioned. In saying the view is threshold partitioned
I assume that there are comparisons among degrees of injustice as well. The principles do
not give much guidance on those comparisons, but it is clear, for one thing, that the
principle of equal basic liberties could be violated more or less severely.
Dictionary: NOSD
Just and Juster 23
(at least not obviously) come in degrees.17 Even a statement that asserts
several things is not partially true if some of them are true; it is false. It is
possible that there are no subpar rankings among degrees of truth. Might
justice be like that? As I explain in the next section, the case of arithmetical
equality illustrates how it could be.
8. THE CASE OF EQUALITY
There is a common though not unanimous view that social justice consists,
at least partly, in an equal distribution across individuals of something or
other. Among the candidates for what is to be distributed equally are:
certain resources, capabilities, opportunities, formal rights and liberties,
and the substantive value of certain rights and liberties. Distributive equality,
of course, is a partitioned concept. But does it admit of rankings within the
range of unequal distributions? If not, then the idea that justice is equality
would entail a partition but no subpar rankings. Justice might be like that.
There are numerous ways to measure and rank species of distributive
inequality.18 Sen himself, in earlier work, lists “coefficient of variation, Gini
coefficient, standard deviation of logarithms, measures of entropy” (Sen 1992,
p. 132). Consider a few simple examples of sets of numbers, without interpreting
them as measures of the holdings or utilities of people, but simply as numbers.
Which column, 1 or 2, is a more equal distribution in the following examples?19
It is not as if we can never “tell.” Consider a first example:
A1 A2
100 200
100 100
17
There is a large literature, however, laying out the problems with the several
proposals for defining a plausible property of one theory being truer or closer to the truth
than another. So even though there is a perfectly achievable standard of truth for a theory
(all its statements and all the implications are true), this may not, as far as we know,
generate any measure of comparative truth at all. The literature began with Popper
(1963). For citations to some of the literature, as well as a contribution, see Forster
(2004).
18
Even perfect distributive equality is not always a simple thing, when the thing
distributed exhibits what Sen calls “internal plurality.” For a brief discussion, see Sen
(1992) p. 131ff.
19
Temkin (1993) discusses related issues, but there is a difference between his question and
ours here. He considers the question of which distributions are worse with respect to equality.
That is not obviously the same as the question I’m briefly considering, which is which
distributions contain more inequality. This latter question could be asked about distributions
of weights rather than wealth or utility, and so it is independent of any questions about value.
Dictionary: NOSD
24 David Estlund
One thing we are looking for is perfect arithmetical equality, if it should be

present. Where it is, there is more distributive equality than in any case
where perfect equality is absent. We can rank these two cases with respect to
relative inequality: A1 is more equal than A2. Now consider two more
examples:
B1 B2
100 100
90 20
10 10
C1 C2
2 200
1 100
A moment’s reflection reveals that we are not immediately sure what we are
looking for as we scan these for degrees of distributive equality. We see that
there is inequality, but the criterion that establishes that fact (namely, that
some of the numbers are larger than others) is very little guidance in
interpreting degrees of inequality. Perfect arithmetical equality is a clear and
simple thing, but except in a small range of cases (see D1, D2) it does not
come in degrees. There is no such thing as the one true purely arithmetical,
non-normative, measure of inequality. In economic contexts, then, as
Atkinson says, “any measure of inequality involves judgements about social
welfare” (Atkinson 1970). The degree of inequality of a distribution of
goods, if it is interpreted arithmetically—that is, without any reference to
normative standards—would be undefined.
Would it follow that, if justice is distributive equality, there is no defined
measure of degrees of injustice? It would not follow. The reason is that full
arithmetical equality also lies at the top of a variety of normative measures
of inequality (Gini, etc.). So when someone says that full justice is full
equality, they could mean that inequality as measured in one of the
normative ways is the content of injustice, and zero normative inequality is
the content of zero injustice.
On the other hand, there is a different position about justice and equality.
Call this arithmetical egalitarianism. It is not about degrees of inequality in
the first instance, but about the clear and salient relation of arithmetical
equality. That, according to a natural view, is the content of justice, and not
in a way that is derivative from some prior normative measure of inequality.
An analogy might be the standard of legality: an action conforms to the law
Dictionary: NOSD
Just and Juster 25
or it does not, and the idea of degrees of legality is undefined. On this view
of distributive justice as arithmetic equality, any degrees of injustice would
have to derive from a standard of degrees of arithmetical inequality. But
since there is no such thing as degree of arithmetical inequality, there is no
such thing as degree of injustice.20
Of course, for many of us, accommodating the firm conviction that some
unjust arrangements are less unjust than others will be among the
touchstones by which we will evaluate a conception of justice. In that case,
arithmetical egalitarianism is presumptively highly implausible. If we are
quite sure that there are sub par comparisons of injustice, then since
arithmetical egalitarianism cannot allow this, we will be quite sure that it is
false (pending, of course, any powerful arguments that might shake up our
initial convictions).
The next closest thing to an interpretation of the idea that justice is
equality would be to defend one of the various measures of inequality as a
measure of injustice. But the idea that justice is equality will be of no use in
recommending one measure over another in that case. None of them counts
as more egalitarian, and so they must be compared as interpretations of
justice on some other—that is, non-egalitarian—grounds.
Equality is just one example. What I say here about equality will transfer
to many other concepts. Briefly, consider sufficiency: sufficiency for all is
binary. But the idea of degrees of sufficiency is not obviously a defined
notion. Summing the number of individuals above the sufficiency level will
not work for comparing different population sizes. It also pays no attention
to anyone’s distance from the sufficiency point, which seems perverse in the
normative context. Degree of injustice in a sufficientarian framework could
only be defined by invoking values other than sufficiency.
Summarizing: in this section I have argued that there is no advantage
for comparativism in the fact that justice might, like some other
standards, admit of comparisons but not a partition. After all, it is also
true that justice might, like some other standards, admit of a partition
but no further comparisons. Arithmetical egalitarianism would be a
bare partitionism of that kind. I grant that, intuitively, most of us will
have firm convictions about some justice comparisons, erecting a certain
presumption against bare partitionism. But equally, most of us will have
firm partitioned convictions, such as that slavery is unjust, erecting a
presumption against comparativism, the view that there are only
comparisons and no partitions.
20
A similar result applies to a view in which arithmetic equality is one among several
ingredients of justice, somehow weighted against each other.
Dictionary: NOSD
26 David Estlund
9. SPARE PARTITIONISM
But, it might be objected, surely there are obvious comparisons of inequality.

Why not grant that there is a kind of proto-theory of degrees of inequality
underlying such judgments? Certainly, it is true that even if the idea of
equality does not immediately suggest any general criterion for subpar
rankings, there are cases in which one of two unequal distributions seems
very clearly to be arithmetically more unequal than another. Consider:
D1 D2
100 100
99 50
Surely, D1 is more equal than D2. Or is it? This much seems clear: if there is
any such thing as a true measure of arithmetical inequality, it must surely
count D1 as more equal than D2. This is an intuitive data point that we
conjecture any adequate comparative measure of arithmetic inequality will
accommodate. We can expect the same to be true in the case of normative
distributive inequality: there will be intuitive subpar comparative judgments
to which any more general theory of normative inequality must conform.
This raises an important point about the epistemology of theory-
building in this area (and in many others). We do not, in the first instance,
learn that D1 is the more equal of the two from seeing that it is implied
by our best comparative theory of arithmetic inequality. Rather, we simply
“eyeball” this case—we expose ourselves to it and form a judgment, pre-
theoretically. There will be many cases that we are inclined to resolve in
this way, and many that we cannot (such as cases B and C). The same goes
for cases of comparative justice more generally. Prior to any theory, there
will be certain cases that seem to us more unjust than others: most will
think slavery to be more unjust than wage discrimination, for example.
Some cases we simply eyeball in that way, but many others we cannot. Is
rampant incarceration more unjust than widespread state appropriation
of property? Is denial of women’s suffrage more unjust than inadequate
procedural protections at trial? These and many other issues will often
defy the simple eyeball method.
In his discussion of the limits of partitionism, Sen gives numerous examples
of widely accepted comparative judgments of relative justice that evidently
neither entail nor derive from an account of full or “transcendental” justice.
Among other things, he names “. . . persistent famines . . . widespread exclusion
from medical access, . . . government-arranged torture of prisoners, . . .
arbitrary incarceration of accused people without access to court procedures”
Dictionary: NOSD
Just and Juster 27
(Sen 2011, p. 104). Importantly, though, these are never shown to be derived

from or explained by any general account of even comparative justice.
Rather, they are offered as plausible or even obvious on their face—the
reader is expected to eyeball them. Distinguish, then, between comparisons
that are accepted by a theory of justice and those that are delivered by it.
Consider the case in which there is a partition, but there are no other
comparisons except those delivered pre-theoretically, by the eyeball method.
Call this a spare partitionism. Spare partitionism is an important class of
views if, as is difficult to rule out, standards such as distributive equality or
sufficiency yield no more subpar comparative content than what can be
eyeballed pre-theoretically. A spare partitionist theory does have some
subpar comparative content, unlike bare partitionism, but it is not a theory
that contributes anything of its own to the problem at hand: identifying
comparisons between alternatives with respect to their degree of justice in
order to facilitate good choice. A spare partitionism has nothing to
contribute for purposes of choice beyond the binary structure of bare
partitionism. More generally, an eyeball comparativism with or without a
partition is not delivering the eyeballed comparisons that it accepts. Our
eyeball knowledge of the additional comparativist content owes nothing to
the theory of justice, but precedes it.
10. RICHNESS FIRST
Return, now, to the prospects for a comparative theory, one without any
appeal to the just/unjust partition. A methodologically comparativist
(hereafter MC) approach might hope, in one way or another, to leverage the
spare comparativist information generated by the eyeball method into
further comparisons not available to the eyeball. Suppose very roughly that
the method involves reasoning with the eyeball cases, along with arguments
from analogy and consistency, and so on. This is often how normative moral
philosophy proceeds, after all. In that case, theory will have delivered new
comparative information of its own. If we begin with a sufficiently rich set
of eyeball comparisons, some such method might hope to leverage these
into a useful number of new and practically relevant comparisons.
A first point is that it should be clear that the eyeball judgments about
lesser levels of arithmetic equality (and truth and consistency) are nothing
but interpretations of explicit and precise partition standards, and have no
place unless the partition has a place. Take away the understanding of
perfect arithmetic equality (which is among the simplest of all arithmetic
ideas, of course), and there is little to go on in trying to rank one case as less
unequal than another. In those cases, “x is more F than y is” means nothing
Dictionary: NOSD
28 David Estlund
but the vague proposition that “x has ‘more’ of the perfection-standard

F than y does.” Arguably, then, on a comparativist theory, in which no use
is to be made of the idea of a partition (such as the partitions in perfection
standards such as equality, truth, consistency . . . ), none of these eyeball
cases can be incorporated into the theory, since they would be meaningless.
A second point is that our pre-theoretic judgments about justice are very
often partition-entailing, even if no particular partition standard is entailed.
We might, for example, think that among the eyeball judgments, we would
find “slavery is unjust,” but recall that this judgment goes beyond what
comparativism can access. It entails a just/unjust partition, and the
methodological avoidance of any such partition is what defines MC as
comparativist. The same goes, of course, for “slavery is severely unjust.” In
order to steer clear of partitionism, the eyeball judgments to be gathered and
leveraged must be either merely ordinal, or merely interval-comparable. So,
the fund of pre-theoretic judgments must not be any kind of interpretation
of a specific partition or perfection standard, and they must not even entail
that there is some partition or perfection standard or other. Obviously, then,
many of our eyeball judgments of justice are impermissible by these
comparativist strictures. To that extent, this shrinks the fund of pre-theoretic
fixed points that might be leveraged by argumentation into a rich theory-
delivered set of comparisons. The question is what rationale there is for this
constraint. What rationale is there, in light of this impoverishing constraint,
for methodological comparativism?21
We are granting that once comparisons are given, no partition information
is of any further use for purposes of choice, and we are supposing, with
practicalism, that this is all that matters. The question at this point is how
the comparisons are to be “given.” The sufficiency of comparisons for choice
is no answer to this. One of our only routes to the comparative information
we want for choice is, as a matter of moral epistemology, to reason with our
considered judgments about justice, and these come, as often as not, in
partition-entailing forms. A theory of justice that engages with (rather than
simply stripping away) such partitioned judgments about justice must, by
21
Consider the possible view that there is nothing wrong with theorizing with
partition-implying judgments, but that doing so does not require theorizing the nature
of the partition itself, which might be more difficult. Steve Wall rightly pointed out to me
that one version of MC could hold that we ought not to bother theorizing the partition—
seeking full understanding of the standard of justice involved—but might perfectly well
draw on our partitioned eyeball judgments for their value in building a comparative
theory. It is hard to see why, if such value is granted to those judgments, it would not be
enhanced by theoretically elaborating and systematizing them in the manner of a theory
of the standard. Since we have seen no strong reason to think such theorizing is either
pointless or hopelessly difficult, there seems to be no point in eschewing the theoretical
understanding of full justice.
Dictionary: NOSD
Just and Juster 29
definition, be non-comparativist. And as we have seen, there is no basis

for MC to hold either (a) that such partitioned judgments are false or
confused (it allows that they might be true), or (b) that they are epistemically
less available or respectable (in what way?). Practicalism plus the sufficiency
of comparisons for choice do not entail the sufficiency of comparisons for
a theory of justice—methodological comparativism—after all. The reason
is that we would be left with either too few eyeball judgments that are
not delivered by theory at all, or those along with what further purely
comparative judgments into which they could somehow be leveraged. There
may be much richer comparisons available if only we could reason with a
fully partitioned conception of justice along with the much larger set of pre-
theoretic judgments this would make available. Without some principled
reason for eschewing the broad set of pre-theoretic but partition-entailing
judgments about justice, the sufficiency of comparisons for choice does not
provide any reason for a comparative theory of justice.
The tempting idea that a theory of “perfect justice” is not only useless in
practice but also a relatively intractable theoretical ambition should not
be taken at face value. Perfect equality is dead simple to define; degrees of
inequality—not so much. What about the case of justice? As for degrees
of injustice, it is, so far, not at all clear how a comparative theory could be
constructed. A theory of perfect justice? It could be a matter of a couple
of principles.
11. COMPARATIVISM AND CONSERVATISM
I turn, finally, to a second aim of Sen’s advocacy of a theory of “juster.” In

de-emphasizing the search for a theory of “perfect” justice, Sen is not only
calling for a comparative theory of degrees of justice. He is also calling for a
theory of realistic, immediate, and incremental improvements in justice.
Why, after all, would we want anything else? This concern is analytically
separate from the concern about the value, if any, of a partition between just
and unjust. The reason is that even without any partition at all, there will be
relatively higher and less realistic reaches on the comparative scale. This
second concern attempts to orient our theorizing more (if not exclusively)
toward the more realistic range.
Let’s call methodological conservatism:
The view that the theory of justice ought to favor the investigation of standards
whose achievement is not too remote in one of the following ways: too far in the
future, too dissimilar to the status quo (or to what has been seen), too unlikely, or too
difficult or costly.
Dictionary: NOSD
30 David Estlund
This is different from practical conservatism:

The view that social changes that are remote in one of the above ways are, to that
extent, less advisable.
Methodological conservatism is a position about the merits of certain kinds

of normative theory, while practical conservatism pronounces on action. I
will briefly question one way of arguing for methodological conservatism
before turning to my main point, which is that even if it were a sound view,
it would not provide the support it might seem to provide for comparativism.
Practical conservatism plus practicalism about political philosophy
might seem to support methodological conservatism. That is, if political
philosophy ought mainly to serve political or social choice, and such choice
ought itself to be conservative, then it may seem to follow (at least in a
rough sense) that so ought political philosophy to be conservative. To reject
that strategy of argument, one must reject either practical conservatism
(I will take no stand for or against it), or reject practicalism about political
philosophy (I reject it in fact, but here I leave this open), or deny that the
conclusion of methodological conservatism would follow from those
premises. I deny that it would follow.
We would need some reason to believe that wise conservative social choice
could not benefit from normative political theory’s investigations of less
conservative possibilities. For just one example of how it might benefit,
consider the attempt to understand just how remote certain practical options
really are or are not (in time, similarity, likelihood, or difficulty). We can
hardly rule out in advance that this might be facilitated by reflection on a
broader set of options, including some that are patently remote. The case is
similar to the failure, noted earlier, of the argument that since comparisons
are sufficient for choice, theory ought to be comparativist. Here I assert the
failure of the argument that since choice ought to be conservative, theory
ought to be conservative. And this is even on the concession for the sake of
argument that political theory ought mainly to serve political practice.
Whatever the merits of methodological conservatism, it is sometimes
thought to count in favor of comparativism about justice. Methodological
conservatism is certainly no argument for comparativism as superior to
partitionism. Suppose that meeting the partitioned standard, the standard
of full social justice is remote (in one or another way). If standards of full
justice are unworthy of theoretical attention because of their remoteness,
then remote reaches of justness on a wholly comparative scale must be
unworthy too. There is no particular challenge to partitionism in that.
However, if there is a standard of full justice and it is remote, then if
remote standards, be they partitioned or comparative, are unworthy of
theory (as says methodological conservatism) then full justice is unworthy
Dictionary: NOSD
Just and Juster 31
of theory. The partition standard itself would be of no theoretical interest on

that methodologically conservative view, being so remote. The objection is
not to partitionism as such in that case, but to wasting time theorizing
about such a remote standard—the standard of full justice. In that case, we
would hope at least to have subpar comparisons with which to work. The
argument does not favor comparativism over partitionism, but favors
theorizing only with the subpar comparisons, and only with those that are
not themselves too remote.
It is true, then, that methodological conservatism’s preference for non-
remote comparisons would indeed count against theorizing with a justice
partition if full justice is a remote standard. However, this is a count against
methodological conservativism itself. As I have argued, reasoning with
partitioned or at least partition-implying comparisons may be the only way to
leverage our intuitive or eyeball convictions about justice into a set of
comparative judgments that have any support in normative theory. If
methodological conservatism rejects that method on the ground that it is
contaminated by the remote (which we can grant for the argument) partition
between just and unjust, then methodological conservativism, for no evident
reason, bans the use of a wealth of ordinary convictions about the nature and
content of not only justice but even of “juster.” The point is exactly like my
central point above about the pointlessness of comparativism’s averting its
eyes from partition-implying judgments. Whether those judgments are
excluded on the ground that they are partitioned, or on the ground that they
are remote, the exclusion appears to me to have no adequate justification.
There is more to be said about each of the several species of methodological
conservatism—those based on remoteness in time, similarity, likelihood,
and difficulty. I believe a start would be to point out that these are often
conflated, as if the changes that would be least difficult are also those that
would be least dissimilar to what has been known, etc. That correlation is
weak at best, and similar questions could be raised about the other possible
pairings. But I leave all this for another occasion.
Bibliography
Atkinson, Anthony B. (1970). “On The Measurement Of Inequality,” Journal of
Economic Theory 2: 244–63.
Cappelen, H. and Lepore, E. (2005). Insensitive Semantics: A Defense of Semantic
Minimalism and Speech Act Pluralism (Oxford: Blackwell).
Estlund, David (2011). “What Good Is It? Unrealistic Political Theory and the Value
of Intellectual Work,” Analyse & Kritik 33: 2, 395–496.
Forster, Malcolm (2004). “Verisimilitude and Likelihood,” at <http://philosophy.
wisc.edu/forster/920/Verisimilitude&Likelihood.pdf> accessed Nov. 15, 2014.
Dictionary: NOSD
32 David Estlund
Gaus, Gerald (2011). “Social Contract and Social Choice,” Rutgers Law Journal
43: 243–86.
MacFarlane, John (2007). “Semantic Minimalism and Nonindexical Contextualism.”
in G. Preyer and G. Peter (eds), Context-Sensitivity and Semantic Minimalism: New
Essays on Semantics and Pragmatics (Oxford: Oxford University Press), 240–50.
Popper, Karl (1963). Conjectures and Refutations (London: Routledge and Kegan
Paul).
Rawls, John (1971). A Theory of Justice (Cambridge, MA: Harvard University Press).
Rawls, John (1993). Political Liberalism (New York, NY: Columbia University Press).
Resnik, Michael D. (1987). Choices: An Introduction to Decision Theory (Minneapolis,
MN: University of Minnesota Press).
Sen, Amartya (1992). Inequality Reexamined (Cambridge, MA: Harvard University
Press).
Sen, Amartya (2006). “What Do We Want From A Theory of Justice?” Journal of
Philosophy 103: 5, 215–38.
Sen, Amartya (2011). The Idea of Justice (Cambridge, MA: Harvard University Press).
Temkin, Larry (1993). Inequality (Oxford: Oxford University Press).
Dictionary: NOSD
PA RT I I
THE MORAL ASSESSMENT
O F S TAT E S
Dictionary: NOSD
Dictionary: NOSD
2
Political Rule and Its Discontents
Niko Kolodny
1. INTRODUCTION
States stand or are said to stand in certain relations of rule to their subjects.1
A state wields “authority” or “power” over its subjects. It “obligates,” or
“coerces,” or threatens, or uses “force,” or violence against them, so as to
compel them to comply with its commands. It claims a “monopoly” or
“exclusive right” to issue and enforce these commands. It taxes. And even
when it does not directly affect its subjects in these ways, it profoundly
shapes their natural and social environment. These relations of rule are
thought to pose a problem. It is often described as a problem of “legitimacy,”
although I will not rest much on that notoriously supple term.
My aim is to understand the problem better, in a common way albeit not
the only way of framing it. This framing has two main elements. First, those
who are subject to the problematic relations of rule are thought to have a pro
tanto moral complaint against those relations. Those relations of rule threaten
their liberty—or, at any rate, some right, interest, or status that they have.
As Pettit defines the “problem of legitimacy,” for example, it is “how to
reconcile . . . political submission with personal freedom” (2012, 147).
1
Earlier versions of this chapter were presented as a paper in Joseph Raz’s seminar at
Columbia Law School, fall 2012; the Ethics Writing Seminar at UCLA, spring 2013; a
graduate workshop on legitimacy with Massimo Renzo and Annie Stilz at the Australian
National University, summer 2013; a colloquium at Brown University, fall 2013; Dick
Fallon and Tim Scanlon’s Law and Philosophy Workshop at Harvard Law School, fall
2013; a seminar at the Center for Ethics and Public Affairs at the Murphy Institute,
Tulane University, spring 2014; the Oxford Studies in Political Philosophy Conference at
the University of Missouri, Columbia, fall 2014; a colloquium at the University of
Arizona, fall 2014; and at the Responsibility Beyond the State Conference at the
University of Virginia, spring 2015. I’m grateful for comments at those events, especially
prepared comments by Jon Quong and Harrison Frye, as well as for correspondence from
Massimo Renzo and Victor Tadros, and painstaking and insightful criticism by two
anonymous readers.
Dictionary: NOSD
36 Niko Kolodny
Second, this complaint is not that the state in question is a bad bargain,
in the sense that it leaves its subjects with a worse distribution of means to
a fulfilling life—or, for shorthand, detracts from the common good. These
means to a fulfilling life include negative goods—such as the absence of
physical invasion, constraint, or impediment of one’s person by other
people—and positive goods—such as food, water, shelter, sanitation,
medical care, and education. We can leave largely open what counts as a
better distribution: whether it is sensitive to aggregation, equality, priority,
desert, or some combination thereof. Suppose, at least for the sake of
argument, that the state in question is a good bargain. By issuing and
enforcing directives, it induces cooperation that would otherwise not take
place, and that cooperation promotes the common good. If the state reduces
each individual’s enjoyment of some means—such as the negative good of
freedom from state interference—then it increases her enjoyment of other
means—such as the negative good of freedom from interference tout court,
as well as many positive goods. These more than compensate, so that the
state puts its subjects in a better position overall to lead worthwhile lives.
Still, this does not answer the complaint. Perhaps, if the alternatives threaten
vastly greater evils, the state should not be dismantled. But, unless more is
said, the relations of rule it involves are “illegitimate” or in some other way
morally imperfect.
What does answer the complaint, if anything does, is either a condition
or a limit. A condition on the relation of rule changes its context or character
so that it is no longer as objectionable. Such a condition might be, among
other things, that those subject to the relation of rule consent to it, or that
it is acceptable to them. A limit removes the relation of rule from a sphere
in which it would be objectionable. Such a limit might be that the state is
minimal, that it acts to secure only negative, not positive goods; or that it is
liberal, that it does not regulate private or self-regarding choices.
A complaint of this kind drives many libertarian views. A more extreme
libertarian position says that, absent consent, the state, like any other agent,
may require compliance only with natural prohibitions on force (Simmons
1979, 2000, 2005). A more moderate libertarian position says that, absent
consent, the state may require only compliance with natural prohibitions
on force, or contributions to schemes that require such compliance; absent
consent, the state must be minimal (Nozick 1974). Why not a more
extensive state, which requires contributions to schemes that provide
positive goods, such as greater literacy or protection from infectious disease?
Few libertarians think that they need to deny the seeming truism that things
would be better if people enjoyed greater literacy or protection from
infectious disease. (Some libertarians may support private charities, or
harbor personal hopes for the victims of natural disasters or command
Dictionary: NOSD
Political Rule and Its Discontents 37
economies, that are predicated on precisely that truism.) Instead, most

libertarians will first answer: “Even if the state has good results, that doesn’t
answer the complaint against the relations of rule that the state involves. For
instance, it doesn’t license the state to coerce people, without their consent,
to bring about those results. The ends don’t justify the means.”
Many who would favor a far more extensive state nevertheless agree that
there is some such complaint against relations of rule, which is overcome
only if certain conditions or limits are satisfied. Rawls’s “liberal principle of
legitimacy” says that because the state “exercises political power,” it must
meet the condition of being (as I will put it) “acceptable” to those subject
to it: roughly, justifiable to them in terms that do not presuppose any
particular religion or philosophy of life (1993, 136–7). Those subject to
such “political power,” it would seem, have some complaint about it, which
must be addressed, if not by their consenting to it, then by its being
acceptable to them. The complaint isn’t answered simply by showing that
the state brings about a better distribution of means to a fulfilling life—or,
specifically, that the “basic structure” distributes “primary social goods”
according to the “two principles of justice.” If it did, Rawls (1971) would
not have needed a sequel.
To take another example, Dworkin agrees that there is a crucial “puzzle
of legitimacy”: “How can anything” supply a general “justification for
coercion in ordinary politics?” (1986, 191). The condition that must be met,
he argues, is that those subject to such coercion comprise a community of
a special and demanding kind—a “community of principle”—which goes
beyond merely having a state that promotes the common good. And the
more recent Dworkin (2011) is sown with thoughts of a similar form: the
government “has no moral title to coerce, unless . . .” (372), “coercive
political organizations undermine the dignity of their members unless . . .”
(319–20), and so on.
Finally, consider a popular argument that economic justice is more
urgent within borders than across them. Certain relations of rule, such as
coercion, are thought to obtain distinctively within borders. And these
relations are thought to provoke a complaint, which is answered only by the
condition of economic justice beyond mere humanitarianism. To answer
this complaint, it isn’t enough to show that state coercion is a good bargain.
No doubt, establishing a minimal state is an improvement over a state of
nature. But the minimal state’s relations of rule provoke a new demand for
justification, not present in a state of nature, which that improvement alone
does not satisfy (Blake 2001; Nagel 2005).
So it seems fair to say that the idea is widespread, at least among political
philosophers: namely, that there is some complaint against certain relations
of rule, which is answered not by a net improvement in the distribution of
Dictionary: NOSD
38 Niko Kolodny
means to a fulfilling life, but instead by conditions or limits of the kind that
we have described. But why, exactly? Imagine that you improve the
distribution of means for a fulfilling life among us, your neighbors. You
drain a stagnant pool in your garden, lowering our risk of mosquito-borne
disease. It would be odd to suppose that we have a complaint against you.
Yet when the state improves our means for fulfilling lives, we are supposed
to have a complaint. Presumably, this is because of the relations of rule that
the state, but not your neighborliness, involves. Which relations of rule
provoke this complaint? Why? And why is the complaint met by and only
by limits or conditions of the kind so often proposed? These are the
questions this chapter asks.
The chapter investigates several familiar candidates for the problematic
relation of rule. With respect to each candidate target of the complaint,
I submit, we find one or both of two things. First, removing the candidate
target doesn’t remove the complaint. That is, if we subtract, in imagination,
the relation of rule in question, we are still left, intuitively, with a complaint
of the kind of which we are trying to make sense. Thus, that relation of rule
cannot be the thing, or at least not the only thing, that provokes the
complaint. Second, the complaint against the candidate relation of rule is
answerable, either by anyone’s lights, or at least by the lights of those who
insist that there is a complaint, even without the conditions and limits that
they invoke.
So what then nourishes this pervasive idea: that some relation of rule
provokes a complaint, which in turn requires special conditions or limits?
I am not sure, but at the end of this chapter, I explore a possibility. It is an
anxiety that in being subject to the state’s decisions, we are subordinated or
put into relations of inferiority to other people. The problem of relations of
political rule, if there is one, is not so much of reconciling such relations
with the liberty of the individual, but rather of reconciling them with an
ideal of equality among individuals.
2. FROM OBLIGATION TO ENFORCEMENT
To sharpen our focus, suppose henceforth that the state against which the
complaint is brought is not only a good bargain, but moreover the best
available bargain. The state is an ideal enforcer: it enforces all and only
violations of its directives. Its police, courts, and so on make no mistakes.
And the state is ideally directive: there is no alternative set of directives that
the state could issue and enforce that would bring about a better distribution
of negative and positive goods (although likely there will be alternative sets
that do equally well). Granted, this ideal state may automatically meet one
Dictionary: NOSD
of the complaint-answering conditions proposed in the last section: namely,

economic justice beyond humanitarianism. But it does not, unless more is
said, satisfy the other conditions such as consent or acceptability. Nor does
it respect the limits of the minimal state; it aims to provide positive, no less
than negative goods.
To begin our search then: which relation of rule provokes the complaint?
No relation of rule has attracted more discussion than “political obligation,”
a moral duty to comply with state directives, as such. Let me briefly explain,
in this section, why I nevertheless turn my attention elsewhere.
We can include under the heading of “political obligation” the alleged
“duty to obey the law,” as well as “political authority,” understood as the
state’s power to create political obligations, by issuing directives. We might
also include under this heading the state’s oft-discussed “moral monopoly”
or “exclusive right”: namely, that where there is a state, it is morally
impermissible for private agents to enforce natural prohibitions. This can be
seen as a special case of political obligation: namely, to comply with the
state’s ban on private enforcement.
Our ideal state issues directives to contribute to schemes to promote the
common good. Such directives will go beyond mere natural moral
prohibitions on the use of force. As we will discuss in section 4.2, such
directives will also go beyond natural moral requirements to promote the
common good. Thus, if there are political obligations to comply with the
ideal state’s directives, we are morally constrained to a greater extent than
we would be if we faced those directives with only natural duties, even on
an expansive view of what those natural duties are. Put another way, if
there are no political obligations, then sometimes disobeying the state’s
directives violates no moral duty whatsoever, not even a duty to promote
the common good.
There is an intelligible complaint against being bound by political
obligations. Of course, this complaint doesn’t grant that there are political
obligations and then rail against Moral Reality for having put us in chains.
Instead, the complaint comes earlier, as a reason why Moral Reality does not
in fact so obligate us. “In general,” the complaint might go, “agents who
would be bound by any putative moral requirement have at least pro tanto
complaints against being so bound. Unless those complaints are answered
by sufficiently important values that the requirement serves, there simply is
no such moral requirement. Since the complaints against being bound by
political obligations are not answered by sufficiently important values, there
are no political obligations.”
Can this sort of complaint against political obligations be the complaint
that we are after? Let us ask two questions. First, does removing the
candidate remove the complaint? Second, can the complaint against the
Dictionary: NOSD
40 Niko Kolodny
candidate be answered, either by anyone’s lights, or at least by the lights of

those who insist that there is a complaint, even without the conditions and
limits that they invoke?
Let’s begin with the second question. Even those who press such a
complaint against political obligations are likely to accept that there are
some natural duties. And there is a pro tanto complaint against natural
duties, as there is against any putative moral requirement. So they accept
that the complaint against natural duties is overcome. Why, then, isn’t the
complaint against political obligations also overcome? Is the difference
that the complaint against political obligations, the objection to being so
bound, is somehow more serious? Political obligations are not, as a rule,
more burdensome. After all, political obligations to refrain from private
enforcement are requirements simply to let the state take a distasteful chore
off one’s hands. Perhaps the trouble is that political obligations are imposed
on us by another person or “will” whereas natural duties are not? But this
is an illusion. The basic principle that when a state issues a directive to us,
we are morally required to comply, if there is such a principle, is not itself
imposed by any state. Rather, the state determines how it applies, as a result
of making certain choices; namely, to issue directives. The same is true of
natural duties. The basic principle that you may not step on my foot is not
imposed by me. Rather, I determine how it applies, as a result of making
certain choices. If I move my foot from here to there, then you may no
longer step there (van der Vossen 2015).2
Perhaps, then, the difference is not that it’s somehow worse to be bound
by political obligations than by natural duties, but instead that there’s
simply less to justify being so bound. What’s to be said for complying as
such, especially when not complying will do just as much good? Granted, if
one has promised to comply, then the value of fidelity argues in favor of
compliance. But if that is the only reason, then something like consent is a
necessary condition for political obligation.
For the purposes of this chapter, I will grant that this is so: that we can’t
answer the complaint against being bound by political obligations without
appealing to a condition like consent. What matters is that this does not
end our search. For consider our other question: does removing the
candidate relation of rule remove the complaint? Imagine (or observe) that
we don’t have political obligations. This means, again, that disobeying the
state’s directives will sometimes violate no moral duty whatsoever. Otherwise,
2
The difference, it might be replied, is that when I move my foot from here to there,
you are not “taking orders from” or “being bossed around by” me, as an inferior by a
superior. If so, then this would be a step in the direction of the Subordination Complaint
of section 7 within this chapter.
Dictionary: NOSD
the state relates to us in the same way. It still issues and enforces its directives
(that is, all of the directives it needs to issue and enforce to promote the
common good to the greatest extent possible). Does the fact that we don’t
have political obligations to comply with these directives, which it
nonetheless issues and enforces, silence the complaint that so many feel? On
the contrary, it would seem to intensify the complaint.
The residual complaint, some might say, is that the state, in issuing its
directives, asserts, falsely, that we have political obligations. But can the
complaint be merely that the state asserts untruths? In any event, imagine
that the state does not assert that we have political obligations. (Is any
imagination required? Do states assert that we are morally required to obey
them?3) For example, although the state claims no moral monopoly on
enforcing natural prohibitions, it nonetheless announces that it stands ready
to imprison anyone else who tries to enforce them. Does the state’s conceding
that we aren’t obligated to comply with these directives, which it nonetheless
issues and enforces, quell the felt complaint? Quite the contrary.
It may seem obvious what the target of the residual complaint is. The
state is enforcing our compliance with its directives. Indeed, it is very often
said that it is the state’s use of “force” or “violence” or “threat of punishment”
or “coercion” that calls for special conditions or limits.4 As Edmundson
(1998, 90) paraphrases the complaint: “The coercive nature of law not only
renders the state presumptively illegitimate, it sets the bar of legitimacy at
a higher level than is normally necessary for the legitimacy of individual
or concerted private activity.” The state’s concession that we are free from
any moral bonds of political obligation to comply with its directives does
nothing to answer this complaint, about what the state still does to us in
enforcing those directives. Indeed, the concession seems to amplify this
complaint. This is especially so if we accept the Duty Requirement: that
only duties may be enforced.5 If there are no political obligations, then, as
3
Raz (1994) suggests that a claim to the “right to impose obligations on . . . subjects”
is constitutive of a legal system. I find this far from obvious (compare Murphy 2014, 86,
115–16). It may be constitutive of the state that it claims, or presupposes, a permission to
issue and enforce directives (which may suffice for Raz’s jurisprudential purposes).
4
See Simmons 2000, 137; Nozick 1974, ix; Rawls 1993, 136–7; Larmore 1999,
605–8; Nagel 1991, ch. 14; Dworkin 1986, 191; Pettit 2012, 147; Ripstein 2009, ch. 2. v;
Huemer 2013, ch. 1.
5
See Nozick 1974, 6; Dworkin 1986, 191; 2011, 319–20; Klosko 2005, 49–50; Quong
2011, 115. This is why Dworkin holds, as noted in the introduction, that justified coercion
requires a “community of principle”: justified coercion requires political obligations, which
in turn require a community of principle. The Duty Requirement also appears to be an
implicit premise in the argument that the state wrongs us by enforcing prohibitions on
private enforcement (Nozick 1974, 24; Simmons 2000, 156). Since there is no natural
duty to refrain from private enforcement, the argument runs, the state violates the Duty
Requirement in enforcing its directives to refrain.
Dictionary: NOSD
42 Niko Kolodny
we observed earlier, there will be cases in which we have no moral duty

whatsoever to comply with even an ideal state’s directives. In such cases,
according to the Duty Requirement, the state wrongs us in enforcing those
directives.
3. DETERRENTS: THE DISTRIBUTIVE COMPLAINT
What then is enforcement? It divides into three categories, which raise

different concerns. To enforce a directive, D, may be to threaten: to prevent
the agent’s violation of D by telling him that he will suffer some consequence
that he seeks to avoid if he violates D.
Next, to enforce D may be to defend: to prevent the agent’s violation of
D by more direct, physical means. Note that “defense” covers a wider range
of cases than it might at first seem. Restitution “after the fact”—such as
returning stolen goods—is often described as a response to a past violation.
But many such responses are forward-looking defense; they aim to prevent
the future violation that would take place if, say, the thief were to remain in
control of the stolen goods.6
Finally, to enforce D may be to impose a deterrent: to follow through on
the threat (whether or not the threat itself was permissibly issued), not to
prevent the violation of D (which has already occurred), but instead to
sustain the potency of future threats to deter the agent or others from
violating instances of the same sort of directive. I use “impose a deterrent”
instead of (the admittedly less cumbersome) “punish,” to stress that it does
not involve condemnation, as punishment, perhaps by definition, does. The
function of following through on deterrent threats is simply to induce
cooperation, and that needn’t involve condemnation.
For reasons that will become clearer as we proceed, I start by looking for
a complaint against the permissibility of the state’s imposing deterrents for
violations of its directives. Suppose that some subject, Violet, has violated a
state directive. May the state impose a deterrent on her? Let us assume that
the deterrent, following contemporary practice, is imprisonment.
Imprisoning her would deter future violations, which sustains cooperation,
which in turn promotes the common good. So what’s the problem?
Needless to say, in order to be effective, the deterrent may need to curtail
radically the goods that Violet enjoys, especially negative goods (not least
6
Some forms of restitution can’t be counted as forward-looking defense. If I destroy
your property, there’s nothing left to defend. However, such restitution-as-compensation
is not a way of enforcing a requirement. Instead, it is a source of requirements (e.g. to
compensate you for your destroyed property).
Dictionary: NOSD
freedom of movement unobstructed by guards). But this is not enough for

a complaint. For by hypothesis, the deterrent provides others with important
goods. By analogy, suppose we don’t save one person from one-month-long
entrapment in a pit, in order to save two others from two-month-long
entrapments in similar pits. We do indeed leave the freedom of movement
of the one person worse than we could have left it. But this is in order to
avoid leaving the freedom of movement of two others worse to a far greater
degree. If the one has a complaint, it seems straightforwardly answered by
observing that the outcome we are bringing about is distributively fair.
It might be replied, however, that Violet’s case is not like this. It isn’t as
though if Violet isn’t imprisoned, two others will be imprisoned in similar
cells for twice the time. Instead, not imprisoning Violet will affect each
other person far more modestly. By hypothesis, not imprisoning Violet will
weaken deterrence. But the effect of this weakened deterrence will be to
leave each other person only a little more exposed to property crime, or
leave each other person with only a little less in the way of public services.
In sum, Violet bears great losses in order to provide others with much
smaller benefits. Now, according to an aggregative principle of distribution,
imposing the deterrent on Violet might make the distribution better, at
least if the little losses that those many others suffer add up to a greater sum
than the severe loss that Violet alone suffers. Similarly, according to a desert-
based principle of distribution, which discounts the suffering of wrongdoers,
imposing the deterrent on Violet might make the distribution better, at
least if she acted wrongly in violating the state directive. But according to
non-aggregative and non-desert-based principles of distribution (or to
desert-based principles in those cases in which Violet has not acted wrongly),
imposing the deterrent on Violet worsens the distribution. Which is to say
that imposing the deterrent on Violet does not promote the common good
after all. Call this the Distributive Complaint Against the State Imposition
of Deterrents, or the Distributive Complaint, for short.
How, if at all, can the Distributive Complaint be overcome? Consent
would presumably do the trick. If someone consents to a smaller share in
order to provide others with greater shares, then he has no complaint about
a lesser share. Why? A natural answer is this: what he had a distributive
claim to was not, strictly speaking, the share of the good. Instead, it was to
the opportunity for such a share: the chance to have that good, if he chose
in a certain way. When he consents to a smaller share, he hasn’t been
deprived of this opportunity. He has simply exercised it in a particular way.
In section 1, I described the state’s aim, the common good, as a just
distribution of goods. In many cases, however, the state may have reason to
provide not goods, but instead opportunities: to put a person in a position to
enjoy a good if he chooses appropriately, but not otherwise. There are many
Dictionary: NOSD
44 Niko Kolodny
possible reasons for this. The recipient’s choice may be the best indicator that
she has reason to have the good. Or the opportunity to choose may facilitate
activities that are possible or valuable only insofar as they flow from “one’s
own,” “free” choices or judgments, such as expression, religious observance,
association, or—as Raz (1986) understands “autonomy”—being the author
of one’s life as a whole. Or the denial of opportunity may be paternalistic. Or,
more importantly for present purposes, the state may face a “more-through-
choice” dilemma. Either the state can provide us with less of a good whatever
we choose, or it can provide us with the opportunity for more of the good
but only if we choose appropriately. Often, although of course not always,
the latter opportunity is what we have reason to want.7
If what is to be distributed is opportunity, then Violet can raise the
Distributive Complaint only if the state provides her with a worse
opportunity than it provides others, or a worse opportunity than anyone
needs to have. Turn the clock back to before Violet’s violation of the state’s
directive. At that point, the state offered her exactly the same overall set of
opportunities—goods conditioned by her choices—that it offered everyone
else. Part of that overall set was this particular opportunity, O1: that if Violet
complied with a certain directive, she would not be imprisoned, and if she
violated this directive, she would be. In imposing the deterrent on Violet, it
might be said, the state isn’t depriving her of this opportunity, O1. What it
does is consistent with her having this opportunity.
Violet might protest: “Yes, I grant that my opportunity was no worse
than anyone else’s, but it was worse than anyone’s needed to be. The state
could have provided everyone a clearly better opportunity, O2: that whether
or not one complies with the directive, one will not suffer the deterrent.”
But then Violet would be mistaken. If the overall set included O2 instead
of O1, then other opportunities in the O2-set would be so much worse that
the O2-set overall would be worse (for each individual). The state faced a
more-through-choice dilemma. Other opportunities in the O1-set depend
on deterrence that is provided only by the inclusion of O1.
If this seems like sleight of hand, compare a case that has nothing to do
with imposing deterrents. Suppose that there is some publicly provided
benefit to be distributed. In order to know how to distribute it, the state
asks people to apply for it. Imagine that if the state had to gather the relevant
information on its own, it would be too costly to provide the benefit. Again,
7
No doubt, there are difficult questions here about what makes one opportunity or
overall set of opportunities better than another opportunity or overall set of opportunities
(what Olsaretti 2009 calls “principles of stakes”). I don’t think the value of an opportunity
can be reduced to the expectation of the effects of possible exercises of it, based on some
probability of those exercises. But I don’t have a general, positive theory to offer. I am just
relying on what seem to me plausible particular judgments.
Dictionary: NOSD
the state faces a more-through-choice dilemma. Dithers chooses not to

apply before the deadline, whereas others do apply. As a result, shares of the
benefit are distributed to those others, but not to Dithers. Dithers protests:
“Since I have just as much of a claim to the benefit as others, it is unfair that
they have more than I have.” The state’s reply is: “What you had an equal
claim to was not the benefit, but the opportunity to receive it if you applied.
And your claim has been honored as fully as the claims of those who applied
and received it.”
Observe that this response to the Distributive Complaint has two welcome
implications. First, it puts pressure on the deterrent of a given severity to be
“necessary.” If a less severe deterrent would have the same deterrent effect,
then a better overall package of opportunity for each person is possible;
namely, one with the less severe deterrent. Second, it puts pressure on the
deterrent of a given severity to be “proportional” to the violation. If the only
deterrent that will deter a given violation is very severe, whereas such
violations have only small effects, then that deterrent may make the overall
package of opportunity worse.
4. DETERRENTS: THE DEONTOLOGICAL COMPLAINT
So much for the Distributive Complaint. Another complaint against the

state’s imposition of deterrents, however, seems to be staring us in the
face. Grant that imposing the deterrent achieves a greater good. Still,
there are certain things that we may not do to a person even to produce
a greater good. We may leave the one person in a pit in order to rescue
two others. But surely we may not push the one into the pit as a means
of rescuing two others. It’s not quite Thomson’s (1985) paradigm of
fatally pushing someone off a footbridge to stop a trolley that would
otherwise kill five, since the numbers and stakes for each are lower and
(arguably) since we are only “removing” not “using” the one. But it still
runs up against similar “deontological” resistance. Likewise, one might
protest on Violet’s behalf that imposing a deterrent on her violates a
deontological constraint on what may be done to a person even to
produce a greater good, such as:
Force Constraint: It is impermissible to use force on someone as a means to, or
foreseeable side-effect of a means to, a greater good (compare Kamm 2006).
And, it might be said, imprisoning Violet subjects her to force as a means
to, or a forseeable side-effect of a means to, a greater good. This, then, is the
Deontological Complaint Against State Imposition of Deterrents, or the
Deontological Complaint.
Dictionary: NOSD
46 Niko Kolodny
4.1 The Forceless State

With earlier candidate complaints, our first move was to show that removing
the target did not remove the complaint. However, it might seem impossible,
in this case, to remove the target. How can a state impose deterrent
imprisonment without force (Huemer 2013, 10)?
In fact, it takes only a little imagination. The state might build a cage
around Violet, while she sleeps in a public park, using materials she does not
own, without laying hands on her (directly or with the use of implements).
Now, it might be said that this still “harms” her by “actively” bringing it about
that she bears the loss that consists in confinement. And “active harming” is a
close cousin to force, subject to similar deontological constraints.
So, for good measure, imagine the Omittite Empire. Their emperor, the
Guardian of the Ladder, does not put violators of his directives in prison or
build prisons around them. He doesn’t need to. This is because each Omittite,
to survive the elements, must descend into his naturally carved hole each
night. Every morning, the Guardian drops the ladder into each hole to
enable its occupant to climb back up. His deterrent is simply to withhold the
ladder, confining the occupant there for a fixed period. Suppose an Omittite,
“Holton,” violates some directive, and so the Guardian, as announced, does
not drop the ladder into Holton’s hole for several months. This isn’t a use of
force or an “active harming,” it is simply a failure to aid.
To be sure, there are deontological constraints on refusals to aid, even for
the greater good. We may refuse to give life-saving medication to the one in
order to have it to give to the five. But we may not refuse to give life-saving
medication to the one in order to learn from the progress of his disease how
to save the five from it (Foot 2002, 28). I take it that this is explained by
something like:
Non-Aid Constraint: If one is otherwise required to aid someone, it is not sufficient
to release one from this requirement that by refusing to aid that person, one can use
or affect that person as a means to a greater good.
But is the Guardian refusing aid so as to use or affect Holton as a means to
the greater good?
The Guardian is withholding the ladder from Holton so that others,
among them Dieter, will be deterred from violating the directive. Dieter is
deterred by the combination of two beliefs. First, the Belief in Credibility:
Dieter’s belief that the Guardian won’t drop the ladder to Dieter, if Dieter
violates. Second, the Belief in Consequence: Dieter’s belief that this is
something for Dieter to avoid. Now, if the Guardian were withholding the
ladder from Holton so as to be able to make a spectacle of his confinement,
so as to sustain Dieter’s Belief in Consequence—as if to say, “Obey, lest ye
Dictionary: NOSD
suffer as, lo, this wretch suffers”—then he would be refusing aid so as to

use Holton as a means. But the Guardian doesn’t need to sustain Dieter’s
Belief in Consequence, and indeed probably can’t have much effect on it.
It’s obvious to Dieter that it will be a bad thing for Dieter if the ladder isn’t
dropped to him. He doesn’t need to be “scared straight.” The Guardian
needs to sustain only Dieter’s Belief in Credibility. And the means to
sustaining that belief is simply not dropping the ladder into Holton’s hole,
as if to say to Dieter: “Look, I mean business. The same will be done in
your case.” Nothing that happens to Holton as a result is part of the
Guardian’s means to the greater good. Put another way, the Guardian’s
deterrent aim would not be thwarted if (contrary to fact) confinement
were a benefit to Holton (if he needed and wanted more than anything
quiet respite without the temptation of escape) or if refusing to drop the
ladder to Holton did not confine or otherwise involve him (if he,
exceptionally, could survive the elements outside or climb out on his own).
Suppose the Guardian’s intelligence officers bring him two complete and
fully accurate dossiers: one on how Holton would be affected by
withholding the ladder, the other on how Dieter (as he believes) would. It
seems the Guardian has no reason to read Holton’s, but every reason to
read Dieter’s. If Dieter believes that confinement would benefit him or that
he would not be confined, then the Guardian’s deterrent aims will be
thwarted. But what will happen to Holton—the contents of his dossier—
are neither here nor there.
In sum, the Deontological Complaint cannot so much as be raised in
Holton’s case. And yet, intuitively, the Omittites’ forceless system of
deterrents seems not very different in its moral character from more familiar
forcible systems.8
4.2 The Natural Duty Argument

Suppose, however, that the state does not have the Guardian’s luxury. It
must use force in its deterrents. This brings us to our second response: to
show that the candidate target is not objectionable—or, rather, that those
8
Libertarians may reply: “whether there is an objection to the regime all comes down
to whether the Guardian owns the ladder. If he wove it from his own hair (and happened
upon the design by chance inspiration and not from any scarce genetic advantage, etc.),
then all’s hunky-dory. He’s just a private citizen going about his business. But if he wove
it from plant fibers (or did so without leaving enough and as good for others, etc.), well
then, he’s an enslaving tyrant.” If the libertarian’s concern turns on such subtleties about
the provenance of the physical instruments of deterrence, then it seems to me a long way
off from any traditional or commonsense concern about relations of rule.
Dictionary: NOSD
48 Niko Kolodny
who press a complaint cannot find it objectionable, compatible with their

other commitments.
Some try to show this with the Natural Duty Argument.9 Even proponents
of the Deontological Complaint must accept that:
1. Each individual has a natural duty to promote the common good.
We are assuming that the state is ideally directive, i.e. that:
2. No alternative set of directives that the state could issue and enforce
would better promote the common good.
So, it follows that:
3. The uniquely best way for any individual to promote the common
good is to comply with state directives.
So, it follows that:

4.
Each individual’s natural duty to promote the common good is
extensionally equivalent to a duty to comply with state directives.
Now, assume:

5.
Duty Permission: The Force Constraint is lifted, for purposes of
deterrence, when the target violates a duty.
Then it follows that:

6.
State Imposition: The Force Constraint is lifted, for purposes of
deterrence, when the target violates a state directive.
I don’t press the Natural Duty Argument. One reason is that it has limited
dialectical reach. First, some might deny Duty Permission.10 In particular,
9
The Natural Duty Argument is suggested by Quong (2011, ch. 4) and Wellman
(1996, 2005). Waldron (1993) argues for something like no. 4 in the argument, but does
not discuss the enforcement of directives.
However, Wellman (1996, 219 n. 13) says that his argument for the permissibility of
state coercion does not rest on anything like Duty Permission. Instead, the claims of the
target to be free from coercion are simply “outweighed” in cases of emergency rescue. But
this seems inadequate. The examples Wellman uses to motivate the claim of “outweighing”
appear to be either of (temporarily) commandeering someone’s property, or of issuing (as
opposed to following through on) threats. But what is presently at issue is something
different: following through on a threat with forcible action on someone’s person. And
it’s not at all intuitive that the Force Constraint is overcome merely because an emergency
rescue is underway. After all, our motivating case, of one person toppling another to save
two, was an emergency rescue.
10
Simmons (2005, 192), who affirms “the natural right of all persons to enforce
morality (by coercion, if necessary),” may accept Duty Permission. But Nozick (1974,
91–3) does not.
Dictionary: NOSD
they might say that the Force Constraint is lifted for violations of natural
prohibitions on the use of force, but not for violations of other duties.
Would they thereby draw an arbitrary distinction? Perhaps, but in advance
of hearing some explanation of Duty Permission, how can we know?
Second, some, especially libertarians, might deny that there is a natural duty
to promote the common good. They may accept only that there are natural
prohibitions on the use of force—or, at most, requirements to provide aid
in extreme circumstances.11
The other, more important reason for avoiding the Natural Duty
Argument is simply that it is invalid. Premise 2 that the state is ideally
directive does not imply 3 that the uniquely best way to fulfill one’s duty to
promote the common good is to comply with state directives.12 Simmons’s
well-known “particularity problem” supplies one reason for this “Gap”: this
divergence between what natural duty requires and what an ideally directive
state directs. Suppose that the natural duty is to contribute to the “global”
common good, and suppose that one can contribute to that at least as well
by complying with the directives of a foreign state. For example, a Swede
might pay Danish taxes instead of Swedish taxes. So her natural duty to
promote the common good does not imply a duty to comply with the
directives of the Swedish government to pay Swedish taxes, only a more
permissive duty to pay Swedish or Danish taxes (Simmons 1979, ch. 6;
2005, sect. 7).
But the Gap does not depend on “particularity,” so understood. Even if
we assumed a single world-state, the Gap would still be there, for reasons
familiar from debates over rule utilitarianism. There is often no way for the
state to carve out an exception for benign or beneficial individual actions
without worse consequences overall. To put it schematically: although it
promotes the common good at least as well for those in condition C to X,
it detracts from the common good for those not in C to X. And there might
be no way for the state to deter the latter without a blanket prohibition of
X-ing, whether or not one is in C. Countless examples fit this schema. In
the case of coordination problems, it might promote the common good at
least as well for those in a condition in which enough others will coordinate
to promote the common good in some other way, although it detracts from
the common good for those in a condition in which not enough others will
coordinate to do so. Similarly, it might promote the common good at least
as well for those in a condition in which they can act competently without
official authorization to act without official authorization, although it
11
Simmons (2000, 137), however, accepts a natural duty to promote the common
good.
12
Compare Murphy’s “basic structural point” (2014, 130).
Dictionary: NOSD
50 Niko Kolodny
detracts from the common good for those who cannot act competently
without official authorization to act without official authorization. Examples
would be skilled and responsible operation of a motor vehicle or practice of
medicine without a license, entry into a secured space without proper
identification, or the revelation of state secrets in the public interest.13
Similarly, it might promote the common good at least as well for those in a
condition in which it is known that their attempts at harmful acts will be
futile to attempt (such as the subjects of an undercover “sting” operation),
although it detracts from the common good for those whose attempts will
succeed to attempt.
In sum, an ideally directive state will have to impose deterrents for the
violation of directives to act in ways that are not required by any natural
duty to promote the common good. Because of this Gap, even if Duty
Permission is true, the state may still violate the Force Constraint in
imposing deterrents for the violation of such directives. Of course, one
might try to bridge the Gap with political obligations. But are there political
obligations? As we noted in section 2, it’s far from clear.
Indeed, instead of answering the Deontological Complaint, the Natural
Duty Argument seems only to reveal its force. Suppose we accept (i) the
Duty Requirement: that the only thing that can lift the Force Constraint,
absent consent, is the violation of a duty. And suppose that we accept
(ii) that there are not, in general, political obligations. Then we must accept
that, in light of the Gap, even an ideal state (unless it is, like the Omittite
Empire, forceless) will routinely violate the Force Constraint in imposing
deterrents for its directives. That is a simple and powerful complaint against
a relation of rule.14
4.3 The Avoidance Principle

The unmet ambition of the Natural Duty Argument was to show that those
who press the Deontological Complaint cannot consistently object to the
13
Compare the cases that Raz (1986, 74) advances to show that (at least as far as his
“Normal Justification Thesis” is concerned) the state’s directives in a given area often will
not have authority over citizens with specialized skills or knowledge in that area.
14
Somewhat surprisingly, Raz may be committed to this consequence, since he affirms
(ii), and there is some evidence that he affirms (i) as well. According to his Harm
Principle, “coercion,” at least, is permissible only to prevent someone from violating a
duty of autonomy, which suggests something like the Duty Requirement. Granted, Raz
(1986, 104, 148) stresses that the fact that we do not have a general duty to obey “even
laws which the government is justified in making” does not mean that the state is not
justified in using force or coercion to “enforce moral duties on those who are inclined to
disregard them.” However, by Raz’s own lights, the state enforces many directives that
citizens have no moral duty to obey.
Dictionary: NOSD
state’s imposition of deterrents for violations of its directives. In the rest of

this section, I will try to show this in a different way. I will argue that one
cannot consistently object to state imposition so long as one accepts an
elementary Lockean idea: that it is at least permissible to enforce the “law of
nature.” More precisely:
Natural Imposition: The Force Constraint is lifted, for purposes of deterrence, when
the target has violated a natural prohibition on the use of force.
To be sure, the ideally directive state’s directives go way beyond natural
prohibitions on the use of force. Again, they include directives to cooperate
to promote the common good in many other ways: for example, to
contribute to police protection and public education in the specific manner
that the state has decided. But I will argue that these differences between
natural prohibitions on the use of force and state directives simply don’t
matter to the permissibility of imposing deterrents for their violation. This
line of argument assumes neither Duty Permission, nor a natural duty to
promote the common good. Nor does it require political obligations. Again,
it assumes only Natural Imposition.
If we accept Natural Imposition, then we need some explanation of it.
Why is it that if some state-of-naturalist, Flintstone, violates a natural
prohibition on the use of force, then the Force Constraint is lifted for the
purposes of imposing a deterrent on him? It doesn’t help to say that by
punishing Flintstone we bring about the good of apportioning suffering to
desert. Even if there is such a good, and even if punishing Flintstone brings
it about, it isn’t goods brought about by punishment that we need to find.
We already have a greater good to be brought about by imposing a deterrent
on Flintstone; namely, protection from force. The “pro” column is already
drenched in ink. What we don’t have is an explanation of why the Force
Constraint, which usually prevents us from using force even to bring about
the greater good, should be lifted in this case.
What lifts the Force Constraint in this case, I suggest, may be captured
by a very simple principle:
Avoidance Principle: The Force Constraint is lifted when and only when the target
has or had adequate opportunity to avoid the use of force (deeply indebted to
Hart 1968 and especially Scanlon 1998, 1999).15
“Adequate” is determined by fairly balancing the two main interests at stake.
On the one hand there is the interest underlying the Force Constraint. This,
I would argue, is the target’s interest in not being subject to force by others
15
Note that even if the Force Constraint is lifted, it may still be wrong to use force.
For one thing, the use of force, unlike the ideal state’s enforcement, may not bring about
a greater good. For another, it may violate other deontological constraints.
Dictionary: NOSD
52 Niko Kolodny
that he does not control.16 On the other hand, there are the burdens that
others may have to bear in order to provide him with control. In some
circumstances, the only control that would count as adequate is the target’s
present consent. In other circumstances, however, weaker control is
adequate, given that the burdens that others would have to bear to provide
stronger control would be too great. In particular, it would burden others
severely to require Flintstone’s present consent, after he has violated a
natural prohibition, in order to impose a deterrent. This would make the
deterrent empty, since one could always escape its imposition by refusing to
consent to it. And others rely on the deterrent to sustain the credibility of a
threat that induces behavior that promotes the negative common good: a
just distribution of freedom from force. Hence, a weaker form of control
seems adequate in Flintstone’s case: the control exercised in complying with
the natural prohibitions. Flintstone’s adequate opportunity to avoid force
was his opportunity not to violate the natural prohibitions.
Why think that “adequate opportunity to avoid” is what does the work
in lifting the Force Constraint? In particular, why not just appeal to the
glaring fact that Flintstone has a duty? First, the fact that Flintstone has a
duty to refrain from force, by itself, is scarcely sufficient to impose a
deterrent on Flintstone, so as to induce others to refrain from force. After
all, if Flintstone had complied with his duty to refrain from force, then it
would be wrong to make him a scapegoat, even if this would be an effective
deterrent. Why? Because he did not have adequate opportunity to avoid
the force.
Second, it’s uncontroversial that, even when someone has no relevant
duty, the mere fact that he consents can lift the Force Constraint. The
Avoidance Principle explains this straightforwardly. Withholding consent
to force, when one had opportunity to withhold consent, is just a special
case of exercising an opportunity to avoid force.
Finally, even when someone has no relevant duty, the fact that he was
given control weaker than consent can, in the right circumstances, intuitively
lift the Force Constraint. Suppose again that we are rushing to save two
people from two-month-long entrapments in pits. In order to get there in
time, we have to forcibly knock Block, who is in our way, into a pit for a
month’s stay. If Block’s just stuck there in our way, then, as noted before, it
seems we can’t do it. But if he could easily step aside, and we make him fully
aware of the situation, and he still refuses, then I think we may knock him
16
I don’t claim that this interest in control explains why the Force Constraint has a
“deontological” or “agent-relative” character. After all, the five whom we do not save from
force might ask why their interests in control do not outweigh the interest of the one.
Why certain kinds of interests should give rise to “deontological” constraints is a difficult
question. My claim is only that this interest in control is among them.
Dictionary: NOSD
into the pit. Suppose, further, that it makes no difference to the success or
cost of the mission whether he is in the way. If he isn’t in the way, then we
don’t need to knock him. If he is in the way, then we do need to knock him,
but doing so is completely effortless. Then Block has no duty to step aside
(at least as far as the rescue mission is concerned). To repeat: his presence
there makes no difference to the success or cost of the mission. It’s not that
he has a duty to step aside, but rather that he cannot complain (at least not
on grounds of the Force Constraint) if, when he doesn’t step aside, we push
him in. This is because he had adequate opportunity to avoid.17
17
The Avoidance Principle captures, I think, the defensible part of a “rights forfeiture”
theory of punishment. See Goldman 1979; Kershnar 2002; Morris 1991; Simmons 1991;
and Wellman 2009, 2012. However, this account differs from rights forfeiture theories in
a number of respects.
First, the account doesn’t imply, as most rights forfeiture theories of punishment
maintain, that one forfeits a right only by violating a right, which is more or less the Duty
Requirement.
Second, this account does not, a fortiori, imply a strict equivalence between the right
violated and the right forfeited (which is what leads to Goldman’s (1979) “paradox”).
“Proportionality” is explained in the way described at the end of section 3 within this
chapter.
Third, this account also doesn’t imply, as some rights forfeiture theories imply, that if
one violates a right, then one forfeits a right for any purpose. It does not imply, for
example, if a sadist secretly inflicts pain on Flintstone without knowing that Flintstone is
a violator, then the sadist does not violate his rights. The Force Constraint is lifted only
for uses of force, such as deterrence, that provide others with goods that are sufficiently
important to justify Flintstone’s reduced control over others’ uses of force. Uses of force
in secret and for private satisfaction don’t provide others with such goods.
Finally, the Avoidance Principle offers a justification for the “forfeiture of rights,”
which rights forfeiture theories tend to leave mysterious. The justification, to put it in
terms congenial to the rights forfeiture theory, is that just as one can “waive rights”
through one’s choices, so, too, can one “forfeit rights” through one’s choices, when the
costs to others of greater “immunity to the loss of rights” would unfairly burden them.
“Waiver” and “forfeiture” are different answers, in different contexts, to the same basic
question: what sort of control over how others treat one is it fair to expect when balanced
against the costs that others must bear to provide one with such control?
Simmons (1991, 335) similarly appeals to fairness to explain why the Force Constraint
is lifted in Flintstone’s case, although, I think, in the wrong way. “[T]o extend such
privileges to those who break the rules,” he argues, “would seem to involve serious and
straightforward unfairness to those who limit their own liberty by obeying the rules.” The
thought appears to be that, if others bear burdens to respect the Force Constraint, but you
don’t bear them, then they are permitted to compensate themselves, and so equalize the
burdens, by not respecting the Force Constraint toward you. How does this compensate
them? Presumably, by providing them with deterrent protection. The trouble is that
unequal burdens borne in respecting the Force Constraint can arise even if no one has
violated the Force Constraint. In such a case, Simmons’s argument would seem to license
scapegoating to equalize burdens. In short, this seems the wrong way to think about fairness
in this context. The relevant question of fairness is how to balance the interests that the
Force Constraint is meant to protect against the interests that would be disadvantaged by
more extensive protection. The Avoidance Principle does this directly.
Dictionary: NOSD
54 Niko Kolodny
The Avoidance Principle, however, might seem obviously vulnerable to

counterexample. First, why suppose that Flintstone’s opportunity to avoid
was adequate? Suppose that the cost of compliance was death. He would
have died from organ failure had he not harvested the vital organs of his
victim. The reply is that “adequate” is a moralized notion. The target may
not cite, as a “cost” of exercising an opportunity to avoid force, that he
thereby had to “forgo” the benefits of wrongful conduct.
Second, suppose Coldfoot consented yesterday, with the best possible
opportunity to withhold consent, in the freest and most informed
conditions, to our pushing him off of a footbridge to stop the (slow but
inexorable) trolley. Today, without anyone having materially relied on his
consent, he says: “I no longer consent to being pushed.” Arguably, we may
not push Coldfoot.18 Or suppose that Hefty, with the best possible
opportunity to avoid doing so, in the freest and most informed conditions,
intentionally, knowingly, etc. steps onto an overpass, despite the sign that
reads, “If you are heavy enough, you may be pushed off to stop runaway
trolleys.” However, mounting the overpass, Hefty clearly announces, “I do
not consent to being pushed.” Again, many will deny that we may push
Hefty. We can’t set up deontology-free zones simply by erecting signage.
In other words, the opportunity to avoid that is intuitively adequate for
Flintstone, namely the opportunity to refrain from violation, is weaker
than the opportunity to avoid that is intuitively adequate for Coldfoot or
Hefty, namely the opportunity to withdraw or withhold present consent.
Why is this? Our point of departure is that others are not overly burdened
by a principle that grants Coldfoot (or Hefty) freedom from force provided
he didn’t consent yesterday (or doesn’t mount the overpass). Given that,
how much more are others burdened by a principle that grants Coldfoot (or
Hefty) more extensive control: that insists, as it were, on a waiting period
on Coldfoot’s gift (or further conditions on Hefty’s)? Not much, it would
seem. By contrast, while others may not be overly burdened by a principle
that grants Flintstone freedom from force provided that he does not violate
a natural prohibition on force, it seems they are significantly more burdened
by a principle that grants Flintstone freedom from force even if he does
violate. That extension of Flintstone’s control deprives them of the deterrent
and its protections. It asks a great deal of others.
If the Avoidance Principle is what explains Natural Imposition, then the
Deontological Complaint against State Imposition collapses. The
Avoidance Principle would seem to justify State Imposition as well. Just as
Flintstone had opportunity to avoid the deterrent, by complying with the
18
This suggests that even one-off, historical consent, of the kind that Locke envisioned,
may not suffice to answer the Deontological Complaint (Huemer 2013, 21 n. 3).
Dictionary: NOSD
natural prohibitions on force, so, too, Violet had opportunity to avoid the
deterrent, by complying with the state’s directives. And just as to provide
Flintstone with even greater opportunity (e.g. to require his present
consent) in order to impose a deterrent would burden others severely, so,
too, to provide Violet with even greater opportunity (e.g. to require her
present consent) in order to impose a deterrent would burden others
severely. Just as others rely on the deterrent in Flintstone’s case to sustain
the credibility of a threat that induces behavior that promotes the negative
common good, so, too, they rely on the deterrent in Violet’s case to sustain
the credibility of a threat that induces behavior that promotes the common
good, negative and positive.19
4.3.1 Replies: Opportunity to Avoid State Imposition is Inadequate

To salvage the Deontological Complaint, one needs somehow to drive a
wedge between Natural and State Imposition, so that State Imposition, but
not Natural Imposition, is ruled out.
One might reply that we can do this even while granting the Avoidance
Principle. While Flintstone’s opportunity to comply with natural
prohibitions is adequate, Violet’s opportunity to comply with state
deterrents is not adequate. So the Avoidance Principle explains why the
Force Constraint is lifted in Flintstone’s case, but not Violet’s.
Indeed, there are grounds for such a reply. We have granted that one
cannot cite as costs of exercising one’s opportunity to avoid that one had to
forgo benefits of wrongful conduct. Since Flintstone has a duty to exercise
his opportunity to avoid—i.e. to comply with natural prohibitions—it
seems fairly easy to explain why his opportunity counts as adequate. But if
19
Indeed, on this view, deterrents may be permissible even when the state is not
ideally directive. Even if the current set of directives is suboptimal, the “stern” message
sent by following through—“If you violate one of these directives, then you will suffer
the deterrent”—may have better effects than the “lax” message sent by not following
through—“If you violate one of these directives, then you may not suffer the deterrent.”
While, by definition, there are patterns of conduct better than general compliance with
the suboptimal directives, there may also be worse patterns of conduct. And the lax
message may only encourage such worse patterns. Assuming that people have had
adequate opportunity to comply with the suboptimal directives, the Deontological
Complaint might be met. Of course, the state should replace its suboptimal directives
with optimal ones. Indeed, it may be acting impermissibly in not doing so. The point is
that, if the state has not yet done so, then the message sent by its not following through
on the threats that it has made may be worse than its following through. Paradoxically
put, it may be permissible for the state to impose deterrents for violations of directives
that it has impermissibly issued and that it is permissible for individuals to violate. This
suggests, incidentally, that relaxing the assumption that the state is an ideal enforcer
makes the Deontological Complaint far harder to answer than does relaxing the
assumption that its directives are ideal.
Dictionary: NOSD
56 Niko Kolodny
Violet does not have a duty to exercise her opportunity to avoid by

complying—i.e. a political obligation to comply with the state’s directive—
it may be more difficult to show that her opportunity was adequate.
But, first, if we can assume a duty to promote the common good, then
this is less likely to present a problem—although, admittedly, this
assumption limits the dialectical reach of the argument. For the situation
will often be as follows. One can promote the common good in way X or
way Y. Neither is markedly more burdensome than the other, but either is
markedly more burdensome than refusing to promote the common good.
The state directive, however, is, specifically, to X. Can one complain, if a
deterrent is imposed for not X-ing, that one did not have adequate
opportunity to avoid? The main “costs” of X-ing were forgoing the benefits
of refusing to promote the common good at all. But, since one has a duty
to promote the common good, one cannot cite these “costs.” The only costs
of X-ing that one could potentially cite are forgoing the benefits of Y-ing.
But since Y-ing is about as burdensome as X-ing, there are no significant
benefits of this kind. Although one is not morally required to X, one cannot
claim that one did not have adequate opportunity to avoid, because all of
the other things that one might have permissibly done would have had the
same cost.
Second, even if there is no duty to promote the common good, complying
with many state directives, such as its ban on private enforcement, carries
little cost.
Finally, if certain familiar features of the “rule of law” are respected, then
there will be better opportunity to avoid state imposition than natural
imposition. Deterrents will be imposed only if they are specifically
announced in advance.
At best, then, this line of reply enjoys piecemeal success. In some cases,
under certain assumptions, there may be worse opportunity to avoid state
imposition than there is to avoid natural imposition. And so, in those cases,
it is less clear that the Avoidability Principle will sanction state imposition
as it sanctions natural imposition. Yet the Deontological Complaint, one
might have thought, was supposed to be more categorical.
4.3.2 Replies: Force May be Used Only for Protection from Force
So how else are we to drive a wedge between Natural and State Imposition?
Perhaps by rejecting or imposing a further constraint on the Avoidance
Principle, in such a way that Natural Imposition remains standing, but
State Imposition does not. But how to do this? One answer might be:
Rugged Individualism: Absent consent, force may be used on S only to protect others
from S’s force.
Dictionary: NOSD
This would rule out State Imposition, since the deterrents that the state
imposes on Violet will very often serve goods other than protection from
Violet’s force, such as protection from others’ force or the “protection” from
ignorance that education provides.
The problem is that Rugged Individualism also rules out Natural
Imposition. Imposing deterrents on S for violations of natural prohibitions
on force cannot be justified, in general, by others’ interest in being free from
S’s force. Suppose that, following his violation, Flintstone is reformed or
incapacitated so that there is no prospect of him using force in the future
(Otsuka 2003, ch. 3). In that case, imposing a deterrent on Flintstone does
nothing to serve the interest of his victim, Vic, in being free from Flintstone’s
force. It may well serve Vic’s interests in being free from another person,
Dieter’s, force, since it reinforces Dieter’s belief that anyone who uses force
on Vic will pay. And, as Locke (1689, §8) assumed, this was much of the
point of punishment: “as may make him repent of doing it, and thereby
deter him, and by his Example others.” But, according to the Rugged
Individualist, Vic’s interest in being free from Dieter’s force cannot justify
imposing a deterrent on Flintstone.20
To be sure, committed Rugged Individualists can deny Natural
Imposition. And they can still allow that nonconsensual force may be used
in defense—which, again, includes some forms of “after the fact” restitution
(Rothbard 1982, ch. 12–13). Again, my argument is directed only against
those who accept Natural Imposition. All the same, there are serious,
perhaps intolerable costs of rejecting it, which it isn’t clear that Rugged
Individualists have squarely faced. If we reject Natural Imposition, then
morality leaves Vic defenseless in cases like those just discussed.21
20
This point is easily obscured by confusing the threat to punish Flintstone, which
aims to prevent Flintstone’s use of force, and so might be justified by Vic’s interest in
being free from Flintstone’s force, with following through on the threat after Flintstone’s
violation, which does not defend against his violation. (Although Quinn 1985 argues that
what justifies the threat justifies following through, I find his argument, for reasons
which there isn’t space to explain here, unsuccessful.) So, for good measure, further
suppose that Flintstone was not even deterred by our threat. In that case, not even the
threat to Flintstone was justified by Vic’s interest in defending against Flintstone’s force,
since it did nothing to serve that interest. All the same, following through on the threat
serves Vic’s interest in deterring Dieter.
21
To my knowledge, Boonin (2008, ch. 5, especially section 5.11) offers the most
resourceful defense of replacing our system of punishment with a system of restitution
against, among other things, the objection that it would provide insufficient deterrence.
However, Boonin relies heavily on the idea that a violator owes restitution to third parties
for encouraging others to violate. But what encourages others is not the violation itself,
but instead the fact that the violator isn’t “brought to justice.” So to apply Boonin’s
approach to our current discussion would amount to including as part of “Flintstone’s
force” negative effects resulting from changes in others’ behavior resulting from
Dictionary: NOSD
58 Niko Kolodny
Compatible with Natural Imposition is the weaker:

Negativism: Absent consent, force may be used on S only to protect others from
anyone’s force: i.e. to promote the negative common good.
However, Negativism is also fully compatible with the minimal state, which
imposes deterrents for violations of directives to contribute to the negative
common good: e.g. to supply service or taxes to support policing and
defense. So Negativism would support the Deontological Complaint only
against a more expansive state.
In any event, Negativism is a far less stable position than Rugged
Individualism, extreme though the latter may be. Rugged Individualism
builds on a distinction that, vague and contested though it is, is accepted in
some form, by most non-consequentialists: a distinction between what S
does to others, regarding which “morality makes relatively strong claims” on
S—either in terms of what morality requires S to do, or in terms of what
morality allows to be done to S—and what merely happens to others (albeit
perhaps because S lets it happen), regarding which morality makes weaker
claims on S. Then Rugged Individualism takes this to an extreme: that
morality makes no claims on S (at least in the sense that morality allows
nothing to be done to S) with regard to what merely happens to others. The
Rugged Individualist doesn’t claim that it isn’t bad or doesn’t matter when
some ill befalls someone without S’s doing, while it is bad and does matter
when some ill befalls someone from S’s doing. “Yes,” the Rugged Individualist
agrees, “it’s worse if your son dies of cholera as a child that S could have
prevented than if S forcibly detains him as an adult, for an indecisive fifteen
minutes, before releasing him. But that isn’t the point. The point is that S is
responsible for what S does (again, in the sense that morality may make
claims on S regarding what S does) in a way in which S is not responsible
for what merely happens.”
But once we deny Rugged Individualism—once we grant that people’s
interest in protection from others’ force, which are not S’s doings, can
justify uses of force against S—how can we defend Negativism—how can
we deny that their interest in protection from ills other than force, which
are not S’s doings, can justify uses of force against S? If we can use force
against S to protect ourselves from the violence of other people, then why
can’t we use force against S to protect ourselves from the ravages of wild
animals? Why then can’t we use force against S to protect ourselves from the
Flintstone’s not suffering a deterrent. But this would make even State Imposition
compatible with Rugged Individualism, since imposing a deterrent on Violet protects us
from “Violet’s force” in the same sense: from negative effects resulting from changes in
others’ behavior resulting from Violet not suffering a deterrent.
Dictionary: NOSD
ravages of microbes? And so on. Here the answer can’t be that S is responsible
only for what S does. Here it indeed begins to look like, in order to defend
Negativism, we do need to assert that it somehow isn’t bad or doesn’t matter
when some ill befalls someone without anyone’s doing, but is bad and does
matter when some ill befalls someone by someone’s doing. And that idea is
lunatic.
4.3.3 Replies: the Duty Requirement

Again, we are trying to find a way to reject or to impose a further constraint
on the Avoidance Principle so that Natural Imposition remains standing,
but State Imposition does not. One last possibility is to insist on the Duty
Requirement, which we can now state more precisely as:
Duty Requirement: Absent consent, the Force Constraint is lifted only when the
target has or will otherwise violate a duty (or, one might add, “infringe” a right
without violating a duty).22
Since natural prohibitions are duties, Natural Imposition is compatible
with the Duty Requirement. But if there are no duties to comply with state
directives, then the Duty Requirement rules out State Imposition.
We have already seen reasons to reject the Duty Requirement. Again,
Block’s case seems a counterexample to it. It is permissible to knock him
into the pit, because he can step aside, even though he has no duty to step
aside (because his stepping aside contributes nothing to the rescue). And we
have a theory of error for the Duty Requirement. What really lifts the Force
Constraint is adequate opportunity to avoid. It’s just that the presence of a
duty to do what will avoid force can contribute to making that opportunity
adequate in a moralized sense.
Is there, then, any reason to accept the Duty Requirement? Why think
that it is true? Of course, it is inappropriate to condemn the target if he
hasn’t done anything wrong. But imposing a deterrent, to achieve its
function of inducing cooperation, need not involve condemnation.
Perhaps one might support the Duty Requirement by expanding Rugged
Individualism into:
Rugged Individualism + Desert: Absent consent, force may be used on S only to
protect others from S’s force or to give S what S deserves.
22
There is a different principle in the vicinity of Duty Requirement, put forward in
the excellent Tadros (2011), namely that the Force Constraint is lifted only when the
target has a duty to bear the costs that the force imposes, or would have such a duty in an
otherwise similar situation where there was something that the target could actively do so
as to bear those costs. I find this view, while ingenious, ultimately undermotivated and
overly constraining. But I don’t have space to discuss it here.
Dictionary: NOSD
60 Niko Kolodny
But this view seems an odd hybrid, linking together personal interests in
freedom from force with the impersonal good of restoring the world to
karmic balance. If these two goods, why not others?
To sum up: If we accept Natural Imposition, then the Deontological
Complaint Against State Imposition collapses. For the resources that we
must invoke to explain why the Force Constraint should be removed in
Flintstone’s case for natural imposition equally explain why the Force
Constraint should be removed in Violet’s case for state imposition. There is
no load-bearing difference between the two cases.
5. IS THERE A DEONTOLOGICAL

COMPLAINT AGAINST THREATS?
In sections 3 and 4 within this chapter, we considered the possibility that

the complaint is against the imposition of deterrents. But might the
complaint be against threatening to impose deterrents, whether or not they
are imposed? By “coercion,” after all, many have in mind coercive threats.
To begin with, let us remove the target. Imagine that tomorrow common
knowledge of dispositions to comply were to emerge spontaneously. And
imagine that, in this Trusting Future, the state stops backing up its directives
with threats. Still, it continues to issue directives, to coordinate our behavior,
and so to shape our natural and social environment profoundly, comprehensively,
and inescapably. Would the complaint disappear? Perhaps. But I suspect that
many would answer that a complaint would persist. Would Rawlsians, for
example, concede that, in our Trusting Future, the state would be permitted to
issue directives that could be justified only by a sectarian doctrine (Bird 2014;
Quong 2013, 271–3)?
Next, we can ask whether the state’s threats are objectionable. If there is
a deontological constraint that applies to the state’s threats, then, it is true,
we cannot appeal to the Avoidance Principle. For even if we have adequate
opportunity to avoid the imposition of deterrents, by complying with the
state’s directives, we have no opportunity to avoid threats of their imposition.
We are all born to such threats.
But is there a deontological constraint that applies to the state’s threats? It
can’t be said that threats are wrong as a rule. Presumably, threats to defend
oneself or to impose natural deterrents are permissible. And threats lie, along
with warnings and offers, on a continuum of announcements that something
will be done if something else isn’t. Not only are many such announcements
perfectly permissible, but also the boundaries between threats and other such
announcements are far from clear. To tell whether there is a deontological
constraint on state threats, therefore, we need to know why threats, or more
Dictionary: NOSD
broadly such announcements, are wrong when they are. Kolodny (ms. a)
explores various possible answers. Here I summarize briefly why none of
these answers would count our ideal state’s threats as wrong.
First, it is often said that threats are wrong just when what they threaten
is wrong. I don’t think this is true. An overwhelming response to a nuclear
first strike is wrong, but permissible to threaten. But, in any event, we are
supposing at this point in the argument that what the state threatens is not
wrong. The state is permitted to impose deterrents.
Second, threats may be objectionable when and because they leave the
opportunity of the threatened person worse than he is entitled to from the
threatener. So long as the state is ideally directive, however, it leaves each person’s
opportunities overall as good as is within the state’s power to leave them,
compatibly with a fair distribution of opportunities for others. How could
anyone be entitled to more from the state? So, again, there is no complaint.
Finally, some threats may be objectionable even though they do not leave
the opportunities of the threatened person worse than he is entitled to from
the threatener. Blackmail and abuse of office are the paradigms. But, in
brief, these involve factors unlikely to arise in the case of state threats.
In summary, just as there was no sound complaint against state force,
there is no sound complaint against state threats.
6. IS THERE A DEONTOLOGICAL

COMPLAINT AGAINST TAXATION?
Perhaps the complaint, then, is to the state’s use of our external property,
not simply in compensation or deterrent fines, but also in taxation.23 To be
clear, I have discussed two other state actions that might be described as “the
use of our property.” First, I have discussed the permissibility of the state’s
use of force in imposing deterrents, which is the state’s use of our bodies,
which might be said to be our property. Second, I have discussed the
permissibility of the state’s inducing us, by threat, to act in ways that
contribute to the common good. The state might induce us to build a well
or stand sentry. This might be described as the use of our labor, which might
be said to be our property. Whether or not such descriptions are accurate or
illuminating, we have already discussed what they purport to describe.
The subject that we have not yet discussed is the state’s use of our property
in the most natural, literal interpretation of the phrase: its use of some
23
We might also include commandeerings of private property, or dispositions of
public property, such as public land, buildings, and equipment—or, more abstractly,
things done with “our flag” or in “our name.”
Dictionary: NOSD
62 Niko Kolodny
object, not itself part of our bodies, that nonetheless belongs to us. It might
be argued that just as there is a deontological constraint on the use of our
bodies, there is a deontological constraint on the use of our external
property. And it might be argued that taxation violates this deontological
constraint. Taxation does something morally akin to invading and removing
parts of our bodies, akin to draining the blood from our very veins.
First, we can remove the candidate target. After all, we did not assume
that the Guardian of the Ladder taxed his subjects. The complaint would
remain, I take it, even if his empire was self-financed.
Second, can those who hold that there is a complaint against the state
legitimately hold that taxation violates a deontological constraint, akin to
an invasion of the body? Only if they accept:
Natural Property: There are rights in property other than those assigned by a system
that reliably secures the common good.
Taxation by an ideally directive state is itself part of a system that reliably
secures the common good. So if Natural Property is false, if people have
property rights only in what such a system assigns them, then taxation
defines rather than violates their property rights. And some who hold that
there is some complaint against some relation of rule reject Natural Property.
So their complaint against the state can’t be that taxation violates property
rights.24 Moreover, even if one accepts Natural Property, the Avoidability
Principle may still license taxation. There might be adequate opportunity to
avoid taxation, just as there is adequate opportunity to avoid the force used
in deterrents.
Suppose, however, that one holds both (i) Natural Property, and (ii) that
there is not adequate opportunity to avoid taxation. Then, I grant, one
could hold that there is a complaint against taxation. But it bears emphasizing
how strange the resulting position would be. For it has already been granted
that there is no complaint against the state’s use of our labor: its directing
us, under threat, to act in certain ways. The resulting position would be that
there is a complaint only against the state’s use of the material fruits of our
labor. And yet one might have thought that the complaint against the state’s
use of our labor had far greater power than the complaint against the state’s
use of its products. Nozick (1974, 169–71), for example, implicitly
acknowledges this when he argues that taxation is objectionable because it
is “on a par with forced labor.”
24
An example is Thomas Nagel, who voices the complaint in Nagel (1991) while
rejecting Natural Property in Nagel and Murphy (2004). Consider also left-libertarians
who view taxation as justified by uses or appropriations of the external world involved in
the production of the taxed property (Vallentyne 2012).
Dictionary: NOSD
7. THE SUBORDINATION COMPLAINT
For all the alarm stirred by the words “force,” “violence,” “power,” “coercion,”
“authority”—for all of the insistence that some relation of rule provokes a
special justificatory demand met only by special limits or conditions—we
have struggled to find a complaint that would account for it. Either the
apparent target of the complaint can be removed without removing the felt
complaint, or there isn’t, on reflection, anything in the apparent target to
complain about consistently (unless one insists on Rugged Individualism or
Natural Property). To put it another way: surely those who insist on consent
or acceptability for political rule don’t require consent or acceptability for
everything. You may drain your stagnant pool, or give a church sermon
without my consent, and you may do so for entirely sectarian reasons,
which I cannot be expected to accept. But, the thought runs, political rule
is different. Something about political rule calls for consent or acceptability.
But we still have not found what this “something” could be.
So I explore a possible answer. What animates the sense of complaint
may be an anxiety that to be subject to political rule is to be put in relations
of subordination or social inferiority to other individuals with whom one
has a claim to stand as an equal.
To explain: I take it that we grasp intuitively the notion of relations of
subordination or social inferiority: that, in virtue of how a society is
structured, some people can be—in a sense that is perfectly familiar, even if
its analysis is elusive—“above” and others “below.” We know the paradigms.
The servant is “subordinate” to the lord of the manor, the slave “subordinate”
to the master, and so on. If asked to place various social groups in a hierarchy,
we do this with ease. The plebian is “lower than” the patrician, the untouchable
“beneath” the Brahmin, and so on. But what is it in the paradigms that
provokes this unease? What are relations of subordination or social
inferiority, exactly? Kolodny (2014b) suggests at least a rough and partial
analysis. At its core is the suggestion that subordination consists in being
exposed to the greater power and de facto authority of another individual.
The greater power need not be of force or violence, but instead of
withholding goods or altering another’s environment. The greater de facto
authority consists in the capacity to issue directives that are generally, if not
exceptionlessly, followed by others (whether because of moral belief, threat,
salience, or mere force of habit). Crucially, subordination consists, at least
in part, in the asymmetric relations of power and authority themselves,
independently of how, if at all, that power and de facto authority are
exercised. Thus, the familiar republican paradigms of the kindly slave-
master, the aristocrat given to noblesse oblige, the colonial administrator
who bears the “white man’s burden,” and so on, count, as they intuitively
Dictionary: NOSD
64 Niko Kolodny
seem to count, as forms of subordination, even though the power and

authority are exercised beneficently and fairly, without violating any
independent deontological constraints.25
To be sure, not all asymmetries of power and authority are political. Nor
do all asymmetries of power and authority constitute subordination (of an
objectionable kind, at any rate). However, asymmetries of power and
authority tend to constitute subordination when they have certain features.
And asymmetries in political power and authority characteristically have
these features.
One such feature is that it is costly or difficult to escape the relations
that are marked by the asymmetry of power and authority. To return
to the paradigms: if one can exit a slave “contract” at will, then it is not
clear in what sense one really is a slave. The point isn’t that while social
inferiority is always a burden, one forfeits one’s complaint when the
burden is self-imposed. It is rather that the freer one is to exit what would
otherwise be a relation of social inferiority, the less it seems a relation of
social inferiority in the first place. And one typically cannot escape
the effects of political decisions at will, or at least not without high cost
or difficulty.
Another feature is that the relevant power or authority is final. This is
because one strategy for avoiding or moderating the subordination that an
asymmetry would otherwise entail is for there to be equality higher up, as it
were, the chain of command: a decision that sets the terms for how the
asymmetric power or authority lower down is to be exercised. However, this
strategy of avoidance or moderation is not available when the power and
authority are final: when, as it were, there is no higher court of appeal. And
such finality is characteristic of political power and authority.
The Subordination Complaint is, then, that in being subjected to the
state’s decisions, we are subordinated to other people, with whom, as
individual persons, we have a claim to equality. This seems to follow from
two very simple and plausible—although, in the end, misleading—ideas.
First, in being subjected to the state’s decisions, we are exposed to the
greater, inescapable, and final power and de facto authority of the state.
25
This concern with subordination may remind many readers of the concern with
“domination” in the revival and development of the republican tradition—most
prominently by Pettit (1999, 2012)—and the concern with “dependence” in the recent
revival and development of Kant’s legal and political philosophy—most prominently by
Ripstein (2009). Indeed, I believe that non-subordination, non-domination, and
independence are rival interpretations of the same underlying concern: different analyses
of the sort of anxiety evoked by republican paradigms like the kindly slave-master.
Kolodny (ms. b) argues that the underlying concern is better interpreted as a concern
about subordination.
Dictionary: NOSD
And, second, the state, once the robes and badges are peeled off, is just
other people.
Can the Subordination Complaint be the complaint that we’ve been
after? It fits its contours remarkably well. First, the Subordination Complaint
would not apply to what is typically held up as the antithesis of relations of
rule: namely, a Lockean state of nature. For, at least when the Lockean state
of nature is depicted attractively, it is a condition of equality, wherein no
man rules over any other.
Second, the Subordination Complaint would apply just as well to the
Omittites. As we have assumed, the Guardian of the Ladder wields his
greater power and authority so as to promote the common good. And he
does so in a way that never actually violates any independent deontological
constraints. But this does not change the fact that every other Omittite is
exposed to his greater, final, and inescapable power and de facto authority.
Third, the Subordination Complaint might similarly survive into our
Trusting Future. Although the state no longer threatens, let alone follows
through on threats, the state still enjoys vastly greater de facto authority,
through its asymmetric capacity to coordinate. It also enjoys vastly greater
power to affect one’s natural and social environment.
Finally, the Subordination Complaint would help to explain why “force”
and “coercion” suggest themselves as targets for the complaint. What makes
force special for political philosophy is not simply the badness of its effects
or the deontological constraints that govern its use. It is instead that the
power to use force typically, albeit contingently, determines the social
hierarchy. The power to use force is usually the final power: the power that
regulates and controls the exercise of other powers. In the ordinary run of
human affairs, an individual enjoys superior powers of other kinds only if
he also enjoys superior powers to use force. (After all, what prevents some
Omittite from seizing the ladder, if not that the Guardian can fend him off?
Of course, we can imagine that the Guardian has no capacity for force.
Instead, he is protected by a magical shield that frustrates any attempt to use
force against him. But things usually aren’t like that.) The power to coerce,
at least when understood as inducing another to act by giving him, in some
sense, “no choice,” is final in a similar way.
If the complaint is the Subordination Complaint, then how, if at all, can
it be overcome? Setting aside whether a retreat to the minimal state, or
consent, or acceptability would suffice to overcome the Subordination
Complaint, none of these is obviously necessary. It is true that in being
subjected to the state’s decisions, we are exposed to the greater power and de
facto authority of the state. And it is true that the decisions of the state are
just the decisions of people. Yet, it does not follow from this that, in being
subject to the state’s decisions, one is subordinated to another individual.
Dictionary: NOSD
66 Niko Kolodny
That would follow only if the decisions of the state represented the
superior power and authority of some other individual. And, at least in
principle, we can imagine democratic arrangements, in which each
person has equal opportunity to influence the state’s decisions or the
delegation of making them. In that case, it could be argued, merely by
virtue of being subjected to the state’s decisions, we would not be
subordinated to any other individual. There would be no other person,
qua citizen, of whom I could say: “Because he has greater opportunity
than I have to influence what the state does, in being subjected to its
decisions, I am subordinated to him.” Much more needs to be said, of
course (Kolodny 2014b), but if there could be a state whose decisions did
not represent the subordination of any individual to any other, then there
would be no need to withdraw to the minimal state or to wheel in consent
or acceptability.26 This is because the complaint that these measures would
be taken to overcome—namely, the Subordination Complaint—would simply
not arise in the first place.27
At this point, one might protest that if the complaint can be addressed by
equal opportunity to influence political decisions, then it can’t count as the
26
This point may illuminate the structure of Estlund’s (2009) defense of democracy:
very roughly, that no alternative to democracy is acceptable. The immediate difficulty,
which Estlund himself raises, is that it isn’t clear that democracy itself is acceptable.
Estlund replies that because democracy does not subject some to “rule by others” in the
same way, it enjoys a kind of default status; it does not need to meet the same conditions
of acceptability as alternatives (36–8). For a time, this reply struck me as ad hoc. But if
acceptability is a response to the Subordination Complaint, then it isn’t ad hoc at all. If
democracy doesn’t subordinate, then it simply doesn’t raise the objection that acceptability
is required to meet.
27
This response to the Subordination Complaint may also respond to the style of
anarchist argument in Huemer (2013). “Your initial intuition,” the argument runs, “is
that you have no objection when the state does what it does to you, although you would
have an objection if your neighbor did the same thing. You would object, for example, if
your neighbor imprisoned you in his basement because you violated the (admittedly
ideal) directives that he issued to improve your local sewer system. However, on reflection,
there is no relevant difference between your neighbor and the state. So to be consistent,
you ought either to deny that you have an objection to your neighbor, or to concede that
you have an objection to the state. Since the thought that you have no objection to the
state can, among other things, be plausibly explained as false consciousness, you ought to
concede that you have an objection to the state.” Our response is that, on the one hand,
if we set aside the concern about subordination, neither what the state nor what the
neighbor does to you is any different from what Flintstone’s fellows do to him, which
many anarchists would find acceptable. On the other hand, if we include the concern
about subordination, then perhaps it explains our initial intuitions. In issuing orders to
you, backed by threats of confinement, your neighbor puts you in a relation of
subordination to him. (“But, like,” you might grumble, languishing in his basement,
“who died and made him king?”) Things are different when the issuing and enforcing of
commands are not his personal fiat, but are instead regulated by a process that is no more
his diktat than one’s own.
Dictionary: NOSD
sort of complaint that we were looking for. We were looking for a complaint
that would persist even for a state that was achieving the common good.
But a state that was achieving a just distribution of means to a fulfilling life
would already be a state that distributed equally opportunities for political
influence. The more concessive response is simply to grant the point. If
we keep accounts carefully, we see that the ideal state is, by definition, not
vulnerable to any complaint, not even the Subordination Complaint, after all.
The less concessive response, which better fits the case, is that the opportunity
to influence political decisions isn’t primarily important as a means to a
fulfilling life (Kolodny 2014a). Equal opportunity for political influence is
more like unanimous consent or acceptability than like a fair distribution of
bodily security or a basic income. It matters principally as a response to a
concern about relations of rule, rather than as fair distribution of means for
a fulfilling life. Including equal opportunity for political influence as part of
the common good would be like including unanimous consent to the state
as part of the common good. It would indeed guarantee that a state that
realized the common good, so defined, was immune to the complaint. But
it would nonetheless be a kind of category mistake.
8. CONCLUSION: A PROBLEM OF EQUALITY,

NOT FREEDOM
The problem posed by relations of rule, it is commonly thought, is the

problem of reconciling them with the freedom of the individual, whether
this is understood as the individual being protected from things being
done to her person or field of choice, or as the individual being enabled to
do things. Negatively, I’ve suggested that if this is the problem posed by
relations of rule, then there isn’t a problem, at least not a problem beyond
simply showing that things are better with the state. If we think of freedom
as a good that state action brings about, then so long as the state is reliably
achieving the common good, then the state is bringing about that good for
each to the greatest extent compatibly with fairness to others. If we think of
freedom as a deontological constraint on what the state does, then the state
need not violate such a constraint.
More positively, I’ve conjectured that if the relations of rule that the state
involves pose a problem, it is a problem not of reconciling those relations
with the freedom of the individual, but instead of reconciling those relations
with an ideal of equality among individuals, understood as not being
subordinated to any individual as an inferior to a superior. Or if this is a
problem of reconciling relations of rule with individual freedom, the notion
of “freedom” in play is quite different. It isn’t being insulated from invasion
Dictionary: NOSD
68 Niko Kolodny
or interference, nor is it being resourced to chart a course through life.

Instead, it’s something like being “without a master.”
I leave open how much we should care about this Subordination
Complaint. Perhaps it is overblown or anachronistic. I also leave open
whether any conceivable state might avoid it. Again, my aims in this chapter
are more preliminary: to draw attention to the influence of a concern about
subordination over our thinking about relations of rule and to suggest that
if relations of rule pose some special problem, this is where it lies.
Bibliography
Bird, Colin (2014). “Coercion and Public Justification,” Politics, Philosophy,
Economics 13: 189–214.
Blake, Michael (2001). “Distributive Justice, State Coercion, and Autonomy,”
Philosophy and Public Affairs 30: 257–96.
Boonin, David (2008). The Problem of Punishment (New York: Cambridge University
Press).
Dworkin, Ronald (1986). Law’s Empire (Cambridge, MA: Harvard University
Presss).
Dworkin, Ronald (2011). Justice for Hedgehogs (Cambridge, MA: Harvard University
Press).
Edmundson, William A. (1998). Three Anarchical Fallacies (Cambridge: Cambridge
University Press).
Estlund, David (2009). Democratic Authority (Princeton, NJ: Princeton University
Press).
Foot, Philippa (2002). “The Problem of Abortion and the Doctrine of Double
Effect,” in P. Foot, Virtues and Vices (Oxford: Oxford University Press), 19–32.
Goldman, Alan (1979). “The Paradox of Punishment,” Philosophy and Public Affairs
9: 42–58.
Hart, H. L. A. (1968). “Prolegomenon to the Principles of Punishment,” in H. L. A.
Hart, Punishment and Responsibility: Essays in the Philosophy of Law (Oxford:
Oxford University Press), 1–27.
Huemer, Michael (2013). The Problem of Political Authority (New York: Palgrave).
Kamm, Frances (2006). Intricate Ethics (New York: Oxford University Press).
Kershnar, Stephen (2002). “The Structure of Rights Forfeiture in the Context of
Culpable Wrongdoing,” Philosophia 29: 57–88.
Klosko, George (2005). Political Obligations (New York: Oxford University
Press).
Kolodny, Niko (2014a). “Rule Over None I: What Justifies Democracy?” Philosophy
and Public Affairs 42: 195–229.
Kolodny, Niko (2014b). “Rule Over None II: Social Equality and the Justification of
Democracy,” Philosophy and Public Affairs 42: 287–336.
Kolodny, Niko ms. a: “What Makes Threats Wrong?”
Dictionary: NOSD
Kolodny, Niko ms. b: “Being Under the Power of Others.”

Larmore, Charles (1999). “The Moral Basis of Political Liberalism,” Journal of
Locke, John (1689). Second Treatise of Government, in P. Laslett (ed.), Locke: Two
Treatises of Government, 3rd ed. (Cambridge: Cambridge University Press, 1988).
Morris, Christopher (1991). “Punishment and Loss of Moral Standing,” Canadian
Journal of Philosophy 21: 53–79.
Murphy, Liam (2014). What Makes Law (New York: Cambridge University Press).
Nagel, Thomas (1991). Equality and Partiality (New York: Oxford University Press).
Nagel, Thomas (2005). “The Problem of Global Justice,” Philosophy and Public
Affairs 33: 113–47.
Nagel, Thomas and Murphy (2004). The Myth of Ownership (New York: Oxford
University Press).
Nozick, Robert (1974). Anarchy, State, and Utopia (New York: Basic Books).
Olsaretti, Serena (2009). “Responsibility and the Consequences of Choice,”
Proceedings of the Aristotelian Society 109: 165–88.
Otsuka, Michael (2003). Libertarianism without Inequality (Oxford: Oxford
University Press).
Pettit, Philip (1999). Republicanism (New York: Oxford University Press).
Pettit, Philip (2012). On the People’s Terms (New York: Cambridge University Press).
Quinn, Warren (1985). “The Right to Threaten and the Right to Punish,” Philosophy
and Public Affairs 14: 327–73.
Quong, Jonathan (2011). Liberalism without Perfection (New York: Oxford University
Press).
Quong, Jonathan (2013). “On the Idea of Public Reason,” in J. Mandle and D. A.
Reidy (eds), A Companion to Rawls (Chichester: Wiley-Blackwell), 265–80.
Rawls, John (1971). A Theory of Justice (Cambridge, MA: Harvard University Press).
Rawls, John (1993). Political Liberalism (New York: Columbia University Press).
Raz, Joseph (1986). The Morality of Freedom (Oxford: Oxford University Press).
Raz, Joseph (1994). “Authority, Law, and Morality,” in J. Raz, Ethics in the Public
Domain (Oxford: Oxford University Press), 210–37.
Ripstein, Arthur (2009). Force and Freedom (Cambridge, MA: Harvard University Press).
Rothbard, Murray (1982). The Ethics of Liberty (Atlantic Highlands, NJ: Humanities
Press).
Scanlon, T. M. (1998). What We Owe to Each Other (Cambridge, MA: Harvard
University Press).
Scanlon, T. M. (1999). “Punishment and the Rule of Law,” in H. Hongju Koh and
R. Slye (eds), Deliberative Democracy and Human Rights (New Haven, CT: Yale
University Press), 257–71.
Simmons, A. John (1979). Moral Principles and Political Obligations (Princeton, NJ:
Princeton University Press).
Simmons, A. John (1991). “Locke and the Right to Punish,” Philosophy and Public
Affairs 20: 311–49.
Simmons, A. John (2000). “Justification and Legitimacy,” in Justification and Legitimacy:
Essays on Rights and Obligations (Cambridge: Cambridge University Press), 122–57.
Dictionary: NOSD
70 Niko Kolodny
Simmons, A. John (2005). “The Duty to Obey and Our Natural Moral Duties,” in
A. J. Simmons and C. H. Wellman, Is There a Duty to Obey the Law? (New York:
Cambridge University Press), 93–196.
Simmons, A. John and Wellman, Christopher Heath (2005). Is There a Duty to Obey
the Law? (New York: Cambridge University Press).
Tadros, Victor (2011). The Ends of Harm (Oxford: Oxford University Press).
Thomson, Judith Jarvis (1985). “The Trolley Problem,” Yale Law Journal 94: 1395–415.
Vallentyne, Peter (2012). “Taxation, Redistribution and Property Rights,” in
A. Marmor (ed.), The Routledge Companion to Philosophy of Law (New York:
Routledge), 291–301.
Van der Vossen, Bas (2015). “Imposing Duties and Original Appropriation,” Journal
of Political Philosophy 23: 64–85.
Waldron, Jeremy (1993). “Special Ties and Natural Duties,” Philosophy and Public
Affairs 22: 3–30.
Wellman, Christopher Heath (1996). “Liberalism, Samaritanism, and Political
Legitimacy,” Philosophy and Public Affairs 25: 211–37.
Wellman, Christopher Heath (2005). “Samaritanism and the Duty to Obey the
Law,” in A. J. Simmons and C. H. Wellman, Is There a Duty to Obey the Law?
(New York: Cambridge University Press).
Wellman, Christopher Heath (2009). “Rights and State Punishment,” Journal of
Wellman, Christopher Heath (2012). “The Rights Forfeiture Theory of Punishment,”
Ethics 122: 371–93.
Dictionary: NOSD
3
Consent and Political Legitimacy
Amanda Greene
1. INTRODUCTION
Political legitimacy rests on the consent of the governed.1 Many theories of

legitimacy accept this basic principle in some form, but they differ on how
consent is to be understood. Today, those who believe that consent matters
for legitimacy typically articulate its significance via theoretical idealizations.
Those idealizations have costs. I will argue that we need not incur those costs,
and indeed we ought not incur them, in a consent-based theory of legitimacy.
Ultimately, I will defend an alternative view that avoids those costs and has
considerable advantages. While I cannot defend a comprehensive account
of political legitimacy here, I hope to say enough to motivate an approach
to legitimacy that is, I believe, superior to the dominant approaches in
political philosophy.
The views I criticize fall into two camps. Those in the contractualist family
believe legitimacy rests on hypothetical consent, where this consent is
imputed based on certain presumptions about the agent’s evaluative stance.
However, these presumptions fail to adequately respect the subject’s interest
in voluntary rule, since legitimacy does not depend on whether subjects
actually consent. On the other hand, those in the voluntarist family believe
1
For helpful comments on earlier drafts of this chapter, I would like to thank Samuel
Asarnow, Mark Budolfson, Lee-Ann Chae, Alan Code, Joshua Cohen, Brian Coyne,
Bryan Cwik, Jorah Dannenberg, David Estlund, Blake Francis, David Hills, John
Horton, Jeff Howard, Abby Everett Jacques, Dhananjay Jagannathan, Richard Kraut,
Cristina LaFont, R. J. Leland, Nethanel Lipshitz, Benjamin Miller, Katherine Meadows,
Jane Mansbridge, Martha Nussbaum, Josiah Ober, Debra Satz, Kevin Vallier, Han van
Wietmarschen, Allen Wood, and anonymous referees. I am also grateful for comments
from audiences at Stanford University, the University of Chicago, Northwestern University,
and the 2014 Oxford Studies in Political Philosophy Workshop. I wish to acknowledge the
generous support of the Stanford Humanities Center, the Institute for Humane Studies,
and the University of Chicago Law School.
Dictionary: NOSD
72 Amanda Greene
legitimacy rests on actual consent, but an individual’s consent renders the

government legitimate only for that individual. However, since the
unanimous consent of all subjects is unattainable, these views must idealize
about the degree of unanimity required to achieve legitimacy overall, or else
accept that no past or present regimes are legitimate. The shortcomings of
both views seem to create an impasse. Neither side can see an alternative
consent-based theory that avoids these costs. I will argue for a new view that
does so, while retaining much that is of value in the other views. A discussion
of these shortcomings will occupy the first portion of the chapter. In the
second part of the chapter, I present the new view and show why it has
significant advantages as a theory of political legitimacy.
In order to explain the shortcomings of current consent views of
legitimacy, I want to begin with some general thoughts about how the
question of legitimacy is to be theorized. One basic question concerns the
objects of our assessments of legitimacy. I take the paradigmatic object of
legitimacy assessments to be political orders broadly construed, which I will
call regimes.2 For the purposes of this chapter, I take legitimacy assessments
to be addressing the question of when a regime has appropriate standing to
exercise power over its subjects.3 It is also possible to assess the legitimacy of
ruling parties, policies, laws, and even particular actions, but in this chapter
I will be concerned with the overall political or constitutional order.
Second, I observe that many theorists subscribe to a strict distinction
between normative and descriptive legitimacy.4 I think this dichotomy is
too strong. Theorists ought to give more consideration to how the term is
used in non-philosophical discourse, where it is more or less used in the
sociological sense first identified by Max Weber.5 In these discourses,
broadly speaking, legitimacy is a social fact about whether the subjects of
a regime believe it has authority. In my view, it is worthwhile to explore
what normative significance this sociological concept may have. Relatedly,
I observe that theorists who are focused on a theory of normative legitimacy
often build in liberal and democratic presumptions. For example, one
2
Assessments of regimes may include abstract institutional features (e.g. constitutional
essentials) as well as concrete instantiations of those features (e.g. how the ruling party or
government is currently exercising power via those institutional features).
3
Because part of my aim is to propose a new account of the relationship between
consent and legitimacy, leading off the inquiry with a more explicit definition would be
either question-begging or self-defeating. For example, “right to rule” approaches tend to
result in accounts of justified coercion (which often cast consent in a contractualist role)
or justified authority (which often sideline consent altogether).
4
(Buchanan and Keohane 2006); (Copp 1999); (Rawls 2001).
5
For a succinct discussion of the contemporary practice of making legitimacy
assessments, see chapter 5 of Buchanan (2013). For Weber’s original discussion, see
Weber (1978).
Dictionary: NOSD
Consent and Political Legitimacy 73
family of views claims that a regime is illegitimate unless it treats all subjects
as free and equal.6 While such a demanding stance might license us to call
illiberal or undemocratic regimes illegitimate, it also runs together several
distinct registers of political evaluation. In what follows, I aim to show how
it is possible and fruitful to keep the assessment of legitimacy distinct from
other modes of political evaluation, such as democracy and liberalism.
Finally, I observe that some theorists advance views of legitimacy that are
quite demanding. If the theories were applied as stated, the result would be
that no regime in the past or in the present would meet the standards of
legitimacy. This approach to theorizing legitimacy has limited usefulness, in
my view, and it also invites a kind of resignation on the part of political
actors. Ultimately, I will argue that the alternative conception of legitimacy
defended in the chapter does better than other consent-based theories at
fulfilling three separate desiderata: articulating a useful conception of
legitimacy, showing how it is distinct from other modes of political
evaluation, and explaining why it is valuable.
2. SHORTCOMINGS OF OTHER CONSENT THEORIES
The two families of consent-based theories of legitimacy that I will discuss

may be formulated as follows:
Voluntarism: A regime is legitimate for each subject insofar as that subject has
actually consented to it.
Contractualism: A regime is legitimate insofar as every subject would consent under
idealized conditions.
To motivate the development of an alternative, I will outline some
shortcomings in these versions of consent theory. First I will address
contractualism, to which I devote more space because it is by far the
predominant philosophical approach to legitimacy.
Contractualism employs the idea of a hypothetical contract or unforced
agreement among free and equal persons.7 The contractualist claims, very
6
Examples abound, but the seminal statement is Rawls’s liberal principle of legitimacy:
“Our exercise of political power is fully proper only when it is exercised in accordance
with a constitution the essentials of which all citizens as free and equal may reasonably be
expected to endorse in the light of principles and ideals acceptable to their common
human reason” (Rawls 1996, 137).
7
While accounts of political community are more often called contractarian (based on
claims of mutual advantage), I deliberately employ the term contractualist in order to
consider just those theories that are based on the “equal moral status of persons,” drawing
on their “capacity for rational autonomous agency” (see Ashford and Mulgan 2012).
Dictionary: NOSD
74 Amanda Greene
roughly, that a regime gains legitimacy by organizing institutions around

principles of justification that all citizens would freely accept under certain
conditions. The construction of the hypothetical choice situation employs a
variety of idealizing assumptions, typically including citizens’ willingness to
cooperate on fair terms, the reasonableness of citizens’ value judgments, the
common interest in membership in a group on terms of equal standing, etc.
In order to isolate the consent element of the view, I will focus on the formal
feature of hypothetical acceptability.8 The relevant aspect of the contractualist
position, then, is the idea that regimes are legitimate (at least in part)
because their justification is hypothetically acceptable to all citizens.9 I will
now discuss two shortcomings of contractualism as a theory of legitimacy:
the problem of involuntary rule, which arises at both the collective and
individual level; and the challenge of partisan evaluation.
2.1 The Costs of Contractualism: Collective Involuntary Rule

First, contractualism faces a difficulty in explaining some basic intuitions
we have about voluntary and involuntary rule. Consider the tendency we
have to associate popular acceptance with legitimacy or to associate
popular rejection with a lack of legitimacy. We seem to think there is
something problematic or amiss when the subjects en masse disavow the
regime that claims authority over them. We might even say that the subjects,
at a collective level, are involuntarily subjected to rule in some sense. To
make this point vivid, consider the following illustration: imagine there are
two political communities that have a similar constitution and government,
but in one case 80 percent of the subjects affirm the regime, and in the
other case, 30 percent affirm the regime. (Assume that both regimes are
justified by publicly acceptable reasons.) It seems natural and intuitive to
say that a regime garnering the support of 30 percent of its subjects is
less legitimate than one garnering 80 percent support. How might this
intuition be explained?
8
Contractualists may reply that justification is not conferred by consent per se, but
rather that idealized agreement is used to model moral reasoning—and especially to
identify the morally relevant features of a relationship. So understood, contractualism
ought not be taken as a consent-based theory. But if the model at all relies for its validity
on the idea of unforced agreement (and, consequently, its value in realizing a certain kind
of respect), then it draws on some ideal of voluntary acceptance of rule. Contractualists
might deny that the notion of hypothetical acceptability exhausts their defense of
legitimacy, but I think they must admit that it comprises their account of the role consent
plays in legitimacy.
9
Of the theorists who think hypothetical acceptability is necessary for legitimacy,
some also think it is sufficient, or that it is sufficient in combination with other factors
related to hypothetical acceptability. See Cohen (1996) and Estlund (2007).
Dictionary: NOSD
Perhaps one explanation is that the 80 percent support is evidence that

some of the government’s policies have superior welfare outcomes (either
aggregatively or distributively). But now, stipulate that both have the same
welfare outcomes: if one regime garners 80 percent support while the other
only 30 percent, it seems there is still something problematic or amiss about
the regime that lacks popular support. One might also argue in this vein
that 80 percent support is evidence of the epistemic superiority of the
governance—that is, those in power know more and thus govern better
(even if, by luck, the outcomes are on a par). Then, one might argue, the
difference in degree of popular support accurately reflects the difference in
likelihood that the regime is governed knowledgeably. Does the intuition
remain? In response I would make a similar stipulation: suppose that the
governments of the two countries exhibit identical epistemic virtues and are
equally likely to govern knowledgeably, but one garners 80 percent support
while the other garners 30 percent. Does the intuition remain? Consider
again the non-supporting portions of the population: in the 30 percent
case, a large majority of the population (70 percent) is involuntarily subject
to a government they do not support—though it may be due to their
lacking the cognitive or epistemic capacities to see that it is worthy of their
support. In the 80 percent case, this subjection is true of only a small
minority of the population (20 percent).
One might still argue that intuitions about the 30/80 illustration reflect
our concerns about the stability of the polity. Though somewhat plausible,
a reference to stability still does not fully explain the intuition, since
stipulating a parity of stability-related consequences nonetheless leaves us
with a residual discomfort. When a group of subjects does not accept their
regime, this situation is to be regretted and avoided if possible, and at the
very least it ought to be minimized. In general, most of us have an aversion
to having a government or regime deliberately forced upon us, and we infer
that others are similarly averse to this treatment. The 30/80 illustration
alerts us to a concern about involuntary subjection to a political order, no
matter the merits of that order.
Theorists of legitimacy, including contractualists, must either deny
that the problem of collective involuntary rule has intuitive force, deny
that its force has relevance for legitimacy, or account for it in some way.
In my view, the problem ought not to be denied, because it raises a moral
concern about involuntary subjection to a political order, even if the order
is worthy of support. Nevertheless, a contractualist may accept the intuition
that something has gone wrong, but still question its relevance for
legitimacy. But this denial is strained. If legitimacy is understood, in the
broadest terms, as appropriate rule, then widespread popular rejection of
such rule seems directly relevant. Contractualism, however, is ill-equipped
Dictionary: NOSD
76 Amanda Greene
for accommodating the way in which actual popular acceptance and

rejection matter, because it locates all the facts relevant for legitimacy
within the hypothetical choice situation.
According to the contractualist, legitimacy arises from the unanimous
consent of parties who are idealized in certain kinds of ways: if a person’s
deliberation were to meet certain stipulated conditions, then she would
consent to a political order with certain features. Thus the stipulations
contained in the hypothetical choice situation are paramount in determining
when and how consent bears on legitimacy. The stipulations mediate,
structure, and constrain the way consent can alter legitimacy assessments.
The difficulty with forcing consent’s significance to be mediated by
hypothetical idealizations is that we begin to attenuate the moral force of
consent, or rather, the moral force of non-consent. In The Social Contract,
Rousseau insists that once the people are acknowledged as being sovereign,
their scope of authority cannot be limited, not even by the hypothetical
contract itself. Popular sovereignty requires periodic referenda by the people
on the form of government, in which it is possible to change the regime
entirely. He says:
In the state there is no fundamental law that cannot be revoked, not even the social
compact. For if all the citizens were to assemble in order to break this compact by
common agreement, no one could doubt that it was legitimately broken (Rousseau
1987, III.xviii).
According to Rousseau, any regime relying on agreement for its legitimacy
is thereby liable to lose its legitimacy if the agreement is revoked. Essentially,
the people’s actual exercise of free choice cannot be constrained by its
hypothetical exercise of free choice.
While Rousseau did not fully spell out the implications of these claims
about revoking the social compact, his analysis underscores a tension in the
hypothetical consent approach to legitimacy. Within the framework of
contractualism, we lack a principled way to integrate the hypothetical
consent of the governed and the actual consent of the governed, though
they both seem to matter for legitimacy. If the contractualist approach does
not accommodate a direct sensitivity to actual consent to rule on the part of
the subjects, then it is hard to see how it can address the concern about
involuntary rule at the collective level.
2.2 The Costs of Contractualism: Individual Involuntary Rule

Contractualist theories of legitimacy face another problem related to the
idea of voluntary rule—this time, the problem is visible at the level of the
individual. Consider the following stylized dialogue between the spokesman
Dictionary: NOSD
for a regime (say, a government official) and a subject. If the official justified
an exercise of authority by claiming that the subjects willed it in a
hypothetical sense, then if a subject wished to object, his objection would
have the following form: “No, I did not will it in a hypothetical sense, and
here is why.” With the dispute having been framed in such a way, the official
and the subject would begin arguing about whether the exercise of authority
meets certain conditions, namely those conditions that would settle whether
he willed it in a hypothetical sense. If the official’s reasoning prevails, then
she will have simultaneously settled both whether the justification is valid
(i.e. whether the specified conditions were met) and whether it possesses the
relevant sort of recognition by subjects.
For a contractualist, it does not matter for legitimacy whether an
individual does recognize (or consent to) rule, as long as he would recognize
(or consent to) rule. In other words, the withholding of recognition of a
regime cannot bear directly on legitimacy, because the normative space
required in order for the individual’s withholding to matter for legitimacy
(i.e. to alter legitimacy assessments) is eliminated whenever the regime’s
own claim to be contractually justified is true. While some may see this
feature of contractualism as its strength, there is nevertheless something
morally troubling when an individual is treated as though her actual non-
consent does not matter for legitimacy, as long as her hypothetical consent
has been obtained (i.e. established through philosophical argument). There
are multiple domains in which such treatment would be obviously troubling.
For example, according to ‘informed consent’ requirements in medicine,
actual non-consent settles the matter even when there are very good
justifications for consent. This sort of moral unease gives us good reason to
separate the validity of a regime’s claims from the recognition of a regime’s
claims, such that we can identify situations in which one has been achieved
without the other.
Preserving the independent significance of recognition requires that
when a subject fails to recognize something even when it is fully warranted,
this recognitional failure matters. When the lack of recognition matters
independently of whether it is warranted, then when recognition is correct
and freely given, it adds something beyond hypothetical recognition: it pays
due respect to the individual’s standing as recognizer. A contractualist
defense of political legitimacy inadequately respects individuals’ standing as
independent recognizers of claims that are made in the justification of rule.
A devoted contractualist may still ask: why is this a problem? It’s a problem
because it doesn’t sufficiently respect an individual’s capacity to evaluate rule
in their non-idealized state. The contractualist might respond that it respects
their evaluative capacities in a different, more generic way. The further reply
is that while this generic kind of respect has value, legitimacy should be
Dictionary: NOSD
78 Amanda Greene
about actual attitudes, because the attainment of voluntary rule should be

sensitive to particularity at the individual level.
In a related line of criticism, A. John Simmons argues that contractualism
collapses the distinction between justification to a hypothetical citizen and
justification to a particular citizen. He points out that legitimacy depends
on historical facts about the relationship between an individual and an
institution, a dimension that is missed in the Kantian approach:
The proper grounds for claims of legitimacy concern the transactional components
of the specific relationship between individual and institution . . . No specific actual
history of morally significant relations between a particular state and each of its
subjects is thought by Kant to be necessary to support the transition from justifying
the state to legitimating a particular state with respect to all of its subjects . . . Therefore,
Kantians have appropriated the justificatory force of voluntarism without having
any respect for people’s actual choices (Simmons 1999, 755–6).10
Here Simmons emphasizes the way in which the historical facts about a
state’s relationship with an individual are relevant to its legitimacy.11 The
notion of “standing-to-recognize” reflects our appropriate concern with the
particularity of the relationship between regimes and individual subjects.
2.3 The Costs of Contractualism: Partisan Legitimacy

I have argued that contractualism incurs theoretical costs in its inability to
register the withholding of voluntary acceptance of rule, both at the
collective and the individual level. I emphasized that this inability is a
failure to give due respect to the actual judgments of individuals and
communities in their real circumstances (as opposed to idealized
circumstances), and I suggested that this failure weakens the claim that
legitimacy corresponds to the attainment of voluntary rule. Now I want to
propose that the contractualist’s need to idealize consent has a further
problematic dimension. The contractualist prejudges which categories of
political facts are relevant for consent, thereby making legitimacy
unnecessarily partisan. Some evidence of this partisan mode is the tendency
for contractualist formulations of legitimacy to involve liberal and
10
As a matter of technical clarification, Simmons’s terminology differs from mine. He
says justification pertains to whether a state has a right to exist, while legitimacy pertains
to the “complex moral right [a particular state] possesses to be the exclusive imposer of
binding duties on its subjects, to have its subjects comply with these duties, and to use
coercion to enforce these duties.” Simmons also equates legitimacy with the individual
obligation to obey, a “logical correlate” I do not accept (Simmons 1999, 746).
11
For a fuller discussion of the way in which our conception of legitimacy ought to
accommodate the fact that it is evaluated differently in different historical circumstances,
see Horton (2012).
Dictionary: NOSD
democratic presuppositions as a matter of definition. These presuppositions

in turn make it impossible for non-democratic and non-liberal forms of
government to achieve legitimacy. Note that the putative illegitimacy of
these regimes does not follow from any separate considerations but rather
from the role of consent in establishing legitimacy. This is a theoretical
entanglement we would do well to avoid.
The contractualist approach I have described consists of two distinct
claims: we each deserve acceptable justification for claims of authority
being made on us, and the provision of such justification is necessary and
sufficient for establishing legitimacy (understood as the consent of the
governed). The contractualist must commit to some conditions being
necessary and sufficient for imputing universal agreement, howsoever the
hypothetical choice is constructed. Thus, contractualism determines ex
ante, and on behalf of everyone, which moral considerations do and do not
matter for legitimacy, and how they are to be assessed. By determining in
advance which moral considerations do and don’t matter for establishing
hypothetical consent, contractualist views draw a line around the acceptable
grounds for the consent of the governed in terms of a particular
characterization of the moral justification of government. Contractualism
must, therefore, rely on a ‘partisan’ view of how consent fixes the facts
about legitimacy.
In many cases, the moral considerations that are adverted to by contr
actualists concern whether subjects are treated as free and equal, in some
sense, by their government. This common formulation embeds liberal and
democratic requirements into the ideal of voluntary rule, the very ideal
underlying consent-based theories of legitimacy—indeed, the securing of
unanimous consent is itself an expression of free and equal treatment.
Contractualists may object that what I identify as a theoretical cost of their
approach is precisely the feature they consider to be its theoretical strength.
But I think they have not taken a wide enough view about the costs that are
incurred. Advocating such a partisan view prevents us from having a theory
of legitimacy that applies, without significant idealization, to the situations
in which we find ourselves today. For instance, it seems plausible that there
could be regimes that are legitimate but not democratic; conversely, there
could be regimes that are democratic but not legitimate. It would be useful
to have a conception of legitimacy that can help us understand why things
such as liberal rights, equality, and democracy contribute to the attainment
of legitimacy (if they do), without building those political ideals into our
very concept of legitimacy. In what follows, I present an alternative
conception of political legitimacy that can be usefully applied to non-liberal
and non-democratic states, thereby avoiding a partisan articulation of the
role of consent in establishing legitimacy.
Dictionary: NOSD
80 Amanda Greene
2.4 The Costs of Voluntarism

I turn now to a discussion of the shortcomings of voluntarism. According
to the voluntarist variant of consent theory, a regime is legitimate insofar as
each subject actually consents to it. This consent then is binding in two
ways: the state is bound to exercise authority within the limits of the
expressed consent; and the individual is bound to obey the authority to
which he has consented. A recent advocate of the view, A. John Simmons,
argues that the legitimacy of the state arises only from the consent of the
individual (Simmons 1999). On this view, the actual consent of the individual
is necessary and (usually) sufficient for the state to possess legitimacy with
respect to that citizen.12 The picture often relies on a notion of natural rights
as pre-political moral entitlements, such that each person has a right to
defend their entitlements against encroachments by others. The state’s
authority arises from the deliberate transfer of this right to a single authority,
which makes determinations and exercises coercion in order to enforce
each individual’s rights. This view construes legitimacy as a property of the
relationship between the state and an individual, meaning that the state can
be legitimate with respect to one citizen and illegitimate with respect to
another. I designate this view a form of voluntarism because it so closely
tailors legitimacy assessments to facts about the will of the individual.
There are several problems with this view, but a major concern is that the
required consent standard cannot be met in a way that issues in legitimacy.
One sense in which it cannot be met is that it is impossible to get the
consent of every single individual subject. If unanimity is impossible, one
might be forced to conclude that legitimacy is virtually unattainable.
Indeed, Simmons bites the bullet and concludes that “no existing states
are legitimate.”13 This problem arises because voluntarism is sensitive to the
moral significance of consent, but it is sensitive in such a way as to make
unanimous consent necessary for legitimacy. Alternatively, one might say
that unanimous consent is not necessary for legitimacy overall, but only
necessary for legitimacy construed as a feature distributed across individuals.
The problem then would be that legitimacy is piecemeal, not a feature of
a regime overall. When we ask whether a regime is legitimate, we are
12
Simmons adds a proviso to whether consent is sufficient (derived from Locke):
“We cannot bind ourselves by consent to immoral arrangements” (Simmons 1999, 746).
I return to this proviso later.
13
He says, “Because I subscribe to political voluntarism as the correct account of these
transactional grounds for legitimacy, and because I believe no actual states satisfy the
requirements of this voluntarism, I also believe that no existing states are legitimate.”
However, Simmons goes on to argue that he thinks states that are neither justified nor
legitimate can nevertheless act in ways that are justified (e.g. to prevent violent crime)
(Simmons 1999, 269–770).
Dictionary: NOSD
concerned with its general standing with respect to the subjects of the
regime. It is not clear how piecemeal legitimacy could address the question
of general standing without abstracting away, through idealization, from
some subjects’ non-consent, thereby treating them differently than the
consenting subjects.
And finally, the voluntarist implicitly relies on an implausible connection
between consent and an ideal of self-rule.14 The voluntarist seeks to derive
legitimacy by construing consent as a kind of delegation of authority that
preserves an ideal of self-rule. The individual avoids a pernicious subjection
to the rule of others because their obligation to obey arises only by their
own consent. In positing such a direct connection between consent and the
avoidance of involuntary subjection, the voluntarist implicitly presumes
that consent evinces some status approximating self-rule. But it is very
difficult to see how granting consent to one’s regime ought to count as self-
rule without idealizing to a significant degree.15 Even in societies whose
government is highly participatory, construing a subject’s consent as
evincing self-rule requires making highly controversial assumptions about
collective will formation. The voluntarist strategy of deriving legitimacy
from the willing transfer by individuals of those rights and privileges
associated with self-rule is not a realistic model for today’s political orders.
Therefore the voluntarist must either stretch the definition of consent so far
that it is imputed on the basis of factors whose connection to self-rule is
highly attenuated and idealized, or he must admit that the required consent
standard cannot be met in a way that is sufficient for legitimacy.
3. AN ALTERNATIVE CONCEPTION OF LEGITIMACY
In the remaining discussion, I defend an alternative view of legitimacy,

arguing that it is superior to other consent-based views of legitimacy. A
basic statement of the view, which I call the sovereignty conception, is that
a regime is legitimate insofar as it achieves actual quality consent to rule
based on positive governance assessments. The view retains a key insight
of voluntarism: actual consent matters. But, since unanimous consent is
impossible, the sovereignty conception focuses on achieving actual consent
among a high proportion of subjects. At the same time, the sovereignty view
14
This problem is most evident in libertarian approaches to legitimacy based on the
idea of self-ownership. For the seminal statement, see Nozick (1974). For a more
contemporary version, see Huemer (2012).
15
For a recent exposition of this problem from the perspective of the neorepublican
tradition, see chapter 3 of Pettit (2012).
Dictionary: NOSD
82 Amanda Greene
retains a key insight of contractualism: not all actual consent matters. The
sovereignty conception employs the notion of quality consent because consent’s
significance is dependent on meeting certain minimal conditions—namely,
being based on an individual’s positive governance assessment. The sovereignty
conception, therefore, stakes out an innovative position in theorizing about
how consent relates to legitimacy: a regime is legitimate insofar as it achieves
consent that is (a) actual rather than idealized, (b) proportional rather than
unanimous, and (c) based on positive governance assessments. In order to
show how this view is superior to other views, I first need to explain these
interrelated aspects of the sovereignty view. The next three sections lay out
the key elements of the view.
3.1 Voluntary Rule

I will introduce the sovereignty conception by returning to the original
ideal underlying consent theories of legitimacy: the ideal of voluntary rule.16
It is plausible to think that each individual, considered in his role as the
subject of a political order, has an interest in being ruled voluntarily. One is
ruled voluntarily when one is willing to remain a subject of one’s particular
political order, all things considered. One may characterize a willing subject
as one who freely consents to the particular system of rule to which one is
subject. An unwilling subject is one who disavows being ruled to such a
degree that one is willing to engage in disruptive and costly resistance (if it
stands a chance of being effective) in order to fundamentally change the
constitutional order, including the basic power-conferring norms of that
order.17 It also seems plausible to think that this interest in voluntary rule is
significant and universal, meaning that it is important for everyone, though
it need not override all other interests at stake in politics.
However, being a willing subject of a regime is not unconditionally good.
Not every voluntary act of an individual has import simply because it is
freely chosen—conditions must be met. Accordingly, consent-to-rule is
only valuable when it meets certain conditions. I will say more about these
conditions shortly, but what is important to note here is that consent’s role
in realizing the interest in voluntary rule is going to be conditional. This
conditionality illustrates consent’s nature as a dependent good, a good
16
Since I think voluntary rule is something that all consent theorists would accept at
some level of generality, I take myself to be elaborating a particular interpretation of
voluntary rule.
17
I cannot give a full definition here, but determining whether any given person is a
willing subject will need to include reference to both attitudes and behavioral dispositions.
The paradigmatic case of an unwilling subject is the revolutionary.
Dictionary: NOSD
whose value depends on other things. I will call instances of consent that
meet the relevant conditions “quality consent.”18
Although the significance of consent is conditional, its value is nevertheless
non-instrumental. How should we understand this kind of good, something
that has dependent, non-instrumental value? Joseph Raz elucidates the
structure of a dependent, non-instrumental good when he discusses the
concept of autonomy:
There is no contradiction in claiming both that autonomy is intrinsically valuable
and that the fact that people's lives are autonomous contributes to their well-being
only to the extent that they engage in worthwhile pursuits. What is intrinsically
valuable can be, and in the case of autonomy is, valuable as a constituent of a good
in itself. Remove other elements of the good in itself and it may turn worthless, or
even bad (Raz 1988, 1228–9).
For Raz, autonomous choice has intrinsic but dependent value: when what is
chosen is valuable (e.g. a musical project), the fact that it is autonomously
chosen contributes intrinsically to its value in an organic way.19 But if the
object of choice is not valuable (e.g. an assassination project), then something
that would otherwise be of value has no value.20 In sum, the value of autonomy
cannot be made instrumental to the value of what is chosen, though its value
depends on what is chosen. In my view, consent-to-be-ruled is analogous to
Raz’s claim about autonomy in that its value may be nullified if certain
conditions are not met, but when present it has a non-instrumental value that
cannot be reduced to the conditions on which its value depends.
3.2 Quality Consent

The sovereignty conception claims that a regime is legitimate insofar as
it achieves stable and widespread actual quality consent to rule among
subjects. I construe quality consent as consent based on positive governance
assessments. At a minimum, a positive governance assessment is the judgment
18
I acknowledge that some readers may be inclined to focus on validity when thinking
about consent, but I focus on its value, i.e. why we should care whether there is consent,
since this account will need to underlie any account of its validity.
19
I wish to avoid claims of intrinsic value, so I claim that it has “non-instrumental” value.
For an extensive discussion of dependent, non-instrumental value, see Sandbu (2007).
20
It is important not to confuse Raz’s point with one that sounds similar, articulated
by Amartya Sen among others, about the role of agency in making something valuable.
Sen (1997) distinguishes between comprehensive outcomes and culmination outcomes,
where the former are valued for some aspect of how they are achieved, e.g. involving
agency. Raz is concerned with how the value of autonomous choice depends on the value
of what is chosen, while Sen’s concern is with how the value of what is chosen depends
on whether agency is involved.
Dictionary: NOSD
84 Amanda Greene
by a particular subject of some sort of governance success. The most basic

claim of government is to benefit its subjects through the exercise of power,
so the relevant recognition of success is a subject’s judgment regarding
whether benefits have arisen from the exercise of power. I call these sorts
of judgments governance assessments, and they are in terms of subjective
welfare, i.e. perceived welfare at either the individual or collective level. On
the basis of what a subject considers relevant and valuable by her own lights,
she forms an overall subjective assessment of governance and, on that basis,
consents to her political order. The view is deliberately permissive about what
counts as a positive governance assessment.
Therefore, on the sovereignty conception, a regime comes to be legitimate
by delivering benefits through the exercise of power in such a way as to be
widely recognized as doing so by its subjects. In this scenario, the rulers seek
and promote the common good as they see it, and the subjects voluntarily
accept their subjection because they think the rulers are advancing the
common good, as each subject sees it. This expansive notion of a positive
governance assessment allows for a wide range of judgments to count as
quality consent. The actual benefits delivered may outstrip or fall short of
the perceived benefits, the perceived benefits may be non-overlapping, and
there may be variable sophistication in terms of the understanding of how
and why these benefits were achieved.21 The notion of a positive governance
assessment allows for entirely idiosyncratic judgments about the goods
government ought to serve, letting a “thousand flowers bloom.”
3.3 Non-Quality Consent

However, not all governance assessments meet the conditions for quality
consent. How is non-quality consent to be distinguished? I will approach
this question through an examination of paradigmatic cases of non-quality
consent. First, consent-to-rule which is brought about by certain kinds of
fear and intimidation does not count: it does not qualify as consent-to-rule
as I have articulated it, which is the voluntary acceptance of a system of
rule.22 There are three further cases of consenting subjects whose consent, at
least intuitively, should not count in favor of legitimacy: the dominator, the
underling, and the culpably ignorant. Any adequate definition of quality
consent ought to exclude these cases:
21
While I cannot give a full account here of the judgment underlying quality consent,
what is essential is that it involves a (perhaps implicit) belief that the political order is
subjectively valuable.
22
Not all sources of fear are the same, and only some disqualify the consent as freely
given (e.g. when a subject would not consent but for the intimidation brought to bear on
her by her regime).
Dictionary: NOSD
a) The dominator is someone who consents to rule on the basis of a

judgment that it is acceptable to deprive others of basic security.
b) The underling is someone who consents to rule on the basis of a
judgment that it is acceptable for himself to be deprived of basic
security.
c) The culpably ignorant is someone whose consent to rule is strongly
insensitive to facts about whether some subjects are being deprived of
basic security.
It seems intuitive that these three cases should not bear positively on
assessments of legitimacy. Note that they are all partly characterized in
terms of basic security, i.e. minimal order and protection from violence.
Why set the bar so low? A variation on the view that may be more
attractive to some readers is to set the bar higher, such as protection of
human rights, or provision of basic capabilities. This variation would
still have the structure of an actual quality consent view. But in proposing
the sovereignty conception, I aim to construct an account of legitimacy
that is more agnostic about the relationship of legitimacy to other
political values, such that it has a wider application than only to liberal
democracies. If we were to examine instances of political orders in various
historical periods—honor-based tribal communities, medieval feudal
systems, fundamentalist theocracies, modern liberal democracies—we
would find that virtually every system of political rule advances the claim
that it (at least) provides for the basic security of those it addresses as its
subjects. Though the regimes may fail to, or not even attempt to, actually
provide for all subjects’ basic security, the regimes nevertheless make this
claim. Insofar as a regime claims to rule, it makes what I call the claim of
minimally competent government (CMCG). All three of the paradigmatic
cases of problematic consent rest on judgments that directly conflict with
this claim.
Therefore, in light of the three paradigmatic cases, I construct the
following filter for determining quality consent. In order for consent to
contribute to legitimacy, it must be based on governance assessments that
meet the following two necessary conditions:
I. The assessments do not explicitly contradict the CMCG. That is, the
governance assessments do not contain judgments that explicitly deny that
each subject’s basic security is to be provided for by the regime.
II. The assessments are weakly reliable regarding whether the CMCG
is being met. That is, the governance assessments exhibit some weak
counterfactual sensitivity to governance outcomes, namely those facts about
whether the regime is providing for all subjects’ basic security.
Dictionary: NOSD
86 Amanda Greene
These conditions suffice to exclude the dominator, the underling, and the
culpably ignorant.
To make clearer the implications of my view, examples of quality consent
would include: the consenting libertarian who thinks he is better off than
he would be otherwise but cares little for distributive justice, the consenting
egalitarian who thinks the least well off are better off than they otherwise
would be, and the consenting religious believer who thinks governance
adequately enables the welfare conditions for spiritual development of his
co-religionists but cares little for outsiders. These cases qualify because their
consent is based on the right sort of governance assessment, namely the
recognition of perceived benefits based on judgments that do not fall afoul
of (I) and (II) above. A more ordinary case that also counts as quality
consent is the simple-minded person who affirms something vague, such as,
“Things around here are going well enough for the most part.”
However, an example of non-quality consent is a consenting racial
supremacist who believes his race ought to dominate others. He subjectively
values rule that aims to compromise the basic security of some subjects. The
principled reason for why his consent is disqualified is that the judgment on
which it is based directly contradicts the government’s claim to be minimally
competent. Nevertheless, the sources of disqualification are more limited
than in other theories; accordingly, being a libertarian or a religionist does
not disqualify a citizen’s consent. This admitted laxity in the ‘quality filter’
is designed to allow even minimal awareness of governance success to
qualify the consent as relevant for legitimacy. This latitude is appropriate in
order to balance out the fact that non-consent bears (negatively) on
legitimacy whether or not it is warranted by the governance facts.
One implication of this approach is that legitimacy largely becomes a
matter of what people believe about outcomes. Strictly speaking, there is no
“welfare outcomes floor” built into the theory, and deliberately so. However,
while it’s true that failures of governance (failure to fulfill CMCG) do not
directly reduce legitimacy, they tend to undermine it, because they make it
much more difficult to achieve instances of quality consent among subjects.
3.4 Achieving Actual Consent Among Subjects

To summarize thus far: the sovereignty conception claims that regimes are
legitimate insofar as they achieve actual quality consent to rule among
subjects.23 Actual quality consent captures the complex idea that individual
23
Though I do not have space to discuss it here, the view presupposes an independent
account of subjecthood, lest a regime seek to achieve legitimacy by manipulating who
counts as a subject.
Dictionary: NOSD
consent to rule is non-instrumentally valuable, but that its value is

nevertheless dependent. Its value is dependent because consent to rule, in
order to contribute to legitimacy, must be based on positive governance
assessments. When it is, it is quality consent, and a regime is legitimate to
the degree that its subjects exhibit quality consent. Two further features of
the view now come into view: the proportional basis of legitimacy, and the
role of aiming at consent.
On the sovereignty conception, legitimacy corresponds to the proportion
of individuals who consent: the greater the proportion of free submission to
“rule” under the authority, the more legitimate is the rule. This proportional
approach makes sense if we imagine the case of a “legitimacy utopia,” a
voluntary association of individuals whose activities are just and welfare-
enhancing. In this imaginary association, to the degree that its activities
require the exercise of power and authority, each individual affirms his
voluntary subjection to the governing order. The unanimous consent in a
legitimacy utopia reflects our intuition that, in a perfect world, citizens’
acceptance of rule reliably tracks whether the rule is just and welfare-
enhancing. As we gradually move away from the utopian case, we may
imagine that fewer subjects exhibit quality consent. The sovereignty
conception claims that the fewer instances there are of quality consent, the
less legitimate is the rule.24
In order to realize the ideal of voluntary rule, a regime must treat each
subject’s interest in voluntary rule as an interest worthy of being promoted
and fulfilled for its own sake (rather than, for example, as a means to
compliance). While the primary interest of a subject is in achieving quality
consent, there are secondary interests in having the opportunity or possibility
of exhibiting quality consent and in being treated as someone for whom
consent-to-rule is non-instrumentally valuable. Therefore a regime may aim
at quality consent by advancing the secondary interests even if the primary
interest is not achieved. When quality consent is aimed at and freely given,
its achievement adds to the fulfillment of the ideal of voluntary rule.
4. ADVANTAGES OF THE SOVEREIGNTY CONCEPTION
In the remainder of the chapter I compare the sovereignty conception to the

other consent-based views in order to show its advantages. With respect to
contractualism, the chief advantage of the sovereignty view is that it better
handles the problem of involuntary rule. Consider two examples in which
24
An upshot of this approach is that legitimacy assessments are scalar rather than
binary. I argue elsewhere that a scalar concept has more theoretical advantages than costs.
Dictionary: NOSD
88 Amanda Greene
the sovereignty conception is more sensitive to the problem of involuntary

rule: a decent hierarchy and a dictatorship of enlightened guardians.
Consider first a decent hierarchy. Imagine that a particular community
contains a small linguistic/religious/cultural minority, wholly located on a
political border shared with neighbors. This minority group is small and
actively disliked by every political community in the region. By remaining
a part of the original community, they believe (say, truly) that they are
better off than in any alternative community, including if they were to
secede. Although their basic security is provided for, they are not treated as
full political equals—say that they lack some substantive and some formal
equal political rights. A contractualist account would typically deny
legitimacy, as hypothetical agreement is premised on some presumption of
recognition of political equality. This denial seems to tacitly assume that
valid consent requires full equality. But, given that we do not require full
equality for consent to be legitimating in other spheres such as commercial
exchange and medical care (though we may impose other requirements),
this assumption is questionable. Many people are not treated equally in
their political orders, but they nevertheless consent to their regime on other
permissible grounds.
The second example is rule by enlightened elites.25 Imagine a meritocracy
in which those who rule are selected by past rulers on the basis of
qualifications related to competence. Imagine they are permanently
appointed and tend to enjoy a higher level of well-being because of their
high social standing. Say that the society meets a very high standard of
justice, which is that everyone is as well off as they could have possibly
been, due to the rulers’ wisdom and virtue. It seems that hypothetical
consent ought to be imputed to everyone, since not only is everyone
better off than some alternative scenarios, everyone is better off than all
alternative scenarios.26 However, say that the people do not believe that
powers of office are assigned for the general good, because the inequality
in welfare tracks relative power, and this welfare-determining status is
assigned non-transparently. And so, say that 95 percent of the subjects do
not consent to rule, and a significant proportion actively seeks revolution.
25
This stylized description arises from the example of the guardians in Plato’s Republic,
but it is also meant to encompass technocratic administration by civil servants in
contemporary societies.
26
One may object that welfare should be taken to include political equality and
participatory rights, in which case everyone is not as well off as they could be. However,
one could include these in the welfare definition and still generate the example, since
even with more participatory government, it is still plausible that there be sufficient
inequality of outcomes to foment rebellion.
Dictionary: NOSD
While there may be some welfare-based or justice-based argument for the

regime, the level of its legitimacy seems very low. This is because the rule is
involuntary at the collective level and at the individual level. These two
examples illustrate that the sovereignty conception fits our intuitions about
voluntary rule better than the contractualist approach does.
This discussion highlights that the sovereignty view advances a kind of
asymmetry between consent and non-consent. Since this may seem
counterintuitive, I want to address why this feature is appropriate. For the
contractualist, in the same way that certain cases of consent (the unreasonable
ones) do not bear on legitimacy, certain cases of non-consent (the
unreasonable ones) do not bear on legitimacy. Contractualism thus subscribes
to a kind of theoretical symmetry in the way consent and non-consent are
“counted.” In contrast, the sovereignty conception defends what may be
deemed an “asymmetric” view. Legitimacy is reduced whenever non-consent
occurs, whether or not it is reasonable or well grounded. Legitimacy is
reduced because all kinds of non-consent set back the interest in voluntary
rule. So the charge of asymmetry is a misnomer, because the sovereignty
view construes consent as a dependent good, one whose value depends on
certain conditions being met. Since its value is conditional, it is fitting that
non-consent and non-quality consent both fail to advance voluntary rule.27
This feature also allows the sovereignty view to avoid fixing the facts about
consent in a way that embeds partisan presumptions, one of contractualism’s
shortcomings.28
27
Claiming that non-consent and non-quality consent are similar in this respect does
not imply that they are otherwise similar—indeed, they are very different in terms of
their political consequences.
28
Earlier I criticized contractualism for relying on idealized versions of consent,
particularly ones that employed partisan assumptions about what are and are not
relevant factors in political evaluation. It may seem that the quality filter also involves
an idealization of consent, insofar as the judgments underlying it must square with
CMCG. To this charge, I reply that there is one sense in which the sovereignty view
clearly fares better than contractualism: it accepts without idealization the significance
of instances of non-consent to rule, admitting that individual instances of non-consent
always undermine legitimacy. Even if the individual’s judgments are unjustified or
grossly immoral, there is no tidying things up with ideal theory. Regarding how to
classify instances of consent: the sovereignty conception’s imposition of a quality filter
does introduce some value judgments about consent’s significance, but these are less
problematic than contractualism in two ways. First, the amount of idealization is
comparatively minor, because it is extremely accommodating of individuals’ idiosyncratic
views regarding the relevant factors in political evaluation. Second, the quality filter
employed by the sovereignty conception does not draw on assumptions about a pre-
political morality of justified coercion. For this reason, it can be used to argue for
the legitimacy of non-liberal states, demonstrating that it is less partisan than other
consent theories.
Dictionary: NOSD
90 Amanda Greene
4.1 Comparison to Voluntarism

The sovereignty view also has advantages relative to voluntarism, but they
emerge on different axes of comparison than the ones pertaining to
contractualism. The two dimensions I will particularly discuss are the
quality filter and the emphasis on individualism.
The first advantage of the sovereignty conception over voluntarism
pertains to the theoretical underpinnings of the quality filter. Earlier I noted
that on the voluntarist view expounded by Simmons, there is a proviso on
whether consent is valid, one that may seem similar in form to the quality
filter: “We cannot bind ourselves by consent to immoral arrangements”
(Simmons 1999, 746). Imposing this proviso is well motivated but
inadequately defended. First, the standard of “immoral arrangements” is
obscure and highly contested. Such a standard entails that in order to apply
the proviso in determining the facts about consent (that, in turn, fix the
facts about legitimacy), we must first settle the necessary and sufficient
conditions for a political arrangement to be moral.29 But it is often precisely
when we face intractable moral disagreement that we most keenly need
legitimacy standards. The second reason the proviso is not adequate to the
purpose is structural: it presupposes and then relies on an independent
principle about the moral circumstances under which consent can authorize
coercion. But the question under consideration is precisely how consent
plays a role in legitimizing political rule, which in turn provides part of the
justification for coercion.30 These are thorny questions that may not ever be
perfectly resolved, but I will now show why the sovereignty conception
addresses them more adequately.
Whereas voluntarism says that the threshold for consent to be
legitimating is when the political arrangements are moral, the sovereignty
view has a lower bar. Essentially, the consent is legitimating when it
does not conflict with the CMCG. The imposition of a quality filter in
terms of a claim to provide for basic security is narrower and clearer
than the vague “immoral arrangements” criterion. Moreover, rather than
drawing on a general principle of permissible coercion that claims to be
objectively true, the sovereignty conception derives the quality filter
29
There also will likely be a problem with determining when political arrangements
are moral independent of their legitimacy, but Simmons does not address this.
30
The presupposition is problematic because it forecloses part of the question under
consideration. Whatever principle Simmons (or Locke) is appealing to in order to justify
applying such a proviso must have the following form: coercion in the presence of consent
(including that done by regimes which claim to be legitimate) is justified only when an
arrangement is moral. But such a principle forecloses debate about the very question
under consideration, which is, what is the role played by consent in establishing
legitimacy?
Dictionary: NOSD
from historical generalities about political orders. Its status as a valid quality
filter arises from the trans-historical claim of minimally competent
government, which is to provide for basic security. It takes as the minimum
aim of all political orders to be providing for basic security, so it need not
take a position about the ultimate aims of government, what counts as
successful government, or even the necessary conditions for successful
government. Therefore, in virtue of its source and its content, the quality
filter on the sovereignty conception is less problematic than the voluntarist’s
proviso.
The second major dimension on which the sovereignty conception is
superior to voluntarism is the emphasis on individualism. Both views
subscribe to an ideal of voluntary rule that gives prime significance to the
actual facts about whether individuals consent. But the sovereignty
conception has much more to say about achieving voluntary rule among all
subjects, considered collectively. In order to support this claim, I must first
clarify several related points of contrast between the views.
The voluntarist approach employs an exclusively individualistic approach
to assessing legitimacy, since legitimacy can be determined for each
individual independent of other subjects’ relationship to the regime. The
sovereignty conception, instead, construes legitimacy as a feature of a regime
with respect to its subjects collectively. To whatever degree a regime is
legitimate, its legitimacy applies equally to all the subjects, whether or not
each has consented in her individual case. Therefore, the sovereignty
conception avoids the problems of assigning legitimacy assessments subject-
by-subject. Accordingly, it is better placed than voluntarism to deny that
unanimous consent is necessary for legitimacy. However, it nevertheless
shares voluntarism’s concern for the individual, as evidenced by tying
legitimacy to actual instances of non-consent.
However, one may be reluctant to give up the exclusively individualistic
approach when considering a scenario in which some individuals or groups
are consistently treated much worse than others. For example, Thomas
Shelby has argued that citizens who are differentially liable to be the victims
of violent crime have correspondingly different political obligations
(Shelby 2007). But the differentially worse outcomes for some individuals
ought not just affect the legitimacy of the regime for those individuals, while
permitting the legitimacy of the regime for other individuals to be unaffected.
Indeed, as this example suggests, the notion of political obligation is better
suited to be the dimension on which individual differences may vary
according to differentially worse treatment. By construing legitimacy at the
collective rather than individual level, and by distinguishing legitimacy
from political obligation, the sovereignty view is able to accommodate this
important consideration.
Dictionary: NOSD
92 Amanda Greene
Ultimately, the sovereignty view handles the tension between the

individual and collective in a way that is more subtle and illuminating than
either contractualism or voluntarism. This is because it eschews unanimity
without embracing exclusive individualism. Rather, it makes room in the
account of legitimacy for the primacy of individual consent, while
maintaining that such consent issues in legitimacy only in combination
with other subjects’ consent. It is a strength of the sovereignty conception
that it neither collapses collective and individual instantiations of voluntary
rule, nor does it hold that they are entirely independent.
4.2 The Value of Legitimacy

Recall that one aim at the outset was to identify a consent-based theory of
legitimacy that showed why it was a concept distinct from other political
concepts and at the same time something we should value. I will now
discuss the sovereignty view in light of these desiderata.
Whether the sovereignty conception of legitimacy can illuminate what is
uniquely valuable about legitimacy depends in part on what counts as
quality consent. If the quality filter is too demanding, then the value of
legitimacy can’t be adequately distinguished from the values of liberalism,
democracy, etc. On the other hand, if the quality filter is relatively
undemanding, then it’s hard to see why quality consent is worth achieving.
When the quality filter rests on claims of objective morality, as it does for
the voluntarist, then one may straightforwardly argue for the significance of
legitimacy on that basis. But since the sovereignty conception does not avail
itself of such a claim, the question becomes: how far can the view move
away from relying on claims of objective morality and still assert that quality
consent is something we should value?31 Fairly far, I hope to show, and to do
so I will to return to the characterization of voluntary rule.
The good that is instantiated by voluntary rule can be construed as a sort
of congruence, or alignment, between the facts about organized power and
the attitudes of those subject to that power. When voluntary rule is achieved,
there is at least some partial alignment between what an individual values
and what goods are promoted by the political order to which he is subject
(though it may also promote other goods, ones valued by other subjects). In
fact, the congruence involves three levels:
(i) The regime claims to benefit its subjects through the exercise of power.
(ii) The regime benefits its subjects through the exercise of power.
31
This question applies in general to theorists in the political realism vein, who take
political moralism to be the only alternative (Williams 2007). Elsewhere I argue that the
view defended here avoids shortcomings of both realism and moralism.
Dictionary: NOSD
(iii) An individual (or group) believes that the regime benefits subjects
through the exercise of power.
When these three statements can be made together, the situation represents
a sort of unforced alignment or congruence between a regime’s governance
and the attitudes towards that governance among subjects.
The instantiation of a voluntary ruling relationship at the level of a
group of subjects represents, I propose, a kind of excellence exhibited by
the regime in question. Therefore, voluntary rule has value at the impersonal
level of subjects considered as a group, as well as at the personal level of
each individual. If the good realized by voluntary rule can be characterized
by a kind of unforced alignment or congruence, then the absence of
voluntary rule may be characterized as a kind of alienation or dissonance.
This alienation from one’s own social or political order arises because one
cannot authentically affirm the regime to which one is subject as in any
sense worthwhile, i.e. as providing at least some benefits that one values
subjectively. The absence of voluntary rule—involuntary subjection—also
instantiates a kind of disvalue, for an individual and for a political
community.
How does the value of attaining voluntary rule compare to other political
values? The sovereignty conception recognizes that our interest in voluntary
rule is one interest among others at stake in being ruled. Accordingly, it
allows that other things matter besides legitimacy for political justification.
It does not commit to the lexical priority of legitimacy over other political
values. The view does not entail that achieving voluntary rule is a necessary
condition for all else to be valuable to an individual, or that it is a necessary
condition for a regime to rule permissibly. Indeed, a loss on the dimension
of legitimacy is sometimes desirable for the sake of other values.32
Having shown how the sovereignty conception is oriented towards the
valuable ideal of voluntary rule, nevertheless there remain two kinds of
subjects who may complain that their interest in voluntary rule is not given
due consideration: the person who does not consent, and the person whose
consent is non-quality consent. I will briefly sketch how a reply to these two
kinds of subjects would proceed.
First, consider the subject who does not consent. If the value of voluntary
rule is in overcoming alienation in order to bring about congruence
and identification between a subject and his regime, then why should a
32
The reader may wonder: what is the upshot of legitimacy? In my view, legitimacy
does not directly entail answers to questions about authority, obligation, right of
revolution, etc., but it does bear on these questions. The conception of legitimacy
proposed here is meant to be compatible with a variety of accounts of authority and
political obligation, and especially to allow those to vary by subject.
Dictionary: NOSD
94 Amanda Greene
non-consenting subject feel that his interest in voluntary rule has been
adequately respected? In other words, why is a regime legitimate even for
the non-consenters? Whereas other views claim that consent of the
individual is appropriately valued only when it is actually obtained, the
sovereignty conception claims that there is a way of valuing and showing
concern for consent that doesn’t require that it be obtained.
According to this characterization, a regime may relate appropriately to a
subject’s interest in voluntary rule in two possible modes. One is for the
regime to actually realize voluntary rule for that individual by providing
benefits that result in a positive governance assessment. Failing this, the
other mode is for the regime to respect, in a non-instrumental way, their
interest in voluntary rule. While respecting that interest is second-best to
realizing it, it is still important. Consider what the absence of this respect
would involve: a regime’s treating the interest in voluntary rule in the wrong
way, i.e. valuing consent only as a means, or not valuing consent at all.
Valuing consent non-instrumentally, even when consent is not obtained,
counts as a form of respect for the interest in voluntary rule.
Another part of the reply is that a particular subject’s interest in voluntary
rule is respected when the rulers aim to achieve the consent of all the
subjects. In my view, this position is analogous to consequentialism with
respect to individual utility. When a decision is made on behalf of a group
with a consequentialist justification, then just because an individual’s utility
is not advanced by that particular decision does not mean that their utility
does not count, or count the same as everyone else’s. Likewise, just because
an individual’s consent is not achieved, that does not mean that his consent
does not count, or count the same as everyone else’s. This kind of argument
is distinct from arguing that it is the possibility or the capacity for consent
which matters.33 The capacity to consent is not its own end; rather, the
capacity is to be understood in light of the value of achieving voluntary rule,
for which it is an enabling condition. Therefore, aiming at consent in a non-
instrumental way is necessarily involved in realizing the value of voluntary
rule. Obviously much more needs to be said, but the response would
proceed along these lines.
Now consider the subject whose consent is non-quality.34 In claiming
that involuntary subjecthood is bad, it is not yet clear why voluntary
33
It fundamentally differs, therefore, from Scanlon’s treatise about the significance of
choice. See Scanlon (1988).
34
One may be skeptical that anything short of quality consent is really consent to rule,
in a sense worth caring about. Indeed, some theorists claim that consent to a dominating
regime is not really consent to rule but consent to a system of mastery, thus violating the
very ideal of voluntary rule. But this stance would render it logically impossible to
consent to a dominating or unjust political order. A theory of legitimacy ought not deny
Dictionary: NOSD
subjecthood is good only in combination with some judgments.35 Earlier I

described “the dominator” and “the underling” as subjects whose judgments
about rule explicitly deny that a regime ought to, at a minimum, aim at
providing for the basic security of all subjects. One may ask, why exactly
isn’t their interest in voluntary rule being advanced? They presumably
identify with the regime and feel at home within it, but they affirmatively
deny that the government ought to be fulfilling the CMCG.
According to the sovereignty view, a subject’s interest in voluntary rule is not
advanced when the judgment on which his consent-to-rule is based on
conflicts with the CMCG.36 According to the above arguments about the
enabling role of aiming at consent, voluntary rule at the group level can only
be realized when a regime treats each subject’s interest in voluntary rule as
non-instrumentally valuable.37 If some subjects explicitly deny this, then
they undermine the regime’s aiming at group voluntary rule. The reason they
undermine the regime’s aim is because their consent is based on treating other
subjects’ consent as either not valuable or only instrumentally valuable. For
example, the judgment on which the dominator’s consent is based explicitly
denies the non-instrumental value of voluntary rule for some subjects. It
then becomes virtually impossible for the regime to aim at the dominator’s
that people consent to the exercise of power and authority under immoral conditions and
for immoral reasons (whether consciously or not). The sovereignty conception allows that
people may consent to dominating regimes (e.g. South African apartheid), and this
consent counts as genuine consent, though it may fail to be quality consent. When other
theorists deny that this counts as consent, they are really denying that the consent has a
certain sort of status (that it is legitimizing in some way). But this comes to the same
thing that a quality filter comes to, except that the sovereignty conception does not claim
as a matter of logic that people can’t consent to rule that is unworthy of consent. There is
a difference between a regime that has consent and a regime that is worthy of consent.
Neither of these can be conceptually reduced to the other, and they are both relevant to
an individual’s interests.
35
The exclusion of some cases will require justification wherever there is a quality
filter, but the explanation will obviously vary according to the filter.
36
Alternatively, one might claim that the interest in voluntary rule is advanced
only by consent to political orders that actually achieve their minimal claim, that is,
provide for basic security for everyone. I do not take this route because it would
require either stipulating that “achieving CMCG” is a constitutive ingredient in rule,
or stipulating that it is a constitutive ingredient in the voluntarism involved in
voluntary rule. The former is plainly false, and the latter has more theoretical costs
than benefits, in my view.
37
One potential problem with this response is that it seems to only require a concern
for enough subjects consenting to rule, rather than a concern for all subjects consenting
to rule. But perhaps this is not such a bad problem, since the changeability of people’s
attitudes requires an ongoing practical concern for the consent of all (in losing the
consent of some, you may need to win some others). So a concern to achieve enough
consent could translate in practice to a concern for the consent of all.
Dictionary: NOSD
96 Amanda Greene
consent while also aiming at other subjects’ consent non-instrumentally.

Therefore, those who consent to rule on grounds that conflict with the
CMCG set back one of the enabling conditions for realizing the value of
voluntary rule, and so they set back the interest in voluntary rule. Again,
much more would need to be said, but the response would proceed along
these lines.
5. CONCLUSION
I have argued that there are considerable shortcomings in the two dominant
approaches to consent-based theories of legitimacy, contractualism and
voluntarism. Although these consent-based theories affirm the ideal of
voluntary rule, they lack appropriate sensitivity to factors that should matter
for voluntary rule—namely, the actual levels of consent and non-consent
among the subjects of a regime. In order to show that these shortcomings
can be avoided, I have defended a different interpretation of the interest in
voluntary rule as the basis of legitimacy. The sovereignty conception bases
legitimacy on the proportion of actual consent among subjects in a regime,
where each individual’s consent is based on her own positive governance
assessment. In articulating the sovereignty conception, I proposed a notion
of quality consent to rule arising from the claim of minimally competent
government. Finally, I argued that the notion of quality consent affords a
theory of legitimacy that shows why legitimacy is valuable without relying
on liberal democratic presumptions. Thus the sovereignty conception, as I
have begun to defend it here, represents a promising alternative to other
consent-based theories of political legitimacy.
Bibliography
Ashford, Elizabeth and Mulgan, Tim (2012). “Contractualism,” in E. N. Zalta (ed.),
The Stanford Encyclopedia of Philosophy. <http://plato.stanford.edu/archives/
fall2012/entries/contractualism/> accessed Oct. 29, 2012.
Buchanan, Allen (2013). The Heart of Human Rights (Oxford: Oxford University
Press).
Buchanan, Allen and Keohane, Robert O. (2006). “The Legitimacy of Global
Governance Institutions,” Ethics and International Affairs 20 (4): 405–37.
Cohen, Joshua (1996). “Procedure and Substance in Deliberative Democracy,” in
S. Benhabib (ed.), Democracy and Difference: Contesting the Boundaries of the
Political (Princeton, NJ: Princeton University Press).
Copp, David (1999). “The Idea of a Legitimate State,” Philosophy and Public Affairs
28 (1): 3–45.
Dictionary: NOSD
Estlund, David (2007). Democratic Authority: A Philosophical Framework (Princeton,

NJ: Princeton University Press).
Horton, John (2012). “Political Legitimacy, Justice and Consent,” Critical Review of
International Social and Political Philosophy 15 (2): 129–48.
Huemer, Michael (2012). The Problem of Political Authority: An Examination of the
Right to Coerce and the Duty to Obey (Basingstoke: Palgrave Macmillan).
Pettit, Philip (2012). On the People’s Terms: A Republican Theory and Model of
Democracy (Cambridge: Cambridge University Press).
Rawls, John (2001). Justice as Fairness: A Restatement, Erin Kelly (ed.) (Cambridge,
MA: Harvard University Belknap Press).
Raz, Joseph (1988). “Facing Up: A Reply,” Southern California Law Review 62: 1153.
Rousseau, Jean-Jacques (1987). Basic Political Writings, Donald A. Cress (ed. and
trans.) (Indianapolis, IN: Hackett).
Sandbu, Martin E. (2007). “Valuing Processes,” Economics and Philosophy 23 (02):
205–35.
Scanlon, Thomas M. (1988). “The Significance of Choice,” The Tanner Lectures on
Human Values 8: 149–216.
Sen, Amartya (1997). “Maximization and the Act of Choice,” Econometrica 65 (4):
745–79.
Shelby, Tommie (2007). “Justice, Deviance, and the Dark Ghetto,” Philosophy and
Public Affairs 35 (2): 126–60.
Simmons, A. John (1999). “Justification and Legitimacy,” Ethics 109 (4): 739–71.
Weber, Max (1978). Economy and Society: An Outline of Interpretive Sociology
(Berkeley, CA: University of California Press).
Williams, Bernard (2007). In the Beginning Was the Deed: Realism and Moralism in
Political Argument (Princeton, NJ: Princeton University Press).
Dictionary: NOSD
4
The Value of Self-Determination
Anna Stilz
The idea of self-determination was popularized by Woodrow Wilson, who

conceived it as an extension of the doctrine—enshrined in the US
Declaration of Independence—that governments derive “their just powers
from the consent of the governed.”1 Since Wilson’s time, self-determination
has come to play a major role in international law. Article 1 of both
international human rights covenants adopted by the UN General Assembly
in 1966 declares that “all peoples have the right of self-determination,” by
virtue of which “they freely determine their political status and freely pursue
their economic, social, and cultural development.”2 Article 1(2) of the UN
Charter, the 1960 General Assembly Resolution Granting Independence to
Colonial Peoples, and the 1970 Declaration on Friendly Relations also give
prominent place to the principle.
In the broadest sense, self-determination refers to a community’s right
to govern itself independently. We can distinguish two aspects of the idea.
The “internal” aspect refers to a people’s right to choose a government that
reflects their values and priorities, while the “external” aspect denotes a
people’s right to be free from outside interference (Cassese 1995, 5–12).3
It is plausible to see the external side of self-determination as parasitic on
the internal one. We have a duty not to interfere with a foreign political
order because that order reflects the will of the people subject to it. Self-
determination holds that the people should choose or authorize their
1
Though Wilson popularized self-determination, the idea dates at least to the early
nineteenth century and was articulated by Mill, among others (Mill 2008).
2
See <http://www.ohchr.org/en/professionalinterest/pages/ccpr.aspx> and <http://
www.ohchr.org/EN/ProfessionalInterest/Pages/CESCR.aspx> accessed August 25, 2015.
3
There is an important question about the degree of independence required for self-
determination. I deny self-determination entails an unconditional right to a sovereign
state: autonomy within a larger state may be an appropriate vehicle in some circumstances.
Dictionary: NOSD
The Value of Self-Determination 99
government, and it is for this reason that outsiders are obliged not to
interfere.4
While self-determination is a cardinal principle of international law, the
proper interpretation of this principle is a matter of some debate. In practice,
the right has been limited to overseas colonies and populations under foreign
military occupation or apartheid government (Cassese 1995, 126–40). Some
commentators invoke a “saltwater test,” which holds that peoples with a right
of self-determination must be separated from the larger state by an ocean
(Philpott 2010, 156). Though there is some movement to widen the scope of
the right—by applying it to indigenous peoples or internal minorities
(Anaya 2004)—this is not yet legally binding.5 It is unclear whether there is
any sound moral case for the “saltwater test,” since territorially contiguous
populations, much like overseas groups, can be subjected to alien domination.
This chapter seeks a moral account of self-determination that might help us
interpret the proper contours of the legal principle. Is there a weighty moral
interest in collective self-governance? If so, what is the nature of this interest?
To which groups does it apply?
A moral theory of self-determination could also illuminate important
political issues. Most accounts of just war hold that a state has a right to use force
in defense of its people’s political independence. A theory of self-determination
might explain this right of national self-defense, especially in cases of merely
political wars that aim at regime change, rather than directly threatening the
lives of civilians (Rodin 2014). A theory of self-determination could also aid us
in thinking about humanitarian intervention. Do we have principled reasons to
tolerate other states’ unjust laws when they reflect its people’s self-determining
decisions? How far should this toleration extend? Finally, a theory of self-
determination could help in assessing challenges to existing political boundaries,
both from inside (by secessionist groups or decentralization advocates) and
from outside (by foreigners). While this chapter does not directly tackle these
political issues, the morality of self-determination is relevant to them.
1. Does Self-Determination Matter?
Some support for self-determination’s moral importance is provided by our

intuitive judgments about colonial rule, military occupation, and annexation.
4
Though self-determination has been used by dictators to defend oppressive regimes,
that justification seems dubious once we attend to the structure of the concept. A people
must be able to shape their political affairs to be meaningfully self-determining.
5
The UN Declaration on the Rights of Indigenous Peoples adopted by the General
Assembly in 2007 attributes the right of self-determination to indigenous peoples (Art. 3),
but is not a binding legal instrument.
Dictionary: NOSD
100 Anna Stilz
Pre-theoretic intuitions are not sufficient evidence, since we may wish to

revise them upon reflection. Still, they provide a useful starting point.
Consider three cases:
Annexation.6 In 1945, the Allies occupied Germany in a legitimate use of force.
Suppose that instead of restoring the territory to the German people, the US had
annexed their zone of occupation, turning it into an additional state of the union.
After annexation, the US governed reasonably justly, protecting the Germans’
human rights and granting them rights of democratic participation in the now-
unified polity. Would the Germans have had a claim to political independence?
Benevolent Colonialism. One historically influential defense of colonialism was the
“civilizing mission”: European rule over non-Western peoples was alleged to better
guarantee liberal rights and good governance compared to indigenous regimes
(Crawford 2002). Suppose that a “civilizing” empire did a reasonably good job of
protecting its subjects’ rights and delivering enlightened governance. Wouldn’t its
colonial subjects still have a morally significant complaint, on grounds that they
were denied the right to rule themselves?
Humanitarian Intervention. Proponents of humanitarian intervention argue that it
is permissible to intervene militarily in another state in cases of genocide, mass
expulsions, or gross violations of basic human rights (ICISS 2002). They believe
foreign rule can be legitimate, on a pro tem basis, in the aftermath of a justified
humanitarian intervention. Yet occupiers are obliged to restore the country to
independence once a decent domestic government can be established. Why do they
have this responsibility?
The best way to characterize our intuitions about these cases, I believe, is to
say that annexed, colonized, or occupied populations have a pro tanto claim
to self-rule: to govern themselves independently, and to order their
institutions as they choose. This claim is defeasible, and may be outweighed
by competing concerns, as in a justified humanitarian intervention. But
where weighty countervailing considerations are not at stake, the claim to
self-rule ought to be respected. Moreover, the claim does not depend on
cultural distinctiveness. If the US annexed English-speaking Canada, the
annexed Canadians would have the same complaint, despite the lack of
significant cultural differences. Self-determination is a political claim, not a
cultural one (for a contrasting view, see Margalit and Raz 1990).
In accepting the claim to self-rule, we view the state’s legitimate authority
as having two distinct dimensions (Simmons 1999; Pettit 2012, ch. 3).
Mainstream liberal theory has focused on a (reasonably just) state’s role in
benefiting its members. As institutional “takers,” individuals have interests
in protection of their rights, distributive justice, or public goods that the
state provides. This “taker” dimension of evaluation focuses on familiar
6
This example is drawn from Stilz (2011).
Dictionary: NOSD
aspects of the basic structure, e.g. “the way in which the major social
institutions distribute fundamental rights and duties and determine the
division of advantages from social cooperation.”7 It is concerned with the
quality of state institutions, independent of anyone’s attitudes toward them.
But on a self-determination theory, a state’s institutional quality is not all
that matters. Instead, people have an additional interest in being the authors
or “makers” of their political institutions. It may be as important that their
political institutions reflect their priorities and values (in their role as
“makers”), as that these institutions be good ones (from their perspective as
“takers”). This thought represents the “intuitive core” of a self-determination
theory, and I believe it can be formulated apart from cultural concerns.
In important ways, self-determination is the international analogue to
the idea of popular sovereignty. Popular sovereignty holds that in addition
to conforming to minimal standards of justice, a legitimate government
must also reflect the shared will of its people. A fundamental challenge to
popular sovereignty theory, however, has always been whether its central
idea of a people sharing a will makes any sense. No citizenry of a modern
state has ever contracted together to form a political community; every
citizenry is riven with diversity. And since groups rarely agree on anything
in politics, it is unclear how “the people” can share a will.
In what follows, I investigate three questions about the idea that a
political order can, and should, reflect the shared will of its people. The first,
which I call the authorization question, concerns the government–people
relation—when is a government authorized by its people? How must a
government relate to its people for it to serve as their agent? The second,
which I call the unification question, concerns the individual–people relation.
When is a population an appropriately unified people, and what makes that
the case? Finally, I also take up a third question, the value question. For
whom is it valuable that the people enjoys independent self-rule? Is that
state of affairs valuable for “We the People,” as a corporate subject? Or is it
instead valuable for the individuals who make up the group? To be realistic,
my answer must recognize that in any territorially defined citizenry, not
every individual will endorse existing political institutions. Is collective self-
determination valuable for these dissenters? If so, how? If not, does that
matter?
Before engaging these questions, I briefly characterize my approach to a
central difficulty: how do we define the “self ” in self-determination? Which
7
Rawls (1999a, 6). I do not imply that Rawls neglects the “maker” dimension of
evaluation; see his remarks on a “well-ordered society.” Among recent works on global
justice, Rawls’s Law of Peoples is notable for the weight it places on self-determination (see
Rawls 1999b, 31, 118).
Dictionary: NOSD
102 Anna Stilz
groups count as “peoples” that enjoy self-determination rights? In my view,

“peoples” are not marked out by pre-political characteristics—such as
language, culture, or shared interests—that suit them for self-rule. Instead,
a people is born only when members engage in institutionalized political
cooperation, and come to value that cooperation. Some structure of
representation is necessary to create groups with sufficient corporate agency
to act as peoples. But in defining peoples, we should not merely reference
existing structures. We also need to ask: do these structures realize self-
determination for those they currently rule? To be self-determining, members
must not only be ruled by minimally just institutions, they must appropriately
see themselves as joint authors of those institutions, through participation in
a cooperative enterprise they value. There is no independent criterion for
delineating “peoples” beyond the fact that existing structures either succeed
or fail at generating this affirmation. On my approach, the constitution of
the people is endogenous to a theory of self-determination.
The affirmation of existing structures matters because it is especially
significant for individuals to be ruled by an authority that reflects, in some
way, their own priorities. Of course, since individuals must share the
political world, no individual’s personal priorities can be mirrored in every
law and policy. Yet there is a second-order sense in which an individual’s
priorities are often reflected in his institutions: namely, when they are
shaped through his participation in a cooperative enterprise that he affirms.
Because a self-determining state reflects its people’s shared will, it has a
special claim not to be interfered with or removed.
Why is the constitution of the people endogenous, on this approach?
Though there is no independent criterion for “peoplehood,” existing
institutions sometimes fall short of guaranteeing self-determination for all or
part of their constituency. When this happens, I believe that the state’s rule
is only partially legitimate. In such cases, the ideal of self-determination may
call on us to reconfigure political boundaries, enabling alienated groups to be
governed by institutions that better reflect their shared commitments. Thus,
self-determination—the importance, to the individual, of participating in a
cooperative political enterprise he affirms—should play an ongoing role in
the boundary-drawing process. I concede that self-determination is one
value among many—it is not an absolute right—and it can sometimes be
outweighed by competing claims. If an existing regime is the only means for
providing decent rule, if reconfiguring it would jeopardize urgent interests,
or would entail unreasonably high costs, then that regime may have
permission to govern alienated constituents, on a pro tem basis. But regimes
that do not realize self-determination are not fully legitimate, and they may
be displaced if alternatives become feasible in the future. So while there is no
pre-political answer to “who are the people?,” still the ideal of self-determination
Dictionary: NOSD
gives us guidance about how to (re)negotiate political boundaries. I return to

this issue in Section 5.
2. Self-Determination as Collective Agency
Let me begin with the authorization question. When does government serve
as the people’s agent, ruling in their name? According to popular sovereignty
theory, a government (particularly its core features, such as the constitutional
principles and the regime form) serves as the people’s agent when it acts on
their shared will. To make sense of this, we must explain how a people might
share a will, and how government might reflect their shared will. It is helpful
to begin with smaller-scale examples: there is often a perfectly straightforward
sense in which everyday cooperative groups fulfill this condition.
Consider a partnership undertaking a joint venture, say running a coffee
shop together. Several philosophers have offered theories of the agency
involved in such cases, arguing that it is undergirded by a structure of joint
intentions. Michael Bratman analyzes shared cooperation in terms of
interdependent intentions of each member that (1) the group act in a
particular way, (2) they do so via the agency of all the members, (3) each
member carries out his role according to a set of meshing subplans, amid (4)
common knowledge that conditions (1), (2), and (3) obtain (Bratman 2007,
291–3).8 Roughly, I share in a structure of joint intention when I think of
myself as doing something because we are doing something together, and
I adjust my contributions accordingly, “playing my part” in the group
endeavor. In making my intentional contribution, I have an awareness of
the relation in which my action stands as a means to a group end, and I would
not perform it otherwise.
Running a coffee shop together normally involves such intentions on the
part of each partner (Bratman 2014). Their shared intention orients
members’ activity, leading them to take steps toward its fulfillment, e.g.
inquiring about a loan, or looking into possible business sites. Each partner
needs to make sure the group intention is consistent with his other plans,
say, by being careful not to schedule appointments that conflict with group
meetings. A partner may also extend support to the others in playing their
parts, perhaps giving them time off for training. Finally, their shared venture
is a cooperative activity, not merely a jointly intentional one, since the
partners are not motivated to participate because of threats or deception.
8
These conditions may seem demanding. But in my view they are best interpreted as
an (artificially precise) philosophical analysis of a tacit capacity for joint action.
Participants are usually not explicitly aware of satisfying these conditions.
Dictionary: NOSD
104 Anna Stilz
As they cooperate over time, this group will likely develop some shared
commitments about how their enterprise should be run. This does not mean
that the partners will converge in all their judgments. More probably, they
will divide on some issues, such as whether or not to stay open on Sundays
or how much vacation time to take. Even when they do not converge,
partners are often able to generate shared commitments—not reducible to
their own judgments—about how their joint venture should go. Indeed, if
we lacked this ability, everyday cooperation would be impossible for us.9
Say that Amy thinks it best that the coffee shop stay open Sundays for the
extra business. But after hearing the views of her partners and noting the
importance they place on a day off, she decides to go along with a Sunday
closure policy. Amy can freely accept her group’s policy, though it differs
from her first-order judgments. She values her participation in this venture,
so she is happy to license the policy, though she does not personally favor
this action. Often shared commitments emerge as compromises in the face
of disagreement or bargaining about how a group should structure their
enterprise. Once shared commitments have developed, the partners will feel
entitled to expect one another to act on them. If Amy is managing the shop,
her partners may expect her to stay closed on Sunday, though she would
prefer to open the business that day.
A group’s shared commitments can be articulated at higher or lower levels
of abstraction. Eventually, the group may develop policies of granting
weight to certain considerations in all decisions of a certain kind. They may
commit to always giving vacation time priority over higher wages, or to
favoring a business space that is comfortable and clean over one that is large
and modern. Because a group can arrive at shared commitments, they can
develop a group standpoint. This standpoint consists of a set of core values
and priorities—perhaps articulated at a high level of abstraction—about
how to organize more specific plans of action. While the group standpoint
will likely be responsive to some degree to the judgments of individual
members, it will not correspond to any particular person’s first-order views
about what to value. Still, a member can accept the group’s standpoint and
share in it, as long as he intends to participate in the joint venture.
Finally, if they value their enterprise, members may feel slighted when
their group’s standpoint is not respected by others. While I often disagree
with my colleagues about whom to hire, I prefer that we make our own
hiring decisions together, even though that means accepting some decisions
9
An alternative perspective sees groups as corporate agents. I do not deny the value of
this perspective. But group agency usually supervenes on the joint intentions of
individuals that they together accept certain procedures, in order to form a group agent.
See List and Pettit (2011, 33–5). It is this “undergirding” structure of joint intentionality
that interests me here.
Dictionary: NOSD
with which I disagree. Indeed, I would consider myself disrespected if the

dean overruled our collective decision, even when the result was to impose
my preferred candidate. Other things being equal, I value outsiders’ respect
for our group standpoint, though sometimes it differs from my own.
I believe there is a straightforward sense in which we can speak of this
small-scale partnership as sharing a will. Their shared will is to cooperate
together in running their coffee shop, and to license the values and priorities
that make up their group standpoint. A shared will is nothing more than an
interlocking structure of cooperative “we-intentions” on the part of each
participant, amid common knowledge that those intentions obtain. It does
not reference any further fact beyond this.
While coffee-shop partners can share a will, many people would object
that the citizens of a modern state cannot. The worry is that the model
doesn’t scale up. However, sustaining the state seems to have features in
common with the joint intentional activity of smaller-scale groups. Of
course, states exhibit formal governance institutions, such as the legislative,
executive, and judicial branches, the police, and the bureaucracy. These
formal institutions typically direct citizens’ activity. But it is the everyday
contributions of ordinary citizens—their intentions to “play their parts”—
that make formal institutions effective and stable.
Ordinary citizens of well-functioning states coordinate their behavior in
many ways that help support their government’s rule. They cooperate with
officials, judges, and the police, and they generally comply with the law.
They respect one another’s legal rights: what it means for something to be
my property, for example, is that my possession of it is generally recognized
by my fellow citizens. By paying taxes, the people contribute to the
institutions that enforce their laws and policies. Finally, when their state
affords them political representation, citizens offer input into the shape of
their political activity, voicing opinions about how their enterprise should
go. So it does not seem wrong to interpret political activity as jointly
intentional—on a grand scale—among the citizenry. Though they do not
know each other personally, citizens still participate in a shared activity
together: they uphold, reproduce, and sometimes direct a common scheme
of law. Indeed, many civic acts—such as voting, or paying taxes—would
make little sense if we did not expect our fellow-citizens to play their parts
alongside us (Kutz 2000; Stilz 2009, ch. 7).
Still, there are differences between the citizenry of a modern state and
partners in a coffee shop. One might object that these differences are so
great as to undercut any assertion that the people’s shared activity is
cooperative. Consider:
(1) A modern state is hierarchical—it is made up of leaders who direct and
followers who comply.
Dictionary: NOSD
106 Anna Stilz
(2) A modern state is coercive—it can punish people who refuse to “play
their parts.”
(3) A modern state is impersonal—no member can know or interact with
all the other members.
(4) A modern state is involuntary—while the coffee shop partners freely
joined their association, most citizens are born into theirs.
These four features do differentiate modern states from small-scale

partnerships. Still, I do not think they show that shared cooperative activity
is an inappropriate model. Indeed, I think many modern states are at least
partially cooperative enterprises, despite the four features listed above. Let
me begin by discussing the first two on the list: hierarchy and coercion.
While all states are reproduced through their citizens’ shared activity, that
activity is not always a form of free cooperation. While the citizens of North
Korea intentionally support their government’s rule, they likely do so because
they would suffer grave harms if they did not. Yet sometimes I believe that
official coercion serves to stabilize citizens’ free cooperation—facilitating the
achievement of their shared aims—rather than imposing a project on them
from outside. Here governing officials are a tool for a self-organizing citizenry
to more effectively carry out their joint purposes. In this case, the state is a
genuinely cooperative institution that reflects its people’s shared will.
Since one may doubt that hierarchy and coercion can be so lightly
dismissed, let me explain how our coffee shop could evolve hierarchy and
coercion, while still remaining a cooperative enterprise (for a similar attempt
to “scale up” from small-scale groups, see Shapiro 2011, 156–70). Suppose
the partners get busier. Due to their business’s increased complexity, it is
now harder to figure out who should do which tasks at what time. So the
partners decide to appoint a manager. He begins making most of the
decisions day-to-day, and he may even be issuing directives to the partners,
about how to divide the labor, or what prices to set. The fact that the
partners have introduced hierarchy into their enterprise does not necessarily
mean they are dominated by the manager. Even if he is largely running the
show, it is possible that the manager furthers the shared aims of the partners.
How would we know? One criterion is procedural: the partners hired him
and if they become dissatisfied enough with his performance, they could
fire him. On a “minimalist” interpretation, the manager’s acts reflect the
shared aims of the partners so long as they haven’t dismissed him yet. (One
worry is that the manager, once installed, might be able to manipulate the
partners to prevent his own dismissal, e.g. by hiding relevant information to
keep them from firing him.)
We should therefore add a substantive criterion: what the manager does
must be a reasonable pursuit of the joint venture’s aims, as these were
Dictionary: NOSD
worked out among the participants themselves. The manager has certain
“constitutional” purposes to pursue: these are defined by the commitments
that constitute the group standpoint. As I said, these commitments may be
abstract, and they may conflict in concrete cases. Part of the manager’s role
is to fill them out with more specific plans. But to meet our substantive
criterion, these specific plans must be “reasonable” elaborations of the group
standpoint. And there will be some acts the manager could undertake that
could not be plausibly interpreted that way. A manager who embezzled the
coffee shop’s funds would not count as pursuing the purposes of the
partners, under any reasonable description of those purposes. Even if he
successfully hid the evidence, or blackmailed the partners and prevented
himself from being fired, we would not want to say that he furthers the
partnership’s aims (Pettit 2005, 2012, ch. 5).
Now let me add in coercion: in addition to the organizational problems,
suppose some partners are concerned that members are shirking their
duties. Perhaps they have added new partners, so the group has become
larger. It is becoming hard to rely on everyone showing up for scheduled
shifts. This threatens to create a culture of “slacking”: since Amy knows that
Bob is often late, she feels she can be late too. Still, the partners value their
enterprise, and they don’t want to see it unravel. So they devise a solution:
from now on, the manager will now have the power to fine anyone who
does not show up for work on time. This will give everyone incentive to
comply, and return the partnership to a culture of efficiency.
Even though the coffee shop now features both hierarchy and coercion,
I submit that the partners are still self-governing through their manager. They
can be self-governing despite hierarchy so long as the manager’s decisions
qualify as reasonable elaborations of the group standpoint, and so long as
there is some way of dismissing him if he begins to abuse his powers. And
they can be self-governing despite coercion so long as they see independent
reason—apart from the manager’s threat to fine—for playing their part in
the shared plan that he now coordinates. I believe the partners do see such an
independent reason. Because they value their venture, they have reason to
show up for work, so long as the others do, and the manager’s coercion
assures everyone that this will occur, stabilizing their enterprise. Under these
conditions, the partnership is self-governing through the manager.
What about the last two features of the modern state—the fact that it is
both impersonal and involuntary? Do they undermine the possibility of
genuine political cooperation? I doubt it. Even when an institution is too
large for all its members to know one another, it can still feature significant
levels of cooperation. Though I do not know most of the other students and
employees at my university, still I “play my part” in it willingly. Many large
groups—think of megachurches, or trade unions—do not exhibit close
Dictionary: NOSD
108 Anna Stilz
personal ties among the participants, yet their activity nonetheless counts as
shared cooperation.
Finally, what about the fact that the state is involuntary? This is the most
compelling of the disanalogies between the state and the coffee shop.
Most citizens were born into their state, and they may have no reasonable
opportunity to leave. Still, I doubt that involuntariness undermines the
state’s claim to be a form of shared cooperation. Some theorists have argued
that our respect for the self-determination of a group cannot rest on respect
for the aims of its members, unless individuals have consented to
membership (Beitz 1979, 78). I reject that view. In the social contract
tradition, consent was valued as a way of ensuring that political institutions
reflected individuals’ priorities. But consent is a red herring here. For a
group to reflect the priorities of its members, its aims must be freely shared
by constituents, but membership need not be consensual for this to be the
case. We do not consent to our families, but not having chosen one’s family
does not mean that one’s family is an alien constraint on one’s life. In good
families, children have a valuable relationship with their parents, which they
care about and endorse. In this situation, though members did not join the
family, it reflects their goals and purposes much as if they had. Many other
social groups—such as workplace and school associations—are only quasi-
voluntary, since our options to join and leave them are highly constrained.
Still, these groups can (and often do) instantiate members’ shared aims. So
the fact that a relationship is not consented to does not show that it is
uncooperative.
Of course, in our coffee shop case, the partners’ shared intentions pre-
exist the establishment of authority structures. Generally, however, citizens’
shared intentions will not pre-exist the establishment of the state. Instead,
the state attempts to create shared intentions by facilitating institutionalized
joint activity among its constituents. Yet I do not think it matters, morally
speaking, whether authority structures or shared intentions come first, so
long as a group exhibits cooperative attitudes. We might imagine that after
a long period, our coffee shop partnership completely changes membership:
all the founding partners leave and new ones join. The current members
have never known a “managerless” enterprise. Yet as long as the partners
willingly play their roles, share commitments, and can revoke their manager’s
authorization, they will remain self-governing. Much the same can be said
of the state. Authority structures such as territorial jurisdiction mark out
individuals as potential members of a political community. However, there
are further “success conditions”—beyond conventional structures—for the
achievement of collective self-determination. One further condition, which
I elaborate in Section 4, is that members must reasonably affirm their
participation. Another is that the community must share an understanding
Dictionary: NOSD
of their constitutional purposes, and must be able to control their

government to pursue these purposes. If these conditions are met, I believe
a political people can be self-determining, despite the fact that their shared
intentions were in part brought into being by authority structures.
3. Institutional Conditions
By analogy to the coffee shop case, a government will reflect the will of its
people when: (a) the people willingly cooperate together to support their
government’s rule; (b) the laws and policies imposed by that government
reflect the people’s shared commitments, as these are worked out by the
members; and (c) there is some mechanism for the people to revoke
authorization of their government if it oversteps its bounds.
I should note here that a people’s shared commitments need not—and
in modern societies, usually will not—rest on commonly held substantive
values or self-understandings. In that sense, my view contrasts with
nationalist approaches to self-determination, which hold that a self-
determining group must share “a common public culture, a set of
understandings about how their collective life should be led” (Miller 2008,
124; Margalit and Raz 1990, 443). Instead, I believe a shared procedural
commitment to a joint venture is all that is required. Political unity need
not be grounded on shared first-order values, but can rest on a second-
order attitude of valuing a process of political cooperation undertaken in
concert with others.
What institutional arrangements might afford us reasonable confidence
that a particular government reflects the will of its people? This is a tricky
question (for useful discussion of these complexities, see Beitz 2009). We
should not assume that popular sovereignty requires modern democratic
institutions such as the right to vote for representatives, to associate in political
parties, and to compete for office. Historically, the meaning of popular
sovereignty was more minimal: it held that there must be some channel by
which the people could set constitutional limits to their government, and
potentially revoke their authorization of it. Locke argued that while a sovereign
people could establish a democratic legislature, they could also authorize an
oligarchy or even a monarchy (Locke 1980, ch. 10). While Rousseau stressed
that the people must retain the power to “determine the constitution of the
state by sanctioning a body of laws,” he too allowed that the people could
institute a “monarchical” or “hereditary” government (Rousseau 1997, III. 8).
In that case, “monarchy itself is a republic,” though this is simply a “provisional
form it gives the administration, until the populace is pleased to order it
otherwise” (Rousseau 1997, II. 6; III. 18).
Dictionary: NOSD
110 Anna Stilz
In a small-scale setting, everyone might simply agree that a particular

individual is wise, virtuous, and shares the community priorities, and defer
to her judgments day-to-day. If that sounds far-fetched, consider that
exactly this often happens in everyday joint ventures. “Which business site
should we purchase?” “Amy used to be a real estate agent, so I’m sure she
knows best.” “Well, let’s do whatever she says.” In a small tribe, collective
self-rule might function much like this: indeed, such groups often feature
informal decision-making by elders.
In any large and complex society, however, institutional controls will
need to be more robust. Centralized coercion is particularly worrisome
here. Where citizens’ joint activity is carried on solely out of fear, it carries
no moral weight. It is possible to freely cooperate with others only where
one is not unduly vulnerable with respect to one’s essential needs and
interests. Otherwise one may be “going along” to satisfy urgent requirements.
Some scheme of guarantees for personal independence—including security,
liberty, and subsistence—is thus a necessary condition for any cooperative
political venture (Rawls 1999b, 65). One can debate about precisely which
guarantees are appropriate—when does one dispose of sufficient options to
prove reasonably invulnerable to others’ domination? But, at least under
modern conditions, a minimal scheme of rights is a precondition of a
cooperative community.
Second, a self-determining people must have mechanisms by which to
express public opinion. Citizens must be able to discuss the shared priorities
that make up their group standpoint, and provide orientation to their
government. This means they must be free to form their own views, to declare
them, and to listen to the views of others. Though the group’s commitments
need not reflect each member’s personal priorities, in order to appropriately
see herself as a “coauthor,” each member must have the opportunity to
contribute her own views to the collective process. Without a “voice” in the
group, one is not a “maker” of the group’s institutions. There must also be
some institutional channel by which governing officials are required to take
notice of public opinion, and respond to it. If the people suspect that the
government is not acting on their shared purposes, then they should be able
to challenge its policies, and require it to provide justification.
Finally, there must be means for the people to revoke authorization of
their government if it oversteps its bounds. One can imagine a range of
mechanisms that could serve this purpose. Lockean popular rebellion seems
antiquated and unreliable, but it might be sufficient in a society where the
power of leaders depends heavily on the cooperation of the people, and
technologies of social control are undeveloped. Mass protests and non-
cooperation movements are a modern version of the same idea. Still, these
methods are insufficient in states that possess armies and secret police.
Dictionary: NOSD
Democratic rights provide robust institutional guarantees that citizens

authorize their regime. But even under contemporary conditions, I believe
it is possible for citizens to authorize a government that is not democratic—
say, a constitutional monarchy—if there is some institutional mechanism
by which they could initiate a process of constitutional reform (Altman and
Wellman 2009, 27–9).
What this discussion suggests is that the institutional forms that
safeguard collective self-determination can vary significantly depending on
the context (Rawls 1999b, 62–78; Cohen 2010). In particular, one should
be wary of dismissing non-Western forms of government in favor of
electoral democracy. Of course, it will be more difficult to know in a non-
democratic context whether the citizenry does authorize its government,
and in disputed cases, this may tell in favor of conducting supervised
plebiscites to allow people to express views about whether to continue with
a monarchy, say, or a tribal regime. But while electoral democracy is one
way of realizing collective self-determination, it may not be appropriate in
all times and places.
4. Why is Self-Determination Valuable?
So far I have argued that rule by the people is a collective phenomenon. It

obtains when “the people” rule their government. Many will be skeptical of
this approach from the beginning, on the basis that groups are not
normatively fundamental. Most philosophers believe that the well-being or
autonomy of individuals matters at the most basic level. So why should one
care about a group’s freedom to choose its own government and institutions?10
The basic answer is that one should care about the self-determination of
a group insofar as it is important for the group’s members. Even non-
consensual groups are often cooperative enterprises, in which members
value their participation and freely endorse shared purposes. But in a non-
consensual context, where individuals cannot join or leave, it is often
difficult to tell to what extent this condition is fulfilled. We need therefore
to reference some connection between the individual and the group that
makes it appropriate for her to see herself—and for others to treat her—as
a partner in a shared enterprise (Dworkin 1996, 23; Cohen 2010, 360). To
specify this connection, I turn to the individual–people relation: under what
conditions does the individual relate to a non-consensual group such as the
10
A worry about Pettit’s account in On the People’s Terms is that he largely ignores this
question. Yet if an individual rejects the people’s shared aims, then he will not enjoy
freedom-as-non-domination when he is governed by that people.
Dictionary: NOSD
112 Anna Stilz
people in such a way that its self-determination is valuable for her? I believe
such a connection exists when a citizen reasonably affirms her participation
in the group.
What is it to reasonably affirm one’s participation? A citizen affirms
participation when, upon reflection, she endorses her intention to “play
her part” in some joint enterprise (Sciaraffa 2009). To “play one’s part” is
to carry out specific tasks associated with one’s role: in the political case,
obeying the law, paying taxes, voting, cooperating with officials, and
contributing to the formation of public opinion. It is important that a
citizen’s affirmation be the product of her own reflective judgments, and
not simply of coercion or manipulation by external agents. This requires
some appreciation of the purpose behind the joint venture, and an
awareness of how one’s own contributions further its achievement. Though
the group’s commitments need not track one’s personal priorities, it is not
appropriate to see oneself as a coauthor unless one has some voice in the
group’s conduct. There must be some channel—in the form of basic
liberties of conscience, speech, and association—by which one’s opinions
can be expressed and taken notice of by other members. Finally, for a
citizen to reasonably affirm her participation, the shared venture must be
genuinely valuable. A shared political project must be at least minimally
just if it is to be reasonably affirmed.
Some may wonder whether reasonable affirmation is a type of consent. It
is not. Consent’s role is to transform the rights and obligations between
people. To do this, consent requires communicative expression: it is not just
a state of mind. Coercion or lack of decent alternatives also invalidate
consent (see Eyal 2011; Kleinig 2010). Yet reasonable affirmation need not
satisfy these conditions: it does not transform rights and obligations, it can
be purely internal, and it can occur against a coercive background where
alternatives are lacking. Like consent, affirmation is concerned with an
individual’s attitudes: there is to that extent a similarity between the
concepts. But affirmation is a weaker condition than consent. It also matters
whether affirmation is reasonable. Unlike consent, which can make a moral
difference even when it is unreasonably given, affirmation lacks normative
significance when it is unreasonable. If it is inappropriate for the individual
to see herself as a participant in a shared enterprise, her affirmation may
show only that she is thoroughly dominated or manipulated.
Imagine a scenario where all citizens reasonably affirm their cooperation
together: I call this the ideal case. In the ideal case, the purposes of the people
count as the citizens’ own purposes, to a significant degree. (I recognize that
no state is an ideal case. I discuss problems posed by more realistic cases in
Section 5. I start with the ideal case, however, because it sharply illustrates
how collective self-determination can be valuable for individuals.)
Dictionary: NOSD
The basic thought is this: in the ideal case, the group’s self-determination
is valuable for individuals because cooperation together as a people is a
shared project of those individuals. But why is it important that political
cooperation count as the citizens’ own project? Why isn’t it instead sufficient
that individuals be brought to sustain a reasonably just state, regardless of
whether they endorse that common purpose? I discuss three individual
interests served when citizens endorse their politically cooperative project:
an interest in stability, in well-being, and in political freedom.
The interest in stability is the least controversial, so let me consider it
first. Widespread affirmation on the part of citizens is often essential to
stabilizing just institutions. As I noted above, as institutional “takers,”
individuals have interests in benefiting from a reasonably just state’s rule,
through protection of their rights, and the provision of distributive justice
and public goods. Yet it is difficult to stabilize even substantively just
institutions when most participants reject their participation in them. The
importance of political stability thus provides us instrumental reason for
valuing citizen affirmation, since everyone’s “taker” interests are better
served when most citizens willingly cooperate to sustain the state. Even
dissenters benefit from the lower levels of insecurity, mistrust, and repression
that widespread affirmation often brings. Citizens’ shared will is valuable,
then, in part because it is instrumental to justice.
While important, these instrumental considerations are not, in my view,
our key reasons for caring about affirmation. I also want to point to two
more intrinsic interests. First, when individuals affirm their participation
together, the collective project comes to matter to them, in a way that
connects the success of the joint enterprise to their personal well-being.
And second, when people participate together in a coercive institution,
their relationship to that institution is different when they affirm their
participation in it than when they do not. Citizens who reasonably affirm
their participation together experience political freedom under coercive
institutions.
How does affirming one’s participation render a collective venture
important for one’s personal well-being? As a number of theorists have
argued, personal well-being depends substantially on the projects and
relationships we adopt (Dworkin 2000; Raz 1986; Scanlon 1998). Once
certain basic physical, economic, and security needs are met, well-being is
largely constituted by our success in leading a meaningful life (Wolf 2010,
26). The endeavors to which a person is committed play an important role
in determining what counts as a meaningful life for him.
Meaning involves both a subjective and an objective component
(Wolf 2010, 9). On the one hand, our actual attitudes play an important
role in determining what is meaningful for us. If we are involved in
Dictionary: NOSD
114 Anna Stilz
objectively valuable projects, yet find ourselves persistently alienated from

them, they lack meaning. Think of two bored teenagers, dragged to the Met
by their parents on a Sunday afternoon. An afternoon at the Met might be
a valuable thing, objectively speaking. Yet it does not contribute value to the
lives of these teenagers, since they lack the appropriate subjective attitudes
to appreciate it. Compare the case of someone who has a loving family, but
is persistently alienated from them. His heart isn’t in it anymore: he finds
himself constantly longing to pursue goals his family does not share. Perhaps
he ought to be more engaged with his family. Still, if he is persistently
alienated, much of the value of the relationship is lost to him (and perhaps
to his family) in an irretrievable way.
Note that the attitude of affirmation or “engagement” required to find
meaning in one’s pursuits is not a sense of pleasure or desire-fulfillment. It
is instead a cognitive attitude: a judgment that a particular endeavor is
valuable and worthwhile. I can be “engaged” with my family or my job even
though it causes me a great deal of stress, as anyone who has stayed up all
night to tend to a sick child or meet an important work deadline can attest.
While my endeavors are sometimes painful to me, still I judge them to be
worthwhile. I stress here, though, that “engagement” is an actual, not a
hypothetical, attitude. While very rational or mature teenagers might value
an afternoon at the Met, if these actual teenagers do not value it, then the
excursion is not meaningful for them.
What about pursuits that are valued but not objectively valuable? Do
these contribute meaning to a person’s life? It seems not. Consider our
reaction to the committed Mafia member, or the dedicated video poker
player. No matter how “engaged” these people are, we do not think they are
leading meaningful lives, the kind of lives we want for ourselves or our
children. Instead, we feel sorry for them: they are tragic figures wasting their
existence on trivial or immoral pursuits. So a goal must meet some threshold
of value to qualify as a constituent of a person’s well-being. We not only
want to be engaged in pursuits we ourselves judge to be good, we also want
other people to be able to appreciate why we care about our goals
(Wolf 2010). At the very least, a valuable goal must not be completely
immoral. And it must not be a pointless goal, one whose accomplishment
is completely worthless.
How might these reflections apply to our political case? When citizens
affirm their cooperation together as a people, they make the success of that
political project into one of their own goals. Insofar as their state is a
reasonably valuable institution, then its success will contribute to the overall
flourishing of their lives. This is one reason why it is pro tanto wrong for
foreigners to invade, occupy, annex, or interfere with a self-determining
people’s institutions. Such interference undermines the flourishing of its
Dictionary: NOSD
members, by preventing the fulfillment of one of their valuable aims. Since

we ought to respect the members of this group, we ought also to respect the
shared projects that they value.
It might be objected here that relatively few citizens conceive their state
as a shared project.11 Most ordinary citizens are not politically active,
knowledgeable, or patriotic. Do these people take any interest in “making”
their state? I deny that citizens must be highly participatory in order to
affirm a civic relationship. Some evidence for this is given by other joint
endeavors. While I value my university, I do not attend college-wide faculty
meetings or wear the university logo, and I cannot name all the deans. Still,
I “play my part” willingly, and I would be upset if my university was
destroyed without our consent, even if my own interests went unaffected
(e.g. if I easily found another job with a higher salary). Likewise, the
importance of citizens’ affirmation may be quite apparent in exceptional
scenarios such as foreign occupation, even if it is not apparent in everyday
political life.
Further, one might think that political disagreement is a barrier to
understanding states as shared projects. Are defeated political minorities
meaningfully part of any joint endeavor? I concede that some subgroups are
so persistently alienated that the civic relationship lacks any value for them,
and I consider their case in Section 5. But I deny that ordinary political
disagreement signals this. Again, we can draw evidence from other joint
endeavors. As mentioned earlier, I often disagree with my colleagues about
whom to hire, yet I prefer that we make our own hiring decisions together.
Much ordinary political disagreement is similarly compatible with shared
commitment to a joint endeavor.
Let me now turn to the third—and most controversial—reason why
reasonable affirmation is valuable for individuals. When a citizen affirms
her participation in the political association, I believe she relates to the
demands coercively imposed by her government in a distinctive way.
Rather than understanding those demands as hostile constraints on her
life, she is able to see them, in part, as self-imposed requirements. Citizens
who affirm their participation together in a self-determining people
experience an important form of political freedom—I call it “maker”
freedom—under coercive institutions. I think this is our most significant
reason for valuing self-determination.
How is affirmation related to maker freedom? Consider that many joint
enterprises—including the modern state—are large and complex. Tasks are
often parceled out among various roles, so each person performs a specific
function in the overall plan. As I noted in the last section, in the state, that
11
The next two paragraphs draw on Stilz (2015).
Dictionary: NOSD
116 Anna Stilz
plan is specified and imposed by a hierarchical and coercive government. As

an occupant of a particular role, it can become difficult to grasp how one’s
role conduces to the fulfillment of a valuable shared purpose. The demands
of the enterprise can seem burdensome, unfulfilling, or pointless. Consider
the mundane tasks of university administration. It is often hard for someone
weighed down with these tasks to see how they fit into any valuable shared
endeavor. When this happens, we can say that a participant is alienated
from the shared enterprise. It has no meaning for her.
To some degree, alienation is to be expected in any complex institution.
And often alienation may not present a problem: if the demands of one’s
professional role seem pointless, one may have the option of pursuing some
other occupation. (Where exit from an occupation is onerous, we should be
more concerned about alienation.) But alienation is a bigger problem when
the institution is a coercive one that pervasively shapes a person’s social
environment, and the role she plays in it is one she cannot abandon. In that
scenario, her duties may seem not just onerous, but imposed by a hostile
force that exercises near-complete control over her life. Since the state
determines so many aspects of our lives—many more than other institutions
such as universities, churches, or corporations—alienation from the state is
an important concern.
In other institutional contexts, often we have a good response to alienation:
a participant can reflect on the way in which her performance of a task
furthers the achievement of a shared purpose that she affirms. Feeling
overburdened by the demands of administration, a professor might recall
that such administration is necessary to running a successful university.
Without this tedious work, the purposes of mentoring students and fostering
an academic community could not be achieved. When we carry out this kind
of reflection, we relate differently to the demands of our roles. Administrative
tasks may still seem burdensome; we do not suddenly acquire a spontaneous
desire to engage in them. But we now see a point to the task—we understand
how this work contributes to a collective goal we value—and so the role’s
demands are imbued with new significance. We see reason to hold ourselves
to this task, even though we have no desire to do so.
Responding in this way to alienation is especially important, I think, in
the context of a pervasively controlling and coercive institution. When one
is alienated from an institution of this kind, substantial aspects of one’s life
can come to seem hostile, threatening, and completely beyond one’s grasp.
Yet suppose that one can reflect on how one’s performance of civic duties
furthers a collective enterprise whose goals one endorses. After that
reflection, these requirements no longer seem threatening; instead they have
new significance. Seeing that connection can give one reason to hold oneself
to these demands. Citizens who affirm their cooperation together as a people
Dictionary: NOSD
can impose these duties on themselves, as part of a meaningful joint project

that they endorse.
Because of this connection with maker freedom, the desire not to be
alienated from one’s political enterprise is more than a mere preference.
Rather, it is connected to a significant human need to be at home in one’s
social world.12 One of the most destructive effects of colonialism was the
coercive imposition, onto a subject population, of a social order that bore
no relation to their own priorities. Those who lived through this experience
tell of a radical sense of powerlessness, and a loss of orientation and control
(for powerful accounts, see Lear 2006 and Standing Bear 2013). Even in the
best imaginable scenario, where colonial institutions are substantively just,
still they treat the colonized as objects to be paternalized on the basis of
commitments they do not share. This imperial authority is inconsistent
with its constituents’ claims to govern their political lives in accordance
with their own priorities.
Alienation is distinct from the other abuses perpetrated by colonial
institutions, and it persists as a legacy of great bitterness and resentment
among formerly colonized populations. Obviously, alienation and affirmation
are matters of degree: one can criticize or reject many aspects of one’s social
institutions without suffering meaninglessness and loss of control. Still, the
coerced imposition of a social order that fails in any way to reflect the
“maker” freedom of those governed by it often leads to serious harms to the
population’s well-being and self-respect. For that reason, persistent alienation
from one’s political institutions is an important concern.
To sum up, then: a people’s collective self-determination will be valuable
for its members if they reasonably affirm their political cooperation together.
Affirmation serves three individual interests. First, it enhances the stability
of just institutions. Second, it contributes to well-being. Third, it enables a
valuable kind of political freedom, the freedom of understanding oneself as
a “maker” of the coercive institutions by which one is governed. In the ideal
case—where a people cooperates together to sustain a reasonably just state
and each member affirms her participation—then the collective self-
determination of that people will be both instrumentally and intrinsically
valuable for all the individuals who make it up.
5. Persistent Alienation
To this point, I have argued that collective self-determination matters

because it is valuable for individuals to be ruled by institutions that in some
12
This is a Hegelian theme. See Hardimon (1994) and Jaeggi (2014).
Dictionary: NOSD
118 Anna Stilz
way reflect their priorities. I stress again that the “priorities” reflected in our
institutions need not be shared first-order values (and in modern states,
they typically will not be). There need be no self-understandings such that
all and only members of this group accept them. Instead, it is enough
that each member of the group values the common political endeavor
and accepts some procedural commitments about how this endeavor is
conducted. When an individual is subject to institutions that are shaped
through his participation in a valued political endeavor, then those institutions
will reflect second-order priorities that he shares. Though this “maker”
interest is ultimately an interest of individuals, it can be furthered through
membership in a political group, to the extent the individual affirms
participation in that group.
But we must recognize that maker freedom is not realized for persistently
alienated dissenters. Their basic rights may be protected, but they don’t
affirm their participation in the state. The people’s shared will does not reflect
their values and priorities, even in a second-order sense.13 My question now
is: to what extent does this matter? Does the persistent alienation of dissenting
subgroups deserve an institutional response? I believe that in a limited range
of cases, alienation does deserve an institutional response.
My discussion here takes us back to the issue, mentioned at the outset, of
defining the “self ” that should enjoy self-determination. I granted there that
peoples are not prepolitically defined: they can be established only through
institutionalized political cooperation. Our discussion of alienation,
however, shows that attempts to create a people can succeed or fail. The
ideal of self-determination holds that each citizen should relate to the
political order by affirming his involvement in the cooperative political
enterprise that undergirds it. Only then can he see himself as a coauthor of
the institutions that govern his life. Yet some citizens fail to relate to their
current state in this way. Where their priorities (a) are consistent with the
provision of basic justice, and (b) can be feasibly addressed through
institutional reconfiguration, I believe we may be required to redraw
political boundaries to afford them greater self-determination. On my
approach, the boundaries of the people are always up for renegotiation. The
“people” is a mutable entity: its claims are contestable, and its constitution
is endogenous to an ongoing process of political cooperation. I stress that
this is not an argument for a unilateral right to secession, but rather an
argument that we should take the value of self-determination seriously, in
an ongoing way, in delineating political boundaries. The best institutional
13
I am not suggesting that alienated dissenters should not be fully included or should
lack full rights. I simply highlight that the self-determination of this particular people
does not further their interests in well-being and political freedom.
Dictionary: NOSD
arrangement for accommodating self-determination will depend on the

specific scenario: absent concerns about oppression or physical security, a
subgroup’s self-determination may be adequately guaranteed through
federalism or internal autonomy.
My argument for this claim emphasizes that our most basic natural
duties of justice themselves support concern with persistent alienation in
some cases. I take a broadly Kantian approach to these natural duties: each
individual has an innate right to external freedom, which requires him to
be independent from the will of other persons. This is a right to be one’s
own master, to make the decisions that govern one’s own life. To fulfill this
claim, each person must enjoy a sphere of self-determination within which
others cannot interfere. The bounds of his sphere of self-determination are
defined by his rights, including rights over his person and property. Thus,
our most basic coercible duties of justice require us to respect the external
freedom of others.
Kant adds the important further claim that many of these natural duties
of justice—particularly the duty to respect others’ property—are mediated
by the state: property requires public law to be made determinate, and so
cannot be established through uncoordinated private acts. Kant also argues
that the state is necessary to fairly enforce all our rights: no private individual
has the claim to wield coercive authority over his fellows, who have a right
not to be subject to his private power. We have a duty to comply with a
minimally just state, then, because that is the only way to fulfill our more
basic natural duty to respect others’ external freedom.
Rather than trying to defend this Kantian argument for the state, I
simply assume it here, since the issue I am interested in arises downstream
from it. It is an issue to which Kant himself did not pay much attention.
Even though a minimally just state may be necessary to guarantee
individuals’ “taker freedom,” still it can conflict with their “maker freedom.”
Let me define these two aspects of freedom more precisely:
(1) Taker Freedom: Political institutions must define and protect—as a
matter of right—certain basic “taker” interests of each member,
including personal security, basic liberties, property, and subsistence.
(2) Maker Freedom: As “makers,” members must actually affirm their own
cooperation in these institutions.
The challenge is that, as “makers,” some people may fail to affirm the
institutions that protect their rights as “takers.” They may lack maker
freedom for either of two reasons: (1) because they deny any duty to
cooperate politically in any state on minimally just terms; or (2) because,
though they acknowledge such a duty in general, they are alienated from
this state in particular.
Dictionary: NOSD
120 Anna Stilz
Such a “maker/taker” conflict arises when the particular state that coerces
the individual fails to reflect his own values, priorities, or judgments. Here
it would appear that the state infringes his innate right to be his own master.
In that case, is he subject to an arbitrary will that threatens his independence?
Sometimes, I think the answer is yes.
To develop the argument, let me return to Kant’s characterization of
our most basic natural duty: that each individual has an innate right to
freedom which everyone else can be coerced to respect. This is the right
to independence from constraint by another’s arbitrary will, in so far as it can
coexist with the freedom of every other in accordance with universal law
(6:238).14 The idea of independence highlights the importance, to freedom,
of being in a position to make one’s own decisions. One is free only insofar
as one is not forced to obey someone else.15 As just noted, it seems possible
that the state itself may threaten this claim to independence. Because the
state controls our lives and coerces us to comply, it risks dominating
alienated dissenters.
However, Kant also held that the claim to freedom-as-independence
is a limited and moralized one: it is bounded by a duty to respect others’
equivalent claims. One person’s freedom can therefore be coercively
restricted in order to protect the rightful freedoms of others (Hodgson 2010).
So while a just state’s use of coercion may threaten the dissenter’s ability to
decide for himself, that coercion is not arbitrarily constraining if it is
required to uphold others’ independence. Individuals have a limited right
not to be coerced on grounds they object to, but only if this coercion is not
necessary to uphold others’ rightful freedoms. Stressing the moralized nature
of the claim, one might argue that coercion of unwilling dissenters by a
minimally just state does not actually threaten their (moralized) independence,
even when it fails to reflect any of their own priorities. For one has no right
to have one’s independence respected in cases where respecting it would
threaten other people’s freedom.
This response is persuasive so far as it goes, but I will argue that it does
not go as far as one might think. Still, it does give us a good reply to a
certain category of dissenters: namely, people who refuse to recognize any
duty to cooperate together in a state on minimally just terms. (Anarchists,
fascists, and theocrats fall into this category, assuming Kant’s argument for
the injustice of anarchism—which I simply take for granted here—is
14
I cite Kant’s writings by the standard German edition, Kant’s Gesammelte Schriften,
edited by the Academy of Sciences (Berlin: Walter de Gruyter, ongoing from 1900).
These numbers are noted in the margins of most English translations.
15
Kantian freedom is similar to the idea of freedom-as-non-domination developed in
Pettit’s work. See Pettit (1997).
Dictionary: NOSD
correct.) To these people we say: your persistent alienation must be

discounted, because greater recognition for your values is incompatible with
protecting the basic “taker” freedoms of others. Groups that are persistently
alienated because they hold unjust aims are not wronged by being denied
self-determination.16
Still, the Kantian argument does not underwrite the thesis that a just
political order may be imposed on unwilling subjects, as in the cases of
annexation or benevolent colonialism with which we began. While coercion
by some minimally just state is necessary for securing freedom, the Kantian
thesis, by itself, is insufficient to decide how many states we should have, or
which particular people should be subject to which states.17 An institutional
response to the persistent alienation of the unwilling is often possible
consistent with protecting others’ essential “taker” freedoms. Where such a
response is feasible, I believe a due regard for Kantian independence requires
us to institute it. Individuals have an innate right to make their own
decisions about their lives, including decisions about which political unit
they wish to belong to, so long as these decisions are compatible with
upholding others’ equivalent claims. This means their actual priorities and
commitments should be extended respect, when these priorities are
consistent with basic duties of justice. So suppose that there exists a feasible
institutional alternative that would: (a) be consistent with others’ essential
freedoms, and (b) more reflective of the priorities of persistently alienated
dissenters. Then I believe we compromise the independence of those
dissenters by continuing to coerce them within existing arrangements. They
reject these arrangements, and while showing that current arrangements
were essential to guaranteeing others’ freedom would be sufficient to override
their objection, we lack that justification here.
The point is that our duties of justice massively underdetermine the
institutional configuration we should adopt. Should we institute a unitary
world state, or a plurality of states? A federal system? Internal autonomy for
indigenous peoples or other minorities? Because there are a variety of just
institutional configurations, the Kantian view cannot establish that coercion
by the current coercer(s) is necessary for background justice to be realized. Yet
16
It might be objected that the aims of anarchists and theocrats are not necessarily
unjust, as long as they don’t impose them on a non-consenting minority. What if
theocrats were able to form a state composed of only like-minded people? My response is
that since states are territorial and intergenerational, it is impossible for them to encompass
only like-minded people. A society of theocrats would be like-minded until the first
newborn arrived: then the justice-based question of how they ought to treat this non-
consenter would immediately arise.
17
Kant does suggest that for rights to be fully realized, a global juridical framework is
required. But which particular units should exist within that framework is indeterminate.
Dictionary: NOSD
122 Anna Stilz
while the alienated have a duty of justice to submit to coercion by some

reasonably just coercer, they may lack a duty to submit to this particular
coercer. If another institution would be equally consistent with others’
external freedom, and more reflective of their priorities, why can’t they
institute it instead?
Consider, then, the situation of dissenters who are persistently alienated,
not because they deny a duty to cooperate in a just state, but because they
seek an alternative configuration of units. These people do not see why this
particular government should be the one ruling their lives. They might be
alienated because their subgroup has a legacy of oppression at their current
state’s hands—many colonized or indigenous peoples fall into this category.
Or they might be alienated because, though they have no significant
history of oppression, they share distinctive political priorities on many
issues, which almost always go unrecognized by the majority. Scotland and
Quebec are persistent minorities of this kind. If these subgroups possess a
structure of representation that can be more readily affirmed, and can
construct minimally just institutions on this basis, then I believe there
is pro tanto reason to grant them self-determination. Allowing self-
determination would better facilitate their sense of participation in a
shared political project, enabling them to enjoy the goods of meaning and
political freedom in their lives. As already noted, self-determination does
not necessarily entail a right to a sovereign state; it may frequently be
satisfied through federalism, devolution, or internal autonomy. I also deny
that a self-determination claim is absolute; it can sometimes be overridden
by countervailing concerns, especially the interests of others in avoiding
conflict or instability. But—where a just alternative is accessible at
reasonable cost—I think persistent alienation generates a pro tanto reason
for institutional reconfiguration.
It might be objected here that the remainder group can become disaffected
if dissenters are allowed to redraw boundaries. The English might be
alienated if they were no longer able to live in Great Britain, in association
with the Scots, rather than in England. Here I reply once again that
freedom-as-independence is a moralized claim. The Kantian argument is
that freedom-as-independence is limited by duties of justice to respect
others’ rights, and that a minimally just state is required for these duties to
be fulfilled. Dissenters can thus be forced to cooperate against their will if
their cooperation is essential to sustaining just state institutions. But there
is a flip side to this Kantian argument: justice is the only reason unwilling
dissenters can permissibly be forced to cooperate. Once their duties of
justice have been fulfilled, individuals have a claim to independence to
pursue their own values. In particular, they should not be coerced to sustain
someone else’s identity. If greater recognition for the identity of the English
Dictionary: NOSD
would involve forcing the unwilling cooperation of the Scots—under

conditions where their cooperation is not required to uphold just
institutions—then English alienation should be discounted. (Much the
same could be said of the widespread disaffection of British imperialists, at
the moment of British withdrawal from India.)
Analogously, in other relationships—compare a marriage or a friendship—
we do not coerce disaffected parties to continue cooperating because their
departure would cause alienation for those they leave behind. Individuals
have claims to lead their own lives, within the boundaries of their duties of
justice to others. Of course, we may delay or place conditions on departure
to ensure that essential duties are discharged. But as long as exit is consistent
with justice, we recognize that people are entitled to pursue the priorities
that matter to them. Using coercion to force a dissenter to uphold someone
else’s identity, without any further justice-based rationale, is wrong.
Of course, not all alienated dissenters can be accommodated. Some will
be forced against their will to cooperate, because their cooperation is
required to sustain a reasonably just state. This is how my account would
explain a permissible military occupation, or a humanitarian intervention.
In cases of civil war, say, there may be no politically cooperative people
currently in existence. As long as the population cannot themselves sustain
a decent domestic government, they may have to put up with alienating
foreign rule, if it is the only means of establishing basic justice. Still, as long
as the occupying administration is not widely affirmed by its constituency,
its legitimacy is only partial. If the time comes when those constituents can
set up a minimally just regime that would better reflect their values, they
may displace their foreign rulers.
A feasible response to the alienation of dispersed ideological minorities—
such as socialists—may also be lacking. My reply to them stresses that our
duties of justice are importantly territorial in nature. Since, on a Kantian
view, the state plays an essential role in establishing property rights,
jurisdictions need to be territorially defined if they are to carry out morally
mandatory tasks. If each person were extended the discretion to sign up for
the regime of property, contract, and tort law of his choice, interactions
between individuals would generate conflicts that undermine their
independence from one another. Groups claiming self-determination must
therefore be capable of territorial organization in representative institutions.
Since dispersed minorities are unable to fulfill this condition, we have
sufficient reason to reject self-determination for them, as inconsistent with
upholding the “taker” freedoms of others.
One might object here that no coercive arrangements are ever strictly
necessary to secure taker freedom. There will always be some alternative legal
regime that could uphold basic justice among us. But I do not have in mind
Dictionary: NOSD
124 Anna Stilz
logical necessity, but rather feasibility in given political circumstances, which

involves a contextual assessment. To be feasible, an alternative institutional
arrangement must: (a) be accessible, through achievable political reforms;
(b) entail acceptable costs of transition; and (c) be likely stable once realized
(Gilabert and Lawford-Smith 2012). Many logically possible alternatives will
not be presently feasible in this sense. So, in concrete cases, we will need to
weigh the claims of the persistently alienated against the countervailing risks
to just institutions, including the potential for civil unrest, instability, ethnic
conflict, or rights violations. Self-determination is not an absolute right, but
rather a weighty moral claim that must be applied with due regard for
circumstances. Sometimes the risks of reconfiguring our institutions will be
so great as to outweigh persistent alienation. Still, institutional alternatives
that are inaccessible at one point in time may become accessible later
(compare the discussion of “context-dependent” moral requirements in
Patten 2014, 24–7). Moreover, it is often immediately possible to allow
greater self-determination at low institutional risk, as in 1934, when the US
Congress passed the Indian Reorganization Act, ending fifty years of direct
federal control over Native American affairs.
Our reasons for granting self-determination—in cases where a just
institutional alternative can be instituted at reasonable cost—are rooted in
the basic Kantian thought that people have an innate right to independence,
and this gives them standing to reject unwanted coercion whenever that
coercion is not necessary to securing the freedoms of others. While self-
determination is important, it stands in a subordinate relation to natural
duties of justice. Self-determination is not an absolute right, and it can be
legitimately denied in cases where its recognition would threaten decent
governance, jeopardize urgent “taker” interests, or entail unreasonably high
costs. Yet I believe self-determination still has very great weight. Only a state
that affords its members self-determination is fully legitimate. Otherwise,
minimally just institutions are subject to the constant possibility of
displacement by alternatives that better guarantee self-determination.18
6.â•‡CONCLUSION
We have come a long way, so I briefly summarize the chapter’s main

conclusions. I believe our regard for self-determination is ultimately rooted
18
â•‡ Here I differ from Allen Buchanan, who argues that self-determination is only a
remedial right against a government that persists in serious injustices. On my view, while
potential injustice is a constraint on self-determination, it is not a necessary condition
(see Buchanan 2004, 351–7).
Dictionary: NOSD
in a fundamental “maker” interest of individuals—an interest in being the

author of the institutions that govern one’s own life. The state’s imposition
of coercive requirements may threaten this interest, if those requirements
fail to reflect the individual’s values and commitments. Though the “maker”
interest is an interest of individuals, it is furthered by an individual’s
membership in a self-determining group, to the extent that he affirms his
participation in that group. Because of the importance of this “maker”
interest, a perfect regime would not only be just, it would also be self-
determining—i.e. it would rest on the shared will of its people. In the real
world, though, the peoples of even reasonably just states fall short of actual
consensus. How seriously should we take the complaints of alienated
dissenters under these conditions? I believe we have an important though
limited reason to take alienation seriously, since unwanted coercion may
threaten Kantian independence where it is not essential to upholding just
institutions. If a reasonably just but less alienating institutional alternative
is available, a due regard for Kantian independence gives us reason to
institute that arrangement.
Bibliography
Altman, Andrew and Wellman, Christopher (2009). A Liberal Theory of International
Justice (Oxford: Oxford University Press).
Anaya, S. James (2004). Indigenous Peoples in International Law (Oxford: Oxford
University Press).
Beitz, Charles (1979). Political Theory and International Relations (Princeton, NJ:
Princeton University Press).
Beitz, Charles (2009). “The Moral Standing of States Revisited,” Ethics and
International Affairs 23(4): 336–8.
Bratman, Michael (2007). Structures of Agency (Oxford: Oxford University
Press).
Bratman, Michael (2014). Shared Agency: A Planning Theory of Acting Together
(Oxford: Oxford University Press).
Buchanan, Allen (2004). Justice, Legitimacy, and Self-Determination (Oxford:
Oxford University Press).
Cassese, Antonio (1995). Self-Determination of Peoples: A Legal Reappraisal
(Cambridge: Cambridge University Press).
Cohen, Joshua (2010). “Is There a Human Right to Democracy,” in J. Cohen, The
Arc of the Moral Universe (Cambridge, MA: Harvard University Press).
Crawford, Neta (2002). Argument and Change in World Politics: Ethics, Decolonization,
and Humanitarian Intervention (Cambridge: Cambridge University Press).
Dworkin, Ronald (1996). Freedom’s Law (Cambridge, MA: Harvard University
Press).
Dictionary: NOSD
126 Anna Stilz
Dworkin, Ronald (2000). Sovereign Virtue (Cambridge, MA: Harvard University

Press).
Eyal, Nir (2011). “Informed Consent,” in E. Zalta (ed.), The Stanford Encyclopedia
of Philosophy (Stanford, CA: Stanford University) <http://plato.stanford.edu/
entries/informed-consent/> accessed Aug. 25, 2015.
Gilabert, Pablo and Lawford-Smith, Holly (2012). “Political Feasibility: A
Conceptual Exploration,” Political Studies 60(4): 809–25.
Hardimon, Michael (1994). Hegel’s Social Philosophy: The Project of Reconciliation
(Cambridge: Cambridge University Press).
Hodgson, Louis-Philippe (2010). “Kant on the Right to Freedom: A Defense,”
Ethics 120(4): 791–819.
ICISS (International Commission on Intervention and State Sovereignty) Report
(2002). The Responsibility to Protect (Ottawa: International Development Research
Centre).
Jaeggi, Rahel (2014). Alienation (New York: Columbia University Press).
Kleinig, John (2010). “The Nature of Consent,” in F. Miller and A. Wertheimer
(eds), The Ethics of Consent (Oxford: Oxford University Press): 3–24.
Kutz, Christopher (2000). Complicity (Cambridge: Cambridge University Press).
Lear, Jonathan (2006). Radical Hope (Cambridge, MA: Harvard University Press).
List, Christian and Pettit, Philip (2011). Group Agency (Oxford: Oxford University
Press).
Locke, John (1980). Second Treatise of Government, C. B. Macpherson (ed.)
(Indianapolis, IN: Hackett).
Margalit, Avishai and Raz, Joseph (1990). “National Self-Determination,” Journal of
Philosophy 87(9): 439–61.
Mill, John Stuart (2008). On Liberty and Other Essays, John Gray (ed.) (Oxford:
Miller, David (2008). Global Justice and National Responsibility (Oxford: Oxford
University Press).
Patten, Alan (2014). Equal Recognition (Princeton, NJ: Princeton University Press).
Pettit, Philip (1997). Republicanism (Oxford: Oxford University Press).
Pettit, Philip (2005). “Rawls’s Political Ontology,” Politics, Philosophy, and Economics
4: 157–74.
Pettit, Philip (2012). On the People’s Terms (Cambridge: Cambridge University
Press).
Philpott, Daniel (2010). Revolutions in Sovereignty (Princeton, NJ: Princeton University
Press).
Rawls, John (1999a). A Theory of Justice (Cambridge, MA: Belknap Harvard
University Press).
Rawls, John (1999b). Law of Peoples (Cambridge, MA: Harvard University Press).
Raz, Joseph (1986). The Morality of Freedom (Oxford: Clarendon Press).
Rodin, David (2014). “The Myth of National Self-Defence,” in C. Fabre and
S. Lazar (eds), The Morality of Defensive War (Oxford: Oxford University Press):
69–89.
Rousseau, Jean-Jacques (1997). The Social Contract and Other Later Political Writings,
Victor Gourevitch (ed.) (Cambridge: Cambridge University Press).
Dictionary: NOSD
Scanlon, T. M. (1998). What We Owe to Each Other (Cambridge, MA: Belknap

Harvard University Press).
Sciaraffa, Stefan (2009). “Identification, Meaning, and the Normativity of Social
Roles,” European Journal of Philosophy 19(1): 107–28.
Shapiro, Scott (2011). Legality (Cambridge, MA: Harvard University Press).
Simmons, A. John (1999). “Justification and Legitimacy,” Ethics 109(4): 739–71.
Standing Bear, Luther (2013). My People, the Sioux (Lincoln, NE: Bison Books).
Stilz, Anna (2009). Liberal Loyalty: Freedom, Obligation, and the State (Princeton,
NJ: Princeton University Press).
Stilz, Anna (2011). “Nations, State, and Territory,” Ethics 121(3): 572–601.
Stilz, Anna (2015). “Decolonization and Self-Determination,” Social Philosophy and
Policy 32: 1.
Wolf, Susan (2010). Meaning in Life and Why it Matters (Princeton, NJ: Princeton
University Press).
Dictionary: NOSD
5
Domination and the Rule of Law
Assaf Sharon
The rule of law is often advocated as a solution to the tension between

authority and liberty. Political liberty, so the claim, is primarily the absence
of domination and the rule of law, though authoritative, does not dominate.
This argument has been articulated most elaborately by contemporary
republican (or neo-Roman) theorists, who identify domination with the
potential for arbitrary interference. I argue,1 first, that their arguments
employ an unsustainable concept of arbitrariness. And, second, that insofar
as individual liberty is concerned, singling out arbitrary interferences is
misguided.
1. Law and Liberty
How can the legitimate wielding of coercive force by political authorities be

reconciled with individual liberty? This is among the most fundamental
questions of political philosophy. A common strategy of reply employs the
rule of law. “As in absolute governments the King is law, so in free countries
the law ought to be King; and there ought to be no other,” Thomas Paine
wrote. Locke famously said that “where there’s no law there’s no freedom,”
and that “where law ends, tyranny begins.” Hayek, to take a more recent
example, claimed that “When we obey laws . . . we are not subject to another
man’s will and are therefore free,” while Rawls asserted that “the rule of law
1
Versions of this chapter were presented at the Law and Philosophy Workshop at the
Hebrew University of Jerusalem and at the Oxford Studies in Political Philosophy 2014
Workshop. I am grateful to participants in both forums for their insightful comments.
Particular thanks to Andrew Lister and Peter Vallentyne for their helpful questions and
suggestions. I also thank Joshua Cohen, David Enoch, Alon Harel, David Hills, Nadeem
Hussain, Debra Satz, and Naomi Sussmann for helpful comments on earlier drafts.
Dictionary: NOSD
Domination and the Rule of Law 129
is obviously closely related to liberty.”2 One can quote Rousseau, Mill,

Blackstone, or Dicey, as well as many other philosophers and jurists
expressing essentially the same idea—hail the rule of law, for it is wedded
to liberty.
Republican political theory continues this venerable tradition.3 The key
novelty of republicanism is meant to be its concept of liberty, supposedly
distinct from either the ancient idea of liberty as active engagement in free
government and from the liberal conception of freedom as lack of external
interferences. One of the main virtues of republican liberty is supposed to
be the fact that it allows for the reconciliation of political authority with
individual liberty, primarily by the rule of law.
Republicans do not say, in the modernist manner, that while the law coerces people
and thereby reduces their liberty, it compensates for the damage done by preventing
more interference than it represents. They hold that the properly constituted law is
constitutive of liberty in a way that undermines any such talk of compensation.4
My aim in this chapter is to analyze and assess this view. I will argue that it
is an indefensible position founded on an incomplete picture both of liberty
and of law. But before looking at the republican argument in support of this
view, let me quickly note the challenge it faces.
2. Law Versus Liberty
On one natural view of law, the reconciliation of law and liberty, let alone
the idea that law is essentially a means for liberty, is a non-starter. “Law,”
Hobbes wrote, “was brought in to the world for nothing else but to limit
the natural liberty of particular men.”5 Bentham similarly claimed that “all
coercive laws are, as far as they go, abrogative of liberty.”6
By its very nature law is a means of control, imposing constraints and
restrictions on the free exercise of individual choice by coercively inflicting
sanctions. Law is nothing if it isn’t an instrument of intervention in the lives
Paine (1997); Locke (1988: §57, §202); Hayek (1978); Rawls (1999: 207).
2
“Republicanism” here refers to the family of contemporary views associated with the
3
work of Quentin Skinner, Phillip Pettit, and others who have followed their lead in the
attempt to resurrect what they view as a political theory that significantly differs from
liberalism. Skinner often uses the term “neo-Roman,” and others use “neorepublican.” To
keep with standard usage I will use “republican,” but this should not obscure the fact that
I’m referring to the contemporary views, not their early modern sources of inspiration.
4
Pettit (1997a: 35, see also 175 ff.); Pettit (2009: 49 ff.); Skinner (2008: 84); Larmore,
(2003: 86); Viroli (2001: 9).
5
Hobbes (1994: xxvi, 8).
6
Bentham (2010).
Dictionary: NOSD
130 Assaf Sharon
of individuals. When a certain domain is brought under the purview of the

law, the state is authorized to restrain, inspect, monitor, discipline, and
punish its subjects’ behavior in that practical domain.
Insofar as the freedom of individuals to do as they like, to conduct their
lives as they see fit, to act on their own judgment, is regarded as valuable,
being constrained by anything other than oneself is undesirable, be it person
or law. In fact, from this perspective being ruled by laws may plausibly seem
particularly pernicious given their typical rigidity and impersonality. But
I should not get ahead of myself. The point for now is just the challenge for
conciliatory views—to explain how law, the very means by which states
restrict individual choice and action, is not only not particularly restrictive
of liberty, but is fully reconcilable with individual liberty.
3. Non-Domination
One approach is to argue that a rule of law provides the best protection
of individual liberty because it guarantees minimal interference, allowing
only interferences that prevent more ominous interferences. As Pettit’s
programmatic comment (quoted in fn. 4) expresses, republicans do not take
this approach, according to which law and liberty remain principally
antagonistic. To forge the desired compatibility between the rule of law and
individual liberty they weaken the connection between liberty and non-
interference. They are not alone in this strategy. As there is little promise in
trying to overcome any and all interference while maintaining the force of the
law, the more prevalent approach has been to identify liberty with the
elimination of only a restricted class of interferences; commonly this is the class
of arbitrary interference. “For republicans,” says Skinner, “the fundamental
affront [to liberty] is always the mere existence of arbitrary power.”7 Republicans
thus identify liberty with ‘non-domination,’ when domination is understood
as the potential for arbitrary interference.8 Non-arbitrary interferences do not
dominate and therefore do not offend against liberty.9
7
Skinner (2008: 96).
8
See, e.g., Pettit (1997a: 52); Pettit (1997b: 115); Pettit (2002: 340); Skinner (1998:
70); Viroli (2001: 35 ff.).
9
This was actually a matter of some contention between the two leading republican
thinkers. Skinner thought that both arbitrary interferences, actual or possible, and
non-arbitrary actual interferences determine the extent of one’s liberty. On this view,
interferences of both kinds—even by non-arbitrary rule of law—limit freedom (Skinner
1998: 82–4). For a restatement of this position see Skinner (2006: 156–70). Pettit resists
this concession to the standard liberal view. For him liberty consists fundamentally in the
alleviation of arbitrary interferences (1997a: 5 and 2012: 43–4). Viroli takes a similar
position: “the central point for classical republican theorists is that dependence is a more
Dictionary: NOSD
The next move is obvious enough. Having established that only arbitrary
interferences jeopardize liberty, republicans proceed to claim—with other
proponents of the rule of law—that it is not arbitrary and therefore not
restrictive of liberty. “While the law necessarily involves interference—while
law is essentially coercive—the interference in question is not going to be
arbitrary.”10 Thus law can be “entirely undominating and entirely consistent
with freedom.”11 In fact, “the model of interference without domination is the
coercion of the will implicit in the imposition of a nonarbitrary rule of law.”12
The difficult question, clearly, is what is to count as arbitrary. It is to the
credit of republican theorists that they have labored to provide a detailed
account of what they mean by “arbitrary interference.” This offers an
opportunity to assess not only the success of the republican version, but also
the prospects of this entire strategy of reconciling authority with liberty and
of justifying the rule of law.
For the argument to succeed it must define arbitrariness adequately, and
it must do so not only once, but twice. To defend the rule of law in virtue
of its non-arbitrariness, it is necessary to establish a principled distinction
between arbitrary and non-arbitrary interferences such that the identification
of only the former with restriction of liberty will make sense. Then we need
painful violation of liberty than interference” (2002: 10). On this view, arbitrary
interferences—both actual and potential—constitute the primary threat to individual
liberty. Non-arbitrary interferences, even if actual, are at most secondary. To be accurate,
interferences of the latter kind, according to Pettit, “condition” liberty, but do not
“compromise” it (1997a: 26; 2002: 342; and see 2012: 28), drawing a corresponding
distinction between being “unfree” and “non-free” (2001: 374 and 2012: 41). In his recent
On the People’s Terms Pettit replaces talk of compromising and conditioning freedom with
a distinction between violation and vitiation of freedom (see Pettit 2012: 38, note 15), but
regardless of the terminology, the basic point remains—non-domination is the key
element of personal freedom. In more recent work Skinner concedes to Pettit that
domination, that is arbitrary interference, is the fundamental affront to freedom
(Skinner 2008: 84; and Viroli 2001: 54). His stated reason for changing his mind is quite
unsatisfying, namely Pettit’s observation “that the capacity to engage in acts of arbitrary
interference depends upon the prior possession of arbitrary power, and thus that the
underlying presence of such power must constitute the fundamental affront to liberty.”
This is obviously an invalid inference. Perhaps part of the motivation for changing his
mind is that his original position narrows the gap between republicanism and liberalism,
thereby undermining its claim to uniqueness (see criticism on this point by Carter 2008,
and Kramer 2008, and the (to my mind unsatisfying) reply in Skinner 2008).
10
Pettit (1997a: 36–7); compare Pettit (2009: 45); Skinner (2008: 86); Viroli (2001:
37 ff., 47).
11
Pettit (1997a: 66).
12
Pettit (2002: 344) and see Pettit (1996: 577). Despite such formulations by
republicans, they should not be ascribed the indefensible view that the rule of law is
sufficient for individual liberty. It does seem to be the case that they regard it as sufficient
for ensuring that legislation does not undermine liberty. This is apparently the implication
they draw from the claim that interference does not conflict with liberty if it is not
arbitrary. It is this claim that is the focus of my discussion.
Dictionary: NOSD
132 Assaf Sharon
a definition of arbitrariness according to which the rule of law is not

arbitrary. Call this the conceptual challenge. The notion of arbitrariness
employed in these parts of the argument must be the same. Otherwise, the
rule of law, though non-arbitrary in some sense, may still conflict with
liberty which requires non-arbitrariness in a different sense. Call this the
non-equivocation challenge. We can think of the republican defense of the
rule of law as an argument from two premises:
1. Interferences are restrictive of liberty when they are arbitrary
2. Interferences subject to the rule of law are not arbitrary, to the
conclusion that:
3. Interferences subject to the rule of law are not restrictive of liberty.
The conceptual challenge concerns the soundness of the premises and the
non-equivocation challenge concerns the validity of the inference of the
conclusion from them.
The second part of the chapter analyzes the notion of arbitrary interference
and argues that it fails to meet either of these challenges. I begin with laying
out theoretical constraints on a proper concept of arbitrariness (Section 4).
Then, I inspect three prominent proposals—defining arbitrary interference
in terms of consent, interests, or control—and show why each of these fails
(Sections 5, 6, and 7). Turning to the rule of law, I show that to the extent
that it can be said to be non-arbitrary, it is so in a different sense from all of
the above, thus failing the non-equivocation challenge (Section 8), and
address possible replies (Section 9).
After presenting the shortcomings of the main proposals on offer, I proceed
in the third part to look at the underlying idea, namely that subjection to
personal rule is more objectionable from the point of view of individual
liberty than non-personal rule. I propose that the best way of making sense of
this idea is in terms of discretionary power (Section 10), and then question the
ideal of eliminating discretion. I first note that eliminating discretion is not
always desirable (Section 11), and then challenge the idea that discretion is the
principal obstacle to political freedom (Section 12). In fact, I argue, focusing
on it draws attention away from other, possibly graver threats to liberty. The
laxness toward non-arbitrary interference, I claim, is based on a narrow view
of political freedom and the challenges it faces.
4. Arbitrariness
The first question to ask is: what makes an interference arbitrary? In one
obvious sense, action is arbitrary when it is not supported by reasons. Thus
Dictionary: NOSD
my choice of one bottle of water over another is usually arbitrary, as both

are identical with respect to my desire to drink water.13 Choosing to purchase
a water bottle is not arbitrary because it is grounded in reasons (my thirst
and the belief that drinking water will quench it). But since I have no
reason to prefer one bottle to another, picking one of them over the other
is arbitrary.
In the case of political intervention, however, the identification of
arbitrary choice with unreasoned choice cannot suffice. Count Almaviva
in The Marriage of Figaro has reason to reinstate the droit de seigneur (or
“right of the first night”) so he can exercise it on the object of his desire,
Susanna.14 But the fact that this intervention would be reasoned on his
part hardly makes it less invasive and indeed arbitrary for Susanna or for
Figaro, her fiancé.
One natural response is to require that interference be based not only on
some reason, but on the right reason. Although the count has his reasons to
reinstate the feudal right, namely his desire for Susanna, these are not valid
or good reasons. Arbitrary interference, according to this suggestion, is not
interference not based on reasons, but interference based on wrong or
inappropriate reasons.
The question is, what kinds of reasons are wrong for political agents to
act on? One answer suggests itself. Acting as the people’s representatives and
from the power vested in them by the people, political authorities can only
appeal to reasons dictated by the terms of their mandate—promoting the
common good.15 A moralistic view of this kind encounters difficulties in
both parts of the argument, that is both as a conception of freedom and as
a defense of the rule of law. One problem is that it entails unintuitive
upshots, such as that being jailed does not conflict with liberty so long as
one was imprisoned by just laws.16 In general, a moralistic view of liberty
13
To adopt the terminology of Morgenbesser and Ullman-Margalit (1977), this is a
case of “picking” as opposed to reason-based “choosing.”
14
The reality of this practice in the Middle Ages (which need not concern us here) has
been contested by historians, see Boureau (1998).
15
Such a moralistic position is associated with Locke, who defines liberty as the ability
to follow reason in the pursuit of one’s true interests (Locke 1988: §58, §59, §63 and
1979: II, xxi), and the rule of law as a regime of laws aiming at the common good. Thus
the rule of law is by definition not only protective of liberty, but actually conducive to it:
“For all the power the government has, being only for the good of the society, as it ought
not to be arbitrary and at pleasure, so it ought to be exercised by established and
promulgated laws” (1988: §57, §137, and see Halldenius 2003: 263). Another version of
the moralistic view is Nozick’s rights-based view of interference (Nozick 1974: 262 ff.).
For another version see Miller (1983).
16
For this criticism see Cohen (2011) and (1995).
Dictionary: NOSD
134 Assaf Sharon
will sanction paternalistic interventions to which defenders of negative

liberty are wholly unsympathetic.17
As a defense of the rule of law, the moralistic or right-reasons view of
arbitrariness faces two additional problems. The identification of the rule of
law with the rule of right reasons divests the notion of its theoretical import:
If the rule of law is the rule of good law then to explain its nature is to propound a
complete social philosophy. But if so the term lacks any useful function. We have no
need to be converted to the rule of law just in order to discover that to believe in it
is to believe that good should triumph.18
A second, related problem for the moralistic view arises from the disparity
between the right reasons for political action and the rule of law. The
moralistic conception of the rule of law as laws based on the right reasons is
a matter of pure stipulation, which doesn’t take us very far theoretically, as
noted by Raz. But if the rule of law is taken as an independent notion—
some version of the idea of equal subjection of all to general, prospective,
and publicly promulgated rules—then its connection with the right reasons
is dubious. The rule of law thus conceived seems neither necessary nor
sufficient for decisions based on the right reasons.
Contemporary republicans are not troubled by the unintuitive upshot
regarding lawful incarceration as they do not share Alan Ryan’s thought that
“a man in jail is paradigmatically unfree.”19 In fact, they regard it a virtue of
their view, announcing that “neither a tax levy, nor even a term of
imprisonment, need take away someone’s freedom.”20 But they do reject the
moralistic interpretation of arbitrariness on account of the paternalist
upshot and its shortcomings as a defense of the rule of law.21
So here is a first general lesson we can draw: To defend the rule of law in
terms of the non-arbitrariness of the interferences it allows, it is necessary to
find a principled middle ground between the no-reasons view and the right-
reasons view.
17
The locus classicus of this criticism is Berlin (2002). Republican theorists share it.
The republican conception of liberty is negative in the relevant respect (see Pettit 1997a:
51, 2002: 342; Skinner 2002: 255 and 261–2).
18
Raz (2009: 210–11).
19
Ryan (1993: 293).
20
Pettit (1997a: 56 n. 3), and see Skinner (2008: 87).
21
See Pettit (2008: 117 and 128 n. 19); Pettit (2009: 40); Pettit (2012: 58–9); Skinner
(2008: 87–8). In light of these later texts it seems clear that Larmore (2003: 91) is wrong
to ascribe to Pettit a moralist view identifying non-arbitrary law with “just law” (p. 85)
and wrong in claiming that his aversion to paternalism and regarding the Rousseavian
idea of forcing people to be free as paradoxical are based on “confusing the absence of
domination with the absence of interference.”
Dictionary: NOSD
5. Arbitrariness and Consent
What is the alternative? Or, to put the question differently, how can the
apparent tension between liberty and coercion be resolved without adopting
a moralized view of liberty? Quentin Skinner defines arbitrariness by the
idea of consent:
What it means to be a free-man under such an association is only that your liberty
is never curtailed by arbitrary power; it is only ever limited by laws to which you
have given your explicit consent . . . so long as you give your consent, the law itself
can be regarded as an expression of your will, as a result of which you may be said to
remain a free-man in obeying it.22
As Skinner acknowledges, this raises the obvious difficulty that consent
does not guarantee freedom (powers erected via consent can be dominating;
individuals can consent to their own enslavement). It is not clear how
Skinner hopes to address this worry. Sometimes it seems that he is assuming
something like direct democracy where every individual can be regarded
as having consented to each and every law to which he or she is subject.23
But this is not a very realistic proposal. But even without this radical
assumption, defining arbitrariness in terms of consent raises familiar
worries. Actual consent is dubious for familiar Humean reasons—there is
no record of contemporary acts of consent and as for the idea of some
historical Ur-consent “besides that this supposes the consent of the fathers
to bind the children, even to the most remote generations . . . it is not
justified by history or experience, in any age or country of the world.”24
Hypothetical consent, on the other hand, is notoriously normatively weak.
The fact that I would have consented to the state’s authority, or even that
I ought to have consented, does not seem to create an obligation for me to
respect its authority.25
It might be suggested that although hypothetical consent does not
suffice for creating political obligation, it may be enough to secure liberty.
But this does not seem very convincing. For one thing, the reason we require
consent to political authority in the first place is the value of individual
liberty. Consent neutralizes offensive interventions, presumably, because it
entails that they are self-inflicted and being restricted by constraints one
imposes on oneself does not limit one’s freedom. As Skinner says, “so long
Skinner (2008: 86).

22
Saying things such as: “we remain free-men if our liberties can be constrained only
23
with our own consent” (Skinner 2008: 88).

24
Hume (1994: 189).
25
For this criticism see Dworkin (1977: ch. 6).
Dictionary: NOSD
136 Assaf Sharon
as you give your consent, the law itself can be regarded as an expression of
your will, as a result of which you may be said to remain a free-man in
obeying it.”26 The appeal to actual consent here is not accidental—what
makes it the case that the law does not restrict your liberty is that you have
in fact consented to it. The hypothetical possibility of your consent does
not have this implication.27
But in the context of supporting the rule of law, there are further
difficulties, beyond the general worries about hypothetical consent. To
succeed, theories of hypothetical consent must specify the conditions
hypothesized, that is the conditions in which the agent would presumably
consent. The question is whether these conditions can be coherently
specified. This general worry about consent theories is particularly pressing
in the present context, as it is not just the act of consent that is hypothetical,
but also the consenting agents themselves. In other words, hypothetical
consent is not just the idea that had the individuals who were party to
the arrangement been asked they would have expressed their agreement
to it.28 This would be a high bar for most political arrangements. The
relevant consent is that of the ideal counterparts of actual agents (or
those agents under—external and internal—ideal conditions). Usually
these conditions are associated with rationality—had one been fully
aware of one’s genuine interests, one would have consented. In the political
context, there is often an added ingredient of impartiality, as what is
required is an overlapping consent of many individuals. What diverse
ideal agents, fully conscious of their interests, would consent to is,
presumably, their shared or overlapping interests, that is, the common
good. So part of the hypothetical conditions of consent to a policy turns
out to be that it track the common good.29 But this takes us back to a
moralistic, right-reasons view, according to which law is justified when
and because it tracks the common good, a view that by their own lights
republicans cannot endorse.
26
Skinner (2008: 87).
27
This is not to say that hypothetical consent never has any normative impact. It
clearly does, for instance, when the agent is not in a position to give or withhold her
consent, as in Dworkin’s example of treating an unconscious, injured individual. As an
anonymous reader for this publication noted, Dworkin also employs the notion of
hypothetical insurance to justify certain forms of redistribution.
28
This is the idea expressed by the standard examples of effective hypothetical consent,
e.g. the drunk being dragged off the road, or the provision of medical assistance to the
unconscious injured.
29
Thus consent does not serve as the source of legitimacy, but as a proxy for reasons
one has independently, or as Dworkin says, we “use the device of a hypothetical
agreement to make a point that might have been made without that device”
(1977: 151).
Dictionary: NOSD
6. Arbitrariness and Interests
Phillip Pettit defines arbitrariness in terms of the interests and opinions of

the interferee:
I think that someone has an arbitrary power of interference in the affairs of another
so far as they have a power of interference that is not forced to track the avowed or
readily avowable interests of the other.30
Arbitrary interference, then, according to Pettit, is not unreasoned
interference, nor interference without consent, but interference that is not
attentive to the subject’s “avowed or readily avowable” interests and
opinions. Legal interference is, according to this view, compatible with the
liberty of its subjects if it satisfies two conditions:
the legal authorities will be entitled and enabled to interfere only when pursuing the
common interests of citizens and only when pursuing these in a manner that
conforms to the opinions received among the citizenry.31
The second requirement—conformity to the opinions of those affected—
comes dangerously close to the requirement of consent. The fundamental
idea behind it is clearly similar to the one behind consent:
Domination, as I understand it here, is exemplified by the relationship of master
and slave or master and servant. Such a relationship means, at the limit, that the
dominating party can interfere on an arbitrary basis with the choices of the
dominated: can interfere, in particular, on the basis of an interest or an opinion that
need not be shared by the person affected.32
Pettit’s choice of third-person singular parlance in this passage is telling. So
long as we think of a single agent, as in the master–slave relation, the
definition may indeed seem quite appealing. But considering a plurality of
subjects, as we must in political contexts, it is not so plausible. Political
disagreements and disputes arise from conflicts of interest and differences of
opinion. In such cases, at least, any interference is bound to address the
interests and opinions of some and not those of others. Ferejohn illustrates
this point nicely with the question of public funding of abortions.
Whichever policy is adopted, either pro-lifers or pro-choicers will be
dominated. “This is because, from the standpoint of the other group, such
an imposition of law is arbitrary in that surely it cannot track their
interests.”33 Given the degree of pluralism in all modern societies, it is
30
Pettit (2002: 341–2).
31
Pettit (1997a: 36–7).
32
Pettit (1997a: 22).
33
Ferejohn (2001).
Dictionary: NOSD
138 Assaf Sharon
difficult to see how any policy can be attentive to the avowable interests
of every individual to whom it applies “according to their own judgments,” as
Pettit says.34
And it is not just the opinions that differ across individuals. The interests
themselves are often different and at odds with one another. If we take
people as they are, that is, there seem to be many ineliminable conflicts of
interest between individuals of any given society. Perhaps, then, we should
not take people as they are, but as they ought to be. The interests that matter
are not those that actually drive people, but those that should drive them,
or that would drive them in ideal conditions. It is easy to see, however, that
such a maneuver would be subject to our earlier objections. The ideal
interests would have to be the real social interests, the common good.35 And
thus we are back to the moralistic view.
We can now draw a second lesson. If overcoming domination requires
attentiveness to people’s actual consent or opinions, then disagreement
entails persistent and ineliminable domination. And if it is people’s real
interests or their consent in ideal conditions that matter, then moralism
cannot be avoided. To escape this impasse, we require a procedure that in
some sense takes agents as they are, their actual interests and real opinions,
but not necessarily their expressed opinions.
7. Arbitrariness and Control
In more recent work it seems that Pettit is trying to do just that, playing
down the concept of arbitrariness and replacing it with the idea of permission
and the notion of control. Interference, he claims, does not entail domination
“when the interferee is in control of the process,” because domination is
defined in terms of “alien control” and not all control is “alien.”
34
Pettit (1997a: 55), and see Pettit (1997b: 115). Another option is to appeal not to
people’s actual or ideal interests, but to the common avowable interests of society so that
non-arbitrary interferences will be interferences justifiable by reference to these interests
(Larmore 2001). On one interpretation, “common avowable interests” is normatively
constrained (McMahon 2005). But, as McMahon argues, since there is wide reasonable
disagreement about what these interests are and about which policies they entail, this
view suffers from deep indeterminacy. Moreover, it takes us back to the moralistic picture.
This is why Pettit insists that it is “the common interests that citizens are disposed to
avow, not those they ought to avow” that matter (Pettit 2006: 276). But then the
argument in the text kicks in—on practically everything that matters in political debate
there are no readily avowable common interests.
35
Pettit (2002: 344–5) betrays this when he writes: “a rule of law is nonarbitrary to
the extent that those who make the law are forced to track the avowable common
interests—and only the avowable common interests—of those who live under the law.”
Dictionary: NOSD
A may interfere in B’s choice and yet not enjoy such alien control, for A’s interference
may be subject to B’s permission. And in that case A’s actual interference with B will
not detract from B’s freedom. It will not impose A’s will on B’s behaviour, being
ultimately an expression of B’s own will.36
If interference requires your permission, then there is a substantive sense
in which it is subject to your control and is therefore not dominating.
Permission, presumably, can be given without explicit expression, e.g. by
inaction.37 If my interference is subject to your permission and you do not
object to it (although you can), you may be regarded as having given your
permission.
But even if permission could overcome the shortcomings of hypothetical
consent, it hardly escapes the anarchist conclusion. For practically any item
of legislation it is easy to imagine individuals who will actively object. In
fact, given even minimal social heterogeneity, it is hard to come up with a
rudimentary set of laws to which everyone can reasonably be expected to
acquiesce that will also be robust enough to meet the minimal requirements
of basic social order. If what Pettit means, on the other hand, is something
short of this, e.g. that individuals have some degree of influence over the
decision to interfere, then this hardly warrants attributing to them control
and does not offset domination.
Pettit might seem to gloss over this worry by mere equivocation. He says
that “interference [that] is subject to the ultimate control of the interferee—
to the extent that interference is in that sense non-arbitrary—it will
represent a form of self-control, not a form of alien control” and talks about
“an impartial government, operating under the control of the citizens.”38
But to the extent that citizens can be said to “control the formation of the
law,” they do so only as negligibly small elements of a large corporate body.39
Thus there is a shift of subject—the corporate agent controlling the law
and the individual agent subject to it are not identical. Pettit is in fact aware
of this discrepancy, but his attempt to mitigate it is not satisfying:
In order for the State’s coercive laws not to be dominating, it must be the case that
the people collectively control the formation of law . . . An equal share in collective
control will give each member the highest possible level of control over the law,
36
Pettit (2009: 47). See Pettit (2008) and (2012: 58).
Though, tellingly, Pettit’s example is one of explicit consent and in which no rule of
37
law is involved, i.e. the sailors bounding Ulysses upon his order (and see 2012: 152).
38
Pettit (2009: 48).
39
The leap from collective self-rule to individual independence recurs in neorepublican
writing, e.g. “we enjoy complete political liberty when we are dependent only on our own
will—that is, when we live in a self-governing polity that permits us to approve or reject
the rules governing the life of the collectivity” (Viroli 2001: 10). But surely no polity
allows each individual to determine whether rules are approved or rejected.
Dictionary: NOSD
140 Assaf Sharon
consistently with no one being given less than that level. Thus it will give members
a level of control such that no one can complain of being treated in a way that
neglects their will, as dominating overtures neglect their will.40
Putting aside some serious concerns—about the plausibility of the idea of
equal share in control, entailing that ordinary citizens can have the same
level of influence over government decisions as seasoned politicians, poor
people the same leverage as the wealthy, etc.—this argument also rests on a
fallacy. An equal distribution of control may be independently desirable.
But having an equal share in control is not the same as having control. In
fact, equal control is consistent with no control—if no one has control over
some decision, then all have equal share in control; namely, no control.41
“The highest possible level of control” may be too little to mean anything.
Sharing equally with others in the control of legislation does not imply no
neglect of individual will. Assuming my share in control over some decision
is minuscule and that others impose their views on me, albeit through a
legitimate majoritarian procedure, it seems that I have a perfectly legitimate
grievance that my will is being neglected and charges of domination may be
in order.42 Even a perfect system of popular control does not amount to
individual control. So the fact that one shares control equally with others
does not entail that one is not dominated.43
The appeal to agreement, then, in either the consent version or the
permission and control version, is unhelpful to offset domination. If non-
arbitrariness requires consent or permission, it is too strong to be workable,
and if it only requires equal share in control it is too weak to neutralize
40
Pettit (2009: 51–3). See Pettit (2012: 167).
41
An anonymous reader raised the concern that since decisions are made by persons,
it may not be possible for no one to have control over them. It seems to me, however, that
decision by lottery demonstrates the possibility of a procedure over which no one has
control in the relevant respect.
42
Such neglect can be systematic—persistent minorities who lack access to emigration
or secession options, for instance. Still worse—the right to vote might seem insufficient
even for equal control, as people may fail to exercise this right, or, even if they do vote,
where there is an established, persistent minority that never gets its way, exercising the
vote might be considered as ineffective share in control. Similar considerations apply to
democratic contestation. Furthermore, even if non-domination can be identified with
having an equal share in the control over decision-making, this does not necessarily
support a rule of law. Equal share in control can be institutionalized in a variety of ways.
It does not rule out retroactive or particularistic legislation, nor does it entail a requirement
that laws be publicly announced—all basic tenets of the rule of law. It is not at all clear,
then, how the desirability of equal distribution of social control justifies the rule of law.
43
In On The People’s Terms Pettit recognizes that “citizens will not be able to think
of themselves as exercising personal control over the state,” but, he claims, “this is a
by-product of the historical necessity of living in a political society” (2012: 167). But if
this is true then the upshot is that individuals cannot have the desired level of control and
therefore cannot avoid domination.
Dictionary: NOSD
domination. Adding further constraints in terms of the interests and

opinions of the subjects doesn’t help. If their actual manifestation is required,
then the bar is set too high for any reasonable, workable legal order to pass.
If the obtaining of hypothetical, idealized consent is sufficient, then right
reasons have been smuggled back into the picture, leading inevitably to an
undesirable moralistic view. So, and this is our third lesson—there doesn’t
seem to be a reasonable middle ground between the no-reasons definition
of arbitrariness and the right-reasons definition.44 The conceptual challenge
has not been met.
8. Arbitrariness and Law
This takes us to the other side of the equation—in what sense is the rule of
law said to be non-arbitrary? Surely it is not the case that laws are non-
arbitrary as such. Even republicans do not deny the interfering potential
of law.45 Indeed, in his more cautious moments Pettit is careful to specify
that domination is eliminated only by “non-dominating laws,” or “non-
arbitrary rule of law,”46 suggesting that in itself the rule of law can be either
arbitrary or not. His considered opinion seems to be that the rule of law is
a kind of rule that can be non-arbitrary. As he says: “There will be systems
of law available, at least in principle, which are entirely undominating and
entirely consistent with freedom.”47 What republicans seem to suggest is
that the rule of law is a kind of rule that is essentially, or is most likely to
be, non-arbitrary.
It is often suggested that what makes law non-arbitrary is its abstractness
and generality: “the law is a universal and abstract command and as such
protects individuals from the arbitrary will of others.”48 This familiar
thought pervades the rule of law literature.49 The idea is something like this:
laws are abstract, or maximally general, in the sense that they do not single
out specific individuals or groups. They must also be universal, which means
44
Another proposed middle ground is to define non-arbitrariness within a public
reason framework. Some have proposed this as a reading of Pettit (e.g. Lovett 2014),
some as an alternative view, thereby highlighting the insufficiency of Pettit’s stated
position (e.g. Bohman 2008: 199 ff.). Either way, this view is saddled with its own set
of difficulties (see Wall 2002; Raz 1990 and 1998; Enoch 2015). And even if these are
met, this would support not the rule of law, but deliberative democracy or some similar
arrangement.
45
See Pettit (1997a: 65).
46
Pettit (1997a: 48); Pettit (2002: 344, 346).
47
Pettit (1997a: 66); Pettit (2002: 347).
48
Viroli (2001: 52, my emphasis, see also 38–43); Pettit (1997a: 174).
49
E.g. Fuller (1964); Hayek (1978); Raz (2009: 213); Waldron (1989: 80–1).
Dictionary: NOSD
142 Assaf Sharon
they apply equally to the legislators themselves. Together these two

conditions seem to entail that any burdens imposed by an item of legislation
will be shared by the legislators, thereby creating a built-in incentive for
them to refrain from oppressive or abusive legislation. The rule of law is
usually also taken to require that legislation be stable, prospective, and
publicly announced. Republicans regard these as further conditions that
eliminate arbitrary power because:
legislators who can make retrospective laws or laws that apply, like the bill of
attainder, to particular individuals or families will be able to interfere more or less
arbitrarily in people’s lives. And similarly, the administrators or judges who can
choose at will to apply unpromulgated laws, or who can exploit the obscurity or
inconsistency of the law for their own purpose, will represent an arbitrary regime.50
Surely any condition imposed on legislation limits the ways in which
legislators can exercise their authority over others. But the question is
whether there is anything special about the rule of law constraints that
serves to eliminate arbitrariness. Clearly, universality alone will not do. If
laws can discriminate between individuals or groups, then the fact that they
formally bind legislators will be futile. But it is doubtful whether the formal
requirement of generality can remedy this. For one thing, it is not clear that
a workable definition of generality exists. This is due to the fact that laws
must make distinctions between people—minors and adults, sane and
insane, citizen and non-citizen, etc. Merely asserting that “true law should
not name any particulars” or “single out any specific persons or groups”51 is
therefore unhelpful. The restriction must be against irrelevant distinctions
between people, yet it does not seem as if the notion of relevance can be
formally fixed. Moreover, laws can be perfectly general in their formulation
and still discriminate, e.g. laws forbidding homosexual activity, or the
famous French legislation forbidding all wearing of religious head covering
in public schools, which was nevertheless tellingly known as “the Hijab
law.” Furthermore, even fully general, non-discriminatory laws can be
oppressive. Think about laws forbidding foreign travel or expressions of
political dissent. Generality then does not eliminate domination, it only
restricts the forms domination can assume. Presumably this is where the
universality condition kicks in—since the law binds the legislators, they will
refrain from such oppressive measures. But, first, this only solves the latter
problem and will not work until the problems of defining and ensuring real
generality are solved. And second, it is not enough that legislators are
officially equally bound by the law. For the restraining effect to work it has
50
Pettit (1997a: 174).
Hayek (1978: 153).
51
Dictionary: NOSD
to be the case that legislators are effectively subject to the same laws and in
the same way as everyone else, which requires interpretation and enforcement
to be equal as well.
But, beyond these worries about universality and generality as effective
restraints on legislators, their connection to non-arbitrariness is dubious. If
non-arbitrariness has to do with attentiveness to the interests and views of
those affected by a law, it is difficult to see how these features entail it. The
laws of the Taliban, let us assume, were fully general (subjecting men and
women alike to the same harsh rules) and universal (binding the mullahs
just like ordinary people). This hardly made them any less oppressive or
any more attentive to the opinions or interests of their subjects. As Raz says,
“many forms of arbitrary rule are compatible with the rule of law.”52
The formal features of the rule of law, then, do not guarantee the
elimination of arbitrariness in republicans’ own sense. Insofar as tracking
the interests and opinions of those affected is what matters, it is not even
clear that law is the best means available.53 Consent and permission clearly
fare no better. In addition to the failure to articulate an adequate, non-
moralistic definition of arbitrary interference (the conceptual challenge),
therefore, the proposals on offer don’t seem to aptly characterize both the
rule of law and political freedom. The non-equivocation challenge has
also not been met and the reconciliation of law and liberty has not been
achieved.
9. Legislative Constraints and

Popular Control
Let me address two possible reactions to the preceding argument. One

reply might be that laws, as Pettit puts it, are typically “principled” as
opposed to “particularistic,” and constrain legislators as they do not allow
abusive, prejudicial exercise of power.54 But as the concern driving the
attempt to limit political power is founded on the assumption of its
corrupting effects, the constraints set by the formal conditions of the rule
of law provide little remedy. These conditions ultimately limit the ways
in which discrimination and oppression may manifest themselves, but
provide no guarantee against them. Given the manipulability of general
Raz (2009: 219).

52
Perfectionist political theorists would argue that education and cultivation of public
53
virtue can be at least as successful. Early-modern republicans often emphasize the

indispensability of civic virtue for the free state, for this very reason (see Skinner 2004).
54
Pettit (1997a: 174–5). Though Pettit there admits that this doesn’t eliminate
arbitrariness, only makes it more difficult.
Dictionary: NOSD
144 Assaf Sharon
formulations, and the ineliminable voluntarist aspects of interpretation

and enforcement, even principled legislation remains subject to the good
will of legislators and officials.55
It is of course possible simply to define the rule of law as non-
discriminatory application of just laws, but in the present context this is
methodologically confused. The rule of law is supposed to solve a practical,
institutional problem, not a normative quandary. The challenge was
not finding what governments have to do in order to protect the liberty
of their subjects, but of getting them to protect it in spite of their
inclination to encroach upon it. The rule of law was supposed to give us
an institutional arrangement to guarantee freedom, not an ideal of its
realization. What is needed is not the identification of an alternative to
discretion, but a mechanism to eliminate it. (This is not to say that the
rule of law is not a good idea or that it doesn’t contribute to the protection
of liberty. Only that it does not have the conciliatory potential republicans
ascribe to it.)
Another possible response to the argument might be that it’s not the rule
of law as such that eliminates domination, but only its combination with
popular control of legislation.56 The formal features of the rule of law are
conducive to predictability, and popular control can be instrumental for
reducing the likelihood of discrimination (though this would probably
depend on the contingent circumstances). Both are undoubtedly desirable
outcomes, and ones liberals wholeheartedly endorse, but, as the previous
arguments have shown, neither condition, nor a combination of both,
eliminate domination. Predictable interferences can be thoroughly
depressing, and subjecting them to popular control, even if it minimizes
discrimination, does not entail that the individuals subject to them are
“dictating the terms” or imposing their “long-term will on [themselves].”57
As long as one did not choose the constraint to which one is subject, or at
least was in a position to do so, one does not control the decision and may
still be said to be subject to the will of another, or to the collective will of
multiple others.
55
A particularly amusing example mentioned by Susan Okin (1989: 11) in a different
context is the “Supreme Court’s 1976 decision that the exclusion of pregnancy-related
disabilities from employers’ disability insurance plans was ‘not a gender-based
discrimination at all’. In a now-infamous phrase of its majority opinion, the court
explained that such plans did not discriminate against women because the distinction
drawn by such plans was between pregnant women and ‘non-pregnant persons’.” For the
general point see Ely (1980: ch. 4).
56
This is suggested by some passages of Pettit (2012). I thank an anonymous reader for
pressing me on this point.
57
Pettit (2012: 57).
Dictionary: NOSD
10. Arbitrariness and Discretion
The republican view consists of three claims: (1) Arbitrary power constitutes
domination; (2) law is non-arbitrary if it is not discriminatory and not
factional and so tracks the “interests and opinions” of its subjects, i.e. the
common good, and if it is subject to their consent or control; and (3) non-
arbitrary power is (essentially) not dominating. My criticism thus far has
been internal—accepting the idea that arbitrariness constitutes the primary
offense against liberty and questioning the republican claim that the rule of
law can eliminate it. Thus my focus has been on (2), which, I have argued,
is false because republican definitions of arbitrariness are inadequate—the
wedding of the rule of law with the common good only makes sense under
a moralistic definition and a robust enough notion of consent or control is
not available. There seem to be no substantive ways in which arbitrary
interferences are especially pernicious and which are eliminated by the rule
of law. But this raises a deeper question: regardless of the details, why are
arbitrary interferences thought to be particularly problematic in the first
place? Insofar as freedom is at stake, what difference does it make if
interference is arbitrary or not? It is to this question, targeting (1), that I
now turn.
Republicans, as I mentioned, are not alone in thinking that there is
something particularly pernicious about interference when it is arbitrary.
Skinner explains:
You will also be rendered unfree if you merely fall into a condition of political
subjection or dependence, thereby leaving yourself open to the danger of being
forcibly or coercively deprived by your government of your life, liberty or
estates. . . . if you live under any form of government that allows for the exercise
of prerogative or discretionary powers outside the law, you will already be living
as a slave.58
To the extent that we care about independence from the domination of
others, however, it seems to make no difference whether this domination is
imposed through laws or by other means. “Being forcibly or coercively
deprived by your government” may occur either outside or inside the law.
Perhaps the point can be put differently. Domination is often identified
as being subject to the desire or whim of another. That one’s fate is in the
hands of another person and depends on an alien will seems to constitute an
unacceptable form of subjection. Here is a typical claim by Pettit:
Being unfree consists rather in being subject to arbitrary sway: being subject to the
potentially capricious will or the potentially idiosyncratic judgment of another.
58
Skinner (1998: 69–70).
Dictionary: NOSD
146 Assaf Sharon
Here is one from Skinner:

The very fact . . . that your rulers possess such arbitrary powers means that the
continued enjoyment of your civil liberty remains at all times dependent on their
goodwill.59
Whether promoting your interests and views or not, then, the potential for
arbitrary interferences offends against liberty by subjecting one to the will
of another. Non-arbitrary interferences, although limiting choice, do not
have this particular consequence.60 This expresses an idea central not only
to republican writings, but also to many other theories of liberty and the
rule of law—there is something particularly offensive about subjection to
another person. Being subject to the control of another agent obstructs
liberty even if one suffers little or no interference.
Sometimes the claim seems to hinge on the psychologically stifling effects
of subjection to another person. Thus Pettit talks about the subject’s fear to
“speak his or her mind without risk of falling out of favour.”61 But in this
respect there is nothing unique about arbitrary interferences. Presumably,
harsh yet non-arbitrary regulation can be equally disturbing. The idea seems
to be that, in contrast with law, the unstable nature of personal control
breeds self-censorship. Skinner rehearses Sydney’s conclusion that “if you
live under such conditions of dependence, this will serve in itself to limit
what you can say and do as an adviser or minister. You will be constrained
in the first place from saying or doing anything liable to give offence.”62 But
to the extent that this is true it would seem to be an upshot of the force
possessed by the superior, regardless of whether this force is employed
arbitrarily or not. It is not at all clear why it is worse to not be able to speak
your mind because you don’t know what might happen to you than to not
be able to speak your mind because you do.
59
Pettit (1997a: 5); Skinner (1998: 70). See also Pettit (1997b: 115). Frankfurt (1998:
45) argues for the opposite view: “the effect of coercion on its victim in virtue of which
the victim’s autonomy or freedom is undermined, is not essentially due to the fact that he
is subjected to the will of another.”
60
“[T]hough interference-minus-domination restricts choice in the same way that
domination does, it does not impose the will of another in the manner of domination. And
that makes, intuitively, for a powerful contrast between the two modes of choice-restriction”
(Pettit 2002: 350). See also Pettit (2009: 49), and Skinner (2006: 164). Pettit (2012: 40)
stresses that what matters for freedom is whether intervention is voluntary, “depending on
your will,” or “when you have no choice” and intervene “not out of a will.”
61
Pettit (2002: 350). One way in which this idea may be understood is as expressing
an instrumental relation between law and liberty—the possibility of arbitrary interference
induces the voluntary adaptation of one’s actions to the will of another (e.g. by what
Elster (1983: 110) called “adaptive preference formation”), even without actual
interference. It is clear that republicans are after more than such an instrumental
connection (see Carter’s (2008) criticism).
62
Skinner (1998: 92).
Dictionary: NOSD
The thought that seems to underlie republican attitudes is that what

makes arbitrary interferences particularly troublesome is the personal,
voluntaristic agency behind them. Non-arbitrary interferences, on this
view, are less pernicious because they are not intentional.63 If choice is
restricted not by an intentional agent, it does not subject one to the will of
another. Law, in this picture, is the paradigm example of constraints that are
impersonal and do not express a particular will. Hence the favored
republican motto: “a government of laws, not of men.”
This slogan, coined by James Harrington, is terribly misleading, however,
suggesting that laws can rule people. But only people rule people, with or
without laws. Law alone is inert. It is, as Hobbes said, “but words and paper
without the hands and swords of men.”64 So it is not as if there is a choice
between human and non-human rule. Rather, the thought must be that
when the law constrains human rulers, citizens are in some sense subject to
the law and not to those who enact or enforce it, the latter acting as officers
of the law, not as “private” individuals. Two challenges arise at this point.
One is the empirical doubt regarding the reality of such constraint. This
worry was voiced in one form by Hobbes who claimed that only power can
limit power,65 and in another by realists who stress the inevitability and
indeterminacy of interpretation. The argument of Sections 8 and 9 within
this chapter raised some of these worries. But even supposing that law can
effectively curb legislators and executives, we need to ask why subjection to
it is less pernicious than subjection to them. The quick answer suggested
above is because only intentional interferences are restrictive of liberty and
impersonal interferences are not intentional. The problem with this answer
is that the entailment from impersonality to non-intentionality is false.66
Intentionality does not require personality, only agency, and impersonal
entities, like corporations, associations, and states, can be agents.67
The key idea of republicans and others who support the rule of law
because it is not arbitrary is that although law is enacted and implemented
by people, there is a sense in which it is not a mere instrument in the service
63
Pettit (1997b: 114); Skinner (1998: 86); and see Pettit (2012: 44).
64
Hobbes (1651: xlvi, 36).
65
“It is men and arms, not words and promises, that make the force and power of the
laws” (Hobbes 1651: xlvi, 35; see Hobbes 1998: vi, 17–18). Austin (1995: 212) reiterates this
view when he writes “Supreme power limited by positive law, is a flat contradiction in terms.”
66
In fact, I think both premises are false—that impersonal interference is necessarily
non-intentional, and also that non-intentional interferences do not restrict liberty—but
I cannot address the latter here.
67
As Pettit himself recognizes, “While laws may emerge as a result of rivalry between
houses of parliament, or as a precipitate of custom and court interpretation, still they are
by all accounts the achievements of a State. And the State is an agent, albeit of a corporate
kind” (Pettit 2009: 51).
Dictionary: NOSD
148 Assaf Sharon
of their will as a gun or a whip might be. Such instruments, so the claim
would go, are wielded purely at the desire of their wielder, whereas law is the
fixed formulation of the consensus or the general will, obligating those who
employ it equally with all others. In a sense, it is they who are instruments
of the law and not the other way around. Officials, judges, and even
legislators are subjects of the law even as they are formulating and enforcing
it, because it is not their own judgments and desires that determine it, but
rather those codified in the law that determine their actions.68 Surely this is
not a feature of law as such, only of a system of legislation that meets the
requirements of the rule of law. According to this line of thought, arbitrary
power is primarily identified with discretion and the rule of law with its
absence.69 The very existence of discretionary power, even if it is not used to
the detriment of those subject to it, is offensive. As one liberal jurist puts it:
Wherever there is discretion there is room for arbitrariness, and . . . in a republic no
less than under a monarchy discretionary authority on the part of the government
must mean insecurity for legal freedom on the part of its subjects.70
Inasmuch as we care about eradicating the subjection of one person to the
will of another, the unrestrained, discretionary Solomon is worse than the
legalistic Stalin.
68
Pettit’s (2012: 57) telling analogy is “a robot that is programmed to satisfy your
instructions.”
69
See, e.g., Pettit (1997a: 65, 173); Skinner (1998: 70, 74); Skinner (2002: 258);
Skinner (2008: 84); Viroli (2001: 49). In his recent On the People’s Terms Pettit defines
arbitrary interferences as “interference that is exercised at the will or discretion of the
interferer” (2012: 58, 165).
70
Dicey (2009: 188). For similar statements see Scalia (1989) and Hart (1983) (I thank
an anonymous reader for drawing my attention to this interesting source). For other
republican expressions of this idea see Larmore (2001: 84) and Viroli (2001: 49).
Skinner (1998: 70) writes: “[I]f you live under any form of government that allows for
the exercise of prerogative or discretionary powers outside the law, you will already be
living as a slave.” Pettit sometimes puts the problem in terms of status: “domination
will also tend to introduce a characteristic asymmetry of status” (2002: 350); and see
Honohan (2003: 183). This hinges on the claim that dominated people cannot command
the respect of others and are deprived of “full standing of a person among persons.”
Vague as his articulation of this idea may be, it is hard to deny the desirability of equal
respect. But it is not clear why loss of status is a loss of freedom. Status and respect are
without doubt important political goods and both can be jeopardized by vulnerability to
interferences by others. Arguably, some measure of individual liberty may be required for
either or both, but neither status nor respect can be simply identified as a kind of liberty.
Rawls (1999: 440) considers “the social basis of self-respect” a primary good, or even the
most important primary good, but it is both conceptually and politically distinct from
liberty, another primary good. Ferejohn (2001), for example, plausibly suggests that
dependence is more an issue of vulnerability and security, and Geuss (2001: 104) describes
essentially the same state as a problem for participation in public deliberation and
political decision-making.
Dictionary: NOSD
11. The Desirability of Discretion
Earlier I argued that discretion is not eliminated by law, nor is law necessarily
the best instrument for minimizing it. But discretion is not only inevitable,
it is also sometimes desirable. Many forms of social interaction and human
relations—friendship, parenthood, neighborliness, love—are founded on
discretion. Thus the elimination of discretion is not a cogent universal social
ideal. We might even say that under a regime that allows no discretion,
people’s freedom to form certain meaningful relations is undermined.
Republicans will counter that this may be true of personal relations,
whereas their standard of non-domination applies to political institutions.71
The distinction between the personal and the political assumed by this
response is notoriously contentious. While I cannot address the general
issue, note that for republicans, the elimination of domination is a moral
goal resting on the value of individual liberty. Why, then, should it apply to
political institutions and not to other social institutions, where liberty
presumably matters as well?
But regardless of this, there are reasons to want discretion in the political
domain. On the level of officials and bureaucrats, it often makes sense to
leave certain issues to experts. Given the complexities of modern society,
economy, technology, and diplomacy, it is hard to imagine proper
administration of many state functions not through the discretion of
experts.72 Furthermore, there are reasons for discretion arising from the very
features of law extolled by republicans, namely generality and abstractness.
For, as Justice Holmes wrote, “general propositions do not decide concrete
cases.”73 Law is a blunt instrument. Without a reasonable measure of
discretion it is misguided and even dangerous. In the words of republican
hero Algernon Sidney himself:
If it be objected that I am a defender of arbitrary powers, I confess I cannot
comprehend how any society can be established or subsist without them . . . The
difference between good and ill governments is not that those of one sort have an
arbitrary power which the others have not, for they all have it; but that in those which
are well constituted, this power is so placed as it may be beneficial to the people.74
71
Pettit (1997a: 172ff.) actually advocates non-domination as a general social ideal,
applying not only to state power (what he calls “imperium”) but also to non-political
social relations (“dominium”). For criticism of this point see Ferejohn (2001) and
specifically about parenthood see Lovett (2001). Pettit seems to recognize this worry, but
his response is not clear (see 1997a: 268).
72
The indispensability and efficiency of bureaucracy and its discretionary abilities was
most famously noted by Weber (see e.g. 1978: 973).
73
Justice Holmes, Lochner v. New York, 198, US 45, 76 (1905).
74
Quoted from Austin (1995: 228). Pettit recognizes the inevitability, indeed the
necessity of discretion, “provided that the discretion is exercised under constraints that
Dictionary: NOSD
150 Assaf Sharon
This observation brings into conflict the two republican definitions of

arbitrariness. As Alan Ryan put it, echoing an observation noted by many
others whether neutrally like Constant or critically like Hayek, “welfare states
confer large discretionary powers upon their politicians and bureaucrats.”75
Since promoting the common good often requires discretion, non-
arbitrariness in one republican sense requires arbitrariness in the other.
But besides consistency, the worry is that identifying the kind of
unfreedom that is politically significant with arbitrariness and specifically
with discretion risks depriving us of much that is significant in both
our personal and our political lives. Eliminating arbitrariness can have
undesirable consequences and may even be regarded as restricting liberty—
the liberty to pursue discretionary activities and relations.
12. Non-Arbitrary Domination
Republicans can retreat to the weaker claim that even if eliminating

discretion comes at a cost, at least it eradicates domination. This can matter
greatly assuming that, though not well defined by republicans, the main
forms of interference that are politically repugnant are those associated with
arbitrary power—personal, discretionary control. This is the third republican
claim. If indeed domination is to be identified with arbitrary interference,
then the ideal, though not perfect, is fundamentally sound.
Is non-personal, non-discretionary control essentially non-dominating?
Is the discretionary Solomon more dominating than the legalistic Stalin?
The republican unflinching positive response is rooted in a specific
philosophical picture according to which the antithesis of liberty is slavery
or monarchical rule. Republicans repeatedly employ the example of the
slave serving at his master’s pleasure as the paradigm of unfreedom.76 This
picture, however, suffers from what Wittgenstein called a one-sided diet of
examples.77
Looking at the most salient contemporary examples of domination,
we find that they often do not involve arbitrariness. In fact, the kind of
control they exhibit is frequently characterized by markedly non-arbitrary
interference. Two clear examples are prisoners and soldiers. Both are obvious
cases of unfreedom and both are subject to an abundance of orders, dictates,
help guard against arbitrariness” (1997a: 175). But then arbitrariness cannot be identified
with discretion.
75
Ryan (1993: 294).
76
See Pettit (1997b: 115); Pettit (1997a: 22, 31–3); Skinner (1998: 39ff.); Skinner
(2008); Viroli (2001: 8–10); Richardson (2002: 29).
77
Wittgenstein (1953: §593).
Dictionary: NOSD
and regulations which serve precisely to limit their freedom. Yet neither
soldiers nor prisoners serve at the pleasure of another individual and the
orders to which they are subject are typically neither whimsical nor personal.
Though it can surely teach us much about liberty, the interpersonal relations
between slave and master do not reveal the whole truth about political
freedom. Focusing solely on this paradigm can actually obscure other,
possibly worse forms of domination.
Reflecting on a broader range of examples, there are reasons for thinking
that law is particularly ill-suited for eliminating certain forms of domination.
Various research programs studying political aspects of group identities,
particularly gender and race, have shown how domination can be
embedded in apparently non-arbitrary systems of social control. Modern
family structure is seen by many feminists as creating a form of domination
of men over women, or sanctioning the domination of children by their
parents. Perhaps most conspicuous—unrestrained capitalism can involve
the domination of workers by their employers and of local producers by
international corporations. In these contexts, non-arbitrariness understood
as the absence of personal control or discretion does not entail independence.
In modern society individuals are dependent on corporate, economic, and
social institutions in a variety of ways, many of which are impersonal
and non-arbitrary. Whether there is domination or not seems to be
primarily a function of discrepancies in the distribution of power, not
the arbitrariness of its employment. Republicans have not given us reasons
to regard personal, discretionary powers as the only, or the primary, threat
to freedom, nor have they explained why political institutions should
address such forms of unfreedom rather than other kinds of control and
coercive intervention.
In fact, once it is recognized that domination takes various forms and
exists in many domains, it becomes apparent that its eradication requires
a powerful state. But this would make government not only extremely
intrusive, but also potentially interfering and hence all the more dominating.
It will require perpetual increase in legislation, which, in turn, expands the
space for arbitrariness. For, as Constant observed:
it is a mistake to hope the proliferation of laws will save us from the tyranny of men.
In multiplying laws you necessarily create more government agents. Consequently
you give a larger number of men power over their fellows and thus double the
likelihood of its arbitrary misuse. This is because however precise these laws, there is
always the possibility of arbitrariness, if only in the more or less severe exactness with
which they are carried out.78
78
Constant (2003: 65).
Dictionary: NOSD
152 Assaf Sharon
In light of these considerations, the abstract, general, and impersonal nature

of law appears as a double-edged sword. While it may provide remedy for
certain kinds of personal domination, particularly those associated with
monarchy and slavery, it is no panacea. Given other contexts and different
social and personal interests, these very features can become not an antidote,
but an accelerant of domination. This, then, is our fourth and final lesson—
the republican claim that the rule of law is not only in general a good idea,
but that it is the answer to the problem of authority and liberty is specious.
The republican defense of the rule of law, identifying unfreedom with
arbitrariness, displays the legalistic tendency discerned by Tocqueville and
later by Weber: “a legal caste, once it had established the ‘rule of law’ securely
against threats from absolutist arbitrariness, was bound to prefer order to
liberty.”79 It is based on a general preference for non-personal mechanisms
rooted in a bias against human judgment and discretion. Overlooking the
dominating potential of such mechanisms exposes them to Tocqueville’s
condemnation of lawyers:
If they prize freedom much, they generally value legality still more; they are less
afraid of tyranny than of arbitrary power.80
Bibliography
Austin, John (1995). The Province of Jurisprudence Determined and the Uses of the
Study of Jurisprudence (Cambridge: Cambridge University Press).
Bentham, Jeremy (2010). “Anarchical Fallacies,” in J. Bowring (ed.), Works of Jeremy
Bentham, Vol. 2 (Charleston, SC: Nabu Press).
Berlin, Isaiah (2002). “Two Concepts of Liberty,” in Four Essays on Liberty (Oxford:
Bohman, James (2008). “Nondomination and Transnational Democracy,” in
C. Laborde and J. Maynor (eds), Republicanism and Political Theory (Oxford:
Blackwell).
Boureau, Alain (1998). The Lord’s First Night: The Myth of the Droit de Cuissage
(Chicago, IL: University of Chicago Press).
Carter, Ian (2008). “How are Power and Unfreedom Related?” in C. Laborde and
J. Maynor (eds), Republicanism and Political Theory (Oxford: Blackwell).
Cohen, G. A. (1995). Self-Ownership, Freedom, and Equality, (Cambridge:
Cambridge University Press).
Cohen, G. A. (2011). On the Currency of Egalitarian Justice and Other Essays in
Political Philosophy (Princeton, NJ: Princeton University Press).
See Shklar (1964: 16).

79
Tocqueville (1969: 266).

80
Dictionary: NOSD
Constant, Benjamin (2003). Principles of Politics Applicable to all Governments

(Indianapolis, IN: Liberty Fund Press).
Dicey, A. V. (2009). Introduction to the Study of the Law of the Constitution (London:
Lightning Source Inc.).
Dworkin, Ronald (1977). Taking Rights Seriously (Cambridge, MA: Harvard
University Press).
Elster, Jon (1983). Sour Grapes: Studies in the Subversion of Rationality (Cambridge:
Ely, J. H. (1980). Democracy and Distrust: A Theory of Judicial Review (Cambridge,
MA: Harvard University Press).
Enoch, David (2015). “Against Public Reason,” in D. Sobel, P. Vallentyne, and
S. Wall (eds), Oxford Studies in Political Philosophy, Vol. 1 (Oxford: Oxford
University Press).
Ferejohn, John (2001). “Pettit’s Republic,” The Monist 84, no. 1, 77–96.
Frankfurt, Harry (1998). The Importance of What We Care About (New York:
Fuller, Lon (1964). The Morality of Law (New Haven, CT: Yale University Press).
Geuss, Raymond (2001). History and Illusion in Politics (Cambridge: Cambridge
University Press).
Halldenius, Lena (2003). “Locke and the Non-Arbitrary,” European Journal of
Political Theory, 2:3, 261–79.
Hart, H. L. A. (1983). Essays in Jurisprudence and Philosophy (Oxford: Oxford
University Press).
Hayek, F. A. (1978). The Constitution of Liberty (Chicago, IL: University of Chicago Press).
Hobbes, Thomas (1994). Leviathan: With Selected Variants from the Latin Edition of
1668 (Indianapolis, IN: Hackett).
Hobbes, Thomas (1998). On the Citizen (Cambridge: Cambridge University Press).
Honohan, Iseult (2003). Civic Republicanism (London: Routledge).
Hume, David (1994). “On the Original Contract,” in Political Essays (Cambridge:
Kramer, Matthew H. (2008). “Liberty and Domination,” in C. Laborde and
J. Maynor (eds), Republicanism and Political Theory (Oxford: Blackwell).
Larmore, Charles (2001). “A Critique of Philip Pettit’s Republicanism,” Philosophical
Issues, 11, 229–43.
Larmore, Charles (2003). “Liberal and Republican Conceptions of Freedom,”
Critical Review of International Social and Political Philosophy 6:1, 96–111.
Locke, John (1988). Two Treatises of Government, 3rd ed. (Cambridge: Cambridge
University Press).
Locke, John (1979). An Essay Concerning Human Understanding (Oxford: Oxford
University Press).
Lovett, Frank (2001). “Domination: A Preliminary Analysis,” The Monist 84,
98–112.
Lovett, Frank (2014). “Republicanism,” in E. Zalta (ed.), The Stanford Encyclopedia
of Philosophy.
McMahon, Christopher (2005). “The Indeterminacy of Republican Policy,”
Philosophy and Public Affairs 33:1, 67–93.
Dictionary: NOSD
154 Assaf Sharon
Miller, David (1983). “Constraints on Freedom,” Ethics 94:1, 66–86.

Morgenbesser, S. and Ullmann-Margalit, E. (1977). “Picking and Choosing,” Social
Research 44, 757–85.
Okin, Susan (1989). Justice, Gender and the Family (New York: Basic Books).
Paine, Thomas (1997). Common Sense (Mineola, NY: Dover Publications).
Pettit, Phillip (1996). “Freedom as Antipower,” Ethics 106:3, 576–604.
Pettit, Phillip (1997a). Republicanism: A Theory of Freedom and Government (Oxford:
Pettit, Phillip (1997b). “Republican Political Theory,” in A. Vincent (ed.), Political
Theory: Tradition, Diversity and Ideology (Cambridge: Cambridge University
Press), 112–31.
Pettit, Phillip (2002). “Keeping Republican Freedom Simple: On a Difference with
Quentin Skinner,” Political Theory 30:3, 339–56.
Pettit, Phillip (2006). “The Determinacy of Republican Policy: A reply to
McMahon,” Philosophy and Public Affairs 34:4, 275–83.
Pettit, Phillip (2008). “Republican Freedom: Three Axioms, Four Theorems,” in
C. Laborde and J. Maynor (eds), Republicanism and Political Theory (Oxford:
Blackwell).
Pettit, Phillip (2009). “Law and Liberty,” in S. Besson and J. L. Marti (eds), Legal
Republicanism (Oxford: Oxford University Press).
Pettit, Phillip (2012). On the People’s Terms: a Republican Theory and Model of
Democracy (Cambridge: Cambridge University Press).
Rawls, John (1999). A Theory of Justice, rev. ed. (Cambridge, MA: Belknap Harvard
University Press).
Raz, Joseph (1990). “Facing Diversity: The Case of Epistemic Abstinence,” Philosophy
and Public Affairs 19:1, 3–46.
Raz, Joseph (1998). “Disagreement in Politics,” American Journal of Jurisprudence
43, 25–52.
Raz, Joseph (2009). The Authority of Law: Essays on Law and Morality, 2nd ed.
Richardson, Henry (2002). Democratic Autonomy: Public Reasoning About the Ends
of Policy (Oxford: Oxford University Press).
Ryan, Alan (1993). “Liberalism,” in R. E. Goodin and P. Pettit (eds), A Companion
to Political Philosophy (Oxford: Blackwell).
Scalia, Antonin (1989). “The Rule of Law as a Law of Rules,” University of Chicago
Law Review 56:4, 1175–88.
Shklar, Judith (1964). Legalism: Law, Morals, and Political Trials (Cambridge, MA:
Skinner, Quentin (1998). Liberty Before Liberalism (Cambridge: Cambridge
University Press).
Skinner, Quentin (2002). “A Third Concept of Liberty,” Proceedings of the British
Academy 117, 237–68.
Skinner, Quentin (2004). Visions of Politics, Vol. II (Cambridge: Cambridge
University Press).
Dictionary: NOSD
Skinner, Quentin (2006). “Rethinking Political Liberty,” History Workshop Journal

61, 156–70.
Skinner, Quentin (2008). “Freedom as the Absence of Arbitrary Power,” in
C. Laborde and J. W. Maynor (eds), Republicanism and Political Theory (Oxford:
Blackwell).
Tocqueville, Alexis de (1969). Democracy in America, G. Lawrence (trans.) (New
York: Anchor).
Viroli, Maurizio (2001). Republicanism (New York: Hill and Wang).
Waldron, Jeremy (1989). “The Rule of Law in Contemporary Liberal Theory,” Ratio
Juris 2:1, 80–1.
Wall, Steven (2002). “Is Public Justification Self-Defeating?” American Philosophical
Quarterly 39:4, 385–94.
Weber, Max (1978). Economy and Society (Berkeley, CA: University of California
Press).
Wittgenstein, Ludwig (1953). Philosophical Investigations, G. E. M. Anscombe
(trans.) (Oxford: Blackwell).
Dictionary: NOSD
6
Elitism
Richard Arneson
Elitism as a political doctrine can take many forms. This chapter considers
the claim that those who know the political truths that determine what
choices of laws and other public policies are correct, and they alone, are
entitled to a share of political rule. A related view is that those whose
political knowledge and judgment are superior to that of others are
presumptively entitled to be political rulers of society, the presumption
being overcome if they are corrupt or otherwise disqualified by political
faults outweighing their political qualifications.
How might a principled advocate of political democracy respond to this
elitist claim? One possible response is epistemic. If knowledge relevant for
political decisions is dispersed widely across the population, then decision
procedures such as majority rule with a single equal vote for each person
might elicit the dispersed information needed for sound public policy
choice and so generate better political decisions than would be generated by
non-democratic decision procedures that confine political power to the
more knowledgeable. This might be a fine response, but in this chapter I set
it aside. (In passing, note that it is not clear why a knowledgeable elite
concerned to base decisions on all relevant information could not gather the
dispersed bits of knowledge possessed by the less knowledgeable by opinion
polls or other social science techniques of information collection.) The
assumption framing this inquiry is that there might be an identifiable group
of political experts, and that confining political power to this group might
reliably bring about better laws and public policies than would otherwise be
attainable.
Another possible response appeals to doubts about feasibility. This says
that in practice no political constitution we could devise and implement in
current conditions could succeed in restricting political power to a select
knowledgeable group that would predictably rule better than political
leaders chosen by ordinary democratic processes. Either the screen that
Dictionary: NOSD
Elitism 157
selects the members of the elite will be inaccurate in selecting the genuine
experts, or the selected group of experts will overrepresent certain segments
of the population (women over men, whites over blacks, rural residents over
residents of cities) in ways that will predictably lead to political decisions
biased in the interests of the overrepresented groups—women, whites, and
rural folk, for example.
Again, skepticism about feasibility might be an adequate reason for
disfavoring any elitist political proposals here and now. However, I suppose
that principled democrats will find it unsatisfactory. The argument appeals to
contingent facts that might in some circumstances fail to hold, and one might
hope to find a more principled and less contingent support for the moral right
of each adult person in any modern society to an equal democratic say.1
Section 1 of this chapter introduces the topic and rebuts two arguments
against the claim that political expertise might entitle one to political rule.
Section 2 presents an argument against the position that each person has a
non-instrumental right to a democratic say and considers some objections.
Section 3 distinguishes two versions of elitism, the claim that the experts
should rule. One appeals to a natural aristocracy ideal, which this chapter
does not endorse. The other says we should put in place whatever form of
government would deliver best results and denies that democracy would
deliver best results. It is noted that even if democracy delivers worse laws
and public policies than some feasible non-democratic alternatives,
democracy might yet bring about better results overall by indirect means.
Section 4 considers and rejects another possible justification of rule by
experts—namely the idea that each of us has a right not to be imposed on
coercively by incompetent persons. Section 5 defends the idea that the
political truths relevant to political decision-making include moral truths
and that there might be moral experts who have better knowledge of these
moral truths than others. Section 6 argues that Rawlsian political liberalism
does not point toward a sound argument that political rule by experts is
morally unacceptable. Section 7 is a brief inconclusive conclusion.
1
I assume that if someone has a moral right to do X, then she is morally at liberty to
do X and others have duties not to interfere with her doing X (the nature of the duties of
non-interference varying with the type of right in question). Having a moral right is
compatible with the right being overridden by other considerations in specific
circumstances so that in those circumstances, all things considered, one is not morally at
liberty to do X and others are not duty bound to refrain from interfering. The objection
to instrumentalist justifications for democracy is not just that they are contingent on
some empirical facts, but that these contingencies do not reflect our convictions as to
when democracy would and would not be justified. Those who regard democracy as an
intrinsically fair procedure and justifiable in whole or in part in virtue of its fairness will
press this objection.
Dictionary: NOSD
158 Richard Arneson
1. CLAIMS OF EXPERTISE VERSUS THE RIGHT

TO A DEMOCRATIC SAY
Suppose that there is a group of A type people in society alongside a group

of B type people. The A people believe, and let us assume they are right to
believe and have epistemic warrant for believing, that on the important
matters that the government is called on to decide, they know better than
the B type people what the government ought to do, what public policies
should be established. In this situation, does superior knowledge confer any
entitlement to rule?
Surely, the answer is not obviously “Yes.” There is a gap here that needs to
be filled in by argument, and perhaps there is no sound argument that fills
the gap. On the face of it, the statement that some identified members of
society have greater political knowledge than others is consistent with the
further statement that these knowledgeable individuals have no special
entitlement to rule. The claim that knowledge confers any sort of moral
entitlement to exercise political coercion elicits a skeptical response from
David Estlund. He calls the argument from possession of knowledge to
legitimacy of rule the expert/boss fallacy. As he puts the point: “You may be
the expert, but who made you the boss?” Along the same line he observes
that even if the Roman Catholic pope has a pipeline to God, that would not
give the pope the moral right to make Roman Catholicism the established
religion and use state power in other ways to favor this particular religious
doctrine over others including Marxism and atheism (Estlund 2008). Many
political philosophers agree.
Philip Pettit takes a stronger stance (Pettit 2015). He argues that
political justice should take priority over social justice. By that he means
that when there is disagreement about social justice—about what the
content of public policies and the shape of basic institutions should
be—we should all agree that the decision about what should be done in
the face of disagreement should be made according to a fair political
process. Of course there will likely be disagreement among citizens
on the issue, over what qualifies as a fair political process, but then this
disagreement, too, must be resolved by a fair and mutually agreeable
higher-order political process. Moreover, once we notice that theories of
social justice are addressed as recommendations to citizens, we must
acknowledge that the theories must presuppose that those addressed are
competent to understand, assess, and decide on the recommendations.
So if we scratch the surface we see that implicit in any sensible theory of
social justice is the qualification that the substantive proposals being
recommended should be implemented only by a fair political process
and that a fair political process must be a democratic process that treats
Dictionary: NOSD
Elitism 159
all citizens as equals and thus gives equal political power by way of an
equal franchise to all.
The relationship of Pettit’s claim to the suggestion that those with
knowledge of the political truths that are required to identify just laws are
entitled to a greater share of political rule than others is that if Pettit is
correct, the suggestion must be wrong. If democracy requires that all are
entitled have an equal share of political power (an equal democratic say),
then an elite in possession of political truths is not entitled to a greater share
of political rule than others.
Estlund is right. From “I know better than you what you should do,” it
does not follow that “I am entitled to decide what you should do and to
bring it about, by threat of force if need be, that you conform in your
conduct to my decision.” Nor is there any readily identifiable plausible
premise that one could insert into this bad argument to render it sound.
One simple explanation of the difficulty we face here is that perhaps it is
false that possession of knowledge about what another person should do
confers a right to control the conduct of another so that it conforms to what
we know it is right for that person to do.
For purposes of this chapter I shall assume that there are moral constraints
on what we may do to advance our ends, even moral ends. So the facts that
I know that B should do X and that the state of affairs that will be brought
about if I force B to do X would be better than the state of affairs that would
result if I refrain from this forcing and indeed better than if I do anything
else instead do not suffice to show that it is permissible for me to force B to
do X. For a start, consider that B’s action may bring about negative
consequences only for herself and others who voluntarily consent to be
involved with her, and perhaps, at least up to a point, we should respect
people’s freedom to live as they choose given that the interests of non-
consenting others are not adversely affected.
Besides being right, Estlund is also wrong. Suppose the pope really does
have a pipeline to God, hence warranted true beliefs of the utmost
importance for all of us. Suppose that each human person can attain eternal
salvation but only if she lives according to the dictates of the Roman
Catholic Church. In this case the pope is surely entitled, and probably
morally required, to coerce the rest of us for our own good, if he happens to
have sufficient military force at his disposal. Too much would be at stake.
What blocks the pope’s entitlements is that, in fact, there is no reason to
believe he has such a pipeline or in other words warranted true beliefs to the
effect that outside the Church there is no salvation.
In a similar way, the argument that political justice demands democracy
and takes priority over social justice falls flat. The idea that the fact of
disagreement about substantive matters of social justice forces us to accept
Dictionary: NOSD
160 Richard Arneson
that we must subordinate our campaigns for social justice to a fair procedure
and that the fair procedure is democracy runs afoul of a point that Pettit
notices: we disagree about which political procedures would be fair.
Ascending to the claim that we need to accept a fair procedure for settling
on first-order fair procedures does not solve the problem, it just initiates
a regress.
Setting that point aside, I submit that it would be making a fetish of fair
procedures to insist that one’s social justice recommendations must always
be advanced subject to the condition that they should be implemented only
if they gain majority vote in fair democratic elections. Consider social
justice recommendations to the effect that we should desist from racist
discrimination in economic life, the oppression of women involving failure
to protect them against sexual violence, and policies of genocide directed at
unpopular ethnic minorities. It would make more sense to flip the priority
and insist that fair political procedures are whatever procedures would best
promote the fulfillment of individual moral rights weighted by their
importance (rights other than the supposed right to a democratic say). But
this issue is somewhat delicate. One might think that among the rights we
possess is a right not to be subject to non-democratic government, so the
formulation just given begs the question against someone who holds that
our basic moral rights include democratic citizenship rights. But this claim
is a far cry from a claim of priority for political rights over all other rights.
Let’s assume at least provisionally that among our moral rights is a right
against being subjected to non-democratic political government. Since this
would be one moral right among many, this assumption still allows the
possibility of conflict and trade-off. Under some conditions fulfillment of
one’s right to a democratic say might be inimical to fulfillment of important
moral rights of other people, so that all things considered, one’s right to
democracy is outweighed by the combined weight of the social justice
rights of others that would be placed in jeopardy by respecting one’s
political right.
If a group of people in a country possesses superior knowledge of the
truths that legislators must know in order to identify the laws and public
policies the establishment of which would best promote human rights
fulfillment, they may also be in a position to know that if they alone control
the legislature, the laws and public policies chosen and implemented will be
morally better than they would be if the legislature were controlled by
majority vote among all adult citizens. If the gap between the moral rights
fulfillment that elite rule would induce and moral rights fulfillment that
democratic rule would induce is sufficiently large, and favors elite rule, then
elite rule is morally preferable even though it involves violation of people’s
rights to a democratic say.
Dictionary: NOSD
Elitism 161
Return to Pettit’s claim that when advocates of social justice theories

address others and present arguments to others in support of their favored
theories, the advocates have to be presupposing that the people they are
addressing are competent, reasonable agents, capable of understanding
assertions and assessing arguments. From this presupposition Pettit thinks
one can tease out the implication that one cannot consistently address other
people and propose social justice theories to them without acknowledging
that the others are competent to rule on the merits of the proposals and
arguments one is making. From there Pettit thinks that with a short
reasonable hop one can conclude that social justice proposals ought to be
instituted across a community of agents only via a fair political procedure in
which all competent adult community members have equal participation
rights.
One problem with this ingenious line of thought is that its initial
presupposition is false. In addressing an agent and presenting arguments
to that agent in favor of some normative conclusion one is not thereby
committing oneself to acknowledging the agent one is addressing as a
fully competent interlocutor in the discussion one is trying to initiate.
Examples come readily to hand. Suppose one has been abducted by pirates
and is about to be sold into slavery, or robbed and then killed. Given
a chance, one proceeds to proclaim to the pirates that what they are
planning to do is morally wrong, and provides arguments in support of
this claim. One’s speech will be pointless if there is no chance one’s words
will be understood by one’s intended audience, but even a chance that
some of the pirates might understand the language in which one is
speaking can give point to one’s making the attempt to be heard. Moreover,
one might have good reason to address the pirates as described even if one
reasonably believes there is zero chance that they will accept one’s
arguments and no chance that even if they did accept the arguments, they
would be motivated to desist from their piratical plans. Nor does
addressing a person and making an argument to that person necessarily
carry the implication that one believes the person addressed is fully
rational and competent and that her opinions and judgments should have
the same epistemic weight as one’s own in the joint determination of what
should be done. One might with good reasoning believe that a life of
piracy causes the moral theory mental faculties to rust, or that the fact
that one has chosen a life of piracy is some evidence that one is less than
ideally competent at moral reasoning, and so on. And whatever negative
judgments one might make about the judgmental competence of the
pirates who are attacking, one can find a parallel in negative judgments
that a member of political society might make about other members of his
political society, especially if they show dull-mindedness, or oddball or
Dictionary: NOSD
162 Richard Arneson
corrupt normative inclinations, or disinterest in questions concerning

what we owe to one another. Of course, these confident judgments of
one’s epistemic superiority might in any given case be false or even
delusional. But they might also be fully correct.
2. AN ARGUMENT AGAINST THE RIGHT

TO A DEMOCRATIC SAY
Section 1 argued that even if there is a right to a democratic say, it might

give way to claims of expertise when the consequences of upholding the
right to a democratic say would be sufficiently bad. This section presents an
argument against the existence of a right to a democratic say and considers
some (inconclusive) reasons to reject the argument (see also Arneson 2009).
Consider people’s fundamental moral rights. Each of us has a duty to
respect and promote these. The duty to promote might not be maximally
stringent; each of us might also have a Scheffler prerogative to live as we
choose, within limits. But arguably the standing duty to rescue others from
suffering violations of their fundamental rights is a duty, within one’s
prerogative, to act effectively to this end.2 Here’s a comparison: in a situation
of disaster in which we all have a duty to pitch in and rescue people from
peril, the duty to rescue might not require heroic measures but does require
that we maximize the saving of people from peril from the sacrifices we have
a moral duty to make. If in a disaster situation someone takes charge and
issues orders, and we can see that this person has expertise and that following
her lead will bring about greater rationally expected saving of people from
peril, we are not at liberty to refuse her commands, and if she sizes up the
situation correctly, she is definitely at liberty and perhaps under a duty to
coerce the recalcitrant and bring about a rationally coordinated rescue
effort. The case of members of a community claiming a right to a democratic
say might be like that of individuals in the set of those who have duties to
rescue protesting to the persons who take charge, “You may be expert, but
who made you boss?” The protest is unjustified.
Here I would appeal to other cases in which one has a moral duty and
must choose a means to fulfill it. One is obligated to fulfill the duty if one
can, not merely make a try at fulfilling it. One is thus under a duty to find
2
The claim in the text here might be wrong. Perhaps one’s duty is to do enough,
achieve a certain amount of good, and whether one does it efficiently or inefficiently is
not a matter of concern. In the plane crash rescue example, perhaps it is acceptable to
decline the orders of those who reasonably take charge provided that by one’s own efforts
one does enough to advance the rescue effort and save lives.
Dictionary: NOSD
Elitism 163
out what means would be best suited to the task, and adopt them, unless
other moral considerations intrude or one is willing to pursue an inefficient
means at greater cost to oneself but will definitely fulfill the duty by one’s
chosen means. When moral rights of others are at stake, it is wrong to insist
on following one’s own lights when there are good and sufficient reasons to
believe that deferring to the judgment of another person will be more likely
to bring about fulfillment of the duty at acceptable cost. Nor would it be
right to resist coercion by another that will bring about greater fulfillment
of the moral rights at stake, so long as the harms of coercion are not
disproportionate to the matter at hand. The same goes when the issue is a
matter of what political regime to establish, I would say. Here is a challenge
to the idea that each of us has a moral right to a democratic say.
As I am using the term, what is at issue is the existence of a non-
instrumental right to a democratic say—a right to be governed by a political
system in which legislators and top public officials are selected by majority
rule in elections with an equal franchise open to all permanent adult
members of society, against a backdrop of freedom of speech and of
association. In contrast, an instrumental right to a democratic say would be
possessed by people just in case according them that right would bring
about better outcomes impartially assessed than not according it.
The argument against a non-instrumental right to a democratic say can
be stated in this way:
1. Each of us has a duty to contribute toward sustaining a political system
that protects people’s moral rights and advances the general welfare.
2. This duty to contribute toward sustaining is a duty to act in ways that
increase, rather than decrease, the extent to which the system functions
well over time to protect people’s rights and advance the general welfare.
3. If one’s claiming and exercising a share of political power would
decrease, rather than increase, the degree to which the system functions
well over time to protect people’s rights and advance the general welfare,
it would be wrong for one to claim and exercise a share of political
power, and wrong for others to press these claims to political power on
one’s behalf.
4. If there were a non-instrumental, intrinsic moral right to a democratic
say, it would not be wrong for one to make the claims for a share of
political power described in premise 3, and it would not be wrong for
others to press these claims on one’s behalf.
5. None of us has a non-instrumental, intrinsic moral right to a democratic
say.
There is a certain looseness in the argument just stated that reveals itself
when we ask for clarification of the duty “to contribute toward sustaining a
Dictionary: NOSD
164 Richard Arneson
political system” that carries out the moral functions proper to states. Is this
duty inherently a duty to support a state in the territory one inhabits, or
rather a duty to contribute to good governance everywhere, which might be
discharged by helping promote good governance abroad rather than where
one lives? This is the problem that in recent literature is called the problem
of “particularity.” I take no stand on this issue and assume that the arguments
advanced in this chapter do not depend on answering it one way rather
than another.
As stated, the argument seems to beg the question against one who
affirms that there is a non-instrumental, intrinsic right to a democratic say.
Premise 2 says one has a duty to help to promote fulfillment of people’s
moral rights, but the right to a democratic say might be among them. If so,
then one has a right to a democratic say, and the duty to promote people’s
rights will be a duty to promote a set of rights that includes the right to a
democratic say. Sometimes to promote rights fulfillment overall one might
be required to sacrifice one right held by some to protect rights held by
others, but this could be true for any moral right, so the claim that the
right to a democratic say (like any right) might be outweighed in some
circumstances by the balance of rights considerations provides no reason at
all to doubt that there is a moral right to a democratic say. Moreover, the
moral rights one has might limit the actions one must take to promote
rights fulfillment, and so actions taken by others to promote overall rights
fulfillment that violate one’s own right to a democratic say might be
impermissible.
In reply: as I understand the argument, it proceeds on the understanding
that initially we are not sure exactly what moral rights people have. We
suppose people have moral rights and that we should both respect them
and promote their fulfillment. We have some incomplete views about the
content of moral rights—for example, that innocent non-threatening
people have a right not to be targeted for physical attack. In the same way,
we believe that we have some duty to advance the general welfare, but we do
not possess a full understanding of what constitutes the general welfare.
In this epistemic context, premise 3 asserts that if we have a serious duty
to establish a system that protects people’s rights, we have a duty to
subordinate whatever interest we might have in having power within that
system to the duty to contribute to making the system effective, given the
moral importance of its goal. In contrast, if premise 3 had stated, if one’s
insistence on defending oneself against predatory attacks by others would
decrease, rather than increase the degree to which the system functions well
over time to protect people’s rights and advance the general welfare, it would
be wrong for one to insist on defending oneself against predatory attack,
and wrong for others to support one’s insistence, or assist in one’s defensive
Dictionary: NOSD
Elitism 165
effort, premise 3 would be much less plausible. Maybe there is some right
to defend oneself against wrongful attack even when doing so would hinder
the overall degree to which people’s rights are fulfilled. In a similar way, in
the rescue example, if it were the case that the rescue effort would be most
effectively advanced if one allowed oneself to be killed, it would not be
implausible to deny that therefore it would be wrong for one to act in
certain ways to protect oneself against being killed even to the detriment
of the successful rescue.
If we accept premise 3, we are accepting a partial specification of what
are, in actual fact, people’s moral rights. The claim is simply that politics
and forms of governance are really just the rescue scenario writ large.
Anarchy is a disaster, from the standpoint of (any credible view of )
fulfillment of people’s moral rights. We have a duty to contribute to escaping
from anarchy and to preventing its return. The duty here is to act with
others and form institutions with others in ways that are effective (without
excessively running a steamroller over people’s uncontroversial, basic moral
rights). The claim to a say in how the process is run or in how the institutions
will function borders on self-indulgence, if regarded as a claim to a non-
instrumental, intrinsic moral right. Regarded as an instrumental claim,
taking democracy to be in many circumstances a large boost to the
effectiveness of government regarded as a mechanism for delivering rights
fulfillment, the claim to a right to a democratic say is plausible and might
be true.
One possible ground for disputing the argument just stated is to hold
that each person is morally at liberty to act to advance her own interests
even at some cost to the interests of others, and others, to some extent, have
duties not to interfere with the individual’s permissible pursuit of her own
advantage. So one might have a moral right to have and exercise a democratic
say in political governance, even at cost to others, using one’s democratic say
to advance one’s own interests (e.g. voting for a change in tax rates
advantageous to oneself but detrimental to the general welfare). This claim
denies premises 2 and 3. Note that everyone’s acting in this way might be
collectively self-defeating overall. Also, one might deny that the prerogative
to favor oneself extends to acting to try to bring it about that coercive state
power is deployed to one’s advantage, in the absence of gain that would
register in impartial assessment.
Another possible ground for disputing the argument appeals to the
deontological distinctions between intended and merely foreseen
consequences and between doing, allowing, and enabling. When we act to
institute or sustain a democratic political system, we aim to enable each
member of society to participate in collective decision-making on equal
terms. We may foresee but do not intend that some will misuse or abuse
Dictionary: NOSD
166 Richard Arneson
their democratic say so that people’s rights are less fulfilled and the general
welfare suffers compared to what would have occurred had they lacked
participation rights. It may in these circumstances be permissible to bring
about as merely foreseen what it would be forbidden to bring about as
intended. This claim might be urged as an objection to premise 3. The
invocation of the distinction between intended and merely foreseen
consequences works in tandem with a satisficing not maximizing
understanding of the duty to contribute asserted in premise 1.
When we establish a barroom, what we are doing is facilitating the
purchase and consumption of alcohol on the premises to achieve conviviality.
Suppose some barroom brawls will ensue. This bad effect is something we
allow, not something we do. In a similar way, one might hold that when we
establish a democracy that upholds each person’s right to a democratic say,
we merely allow that some persons will misuse or abuse their participation
rights. It might be permissible to allow improper democratic participation
even though it would be wrong to do or perpetrate improper participation.
Suppose doing something harmful is initiating or sustaining a harmful
causal sequence. In contrast, one enables harm by removing an obstacle that
is preventing an already existing causal sequence from issuing in harm. If
one holds that the lack of competence and lack of appropriate motivation
in potential democratic voters are like a coiled spring, a causal sequence
underway but blocked if the opportunity to vote is not present, then
perhaps initiating and sustaining a democratic political order that gives each
member of society a democratic say are better described as enabling not
merely as allowing harm. On some views, enabling harm is morally on a par
with allowing harm, so whether initiating and sustaining democracy is
enabling or allowing makes no difference to the moral status of these
democracy-upholding actions (Rickless 2011; Barry and Overland 2012,
10–116). But on some views, enabling harm either should be assimilated to
doing harm or has an intermediate status.
Moreover, in some circumstances, initiating and sustaining a democratic
order might better be understood as doing harm than as merely enabling
harm. Suppose promulgating the idea that people have a non-instrumental
intrinsic right to a democratic say and facilitating the establishment and
continuance of political regimes in which such a claimed right is upheld
encourage and nurture bad tendencies in subpar voters. Believing they have
a non-instrumental intrinsic right to a democratic say, they are less likely
to believe that their incompetence or questionable political participation
motivation are morally problematic and more likely to participate in
democratic politics in harmful ways than they would be in a political regime
in which no such instrinsic non-instrumental right was promulgated and
upheld. In these circumstances, initiating and sustaining a democratic political
Dictionary: NOSD
Elitism 167
order in which a non-instrumental intrinsic right to a democratic say is

celebrated and upheld qualify as doing harm not merely enabling harm.
The upshot: initiating and sustaining a fully democratic order, if the
order brings about harmful consequences, bring about harm as merely
foreseen not intended. But in some circumstances the agency involved is
enabling not merely allowing, and in some circumstances doing harm not
merely enabling harm.
3. EXPERTISE, NATURAL ARISTOCRACY,

AND BEST RESULTS
The elitist who affirms that the politically knowledgeable and they alone
should rule might appeal to the argument against the existence of a non-
instrumental right to a democratic say just considered. Rejecting this right
to a democratic say, she urges that (1) we ought to institute and sustain that
form of government the long-term maintenance of which would lead to
best results, and (2) some non-democratic form of government (in actual
and likely circumstances) would lead to best results. There are other forms
elitism might take that involve no appeal to anything like points 1 or 2. This
section briefly considers one such version of elitism and then notes a
significant wrinkle in the argument that appeals to 1 and 2. The following
section introduces and examines yet another distinct type of elitism.
To clarify the argument from a claim of superior competence to a claim of
a right to rule, consider two different ways in which it might be filled out.
Suppose it is pointed out that even though the A people are more politically
knowledgeable than the B people, and even on the assumption that the
A people are nice folk who would rule conscientiously, nonetheless, the
consequences of instituting elitist A rule, impartially assessed, would be worse
than the consequences of instituting democracy with an equal vote for all
adult citizens. The advocate of rule by the knowledgeable might be unfazed,
and retort that at least up to a point, such bad consequences would not defeat
the entitlement of the knowledgeable to rule, since it is a basic moral truth
that the knowledgeable and wise should rule the ignorant and unwise.
So understood, the claim of expertise is an unvarnished appeal to a norm
of natural aristocracy, which in modern times many would reject. Dennis
McKerlie interprets Aristotle as embracing the natural aristocracy view.
A person of extraordinary virtue, greater than others possess, deserves to
be king:
It is not simply that the king should rule for the sake of the citizens. For Aristotle, it
would be absurd to expect such a person to be subject to the authority of his
Dictionary: NOSD
168 Richard Arneson
inferiors. At 1284b 28–34 Aristotle compares those who would oppose his authority
to rule to presumptuous mortals demanding to take turns with Zeus in ruling the
world (McKerlie 2001, 128–9).
Given the concerns of this chapter, let us just focus on one type of virtuous
person, the wise expert (in other words, one who knows the political truths
and if given the opportunity, will choose the public policies they justify).
McKerlie’s comment on Aristotle suggests two versions of a natural
aristocracy view. One holds that all persons are fundamentally morally
equal, and share the same moral status, so their comparable interests count
the same in the determination of what should be done (see Arneson 1999
and the essays including Arneson in Steinhoff 2015). Moreover, some of
these fundamentally morally equal persons deserve to rule just on the
grounds of their superior virtue. An alternative version holds that just as
humans are not in any sense equally as worthy as Zeus and the other gods,
so, too, some humans are fundamentally morally superior to others, have a
superior nature, deserve more concern for their interests, are entitled to
fundamentally greater respect simply for being the kind of entity they are
essentially. The morally superior (the natural aristocrats, the wise experts)
are simply on that basis entitled to be rulers of others. In light of the
comment about Zeus, Aristotle looks to be an adherent of the latter type of
natural aristocracy view. Be that as it may, this chapter sets aside all natural
aristocracy views without further comment. I assume they are false (though
no doubt interesting and worthy of exploration).
Return now to the argument for elitism that appeals to a background
norm that political arrangements should be set to bring about best
consequences over the long run. The idea would be that putting power in
the hands of political experts is part of the best strategy toward this end.
Establishing expert rule might be causally efficacious in producing better
laws and other public policies than would otherwise be obtainable, and
these better laws and policies might in turn work to promote morally good
outcomes—outcomes better than could otherwise be reached. Regarded
this way, the argument makes no appeal to meritocratic or natural aristocracy
notions.
Of course, it does not automatically follow that if you install the political
experts in power, the laws and other public policies that result from their
rule would turn out to be superior to what would be reached under a
democratic political constitution. Any number of contingencies could
prevent the rule of the wise from issuing wise laws. But this is not an issue
for the argument for elitism as we are now construing it. If establishing elite
rule leads to bad laws and hence bad outcomes for society, the argument for
elitism condemns establishing elite rule.
Dictionary: NOSD
Elitism 169
Assume for the sake of the argument that there are political experts and
they can be identified and that installing them in power would not (unduly)
corrupt them but would lead to enactment and enforcement of more just
laws and other public policies than those that any other defeasible regime,
including democracy, would produce. That is obviously not the end of the
story. The operation of a democratic regime might have beneficial effects
other than production of just laws that would contribute to the justice of
the society. One obvious possible effect is that having the opportunity to
participate in egalitarian political institutions and actually participating in
such institutions might have an improving effect on people’s sense of justice,
on their spirit of solidarity and cooperation, and on their initiative and
drive. Put another way, being a peon in an autocracy might dull the soul.
This is speculation, but maybe in some circumstances and in some respects,
a good empirical case can be made for this claim or some component of it.
John Stuart Mill, following in the footsteps of Alexis de Tocqueville,
confidently asserts this argument for political democracy, or for some
approximation to it, as Mill is not a wholehearted democrat (Mill 1977;
originally published 1861).
The proponent of the argument for elitism should accept this qualification
of it. We should not make a fetish of ideally just laws and public policies. If
in given circumstances morally better outcomes are obtainable if we eschew
just laws and public policies and instead install some alternative package of
arrangements, we should happily eschew just laws and public policies. The
argument for elitism is an instrumental argument: elitist political constitutions
are to be preferred if and only if they bring about the best outcomes we can
reach. Conceivably this consideration could go the other way: it could be the
case that a democratic political constitution would bring about more just laws
and public policies than any feasible elitist political arrangements could reach,
but this same democratic political constitution would cause a souring of the
public culture and private lives of individuals, so that more justice, morally
better outcomes overall, would be reached under elite rule, rule by competent
experts. In this case we should go with rule by the experts even though doing
this would bring about worse governance.
4. ANOTHER VERSION OF ELITISM
Another possible version of elitism supports the claim that experts should
rule by appealing to a moral right possessed by each of us not to be subjected
to coercion at the hands of incompetent people. This right does not require
that anyone subjects anyone else to coercion in the regular and stark ways
that any government standardly does. Maybe there should be no imposition
Dictionary: NOSD
170 Richard Arneson
of governmental coercion. But the asserted right implies that if there is to be

government, its operation must be set so that it does not violate the right
not to suffer coercion at the hands of the incompetent. For this line of
thought I am indebted to Jason Brennan. His argument merits scrutiny.
Jason Brennan urges a right to a competent electorate (if there is a
democracy in place): “Citizens have a right that any political power held
over them should be exercised by competent people in a competent way”
(Brennan 2011). Since many citizens in existing modern democracies lack
political competence at a threshold level that would plausibly qualify them
for the role of holding power over others, many citizens should be denied
the franchise. Hence (if we assume some form of government should be
instituted) an elite electoral system should be instituted in these countries,
replacing the existing one-person-one-vote regimes.
The considerations marshaled so far in this chapter provide grounds for
rejecting the right to a competent electorate. In some situations, perhaps
the circumstances of actually existing democracies, placing some
incompetent citizens in a position of small power by according them the
franchise predictably improves the degree to which the society operates in
ways that predictably protect people’s rights and promote the general
welfare. At least, restricting the franchise, bringing it about that only
competent people hold any political power, would predictably lower the
degree to which the society operates in a way that fundamental moral
principles would accept as satisfactory. (Here the idea is that the society
does more to promote fulfillment of rights and other relevant values leaving
aside the disputed right to a democratic say, which Brennan is challenging.)
Perhaps there are reasons to expect that restricting the class of those eligible
to vote will increase the competence of the voters and the competence with
which the voting task is carried out but will nonetheless lead to worse laws
on the whole. Perhaps those who would be excluded are biased and tend to
vote in their own interests, but their interests happen to coincide better
with the laws that justice properly understood recommends than would the
laws that the competent voters would select. Perhaps the laws that would be
selected in some feasible elite electoral system would be superior from the
standpoint of justice to the laws that one-person-one-vote democracies tend
to produce, but the system that grants universal suffrage might have other
effects, as Mill speculates, such that the good side effects of having universal
suffrage morally outweigh the poor quality of the laws and public policies
the system spawns.
The advocate of the right to competent governance might be unmoved
by these appeals to good consequences, and insist that one does not have a
right to treat others unfairly merely because that would promote good
consequences, even good consequences in the currency of rights fulfillment,
Dictionary: NOSD
Elitism 171
for bystanders. Of course, the issue is not whether it is acceptable for voters
unreasonably to harm others by their votes. Let us suppose the voters’
actions are unacceptable. The issue is whether there is a right against the
installation of a system of governance that allows this to happen. Brennan
gives this more expansive formulation of what he calls the competence
principle: “When a decision is high stakes and involuntarily imposed
through force on others, it must be made by reasonable and competent
people in a reasonable and competent way.” There is a duty owed to those
on whom one is imposing, not to impose unless one is reasonable and
competent and behaving reasonably and competently. The right against
incompetent imposition is not a right against incorrect imposition. Mistakes
might be made by the reasonable and competent acting reasonably and
competently.
As stated, the competence principle sweeps too broadly. Consider a state-
of-nature conflict. I have stolen my neighbor’s cow. My neighbor is an
unreasonable cuss, and he acts impulsively in response, but what he does
happens to be eminently reasonable: he forces me to give back the cow I
stole. In this situation I have no right against being imposed on by the
unreasonable and incompetent, unreasonably and incompetently, in ways
that happen to be right. The same might be true if I claim to be a victim of
a majority of unreasonable and incompetent voters, acting unreasonably
and incompetently, but imposing on me in ways that just happen to be
right, or close enough to right.
Perhaps only the innocent possess a right against incompetent imposition.
Against this suggestion, consider a variant of the state-of-nature example,
in which the only way the aggrieved incompetent can recover his stolen
cow is by forcibly crossing your private property in order to waylay me.
You are an innocent bystander, and you are being treated unreasonably and
incompetently (in this type of situation, the aggrieved person would be just
as likely to trample on the rights of bystanders violating due proportionality),
but in this case it happens that no harm is being done to the forcibly imposed-
upon bystander. Appealing to the competence principle to assert that the
aggrieved incompetent is violating your rights, you would be wrong.3
Perhaps the right against incompetent imposition should be revised
along the following lines: when a decision is high stakes and involuntarily
imposed through force on others, and the decision happens to be, in fact,
morally wrong, those imposed on have a right that the decision imposed on
3
What is stated in the text is consistent with the claim that disagreement in a state-of-
nature regarding who is really at fault in scenarios in which rights appear to have been
violated is a source of conflict and breakdown of cooperation and thus a justification for
establishing a state. Moreover, disputes about who is competent and entitled to rule can
render rule by the competent unfeasible or unsuccessful.
Dictionary: NOSD
172 Richard Arneson
them not be made by unreasonable and incompetent people and not be

made by people acting unreasonably and incompetently. In other words,
when a resolution of a dispute is imposed on someone against his will, the
resolution must either be substantially right or at least issue from what is in
these particular circumstances a reliable procedure.
We should still reject the revised principle of competence. To see why,
consider cases in which a political procedure is instituted that expectably
both leads to violations of the revised principle of competence and also
expectably works in other ways to bring it about that people’s rights overall
are better protected and the general welfare better promoted than would be
the case if we refrained from instituting such a political procedure. Here the
losses imposed on those who become victims of incompetent political rule
might be outweighed by those who gain in other ways, at least if the ratio of
gains to losses is sufficiently favorable. We should not accept the revised
principle of competence as an absolute exceptionless constraint on political
procedure selection.
Nevertheless, even if we reject the principle of competence, we might still
accept that sometimes installing and maintaining a political regime that
empowers the incompetent might be wrong even if it brings about overall
better consequences than any feasible alternative, because the regime
predictably imposes wrongfully on some, in ways we are not permitted to do.4
The claim that one has a right to competent governance prompts the
question, “how competent?.” Suppose the members of society could be
rank-ordered from best to worst by their degree of political competence and
reasonableness. If this could be done, then for any proposal to restrict the
suffrage, to block the less competent from exercising even a little political
power, there would always be available a further proposal, to limit the
suffrage even more. At the limit, the demand for competence would require
dictatorship. Perhaps even rule by the very most competent and reasonable
would not satisfy the competence principle, if the standard of acceptable
competence is set sufficiently high. Of course, beyond some point,
restriction of the suffrage might predictably lead to worse choice of laws.
But Brennan’s principle of competence is a procedural right, a right not to
4
One might hold that one has some Brennan entitlement not to be imposed on by
the incompetent even when the system that does the imposing is the best system that can
be installed as assessed by the standard of protecting rights fulfillment like the rights that
are at stake in the conflict that triggers the imposition. But when the system yields
morally good results in this way, one might deny there is a right not to be imposed on
except competently just as one might deny there is a right not to be imposed on except
democratically. A best-results account of choice of political governance regimes need not
go with acceptance of consequentialism across the board. Non-consequentialists can
accept it. For a broadly instrumentalist justification of democratic political rights, see
Ronald Dworkin, Sovereign Virtue, ch. 4.
Dictionary: NOSD
Elitism 173
be imposed on by the incompetent, not a right to better rather than worse

decisions and outcomes.
An alternative view is that procedures should be assessed by their role in
producing better or worse outcomes. Procedural rights may be of utmost
importance, but their significance is instrumental, so if it is specified that the
procedure that looks better by some standard will produce worse outcomes
than the alternative, concern about procedures should go by the boards.
The claimed right of an individual not to be imposed on forcibly by
people acting unreasonably and incompetently, at least when the stakes are
high, might be decomposed into a claimed right against constitution-makers
that they not install a procedure that allows incompetent, unreasonable
imposition and a claimed right against those who exercise power unreasonably
and incompetently, imposing on innocent victims. Against the first, I claim
that a sufficient response defending the constitution-makers would be a
showing that the challenged political regime does much better than any
feasible alternative to protect people’s rights and advance the general welfare.
Against the second, the issue is complicated by the fact that an individual
voter doesn’t make a difference, any difference at all, to morally significant
features of outcomes. If my vote is really inconsequential, then that is a
good excuse for my failure to become well informed, deliberate carefully,
and vote wisely. This issue of duties of voters takes us away from the concerns
of this chapter, and I shall not pursue it further here.
5. EMPIRICAL EXPERTISE AND MORAL EXPERTISE
One possible way to deny that superior political knowledge could ever
suffice to render authoritarian political rule morally legitimate would start
by dividing political knowledge into empirical knowledge and moral
knowledge. Let it be granted that ordinary citizens often lack empirical
information that is relevant to reasonable policy choice: how to build a
bridge, for example, or what amount of funds would be required to build a
bridge across a particular river at a specified spot. So perhaps we could
somehow restrict the proper scope of democratic decision-making to choice
of goals, which would require only moral knowledge, which we might
suppose any cognitively normal human adult who applies himself to the
task might gain. Given democratically chosen goals, we might accept
political decision-making rules that insisted on deference to scientific
expertise by legislators in the formation of public policies.5
5
Something along this line was advanced once by Thomas Christiano. See his The
Rule of the Many.
Dictionary: NOSD
174 Richard Arneson
This won’t do as it stands. Sensible choice of goals is virtually always

conditioned on many factual beliefs. Adopting the goal of erecting a bridge
over a particular water barrier depends standardly on many assumptions
about the consequences of building a bridge versus not building it and of
alleviating the transportation needs that suggested bridge-building is hoped
to solve in some other way, or of deciding to kick the can down the road and
not worry about these problems at least for the near future.
But one might still appeal to the special nature of moral knowledge to
resist the presumption that a fraction of people possesses the knowledge
that is relevant to proper choice of laws and public policies and they alone
have the right to rule. Even if empirical premises are interlarded with
evaluative and moral premises in arguments for choice of public policy
goals and public policy programs, still it remains so that valid arguments
to conclusions that this or that public policy should be selected will always
involve evaluative and moral premises. But there is no such thing as special
technical expertise in the identification of evaluative and moral truths.
For short, let’s call the relevant non-empirical truths the moral truths. If
it is at all possible to pull off the trick of identifying true moral claims,
any person can do so. So no argument to the conclusion that a particular
public policy ought to be established can be developed without appeal to
moral premises on which no one has special expertise and any normal
human person has authority to pronounce. One’s pronouncements might
well be wrong, but anyone’s pronouncements might be wrong. And every
normal person will have an extensive stock of everyday empirical lore
and common-sense empirical knowledge gained just in the normal course
of growing up and finding one’s way about the world as one lives one’s
life. Plus, the question to what extent it is reasonable to defer to experts
on empirical matters is itself a normative claim, regarding which the
empirical experts have no special knowledge. All in all, deference to
expertise is not rationally required when one is engaged in figuring out
how to vote, and there is no case that ordinary citizens are disqualified
from a share in political rule by some argument asserting the special
claims of expertise.
This line of thought is open to challenge. What justifies the claim that
there is no such thing as moral expertise? Reason goes where it goes. The
moral principle that determines what ought to be done in our circumstances
might take the form of a complex equation, which you can understand and
solve, and I cannot.6 In these circumstances my claim to a moral right to
6
In this connection see R. M. Hare’s distinction between two ideal types of moral
reasoners, proles and archangels, and two types of moral reasoning, critical and intuitive.
Hare himself suggested that we are all variously prolish and archangelic in different
Dictionary: NOSD
Elitism 175
have a share in political power that will issue in legislation or other public
policy affecting many people is a claim to influence decisions when I have
no reasoned basis for deciding one way rather than another. Such a claim is
unreasonable.
Suppose there is an identifiable group in society that has better knowledge
of the political truths relevant to choice of laws and public policies. Suppose
further that if this group had political power, the group would not become
corrupt or lazy, but would bring it about that better laws are enacted than
would be enacted under any other regime we could establish. Suppose further
that this elite rule would not bring about bad consequences in some indirect
way, but would indeed lead to best results impartially assessed. We might
wonder what more would be needed to deliver a sound argument for elite rule.
6. EPISTOCRACY AND RAWLSIAN

POLITICAL LIBERALISM
The argument for elite rule is open to objection from the standpoint of
Rawlsian political liberalism, a prominent trend in contemporary political
philosophy. These objections cast doubt on Rawlsian political liberalism
rather than on the argument for elite rule. This argument can be stated as
follows:
1. We are obligated to institute and sustain a political order that adequately
protects people’s basic moral rights and advances the general welfare,
provided there is a morally permissible way that we can establish such
an order.
2. If the political order assigns political power to competent experts, the
political order adequately protects people’s basic moral rights and
advances the general welfare.
3. If the political order does not assign political power to competent
experts, the political order does not adequately protect people’s basic
moral rights and advance the general welfare.
4. A democratic political order, in which every normal adult citizen has an
equal democratic say, does not assign political power to competent experts.
5. We are obligated to institute and sustain a non-democratic political
order that assigns political power to competent experts, provided there
is a morally permissible way that we can establish such an order.
decision contexts, but one might hold also that some people are overall more archangelic
than others, more competent with respect to moral reasoning and judgment. See Hare
(1981).
Dictionary: NOSD
176 Richard Arneson
Some clarifications: I take no stand on the plausibility of premises 2, 3, and

4. I’m interested roughly speaking in the question—if 2, 3, and 4 are
true—are we morally bound to institute some form of elite rule (rule by
competent experts), provided that doing so would be feasible and would
not require all-things-considered wrongful actions such as murdering (too
many) innocent people?
In premise 1, the obligation to establish and sustain a political order that
adequately protects people’s rights and so on is said to be qualified—there
must be a feasible path to this goal that would be morally permissible to
traverse. The moral permissibility qualification would come into play if we
could only establish a regime that adequately protects people’s rights by
violating a moral constraint that in the circumstances outweighs the
obligation to establish the regime, all things considered. For example, if we
could establish a rights-protecting regime that would last for at least five
years only by slaughtering a million innocent people, doing so might be
deemed impermissible all things considered even though establishment
would in the long run reduce the overall incidence of slaughter.
Suppose that in some circumstances it is not possible to establish a
political regime that adequately protects people’s basic moral rights and
advances the general welfare. If it is not possible to establish a tolerable
regime, we cannot be obligated to do so. But the argument from elitism
still casts a shadow on this situation. If establishing some non-democratic
political regime that assigns political power only to competent experts
would establish a political regime that succeeds better at protecting
people’s basic moral rights and advances the general welfare than any
democratic regime we could establish, even if neither type of regime
would adequately succeed at this basic governance task, we are obligated
to establish the rule by experts, unless the moral costs of transition to this
regime outweighed or preempted the expectable long-run moral gains of
making the transition.
A political order adequately protects people’s basic moral rights and
advances the general welfare just in case rights are protected and welfare
advanced to an extent such that individuals who are not themselves
violating the rights of others are morally at liberty to pursue their own
projects and aims and are not required to devote further personal resources
toward improving the political order. In other words, individuals are not
obligated always to act in ways that produce the impersonally best
attainable outcomes. They are morally at liberty to act in ways that lead to
less than best outcomes, provided the shortfall between the value of the
outcomes they reach and the value of the best they might have achieved is
not excessive. Individuals have a Scheffler prerogative to act as they prefer,
Dictionary: NOSD
Elitism 177
within limits.7 In this same vein, individuals have obligations to work

together to produce a tolerably good political order.
I make no effort to identify what level of government performance
should count as adequate, or tolerable, or sufficiently good. Nor do I make
any attempt to specify how rights protection and promotion of good
quality lives for people might be appropriately balanced in an overall
standard of government performance. This matter is left to intuitive
judgment in wide reflective equilibrium. Some might argue that the good-
enough level is the maximal level: suboptimal performance is unacceptable.
A system of government that could be better designed and implemented,
to do a better job of fulfilling people’s rights and promoting their welfare,
should be better designed and implemented. In many situations a
government system might show obvious flaws and deficits, but yet the
government is no worse than anything else that might be implemented in
the circumstances. If we tried to reform the government structure and
political constitution, we would unavoidably make things worse. In these
circumstances satisficing is maximizing.
Some might argue that a democratic political order that takes the form of
a representative democracy tends to assign political power to competent
experts. Voters by majority will elect competent experts to be legislators
and top public officials. Obviously a democratic system, to be acceptable,
need not perfectly screen political candidates so that only competent
experts succeed in electoral contests. The democratic order to be acceptable
needs only to do well enough, so that a tolerable threshold of government
performance is attained, a good-enough level of rights protection and
advancement of the common good. If an argument to this effect succeeds,
premise 4 in the argument for elitism fails and the argument does not
succeed in establishing its conclusion.
John Stuart Mill sometimes suggested that a representative democracy
system might do a good-enough job at placing political power in the hands
of individuals competent to exercise the responsibilities of political rule (Mill
1977, chs. 3 and 4; Thompson 1976). In a well-functioning representative
democracy, voters choosing among candidates would not need to master
the intricacies of the complex empirical and moral issues, the understanding
of which is necessary for making reasonable choices among alternative
public policies. Voters need only assess the good character and intelligence
of those running for office, and can leave selection among policies to elected
legislators and executive branch leaders. However, Mill’s hopes for this
7
This idea is introduced in Samuel Scheffler, The Rejection of Consequentialism
Dictionary: NOSD
178 Richard Arneson
division of political labor may be unrealistic. Potential political leaders of

superior intelligence and estimable moral character might yet be wedded to
bad political programs and unjust but nice-sounding principles. There
may be no substitute, in a representative government, for the wisdom,
intelligence, and knowledge of the voters who select the representatives.
If these are lacking, premise 4 may become true.
The track record of non-democratic government in modern times is, to
put it mildly, not very good. Recall Hitler, Stalin, and Pol Pot. But of course
the history of disastrous non-democratic political regimes is irrelevant to
the soundness or lack of soundness of the argument for elitism. Even if
there are many actual and millions of possible non-democratic forms of
governance that are expectably worse in terms of promotion of rights
protection and welfare advancement than the range of democratic
governments we have observed from the eighteenth century down to the
present, that does not gainsay the possibility that some identifiable and
implementable form of non-democratic governance would fulfill the
fundamental tasks of government better than any feasible form of democratic
governance.
The argument as presented addresses what should be done if certain facts
obtain and does not tell us what we ought to do in the face of uncertainty
about what the facts are and about what outcomes would result from the
adoption of any of the various policies we might choose. In present and
likely future circumstances we will face, decisions about whether to support
a proposed political constitution or political regime will be decisions under
uncertainty. I assume that the argument for elitism could be reconfigured as
an argument about what to choose under uncertainty. Suppose it is not
known with certainty, but very likely, that if we institute a non-democractic
system, rights will be respected and welfare promoted to a good-enough
degree, and that if we institute a democratic system, fulfillment of rights
and welfare advancement will not be promoted to a good-enough degree.
Then relative to the available evidence, it will be right to institute a non-
democratic system.
However, absence of certainty about the consequences of adopting
one or another political system complicates the argument that would
have to be made. Perhaps concentrating political power in the hands of a
few political experts might tend to produce good consequences across
a wide swath of likely conditions but would lead to catastrophe under
some circumstances, whereas maintaining a democratic regime would
produce generally mediocre results but would avoid catastrophe under all
circumstances. Perhaps in selecting political regimes we should give priority
to avoidance of catastrophe (on disaster avoidance, see Kavka 1986). By
focusing only on the case in which the consequences of adopting one or
Dictionary: NOSD
Elitism 179
another of the available alternatives are known with certainty, we avoid

these complications.
The argument for elitism just presented does not assume that there is
no moral right possessed by each member of society to be ruled only by
democratic procedures. Suppose there is such a right to a democratic say. So
long as this right to a democratic say can be overridden when fulfilling it
conflicts with the fulfillment of other rights and would have to be purchased
at too great a cost to the general welfare, the argument for elitism can still
go through.
It seems to be the case that David Estlund is committed to rejecting the
argument for elitism (Estlund 2008, chs. 2, 3, and 11). That is, he is
committed to denying that we should ever accept its conclusion, that we are
obligated to institute and sustain a non-democratic political order that
assigns political power to competent experts, for the reasons contained in
the premises of this argument.
The background moral framework here is Rawlsian political liberalism,
an attractive doctrine. For our purposes its key element is a liberal
principle of legitimacy, which Rawls formulates as follows: “Our exercise
of political power is fully proper only when it is exercised in accordance
with a constitution the essentials of which all citizens as free and equal
may reasonably be expected to endorse in the light of principles and ideals
acceptable to their common human reason” (quoted from Estlund 2008,
at 43). Let us say political coercion cannot be morally acceptable unless
imposed by a state under a form of government that all reasonable citizens
accept (there is no reasonable ground for its rejection). This liberal principle
of legitimacy gives expression to our conception of citizens as free
and equal.
Exploring the relation between liberal legitimacy and democracy, it
makes sense to focus attention on the views of Estlund, who accepts liberal
legitimacy, deploys it to argue for democracy, and also accepts that part of
the justification of democracy consists of its epistemic virtue—its tendency
to select substantially just public policies. He thus recognizes a threat that
selection of a form of government for its epistemic virtues might lead not
to affirmation of democracy but of some scheme of rule by experts.
He considers what have been called “plural vote” schemes as posing
a form of the threat that is hard to resist. Suppose we cannot reasonably
reject the proposal that educating citizens, training them to be literate and
numerate and so on, would produce a better set of democratic voters. How
then can we resist the proposal to give extra votes (extra voting power) to
those who are more educated than others? He notes that the latter proposal
involves invidious comparisons, claims that some persons are wiser or
more politically knowledgeable than others, and so empowering them to
Dictionary: NOSD
180 Richard Arneson
rule will lead to more just public policies, whereas the proposal to introduce
education for all does not. The proposals that involve invidious comparisons
will all be subject to some reasonable objection, so will fail the liberal
legitimacy test.
Estlund raises what he calls a “demographic” challenge to any such
proposal. Those who are picked out as more educated may differ from the
rest of the population in ways that will tend to worsen their decision-
making to an extent that outweighs whatever benefits might be expected to
come with having a more educated political class. The more educated might
be predominantly white males, for example. For any demographic biases
we can identify, we could correct for the bias by selecting a subset of the
qualified experts to have the franchise, so that the group is a cross section
of the population according to the descriptions we identify. But Estlund
supposes it would always be reasonable to object to any such demographically
adjusted version of the elite rule scheme by postulating that there might be
hidden or latent biases not empirically discoverable.
Question: why isn’t democracy itself open to reasonable rejection, on the
ground that it is reasonable (meaning not that this judgment is correct, but
that it is somewhat reasonable or is reasonable enough), that universal adult
suffrage enfranchises too many incompetent voters, and hence leads to
choice of substantially unjust decisions that suffrage restriction or assignment
of extra votes to the better educated would mitigate? Estlund’s answer:
universal suffrage does not involve invidious comparisons.8 In contrast,
“Under unequal suffrage, some people are formally and permanently
subjected to the rule of certain others” (Estlund 2008, at 37). The claim
is not that democracy must be agreed by all reasonable citizens, or from
all reasonable points of view to be epistemically superior to any feasible
alternative, but that of the political procedures that are not subject to
reasonable rejection, democracy is epistemically best (beats the proposal of
choice of public policy by random coin flip, for example).
In response: the mere claim, in the absence of empirical evidence, that
some particular plural vote or unequal suffrage scheme that corrects for all
known demographic bias might conceivably harbor some undetected latent
source of bias, is not a serious, significant objection. This conceivable bias
problem would have especially little weight if we implemented unequal
8
In passing I note that it is not clear why invidious comparisons should be especially
vulnerable to reasonable rejectability. The constitutional proposal that all political
decisions be made by consulting astrological signs, with the stipulation that no one has
special astrological competence and the determination of who will do the astrological
consulting to decide any particular issue is to be made by coin toss, makes no invidious
comparisons but is reasonably rejectable on the grounds that it is a bad idea. If some
argument for elite rule has independent plausibility, then democracy is a bad idea.
Dictionary: NOSD
Elitism 181
suffrage in some political jurisdictions and the results proved to be

impressive, as judged by the epistemically best standards of justice we can
identify in present circumstances. If you lower the bar of reasonable rejection
so this de minimis objection passes it, then you are trading off expectable
substantive justice in outcomes to gain liberal legitimacy at a clearly
unacceptable discount rate.
One can appreciate the concerns that lead Estlund down this path of
argument. However, we should regard the stern suspicion of elite rule that
this path of argument protects as constituting a reductio ad absurdum of the
political liberalism doctrine that carves out this path. All of the objections
that the Estlund maneuvers discover are compatible with there being good
and sufficient evidence and argument for the truth of each of the premises
in the elitist argument. Even if the premises of the argument are all true and
well supported, they are, according to political liberalism, not available as
justifications of a form of governance as morally acceptable, because the
premises might still be subject to reasonable rejection. This is possible
because in the political liberalism doctrine, a reasonable objection need not
be one that a fully reasonable and rational person would accept, just one
that an ordinary person reasoning in ordinary ways might accept.
This standard becomes a very powerful constraint on the legitimacy of a
political system if “reasonably” is interpreted in a relaxed way, as proponents
of the doctrine affirm. The weaker the idea of reasonable rejectability, the
stronger the constraints on what can qualify as legitimate political rule. The
constraint becomes too powerful, as the argument for elitism shows.
Maybe the argument for elitism does not justify elite rule in actual
circumstances, because in our circumstances one or more of the premises in
the argument is false. What should not be acceptable is to hold that even if
the premises are correct and well supported, and it really is the case that we
can adequately protect people’s basic moral rights and advance the general
welfare only by scrapping democracy and instituting some non-democratic
form of political rule, this course of action would nonetheless be morally
unacceptable and should not be implemented because some member of
society has some partially, but not fully reasonable rejection to it.
Maybe what I have reported as Estlund’s view is inaccurate. Surely an
advocate of political liberalism should allow that circumstances might arise
in which the performance of any feasible democratic regime we could
establish and sustain would be so abysmal compared to the performance of
an identifiable feasible non-democratic alternative involving assignment
of political power to competent experts that in those circumstances the
proposal that we ought to establish the non-democratic alternative would
not be reasonably rejectable. The outcomes of any democracy we could
sustain might be beyond the pale, unacceptable from any reasonable moral
Dictionary: NOSD
182 Richard Arneson
standpoint. The proposal to install non-democracy in these circumstances

would be justifiable by the principle that if there is some feasible political
system that would uncontroversially fulfill the basic moral functions of
government to a tolerable, adequate extent, and no other feasible political
system that would do the same, we are obligated to establish and sustain the
uniquely, uncontroversially tolerable system. This principle would not be
reasonably rejectable.
Grant that in the extreme circumstances described above the political
liberalism advocate following Estlund’s interpretation of the doctrine could
hold that acceptance of rule by experts could be justified. My claim then
would be that political liberalism’s acceptance of rule by experts comes too
late. A political regime that was not so horrendous that from no semi-
reasonable standpoint could it be found acceptable could still be horrendous.
If the best available democracy would really deliver an intolerably bad
record of failure to protect people’s rights and advance the general welfare,
as assessed by the correct view of these matters, and some identifiable form
of rule by experts would deliver a substantial improvement in terms of this
measure, the further fact that this claim would be subject to reasonable (or
more exactly to “reasonable”) objection would be neither here nor there.
The fact that some conscientious person trying to figure out what is right,
even if he is following the lead of some major religious or ethical tradition,
would incorrectly judge that the record of a horrendous democracy does not
qualify as horrendous, and would anyway be inferior to some form of rule
by experts, is insufficient to support the claim that democracy is tolerable,
much less morally claims our allegiance.
7. CONCLUSION
No decisive conclusion emerges from this discussion of elitism.

Considerations introduced in Section 2 stand in the way of acceptance of
the Section 2 argument against the claim that each of us has a non-
instrumental, intrinsic right to a democratic say. I don’t yet see a positive
argument for the claim, however. My own view is that if we should favor
democracy, we should favor democracy purely on instrumental grounds.
But even if you aren’t fully persuaded that this instrumentalism is the
right stance to take, you should still accept the weaker conclusion, that
even if there is some sort of qualified non-instrumental right to a
democratic say, it can be overcome when citizens’ lack of political
competence brings about an excessive shortfall in fulfillment of people’s
fundamental moral rights.
Dictionary: NOSD
Elitism 183
A final thought: Another possible ground for disputing the argument

to the conclusion that there is no non-instrumental intrinsic right to a
democratic say appeals to democratic equality considerations. Samuel
Scheffler writes: “we believe that there is something valuable about human
relationships that are—in certain crucial respects at least—unstructured by
differences of rank, power, or status” (Scheffler 2005; see also Kolodny 2014).
How we might develop the claim that social equality is valuable into an
argument for a democratic say, and how we should assess this line of
argument, are good topics left for future discussion. My hunch is that this
exercise would turn up empty pockets, but hunches are not arguments.
Bibliography
Arneson, R. (1999). “What, If Anything, Renders All Humans Morally Equal?,” in
D. Jamieson (ed.), Peter Singer and His Critics (Oxford: Blackwell), 103–28.
Arneson, R. (2009). “The Supposed Right to a Democratic Say,” in T. Christiano
and J. Christman (eds), Contemporary Debates in Political Philosophy (Oxford:
Wiley-Blackwell), 197–212.
Arneson, R. (2010). “Democratic Equality and Relating as Equals,” Canadian
Journal of Philosophy supp. vol. 36: 25–52.
Arneson, R. (2015). “Basic Equality: Neither Rejectable nor Acceptable,” in
U. Steinhoff (ed.), Do All Persons Have Equal Moral Worth? On ‘Basic Equality’ and
Equal Respect and Concern (Oxford: Oxford University Press), 30–52.
Barry, C. and Overland, G. (2012). “The Feasible Alternatives Thesis: Kicking
Away the Livelihoods of the Global Poor,” Politics, Philosophy, and Economics
11: 97–119.
Brennan, J. (2011). “The Right to a Competent Electorate,” Philosophical Quarterly
61: 700–24.
Christiano, T. (1996). The Rule of the Many: Fundamental Issues in Democratic Theory
(Boulder, CO: Westview Press).
Christiano, T. (2008). The Constitution of Equality: Democratic Authority and Its
Limits (Oxford: Oxford University Press).
Dworkin, R. (2000). “Political Equality,” in R. Dworkin, Sovereign Virtue: The
Theory and Practice of Equality (Cambridge, MA: Harvard University Press),
ch. 4.
Estlund, D. (2008). Democratic Authority: A Philosophical Framework: Democratic
Authority and Its Limits (Princeton, NJ/Oxford: Princeton University Press).
Hare, R. M. (1981). Moral Thinking: Its Method, Levels, and Point (Oxford: Oxford
University Press).
Kavka, G. (1986). Hobbesian Moral and Political Theory (Princeton, NJ: Princeton
University Press).
Kolodny, N. (2014). “Rule over None II: Social Equality and the Justification of
Democracy,” Philosophy and Public Affairs 42: 287–336.
Dictionary: NOSD
184 Richard Arneson
McKerlie, D. (2001). “Aristotle’s Theory of Justice,” Southern Journal of Philosophy

39: 119–41.
Mill, John Stuart (1977). Considerations on Representative Government (originally
published 1861), in J. S. Mill, Collected Works 19, J. M. Robson (ed.) (Toronto:
University of Toronto Press), 371–577.
Pettit, P. (2015). “Justice, Social and Political,” in D. Sobel, P. Vallentyne, and
S. Wall (eds), Oxford Studies in Political Philosophy, Vol. I (Oxford: Oxford
University Press).
Rickless, S. (2011). “The Moral Status of Enabling Harm,” Pacific Philosophical
Quarterly 92: 66–86.
Scheffler, S. (1982). The Rejection of Consequentialism (Oxford: Oxford University
Press).
Scheffler, S. (2005). “Choice, Circumstance, and the Value of Equality,” Politics,
Philosophy, and Economics 4: 5–28.
Steinhoff, U. (ed.) (2015). Do All Persons Have Equal Moral Worth? On ‘Basic
Equality’ and Equal Respect and Concern (Oxford: Oxford University Press).
Thompson, Dennis (1976). John Stuart Mill and Representative Government
(Princeton, NJ: Princeton University Press).
Dictionary: NOSD
7
Is the Gendered Division of Labor
a Problem of Distribution?
Gina Schouten
Over the past several decades, women have significantly increased their
participation in the paid labor force and we have witnessed a corresponding
increase in the number of dual-earner families.1 Still, household divisions
of labor remain highly unequal, with women in every industrialized
country continuing to do the vast majority of housework and childcare
(Coltrane 2000; Gershuny and Sullivan 2003).2 This persistent gendered
division of labor (hereafter, GDL) has harmful consequences—not only for
women, but for men and children as well (Glass 2004; Gornick and
Meyers 2009; Zippel 2009; Coltrane 2009; Correll et al. 2007; Breen and
Cooke 2005; Kluwer et al. 1996 and 2000; Kluwer 1998; Poortman and van
der Lippe 2009; Milkie et al. 2002; Barnett and Shen 1997).
Many theorists have called for political interventions to remedy the
GDL, and a growing body of empirical research suggests that certain policy
initiatives could increase the time men spend on domestic work and enable
women to work more outside the home (Gornick and Meyers 2009). Three
of the most widely endorsed initiatives are: family leave provisions (including
those with mechanisms to induce more men to take leave); work time
regulation (including interventions to limit working hours and encourage
employers to provide proportional benefits for part-time work); and social
1
I have been fortunate to receive helpful feedback on this chapter during many
presentations and discussions. I would like especially to thank the participants of the
2014 Oxford Studies Workshop for Political Philosophy, held in Columbia, Missouri.
I am much indebted to Peter Vallentyne, Cynthia Stark, and two anonymous reviewers
for thoughtful feedback on earlier drafts of this chapter, as well as to Harry Brighouse and
Jeff Behrends for their willingness to read drafts and discuss these ideas with me over the
course of several years. My work on this topic owes so much to those discussions.
2
Extremely highly educated professional men might be an exception to this trend
(Klumb et al. 2006).
Dictionary: NOSD
186 Gina Schouten
supports for childcare. These “gender egalitarian” political interventions

target labor markets and childcare arrangements as intermediaries through
which to influence domestic divisions of labor. They are based on the
premise that the GDL is sustained in large part by a labor structure and
social policies that, failing to respond to changing social and economic
conditions, have remained predicated on the “breadwinner/homemaker”
model of domestic cooperation (Gornick and Meyers 2009, p. 4). The
breadwinner/homemaker model presumes that domestic partners specialize
in either paid labor or caregiving labor, so that each is relatively free of the
work responsibilities primarily borne by the other. On the basis of this
model, employers have come to assume that positions will be filled by
employees who can heavily prioritize the demands of paid labor, and social
policies have not developed to ease the burdens faced by caregivers who
cannot prioritize caregiving to the exclusion or near exclusion of paid labor.3
Clearly, the breadwinner/homemaker model is outdated: economic
realities have increased the labor burdens faced by many families; women
increasingly aspire to work outside the home; and men increasingly want
more involvement in the lives of their children. As a result, we have grown
into a predominately dual-earner society. Those with serious caregiving
responsibilities are often not free from paid labor obligations, and those
with serious paid labor obligations often assume serious caregiving
responsibilities as well.
Overburdened, time-poor families are the consequence, and both men and
women bear the harms. But women’s performance of paid work has increased
much more than has men’s performance of unpaid domestic work. For this
reason, the harms of presumed breadwinner/homemaker specialization fall
disproportionally on women, who struggle to balance caregiving responsibilities
with increasingly serious work commitments outside the home. In anticipation
of this struggle, many women opt against taking on demanding commitments
to paid work in the first place. Others try, but lessen their work commitments
when family well-being seems threatened or seek jobs with greater flexibility—
and correspondingly less pay and status and fewer prospects for advancement.
Meanwhile, women’s greater share of caring work depresses their job prospects
through mechanisms of statistical discrimination, whereby employers use
aggregate group characteristics—including the datum that the average woman
is less likely to make paid work a first priority than the average man—to assess
individuals’ job qualifications (Robeyns 2007; Glass 2004; Correll et al. 2007).
3
Joan Williams has documented the growing ubiquity of extraordinarily demanding
workplaces and what she refers to as the “ideal worker” assumption: the expectation of
employers that they are entitled to demand that their employees work very long hours
and take little or no time off for caregiving work (Williams 2000).
Dictionary: NOSD
Is the Gendered Division of Labor a Problem of Distribution? 187
Gender egalitarian interventions could expand families’ opportunities to

share paid and caring work more equally by limiting paid labor expectations
and enhancing support for caregiving. By changing childcare and labor
markets, these interventions could remove obstacles to gender egalitarianism
and lessen its costs in the short run. And by lessening the costs and increasing
uptake of gender egalitarianism in the short run, they could eventually
erode the gender norms that sustain the GDL under the status quo. Many
of the most effective interventions accomplish this by positively incentivizing
gender egalitarian choices. For example, some theorists support making
parental leave non-transferrable between partners, so that each partner’s
leave is awarded on a “use-it-or-lose-it” basis (Gornick and Meyers 2009);
others support allocating leave to mothers only on the condition that leave
also be taken up by their male partners (Brighouse and Wright 2009).
Other interventions, such as work time regulations, less obviously involve
positive incentives. But all gender egalitarian interventions effectively
subsidize gender egalitarianism relative to the status quo: they lessen its costs
and, by changing gendered social norms, they raise the costs of maintaining
a traditional GDL.
There is good evidence that carefully composed packages of these
interventions could effectively make our society more gender egalitarian
(Bianchi et al. 2000; Ciscel et al. 2000; Estes et al. 2007; Sayer 2005;
Crompton 2009; Coltrane 2009). Still, these interventions need another
kind of justification, in addition to the truth that they will help erode the
GDL. Questions of legitimacy arise because reasonable citizens can disagree
about what makes for a good life or a good society, and because a fundamental
commitment of liberalism is to limit political intrusion into the lives of
citizens so as to allow considerable space for them to act on their own
answers to those questions. Legitimacy concerns the constraints by which
we must abide when we use our collective political power to enact change
in society, given that exercises of our political power influence lives and
constrain choices in ways that many will find unwelcome.
Establishing that gender egalitarian interventions are legitimate cannot
come by way of some easy maneuver of categorization, for example by
maintaining that gender egalitarian interventions do not subsidize but
merely remove obstacles. It is true that, under the status quo, gender
egalitarianism is costlier than enacting a traditional GDL, and so
interventions might be thought necessary to equalize the accessibility of the
options. Indeed, if we could arrange social institutions to make every
reasonable lifestyle choice equally costly, and if gender egalitarian
interventions brought us nearer to that arrangement, then we might regard
those interventions not as subsidies but as mere removal of obstacles, facing
correspondingly diminished burdens of legitimacy. But social institutions
Dictionary: NOSD
188 Gina Schouten
invariably influence the relative costliness of lifestyle choices, and no feasible

default exists wherein all reasonable choices are equally costly.4
Moreover, the status quo GDL is sustained by choices individuals make
about how to allocate paid and unpaid work within cooperative domestic
relationships. While these choices are due largely to social structures whose
design takes breadwinner/homemaker specialization for granted, the causal
relationship here is, plausibly, bidirectional. Employers have been able to
enforce their expectation that employees will be largely free to prioritize
paid labor precisely because workers’ deviation from the breadwinner/
homemaker specialization model—and their non-compliance with the
workplace expectations that model perpetuates—has been incomplete. In
short, the GDL persists because individuals choose, in sufficient numbers,
to make gender a salient consideration in determining how to allocate labor
within their domestic partnerships.5 Because the status quo is maintained
by choices individuals make about what packages of paid and unpaid labor
to take on, and because we cannot feasibly make every lifestyle choice
equally costly, it makes little difference whether we regard gender egalitarian
interventions as merely removing obstacles to particular lifestyle choices, or
more strongly as subsidizing those choices.6 Gender egalitarian interventions
aim to influence individuals’ behavior by changing the relative costliness of
different lifestyle choices. Proponents of those interventions must meet the
justificatory burden of showing that they are legitimate: that we can regard
the new arrangement of relative costs as more just than the old, and thereby
move to supplant the old with the new, even though many in our diverse
justificatory community reasonably prefer the status quo.
Within the tradition of political liberalism, the constraint on political
legitimacy is embodied by the Criterion of Reciprocity and its animating
idea of public reason: “[O]ur exercise of political power is proper only when
we sincerely believe that the reasons we would offer for our political
actions . . . are sufficient, and we also reasonably think that citizens might
also reasonably accept those reasons” (Rawls 1993b, 446–7). The Criterion
of Reciprocity requires justificatory neutrality: political power is legitimate
only when its exercise can be justified on the basis of reasons that are shareable
among citizens. In order for reasons that support political intervention to be
shareable among citizens, those reasons must be neutral among the
4
Nor, if there were, would we necessarily want to bring it about. We often make
certain choices less costly than alternatives for very good reasons—and sometimes for
uncontroversially good reasons.
5
In due course, I will explain why these choices must be regarded as voluntary.
6
Framing the argument as a justification for a subsidy has the strategic advantage of
clarifying that the task is to argue, rather than assume, that gender egalitarianism ought to
be more accessible than it is under the status quo.
Dictionary: NOSD
conceptions of the good that citizens may reasonably embrace. Interventions

aimed at influencing families’ allocations of work appear to be challenged
by this neutrality constraint, because many citizens consciously enact and
even celebrate traditionally gendered domestic arrangements.7 The good of
gender egalitarianism seems not to be a reason that is shareable among all
reasonable citizens; it therefore seems an illegitimate basis on which to
justify gender egalitarian interventions. Social policies that aim to induce
gender egalitarian choices appear to be suspect, then, unless we can muster
some neutrality-abiding reasons in their favor.
The neutrality constraint is widely accepted by both proponents and
opponents of gender egalitarian interventions. Proponents of intervention
who accept the neutrality constraint have devised elegant arguments to
justify those interventions, and claimed that these arguments abide by the
neutrality constraint.8 If these arguments are sound, and if they rely only on
premises that can be neutrally affirmed, then gender egalitarian interventions
would be legitimate exercises of political power, despite initial appearances
to the contrary. Many such arguments deploy a common strategy, which I
call the “mal-distribution strategy.” My project in this chapter is to critique
that strategy. Whether or not gender egalitarian interventions can ultimately
be shown to be legitimate by the lights of the neutrality constraint, I raise
doubts that the mal-distribution strategy can do the job. If I am right, then
proponents of gender egalitarian interventions who seek to work within the
constraint of neutrality must seek a new route to establishing the legitimacy
of those interventions. In Section 1, I introduce the mal-distribution
strategy. In Section 2, I elucidate the daunting argumentative burden that
strategy must meet if it is to be executed within the justificatory constraint
of neutrality. In Section 3, I argue that the GDL is not essentially a problem
of distribution, and that the mal-distribution strategy therefore
fundamentally misdiagnoses the injustice of the GDL.
1. The Mal-Distribution Strategy
Many theorists who find the GDL objectionable diagnose the injustice in
distributional terms. In her seminal work on the GDL, for example, Susan
Moller Okin explains the injustice this way:
I assume here that they are not necessarily unreasonable in doing so.
7
I count myself as a proponent of gender egalitarian intervention; I am a somewhat

8
less committed proponent of the neutrality constraint. I think we should endeavor to

articulate our case for interventions to promote gender justice so as to abide by the
constraint, and I think it can be done. But if it cannot, I would deem that a reason for
rejecting the constraint is too restrictive. More on this in Section 3.
Dictionary: NOSD
190 Gina Schouten
When we look seriously at the distribution between husbands and wives of such
critical social goods as work (paid and unpaid), power, prestige, self-esteem,
opportunities for self-development, and both physical and economic security, we
find socially constructed inequalities between them, right down the list (Okin
1989, 136).
In one sense, the GDL obviously involves distributions: women do less

socially remunerated labor outside the home than do men, and men do less
caring labor than do women. As a first pass, we might say that those who
endorse gender egalitarian political interventions on the basis of the mal-
distribution strategy argue that the GDL is objectionable because it causes
individuals to have an unfairly small or inadequate share of something. Of
course, not every unequal distribution is objectionable.9 There may be
goods the distribution of which is not susceptible to considerations of
justice; there may be holders of goods among whom certain distributional
concerns are suspended; and there may be types of inequalities that are
not unjust.
On the face of it, neither paid work nor caring work appears to be
a plausible candidate for the mal-distributed good. Because unequal
distributions are objectionable only when they are distributions of something
valuable, types of work can serve as mal-distributed goods only if the
distribution renders the holdings of one group less valuable than the
holdings of another. Because neither paid nor caring work appears, on their
face, to be intrinsically more valuable, imbalances in the proportions in
which they figure into men’s and women’s total workloads do not themselves
render one group’s holdings less valuable.10
Another inequality comes readily to mind: women’s greater share of
caring work combined with their increasing performance of paid labor
renders their overall workloads larger, and as a result they enjoy less leisure
time than men (U.S. Department of Labor American Time Use Survey 2011).
But this inequality would constitute an injustice only if leisure time is
a good the distribution of which, taken in isolation, is susceptible to
considerations of justice. This possibility for justifying gender egalitarian
interventions further elucidates the tension between those interventions
and the liberal commitment to neutrality. It will be helpful, therefore, to
9
For ease of exposition, I will refer to the distributive ideal as one of equality.
Nothing hangs on this simplification, I think, and other distributional principles could
be substituted without lessening the force of my challenges to the distributive paradigm.
10
Perhaps caring work is more intrinsically rewarding than paid work, but this
imbalance would indicate that men, as a group, enjoy a less valuable share, because they
have less intimate involvement in the care of loved ones. This diagnosis of the GDL
implies no claims of justice on behalf of women, who are supposed to be the parties most
straightforwardly harmed by it.
Dictionary: NOSD
pause briefly to consider the neutrality constraint and the considerations in

favor of imposing it.
Because citizens living in a diverse, pluralistic society will reasonably
come to affirm different conceptions of the good, political liberals have
argued that a just and stable regime should be neutral among the reasonable
conceptions of the good its citizens may affirm.11 Exercises of political
power are neutral only if they are justifiable without recourse to particular
views about what constitutes a good life. In this chapter, I make no argument
to support imposing the neutrality constraint. My task in the next section is
to defend the conditional claim that if neutrality does indeed constrain
political legitimacy, then the prospects for the mal-distribution strategy
look bleak. For now, the task is just to clarify the burden that constraint
would impose.
If neutrality does indeed constrain political legitimacy, then any exercise
of political power is legitimate only insofar as it can be fully justified on the
basis of reasons that are shareable among all reasonable citizens. We can see,
now, why the defense of gender egalitarian interventions cannot proceed
straightforwardly from an unequal share of work and leisure between men
and women. Because individuals may reasonably disagree about the relative
value of work and leisure—and indeed, of paid and caring work—
inequalities in kinds or amounts of work cannot straightforwardly be
identified as objectionable distributions without violating the constraint of
neutrality. These inequalities do not generate reasons for remediation that
are shareable among reasonable citizens, because reasonable citizens can
reject the presumption that different compositions of work and leisure are
unequally valuable.
We can refine our understanding of the mal-distribution strategy to
account for this nuance: those who pursue this approach argue for gender
egalitarian interventions on the grounds that the GDL is objectionable
because it causes individuals to have an unfairly small or inadequate share
of something the value of which can be affirmed without violating the
constraint of neutrality.
There is another way of putting this, which may help to clarify. To abide
by the neutrality constraint, the reasons offered by the mal-distribution
strategist to justify intervention must be political: they must invoke only
values the good of which can be discerned without recourse to particular
conceptions of the good. Neutrality dictates that the state cannot legitimately
intervene for the purpose of promoting a controversial conception of what
11
I formulate the constraint of neutrality using John Rawls’s political liberalism
(Rawls 1993). Other defenses and elaborations of political liberalism include Ackerman
1980 and 1994; Larmore 1987 and 1996; and Moon 1993.
Dictionary: NOSD
192 Gina Schouten
is good for someone, or to prevent harm as defined by that conception’s

notion of what is bad for someone. The state cannot, for example, intervene
to encourage an atheist to embrace the tenets of Protestantism on the
grounds that doing so can prevent the harm of her eternal damnation. That
harm cannot be cashed out using only political values; it requires that we
invoke a particular comprehensive value system, which not all citizens can
be expected to affirm. Similarly, if some distribution of goods is to ground
governmental policies aimed at changing what goes on within families, then
those goods must be politically valuable: affirmable from within any
reasonable value system. While some package of goods that includes leisure
time may be so affirmable,12 the preference for leisure over work within that
package is not.
Still, the mal-distribution strategy has been pursued widely by those
writing about gender justice, and many of its proponents take great care to
abide by the constraint of neutrality.13 We can consider one variant to
illustrate the general approach. Some theorists argue for gender egalitarian
interventions on the grounds that the GDL results in an objectionable
distribution of positive health outcomes, with women’s inferior share owing
to their performance of such a comparatively large share of caring labor
(Lloyd 1998; Baehr 2008). This justification for gender egalitarian political
interventions appears to bypass substantive value considerations and rely
only on values that all citizens can recognize as such, like the value of certain
positive health outcomes that are allegedly desirable regardless of one’s
12
For example, leisure time figures among Rawlsian social primary goods.
13
S. A. Lloyd, Amy Baehr, Ingrid Robeyns, Anca Gheaus, and Martha Nussbaum
have all pursued mal-distribution strategies for justifying gender egalitarian political
interventions. Lloyd and Baehr argue that their conceptions of the goods of justice can
be grounded in public reason and can therefore form part of a legitimately neutral
conception of justice. Lloyd defends a principle calling for rough equality in citizens’
share of certain publicly recognized values, including liberty, equality, fairness, reciprocity,
stability, security, opportunity, and public health (Lloyd 1998). Baehr interprets Lloyd’s
criterion of sexual equality as “a robust principle of antidiscrimination” and defends it as
a legitimate component of a political conception of justice on the grounds that individuals
behind the veil of ignorance would endorse it (Baehr 2008). Robeyns argues for a
conception of gender justice that counts socially caused inequalities in men’s and women’s
capability sets as unjust (Robeyns 2007). Gheaus argues for a conception of gender
justice that counts society as unjust so long as central components of good lives are
unequally costly to men and women (Gheaus 2012). Martha Nussbaum understands the
harm of gender inegalitarianism as a mal-distribution of human capabilities (including
life; bodily health and integrity; senses, imagination, and thought; emotions; practical
reason; control over one’s environment; and affiliation). Because a conception of the
person as exercising various human capabilities is one that “corresponds to human
experience,” Nussbaum argues that “there is good reason to think that it can command a
political consensus in a pluralistic society” and therefore “form the core of a political
conception that is a form of political liberalism” (Nussbaum 2000, p. 56).
Dictionary: NOSD
conception of the good. If positive health outcomes are indeed a good that
can be affirmed without violating neutrality, then we could invoke that good
to build a case for interventions that compensate caregivers for health deficits,
or that arrange paid and unpaid labor environments so that the deficits do
not occur in the first place. More generally, if the GDL causes an unfair
distribution of a good that can be recognized as such regardless of one’s
conception of the good, then that good might serve as a basis on which to
build a neutral justification for gender egalitarian political interventions.
2. Is the Mal-Distribution Strategy Neutral?
Many mal-distribution strategists are careful to argue that the goods they
allege to be unfairly distributed are goods the value of which can be affirmed
without violating the constraint of neutrality. In this section, I argue that we
should be skeptical of such assurances. I argue, further, that even if we find
a neutrally affirmable good, we should remain skeptical about the mal-
distribution strategy. Identifying a political good does not suffice to show
that the mal-distribution strategy can be executed fully without violating
neutrality, and the other neutrality-based barriers to the enactment of this
strategy are still more difficult to surmount.
I do not argue definitively that it is impossible for a mal-distribution
strategy to constitute a neutral defense of gender egalitarian interventions;
rather, I lay out what I take to be powerful reasons for pessimism. And I do
not claim that a defense of gender egalitarian interventions must be neutral;
rather, I argue conditionally that if that constraint applies—a constraint
commonly accepted by liberals who support gender egalitarian interventions
and those who oppose them—then the burden it generates constitutes
reason for skepticism about the prospects for successfully deploying the
mal-distribution strategy.
First, while general goods such as positive health outcomes may indeed be
affirmed from within any conception of the good, this does not yet establish
that the particular health outcomes of which the GDL causes mal-
distributions are so affirmable. It may be that the particular health outcomes
of which women enjoy a less favorable share are outcomes the value of
which cannot be affirmed from within all reasonable conceptions of the
good, even though health more generally can be so affirmed. This might be
the case, for example, if women’s overall inferior share of health outcomes is
caused in part by the pain associated with giving birth—a health deficit that
some reasonable citizens would not recognize as a harm.
This would constitute no problem for the mal-distribution strategy if
inequalities in the particular health outcomes resulted in an inequality in
Dictionary: NOSD
194 Gina Schouten
the political good of positive health outcomes generally. But it may be that
other differences between men and women have the consequence of
rendering women healthier than men, thereby offsetting the health costs of
the GDL and preempting the purported claim to redistribution. This may
be the case, for example, if women’s greater longevity counterbalances
whatever health deficits they face in other domains. If the particular goods
of which women enjoy a less favorable share cannot be neutrally affirmed,
and if the health deficits caused by the GDL are outweighed by other
differences such that women are not sub-equally healthy, all things
considered, then the mal-distribution strategy will not have established that
a legitimately politically actionable distributional unfairness exists.14 To
justify political remediation, it does not suffice to show that good health
generally is a politically valuable good, and that some particular health
outcomes are unequally distributed. The mal-distribution strategist must
establish that some particular health outcomes are politically valuable and
unequally distributed between women and men, or that the general good of
positive health is, all things considered, unequally distributed between
women and men. Neither of these possibilities has, to the best of my
knowledge, been executed.
Assume that this burden can be met. A second hurdle confronts the mal-
distribution strategy: in order to pursue it successfully, we must establish
not only that the goods to be redistributed are neutral, we also must show
that the value of the distributional paradigm can be affirmed from within
any conception of the good. If we think women’s share of the relevant goods
is objectionable because it is sub-equal, for example, then we must establish
the value of equality without violating the constraint of neutrality. I am
optimistic that a fairly demanding egalitarian distributional paradigm can
form part of a neutral conception of justice.15 But I am less optimistic that
a neutral good falls within the jurisdiction of a neutral distributive principle
in such a way that the principle actually registers the purported mal-
distribution of the GDL.
14
Another problem lurks here. See Hartley and Watson 2010 for an argument that the
goods invoked by mal-distribution strategists are not goods whose value is established by
the commitments of political liberalism as such, but are, at best, goods recognizable as
such from within some political conceptions of justice. If Hartley and Watson are right,
then the legitimacy of the interventions will depend on the outcomes of political
processes rather than on straightforward considerations of political legitimacy.
15
Lloyd 1998 and Baehr 2008 argue for equality as a political value; Nussbaum 2000
seems to assume that some fairly egalitarian distributional paradigm can be defended on
political grounds. Gheaus grounds her theory of gender justice in “a robust understanding
of equality, according to which it is . . . unjust for some people to be worse off than others
on account of their sex” (Gheaus 2012, 12).
Dictionary: NOSD
To illustrate the difficulty, consider another good purportedly mal-

distributed on account of the GDL: opportunities for meaningful lives
(Lloyd 1998; Baehr 2008). Assume that the value of equal opportunity can
be affirmed neutrally, and can thus legitimately be invoked to justify social
policies within a political conception of justice. Nevertheless, that conception
may not register the particular ways in which (on average) men’s opportunities
differ from women’s. It is implausible to think that one’s opportunity set is
always more valuable insofar as she has numerically more opportunities
available to her; the relevant inequality, therefore, must be qualitative. The
mal-distribution strategist must show that women’s opportunity sets are, on
average, less valuable than men’s opportunity sets. And the metric we deploy
in assessing the relative value of opportunity sets must itself be comprehensible
in terms of political values. What metric could render the verdict that
women’s opportunity sets are sub-equally valuable without running afoul of
neutrality? It is plausible that women enjoy fewer opportunities, on average,
for leisure or for satisfaction and achievement in the world of paid work or
in politics. But that difference renders their opportunity sets less valuable, all
things considered, only if the value lost due to their relatively small share of
these opportunities isn’t outweighed by the value gained from the
opportunities of which women enjoy a more favorable share: opportunities
to care for dependents, for example, and to form the bonds of intimacy that
caring work generates. Without arguing that enactments of these opportunities
are unequally valuable—which the neutrality constraint rules out—I do not
see how we can show that the opportunities are themselves unequally
valuable, or that the differences render the overall opportunity sets unequally
valuable.
To be sure, women lack certain opportunities to arrange their lives
according to their own judgments regarding the choice-worthiness of the
activities available. Women’s opportunity sets are less valuable because social
norms and institutions reduce their opportunity to combine caregiving and
paid labor, or to prioritize paid work to the exclusion of caregiving, or to
prioritize leisure over other pursuits. These constraints are unfair. But
because they do not constitute mal-distributions, the mal-distribution
strategy cannot diagnose their unfairness. The constraints do not constitute
mal-distributions for the simple reason that men face corresponding
constraints on their opportunities to arrange their lives as they choose. Just
as women are constrained in prioritizing paid work over caring work, men
are constrained in prioritizing caring work over paid work, and both are
equally constrained in combining paid and caring work. While women’s
opportunity sets are differently constrained, it is far from obvious that they
are more constrained. If the constraints on women’s opportunities were a
mal-distribution, then there would have to be an identifiable group who
Dictionary: NOSD
196 Gina Schouten
faces too few constraints while women face too many. Unlike problems of
scarcity, problems of distribution arise among parties—some have too little
of something, while others’ share is unfairly large. Both women and men are
unfairly constrained by the GDL, and so the distribution of constraints
between them cannot help us diagnose the unfairness of the GDL. This is
not to say that we shouldn’t be especially concerned about the constraints
women face. It is to say, rather, that the mal-distribution strategy does not
explain why we should be.
A second reason not to regard the constraints on women’s opportunities
as a mal-distribution is that lifestyle choices cannot all be made equally
accessible. As we have seen, political institutions inevitably influence the
values of those living under them. As Rawls puts it:
it is surely impossible for the basic structure of a just constitutional regime not to
have important effects and influences as to which comprehensive doctrines endure
and gain adherents over time; and it is futile to try to counteract these effects and
influences .
.
.
We must accept the facts of commonsense political sociology
(Rawls 1993, 193).
There is no neutral state under which all reasonable conceptions of the good
are equally accessible and equally costly; thus, we cannot rest the case for
gender egalitarian interventions on the mere fact that gender egalitarian
lifestyles, wherein partners engage fully in both paid work and caregiving,
are costlier than traditional gendered lifestyles under the status quo. We
must offer some neutral reason to make gender egalitarian lifestyles less
costly. The unequal costliness does not itself constitute such a reason, as
different choice-worthy lifestyles will, inevitably, be unequally costly.
There is a final complication worth noting. Under some circumstances,
inequalities arising from individuals’ own voluntary choices—even
inequalities in their holdings of neutrally affirmable goods—must be taken
as immune to considerations of distributive justice. To be sure, many
theorists reasonably question the degree to which gender inegalitarian
choices are voluntary (Walker 1995; Nussbaum 2000; Okin 2004; Hartley
and Watson 2009; Gheaus 2012; Kernohan 1998; Williams 1989;
Cudd 2004).16 If families choose to comply with the GDL because of
internalized gender norms or because institutional constraints render more
egalitarian alternatives very costly, then those choices may more
appropriately be attributed to background conditions than to genuine
choices. Responsibility attribution is profoundly difficult. But within a
16
For a defense of the claim that liberals must respect even adaptive preferences, see
Levey 2005.
Dictionary: NOSD
political conception of justice, neither internalized social norms nor the

high cost of alternatives can categorically excuse individuals from bearing
the costs of their choices. Classifying social norms or costly alternatives as
voluntariness-undermining mechanisms requires a very robust standard of
voluntariness that cannot be justified without violating the constraint of
neutrality.
To illustrate, consider how such a standard of voluntariness would
adjudicate an agent’s choice to take on a particular religious commitment.
This is a type of choice that we think social institutions ought to regard as
voluntary, even in cases where the choice is responsive to deeply internalized
social norms, or to the difficulty of opting against that religious commitment;
it is a choice the costs of which we rightfully expect individuals to internalize.
Insofar as GDL-compliant choices are attributable to background conditions
in the same way as religious commitments often are, we cannot categorically
excuse individuals from the responsibility to bear the costs of those choices.
A notion of voluntariness suitable for a political conception of justice must
be one that attributes to individuals the capacity voluntarily to make both
very costly choices and choices that comply with social norms.
I have said that the kind of non-voluntariness that applies to GDL-
compliant choices cannot categorically excuse individuals from bearing the
costs of their choices. There may be other reasons that tell in favor of
making gender egalitarian choices less costly.17 For example, GDL-
compliant choices might presently be attributable in large part to unfairness
in the institutional environment that constrains individual choices. But
unless the justificatory burdens set forth earlier can be met, that unfairness
will need to be diagnosed independently of the mal-distributions to which
it gives rise; it cannot be derived straightforwardly from the claim that the
costliness of gender egalitarianism renders GDL-compliant choices non-
voluntary.
In short, in the standard range of cases, the GDL cannot straightforwardly
be classified as non-voluntary, because such a classification would invoke a
notion of voluntariness too demanding to be permissible within a political
conception of justice. In arguing that women should be compensated for the
mal-distributions resulting from the GDL, then, the mal-distribution strategist
must overcome a presumption in favor of holding individuals responsible for
their voluntary choices. To be sure, some of the harms of the GDL are borne
17
In Schouten (forthcoming), I argue that there are. In concluding this chapter, I
briefly describe the strategy I favor for defending social policies designed to make gender
egalitarian choices less costly than they are under the status quo (and I even more briefly
explain why I think that strategy can abide by the neutrality constraint).
Dictionary: NOSD
198 Gina Schouten
by those who do not themselves choose to comply with the GDL.18 But not all
are, and this burden must be faced by mal-distribution strategists who favor
comprehensive, systemic solutions to the GDL such as work time regulation,
family leave provisions, and social support for caregiving.
These complications do not undermine the mal-distribution strategy, but
they do paint a rather daunting picture of the burden it assumes. The mal-
distribution strategist will have to argue that the goods to be redistributed
are politically valuable, that the distributional paradigm invoked is
politically valuable, and that the goods and the distributional paradigm can
be operationalized together in ways that register the mal-distributions of the
GDL. Further, she must argue that systemic remedies are called for despite
the fact that the GDL is sustained, in large part, by decisions that a political
conception of justice must classify as voluntary. Schematic descriptions of
alleged political values such as health and opportunities make the mal-
distribution strategy seem a promising way forward for those of us who seek
to justify gender egalitarian social policies. I worry that, upon closer scrutiny,
much of the promise is illusory.
3. Is the GDL a Problem of Distribution?
Daunting argumentative burdens have been met before and gender

egalitarians could continue to develop the mal-distribution strategy in
hopes of surmounting the hurdles I have drawn attention to. But a deeper,
conceptual problem lurks: it is implausible in the first place to analyze the
harms of the GDL in distributional terms.
To be sure, the problem sounds distributional. Women perform a
comparatively large share of unpaid domestic work and a smaller share of
socially remunerated work outside the home, and many women and men
would be better off if their time were allocated differently. Unequal
distributions of labor can also give rise to further unequal distributions,
such as unequal opportunities for scarce jobs and unequal pay within similar
jobs. The moral problem with these inequalities may be difficult to articulate
within the justificatory constraints considered above, but I have not argued
that they cannot be so articulated.
In this section, I try to elucidate a deeper problem for the mal-distribution
strategy. While certain unequal distributions might trigger moral concern
because they illuminate a wrong to which social institutions should respond,
18
Consider, for example, a single woman who is statistically discriminated against
because she is a woman, though she herself has no more caregiving responsibilities than
the single men with whom she is competing for employment and advancement.
Dictionary: NOSD
those distributions do not constitute that wrong. What is wrong with the
GDL is not primarily a problem of distribution. Why are inequalities
between men and women different than inequalities that we take to be
just—for example, inequalities in average income between retail managers
and (fairly paid but) struggling artists? The fact that the inequalities due to
the GDL are mal-distributions as opposed to morally neutral inequalities is
symptomatic of a conceptually prior injustice in the norms and institutional
arrangements that sustain the GDL. We can see that the mal-distributions
are not themselves what is morally objectionable about the GDL by
considering the institutional response that would be called for if they were.
An objectionable distribution can, at least in principle, be remedied by
way of compensation. If the GDL were a problem of distribution, there
would be no principled problem with addressing the injustice through
mechanisms of redistribution. Consider again the variant of the mal-
distribution strategy that diagnoses the harm of the GDL as an unfair
distribution of positive health outcomes between men and women. We
might opt to confront this unfair distribution by way of the kinds of
structural interventions that gender egalitarians have lately argued for:
interventions such as work time regulation, family leave provisions, and
social supports for caregiving. If the GDL is a cause of poor health among
women, then there is good reason to think that strategies to equalize
household labor allocations would also enhance women’s health outcomes.
But alternatively, we might opt for a strategy that more directly targets the
diagnosed distributional injustice by investing resources to enhance women’s
health outcomes. If the mal-distribution of health outcomes were the
injustice of the GDL, there would be no principled reason to prefer the
former, more comprehensive policy to the latter, more targeted policy.
This is a problematic consequence, because the policy that specifically
targets women’s health outcomes will do little, if anything, to change the
fact that women and men are systematically socialized to have gendered
preferences and aspirations and to allocate work in their domestic
partnerships according to gender. And it will do little, if anything, to change
the fact that social institutions such as labor markets assume and reinforce
gendered choices about work. Notice, too, that the problem persists whether
we think of the more targeted distributional policy as one that compensates
women for health deficits ex post, or one that arranges institutions to
prevent (some or all of ) the deficits from arising in the first place, for
example by subsidizing wellness courses designed to ameliorate the health
costs of childbearing and caregiving. Nor is the problem the practical one of
finding the right distribuenda. Rather, the problem is that redistributing
holdings of goods among men and women does not require tackling the
norms or reforming the institutions that sustain the GDL. Simply improving
Dictionary: NOSD
200 Gina Schouten
women’s health, or enhancing their share of some other good or combination

of goods, will do little to undermine the GDL.
The mal-distribution strategist might respond to this challenge in either
of two ways. First, she might argue that there are good reasons to opt for
comprehensive remedies such as work time regulation, family leave
provisions, and social supports for caregiving, even if the real injustice
resides in the mal-distribution and could thus be treated more directly. She
might point out that policies need not always tackle social problems at their
source. We sometimes treat symptoms, though the root cause is where the
real problem resides. And we sometimes endeavor to remove the root cause
of a problem, though the cause is not itself objectionable. In the spirit of
this latter set of possibilities, the mal-distribution strategist might argue that
we should adopt the more comprehensive approach to the GDL, even if the
injustice—and thus the grounds for intervention—resides in the mal-
distribution the GDL causes.
The problem with this response is that it provides no principled case for
preferring the comprehensive approach. Our diagnoses of social problems
should provide (defeasible) principled reasons to address the problems
themselves. According to the diagnosis of the GDL at the heart of the mal-
distribution strategy, the injustice resides in the downstream distributional
consequences of the GDL. These consequences can be remedied without
dismantling the GDL itself. If the distributional consequences are not what
is ultimately objectionable about the GDL, and if the mal-distribution
strategist’s diagnosis offers no principled case for tackling what is ultimately
objectionable, then it is the wrong diagnosis. The mal-distribution strategy,
which relies on that diagnosis, is therefore liable to issue misguided
prescriptions for how to remedy the GDL.
Furthermore, given the extreme difficulty of undermining the entrenched
social norms sustaining the GDL, and the relative ease of redistributing the
goods that are distributed unfairly on account of the GDL, it is hard to see
even what practical considerations could plausibly be mustered in favor of
the comprehensive policy. Thus, the case for comprehensive policies appears
to rest on the possibility of diagnosing norms and institutions that sustain
the GDL as themselves morally objectionable. The mal-distribution strategy
offers no resources for doing so.
There is a second line of response open to the mal-distribution strategist.
She might object that I have too quickly dismissed the appeal of the more
targeted distributional policies approved by that strategy. Even if
redistribution cannot solve all the problems with the GDL, improving
women’s shares of certain valuable political goods would make some
progress in mitigating the harms of the GDL.
I do not think we should be satisfied with this response. When we
intervene to remedy a mal-distribution, we do not only compensate those
Dictionary: NOSD
with sub-optimal shares. We finance that benefit by redistributing away from

those with super-optimal shares. To be sure, goods like positive health
outcomes need not always be a zero sum game, and some social policies
presumably would enhance health outcomes across the population. But
generally, a gain to one group is purchased at least partially at a cost to
another group, even if that cost is only an opportunity cost, as when
compensatory resources might otherwise have been directed to the benefit of
the advantaged group. In the case of the GDL, we might assume that women
would be the compensated group and men the advantaged group. If so, a
distributional diagnosis of the GDL suggests that we respond by benefiting
women, and that we are permitted to do so even at a cost to men. But the harms
of the GDL do not accrue uniquely to women. Evidence suggests that men
and children are also made worse off by the GDL (Glass 2004; Gornick and
Meyers 2009; Zippel 2009; Coltrane 2009; Correll et al. 2007; Breen and
Cooke 2005; Kluwer et al. 1996 and 2000; Kluwer 1998; Poortman and van
der Lippe 2009; Milkie et al. 2002; Barnett and Shen 1997). We could
constrain our policies so that we compensate for health deficits only when we
can enhance the health outcomes across the population.19 But on a
distributional diagnosis of the GDL, there is no principled reason to impose
that constraint. The distributional diagnosis implies that there are some
individuals on whom, at least in principle, we may justifiably impose a cost
in order to finance a benefit to some others. Because women and men are
harmed by the GDL, there seems to be no such group.
Health outcomes are only one purportedly political good, and gender
egalitarians have executed mal-distribution strategies that invoke others. But
it is far from obvious that any of the candidate goods could be redistributed
from men to women—or among any set of the relevant parties—in a way
that would remedy the harms that befall one group without imposing
unfairly on another group. Consider another prima facie plausible candidate
for the unjustly distributed good: valuable opportunities for good or
meaningful lives. The problem for a mal-distribution strategy that takes
opportunities as the good to be redistributed is that, if women’s opportunity
sets are compromised by the GDL, then there is a good case to be made that
men’s are too. Just as many women would have benefited from increased paid
labor opportunities, many men would be better off if they had opportunities
to develop the kinds of intimacy with their children, spouses, and aging
parents that caring labor enables. It is no solution to the problem, then, to
constrain men’s opportunities (or to divert resources that might otherwise be
spent enhancing them) so that women’s opportunity sets may be improved.
19
One might think we could settle for a weaker constraint, permitting us to compensate
only when doing so does not harm anyone else. But this constraint is too weak, given the
opportunity costs to one group when resources are used to benefit another group.
Dictionary: NOSD
202 Gina Schouten
Perhaps there is yet some good whose distribution is affected by the GDL
and whose redistribution would remedy the harms of the GDL. But when
we consider the most plausible candidate goods, it seems that redistribution
does not offer a genuine remedy. The harms of the GDL do not uniquely
befall members of some identifiable social group that we can benefit at a
cost to other groups. On any plausible redistributional scheme, many of
those who would bear the cost of the benefit are themselves made worse off
by the GDL.
Even independently of the remedies the mal-distribution strategy calls
for, the distributional diagnosis of the GDL rings false. What is unjust
about the GDL is not, in the first place, its distributional consequences.
Some of those consequences—such as, plausibly, the inequality in post-
divorce poverty rates between men and women—are unjust independently
of their relation to gender norms, and adequate general principles of justice
will identify them as such. Other distributional consequences of the GDL
are morally objectionable only because of problems with the broader
structural context in which they arise. This suggests that the fundamental
injustice of the GDL resides not in its distributional consequences, but
rather in the social backdrop against which individuals make decisions
about what paid and caring labor commitments to undertake: the GDL is
unjust because the individual choices that sustain it are constrained in a way
that would be unjust no matter the distributional consequences. Gender
norms and gendered social institutions explain the injustice in the resultant
distribution; not the other way around.
My hunch is that the distributional diagnosis of the GDL has been so
seductive because of its apparent promise of justifying gender egalitarian
policies without making any judgments as to the relative value of gender
egalitarian and gender inegalitarian lifestyles. By locating the problem in
the distributional consequences of a certain set of choices, we might think,
we can avoid making value judgments about the choices themselves, and
thus avoid running afoul of the neutrality constraint. I have argued that the
appeal of this strategy is illusory. On the other hand, I think we should not
yet give up on the possibility of affirming the value of gender egalitarianism
itself while still abiding by the neutrality constraint and acknowledging that
individuals should retain the option to choose a traditional GDL.
While I endeavor to develop this alternative diagnosis of the GDL
elsewhere,20 I limit myself here to these few brief remarks. The diagnosis I
favor locates the injustice of the GDL within the reciprocal interaction of
gendered social norms and social institutions built on those norms—an
interaction that constrains choices regarding the allocation of paid and
20
Schouten (forthcoming).
Dictionary: NOSD
caring labor in a way that is objectionable from the perspective of citizenship.

On the basis of a neutral conception of citizenship, I argue that there is a
citizenship interest in preserving a robust opportunity to choose gender
egalitarianism. Because our status as citizens is something that we all
share—something that is not unique to any particular value system—the
interests of citizenship constitute neutral, political reasons. If this is right,
then when the robust opportunity for gender egalitarianism is obstructed,
political power can legitimately be used to make gender egalitarian lifestyles
more accessible and less costly by subsidizing those lifestyles.21
My project here has been to defend a modest, critical conclusion:
distribution does not account for what is fundamentally wrong with the
GDL, and redistribution cannot fix it. The distribution of opportunities,
health, and other valuable outcomes that result from the GDL is unjust—
when it is unjust—because it results from an unjust set of social constraints.
The mal-distribution strategy misplaces this injustice, and issues
correspondingly misguided prescriptions for how to fix it.
Bibliography
Ackerman, Bruce (1980). Social Justice in the Liberal State (New Haven: Yale
University Press).
Ackerman, Bruce (1994). “Political Liberalisms,” Journal of Philosophy XCI: 364–86.
Baehr, Amy R. (2008). “Perfectionism, Feminism and Public Reason,” Law and
21
It might be helpful to consider the dissimilarity between two kinds of societies: first,
a society (characterized by a GDL) that abstains from enacting gender egalitarian
interventions; second, a society that abstains from subsidizing expensive religious
commitments. In both societies, a certain lifestyle choice is costlier than its would-be
practitioners would like. In the first society, the costly lifestyle is gender egalitarianism; in
the second, the costly lifestyle is the practice of the expensive religious commitments. In
neither society can interventions legitimately be enacted to lessen the cost of the costly
lifestyle unless that lifestyle can be shown to be politically valuable—that is, valuable
without reliance on any particular comprehensive conception of the good. In both cases,
the prospects for establishing the value of the costly lifestyle using only political values
look bleak. My proposal, which I merely allude to here, is that under certain social
circumstances—circumstances that in fact obtain in the contemporary United States—
gender egalitarian lifestyles are politically valuable, despite appearances to the contrary.
Whether this defense of gender egalitarian interventions abides by the constraints of
neutrality depends on whether it can indeed be established that gender egalitarianism is
politically valuable. If it cannot—and if the alternative strategies for grounding such
interventions either rely on implausible diagnoses of the injustice of the GDL or also fail
to abide by the constraint of neutrality—then that strikes me as a good reason for
rejecting that constraint and opting for a more permissive liberalism.
Dictionary: NOSD
204 Gina Schouten
Barnett, R. and Shen, Y. (1997). “Gender, High- and Low-Schedule-Control

Housework Tasks, and Psychological Distress: A Study of Dual-Earner Couples,”
Journal of Family Issues 18, 4: 403–28.
Bianchi, S. M., Milkie, M., Sayer, L., and Robinson, J. (2000). “Is Anyone Doing
the Housework? Trends in the Gender Division of Household Labor,” Social
Forces 79, 1: 191–234.
Breen, R. and Cooke, L. P. (2005). “The Persistence of the Gendered Division of
Domestic Labour,” European Sociological Review 21, 1: 43.
Brighouse, Harry and Wright, Erik Olin (2009). “Strong Gender Egalitarianism,” in
E. O. Wright (ed.), Real Utopias: Institutions for Gender Egalitarianism: Creating
the Conditions for Egalitarian Dual Earner/Dual Caregiver Families (Madison:
University of Wisconsin Press).
Ciscel, D., Sharp, D. C., and Heath, J. A. (2000). “Family Work Trends and
Practices: 1971–1991,” Journal of Family and Economic Issues 21: 23–36.
Coltrane, S. (2000). “Research on Household Labor: Modeling and Measuring the
Social Embeddedness of Routine Family Work,” Journal of Marriage and Family
62, 4: 1208–33.
Coltrane, S. (2009). “Fatherhood, Gender and Work-Family Policies,” in E. O. Wright
(ed.) Real Utopias: Institutions for Gender Egalitarianism: Creating the Conditions
for Egalitarian Dual Earner/Dual Caregiver Families (Madison: University of
Wisconsin Press).
Correll, S. J., Benard, S., and Paik, I. (2007). “Getting a Job: Is There a Motherhood
Penalty?” American Journal of Sociology 112: 1297–338.
Crompton, R. (2009). “The Normative and Institutional Embeddedness of Parental
Employment: Its Impact on Gender Egalitarianism in Parenthood and
Employment,” in E. O. Wright (ed.), Real Utopias: Institutions for Gender
Egalitarianism: Creating the Conditions for Egalitarian Dual Earner/Dual Caregiver
Families (Madison: University of Wisconsin Press).
Cudd, Ann (2004). “The Paradox of Liberalism Feminism: Preference, Rationality,
and Oppression,” in A. Baehr (ed.), Varieties of Feminist Liberalism (Lanham:
Rowman and Littlefield Publishers), 37–61.
Estes, S., Noonan, M., and Maume, D. (2007). “Is Work-Family Policy Use Related
to the Gendered Division of Housework?” Journal of Family and Economic Issues
28, 4: 527–45.
Gershuny, J. and Sullivan, O. (2003). “Time Use, Gender, and Public Policy
Regimes,” Social Politics, 10, 2: 205–28.
Gheaus, Anca (2012). “Gender Justice,” Journal of Ethics and Social Philosophy 6, 2: 1–24.
Glass, J. (2004). “Blessing or Curse? Work-Family Policies and Mothers’ Wage
Growth Over Time,” Work and Occupations 31: 367–94.
Gornick, J. and Meyers, M. (2009a). “Institutions that Support Gender Equality in
Parenthood and Employment,” in E. O. Wright (ed.), Real Utopias: Institutions
for Gender Egalitarianism: Creating the Conditions for Egalitarian Dual Earner/
Dual Caregiver Families (Madison: University of Wisconsin Press).
Gornick, J. and Meyers, M. (2009b). “Further Thoughts,” in E. O. Wright (ed.), Real
Utopias: Institutions for Gender Egalitarianism: Creating the Conditions for Egalitarian
Dual Earner/Dual Caregiver Families (Madison: University of Wisconsin Press).
Dictionary: NOSD
Hartley, Christie and Watson, Lori (2009). “Feminism, Religion, and Shared
Reasons: A Defense of Exclusive Public Reason,” Law and Philosophy 28: 493–536.
Hartley, Christie and Watson, Lori (2010). “Is a Feminist Political Liberalism
Possible?” Journal of Ethics and Social Philosophy 5: 1–21.
Kernohan, Andrew (1998). Liberalism, Equality, and Cultural Oppression (Cambridge:
Klumb, P., Hoppmann, C., and Staats, M. (2006). “Division of Labor in German
Dual-Earner Families: Testing Equity Theoretical Hypotheses,” Journal of
Marriage and Family 68, 4: 870–82.
Kluwer, E. (1998). “Responses to Gender Inequality in the Division of Family Work:
The Status Quo Effect,” Social Justice Research 11, 3: 337–57.
Kluwer, E., Heesink, J., and van de Vliert, E. (1996). “Marital Conflict about the
Division of Household Labor and Paid Work,” Journal of Marriage and Family 58,
4: 958–69.
Kluwer, E., Heesink, J., and van de Vliert, E. (2000). “The Division of Labor in
Close Relationships: An Asymmetrical Conflict Issue,” Personal Relationships 7,
3: 263–82.
Larmore, C. (1987). Patterns of Moral Complexity (Cambridge: Cambridge University
Press).
Larmore, C. (1996). The Morals of Modernity (Cambridge: Cambridge University
Press).
Levey, Ann (2005). “Liberalism, Adaptive Preferences, and Gender Equality,”
Hypatia 20, 4: 127–43.
Lloyd, S. A. (1998). “Toward a Liberal Theory of Sexual Equality,” Journal of
Contemporary Legal Issues 9: 203–24.
Milkie, M., Bianchi, S., Mattingly, M., and Robinson, J. (2002). “Gendered
Division of Childrearing: Ideals, Realities, and the Relationship to Parental Well-
Being,” Sex Roles 47, 1/2: 21–38.
Moon, J. D. (1993). Constructing Community: Moral Pluralism and Tragic Conflicts
(Princeton: Princeton University Press).
Nussbaum, Martha (1999). “The Feminist Critique of Liberalism,” in M. Nussbaum,
Sex and Social Justice (New York: Oxford University Press), 55–80.
Nussbaum, Martha (2000). “The Future of Feminist Liberalism,” Proceedings and
Addresses of the American Philosophical Association 74: 47–79.
Okin, Susan Moller (1989). “Reason and Feeling in Thinking about Justice,” Ethics
99, 2: 229–49.
Okin, Susan Moller (2004). “Justice and Gender: An Unfinished Debate,” Fordham
Law Review 72: 1537–67.
Poortman, A. and van der Lippe, T. (2009). “Attitudes toward Housework and
Child Care and the Gendered Division of Labor,” Journal of Marriage and Family
71, 3: 526–41.
Rawls, John (1999a). A Theory of Justice, rev. ed. (Cambridge, MA: Harvard
University Press).
Rawls, John (1999b). “The Idea of Public Reason Revisited,” in Samuel Freeman (ed.)
Collected Papers (Cambridge, MA: Harvard University Press), 573–615.
Dictionary: NOSD
206 Gina Schouten
Robeyns, Ingrid (2007). “When Will Society Be Gender Just?” in Jude Browne
(ed.), The Future of Gender (Cambridge: Cambridge University Press).
Sayer, L. C. (2005). “Gender, Time, and Inequality: Trends in Women’s and Men’s
Paid Work, Unpaid Work, and Free Time,” Social Forces 84, 1: 285–303.
Schouten, Gina (forthcoming). “Citizenship, Reciprocity, and the Gendered
Division of Labor: A Stability Argument for Gender Egalitarian Political
Interventions,” Politics, Philosophy, and Economics.
Walker, John D. (1995). “Liberalism, Consent, and the Problem of Adaptive
Preferences,” Social Theory and Practice 21, 3: 457–71.
Williams, Joan (1989). “Deconstructing Gender,” Michigan Law Review 87: 797–845.
Williams, Joan (2000). Unbending Gender: Why Family and Work Conflict and What
to Do about It (Oxford: Oxford University Press).
Zippel, K. (2009). “The Missing Link for Promoting Gender Equality: Work-
Family and Anti-Discrimination Policies,” in E. O. Wright (ed.), Real Utopias:
Institutions for Gender Egalitarianism: Creating the Conditions for Egalitarian Dual
Earner/Dual Caregiver Families (Madison: University of Wisconsin Press).
Dictionary: NOSD
Pa rt I I I
Issues in Social
R e l at i o n s
Dictionary: NOSD
Dictionary: NOSD
8
Terms of Trust
Daniel Attas
The importance of trust has never gone unnoticed. A climate of trust makes
life simpler, safer, and generally more pleasant. It is conducive to enormous
gains such as social capital1 and other benefits of cooperation. Though one
can sensibly talk of trust in oneself, in God, or in inanimate objects (such as
the weather), here I shall focus solely on interpersonal trust. Even more
specifically I shall focus on trust in situations of fundamentally opposing
interests, such as in business transactions, industrial relations, and conflict
resolution. These are situations that, presumably, without some degree of
trust, will fail to advance. More specifically still, I shall consider war and the
achievement of peace through a series of reciprocal steps as the primary
setting in which to examine trust, its terms, and its limits.
The primary purpose of this chapter is to work out some practical
implications concerning the terms of trust and how these may be promoted.
A surprising point I aim to show is that, in the attempt to resolve situations
that are fundamentally conflictive, there is often a tendency to take steps
to promote something else, superficially resembling trust, sometimes
counterproductively at the expense trust.
The structure of the chapter is as follows: Section 1 offers an analysis of
trust consisting of three elements and a presentation of the problem of trust
as the idea of overcoming the “trust gap.” Section 2 explains how overcoming
the gap may be achieved either by narrowing the gap or by bridging the gap.
I suggest that only the latter is strictly a case of trust. I then proceed to
specify three ways in which the gap may be bridged, corresponding to the
three elements of trust. Section 3 pulls all the strings together to explain the
possibility of trust based on an idea of reasonableness, rather than
instrumental rationality, sentiment, or faith. Finally, Section 4 explains how
1
Among the most influential see Putnam (1994) and Fukuyama (1996). For a differing
view see Hardin (2002).
Dictionary: NOSD
210 Daniel Attas
narrowing the gap might detrimentally affect bridging the gap, and tentatively
proposes some policy implications.2
1. The Concept and Problem of Trust
Trust is a complex phenomenon. Cognitive, expressive, and active elements

must all come together for a full account of the concept. I offer the following
definition of trust and believe this definition captures the most important
aspects of the most central uses of the concept of trust. I shall not, however,
present a full defense of this definition, though I consider some of the
discussion throughout this chapter to support its plausibility.
P (in doing A3) trusts R (to do B) if and only if:
(1) Vulnerability: Doing A expresses P’s inclination to put himself in a
vulnerable position with respect to R.
In other words, if A is done, this would put P in a vulnerable position
with respect to R. “Inclined” is used here not in the sense of a general
tendency to act in similar situations in a similar manner, but in the sense
of a willingness to act in a concrete situation. Thus, P is prepared to do A,
in full awareness that A makes him vulnerable. In so far as trust is expressed,
it is the actual doing of A that expresses, and thus realizes the attitude
of trust.
There is always an element of risk involved in cases of trust, since the
vulnerability may be exploited or negligently ignored.4 But it is important
to stress here, and will become plain shortly, that this is not a simple
calculated risk whereby taking a chance is the best way to optimize one’s
interests. In circumstances where trust is required or hoped for, it is often
the case that the probability of a trustworthy response can’t be assessed prior
to the act of trust. For trustworthiness, I take it, is primarily a property of
2
I would like to thank participants at the annual Workshop for Oxford Studies in
Political Philosophy (OSPP) and at the Jerusalem Political Philosophy Forum for helpful
discussions. I am also particularly grateful to Eran Fish, Steve Hopgood, Karen Jones,
Tom Parr, Peter Vallentyne, and an anonymous reviewer for OSPP for written comments
on earlier drafts of this chapter.
3
It might seem that the issue of trust may arise in the absence of any particular action.
I may, for example, trust that you won’t steal things from my house. But even in cases
such as these, I refrain from taking certain precautions: I give you a key to my house and
I leave you on your own for a long period of time. I take trust to be more than merely a
state of mind. It must be actualized by some action.
4
This is an aspect of trust that is perhaps the most widely accepted in the philosophical
literature. See Baier (1986); Dasgupta (1988); Becker (1996).
Dictionary: NOSD
Terms of Trust 211
the response elicited by trust and only derivatively a character trait or

disposition.5
(2) Benefit: P does A for the purpose of some gain that will befall P if R
does not exploit this vulnerability.
The gain needn’t be P’s own welfare. I may trust my daughter’s teacher to
take care of her during school hours. To the extent that my daughter’s
welfare is part of my interest, I am taking a risk and it is not only her
vulnerability but mine too. A special case would be when P does A, with the
hope that R does B, for the sake of some mutually beneficial arrangement.
These kinds of cases are of primary interest in this chapter. A commercial
exchange, a functioning workplace, or peace are presumably beneficial for
both parties to the agreement.
(3) Reciprocity: P does A with the expectation that R will be sufficiently
motivated to reciprocate6 (i.e. to do B) at least partly due to the fact that P’s
behavior is informed by such an expectation.7
To say that R will be sufficiently motivated doesn’t mean that R will
actually behave as expected. It means that P believes that to the extent
that it is within R’s power, R will do B. R has to be sufficiently motivated,
but not just any motivation will do. For example, she may be motivated
by her expected greater gains, by the thought that she might be rewarded
for her efforts, or by the threat of punishment if she fails, and so forth. If
this is the kind of motivation that P attributes to R then it would not be
accurate to say that P trusts R, but that he is confident, or has sufficient
assurance that R will behave as expected. R could possibly have these
motivations too, but they would not be the basis of P’s trust, that is to
say, it is not such expectations that would inform P’s trust in R. For trust
5
Of course we could say of one who tends to act trustworthily that he is trustworthy—
but this could be understood either as a generalization over his past actions, or referring
to something deeper in his dispositions that explain why we can expect him to act in a
trustworthy manner in a particular situation. Neither of these is very helpful in the case
of a concrete case of trust: when we want to know whether to trust a person, we want to
know whether in this particular situation he is likely to respond appropriately, and this
depends, among other things, on how responsive he is in general to (moral) reasons—
which is much wider than merely being trustworthy—but also (importantly) on how
likely he is to appreciate the context and the fact that he is being trusted, which goes way
beyond any character trait since it appeals primarily to particular circumstances of how
competently the fact of trust is communicated by the truster, and the extent to which
conditions are favorable to an adequate reception of the message. For a differing view see,
for example, Jones (2012: 72).
6
“Reciprocity” may suggest responding to trust with trust, whereas here trust is being
responded to by the performance of that which one was being trusted to do. Perhaps
“respond appropriately” would be more accurate. Nevertheless, with this clarification,
I shall stick to the term “reciprocity.”
7
See Jones (1996).
Dictionary: NOSD
212 Daniel Attas
P Trust R Reciprocate
Cooperation
U(tt), V(tt)
(peace)
Distrust Defect
U(dd), V(dd) U(td), V(td)

Status quo Mistrust
(conflict)
Fig. 8.1 Trust game
to exist, P must believe R to be motivated by P’s actual behavior (doing

A). Yet it is not simply the fact that P behaved as he did that should
motivate R (P may have been foolish, inattentive, and so forth) but that
he did this with the expectation that R would reciprocate. I expect
you to do B because I did A with the expectation that you do B. Such
expectation as used here is a special kind of belief to which I shall return
later on.
The problem of trust that I address can be best explicated by the following
interactive decision situation (known in the experimental literature as
the “trust game”). It is, of course, an abstract, idealized representation
and therefore subject to all the limitations and qualifications typical of
such models. In this model, one party (P) has to make a concession to
which the other party (R) agrees to reciprocate8 (see Figure 8.1). So P’s
choice is between trust and distrust. If P chooses to distrust, believing
perhaps that R will fail to reciprocate, the benefits to P and R are U(dd)
and V(dd) respectively, and the interaction ends in status quo (conflict).9
If P chooses to trust, the trustee, R, has to choose between reciprocating
that trust and defecting. Reciprocating ends the interaction in cooperation
(peace, or at least one stage up towards peace where the parties face a
more advanced similar decision situation) with benefits U(tt) and V(tt).
Defecting ends the interactive decision situation in mistrust with benefits
U(td) and V(td).10
8
Though the discussion is supposed to be general it may be helpful to think of a
concrete case. Specifically, I am thinking of the Israeli–Palestinian never-ending peace
process. So, say P has to dismantle settlements and withdraw military presence from
certain territories and R is expected to reciprocate by curbing violent attacks against P.
Some readers may find such allusions distracting or confusing, and I definitely do not
mean to make any empirical claims about this or any other state of affairs, so all such
references shall be relegated to footnotes.
9
In our concrete example: a continuous cycle of violence.
10
Further violence, taking advantage of P’s now more vulnerable situation.
Dictionary: NOSD
Terms of Trust 213
The problem, what makes this a case that necessitates trust, follows from
the perceived magnitudes of the benefits to both parties to the trust
situation. Call the following pair of inequalities the Trust Gap:
U(tt) > U(dd) > U(td)

V(td) > V(tt) > V(dd)
Now, since R’s expected benefit from defecting V(td) is greater than the
cooperative outcome V(tt), R’s decision to defect dominates the decision to
reciprocate; and, given this, since P’s expected benefit from status quo U(dd)
is greater than the mistrust outcome U(td), distrust dominates his decision
to trust. In other words, on prudential or self-interested considerations
alone cooperation (peace) cannot be attained.
On the other hand, trust enables the parties to reach a situation that is
manifestly preferable to both than the continuing state of conflict. For the
benefits of cooperation which trust enables are greater for both parties, and
are so perceived, than the expected status quo outcome. This suggests that
trust in a process of conflict resolution characterized by a series of such
decision situations is something we would want to promote and nurture for
the process to get off the ground. Nevertheless, with only self-interest to
guide our predictions and prescriptions, such an endeavor seems doomed.
So the problem is this: given the trust gap, and the rational-prudential
basis to distrust—what could, after all, enable trust? And were there a basis
for (rational) trust, why is trust in such situations merely possible and not
necessary? If trust is rationally justified11 why is trust not always shown in
these circumstances? One answer could be based on irrationality. To
be sure, human beings are not always rational. Our behavior might be
emotionally biased or even capricious. If that means we need to count on
arbitrary factors to explain trust, we could not hope to predict, facilitate,
or encourage trust.
The explanation of the problem of trust I will propose will be based
instead on diverging conceptions of fairness by the parties to the trust
situation, on their differing perceptions of the good, and primarily on their
incongruent appreciation of context. There need be no facts of the matter
that the parties are unaware of to explain why trust fails to materialize. In
particular, the parties may be fully aware of the payoffs that constitute the
trust gap and the full benefits of cooperation. Furthermore, they need not
be governed in their choices by emotions or other biases, neither by a
suspicious nature, nor by blind faith. It is merely their divergent perceived
views that may explain when trust is more likely or less so to ensue.
11
As in Hardin-type analyses of encapsulated interest (Hardin 2002).
Dictionary: NOSD
214 Daniel Attas
2. Overcoming the Gap
So the problem of trust is the problem of overcoming the trust gap. There
are two ways this can be done. Metaphorically speaking, we can narrow the
gap, making the inequalities less alarming or worrisome, even if not
eliminating them all together. Thus, narrowing the gap can encourage
cooperation on the basis of self-regarding prudential reasons, at least in the
sense that they diminish the pull against cooperation, and make the player
more willing to take risks. The second way of overcoming the trust gap is by
bridging the gap. That is to say, leaving the inequalities as they are while
procuring non-prudential resources to encourage cooperation. Only the
latter can be truly classified as trust.
First, it would be helpful to examine the prudential resources that may
narrow the gap and thus might encourage cooperation.
The first determinant of the gap that might inhibit trust is P’s risk of
mistrust, how much P is likely to lose in case of mistrust. The measure of this
is the difference between his mistrust benefits U(td) and his status quo
benefits U(dd). The smaller such a potential loss, the more likely P is to
place himself willingly in a vulnerable position with the expectation of some
gain later on. Thus minimizing the difference, either by reducing the value
U(dd) of status quo for P or by increasing the value U(td) of mistrust, other
things being equal, would be one way to narrow the gap, motivate P, and
encourage cooperation. In particular, we assume that, though values have
been altered, the probability (whatever it may be) of R reciprocating remains
unchanged. After all, the values of mistrust and cooperation for R have not
changed.12
Another way to affect the inequalities and to motivate P is by increasing
his gains of cooperation; that is to say, by enlarging the difference between
the value U(tt) of cooperation for P and his status quo benefits U(dd).
Either by reducing the value U(dd) or increasing U(tt), and again leaving
probabilities of reciprocation unaffected, taking a risk becomes more
expedient for P in so far as, on the condition that R reciprocates, he is likely
to gain so much more.
The second determinant of the trust gap is R’s temptation to defect, how
much she is likely to gain from mistrust compared to the expected gains of
cooperation. The measure of this is the difference between her mistrust
benefits V(td) and her cooperation benefits V(tt). The greater such a
potential gain, the less likely R is to reciprocate. Knowing this, the less likely
P is to take the risk of placing himself willingly in a vulnerable position.
12
As should later become clear, this is no guarantee for R’s likelihood to reciprocate to
remain unchanged.
Dictionary: NOSD
Terms of Trust 215
Thus minimizing the difference, either by reducing the value V(td) of

mistrust for R or by increasing the value V(tt) of cooperation, other things
being equal, would be one further means by which to narrow the gap,
motivate R to reciprocate, and encourage cooperation.
This kind of thought, focusing on prudential reasons, often leads to the
proposal of certain measures to be taken by third parties to narrow the gap.
One such measure provides incentives to either party or both aimed at
boosting the gains of cooperation, increasing U(tt), V(tt), or both, and thus
encouraging P directly by making the risk worthwhile and indirectly by
diminishing R’s temptation to defect. A second measure is the imposing of
sanctions on either or both of the parties aimed at reducing the attractiveness
of non-cooperation, decreasing U(dd), V(td), or both, thus discouraging P
from choosing the status quo and R from defection. Thirdly, guarantees
provided to P in case R will fail to reciprocate, compensating for his losses
effectively increasing U(td) minimizing the risk of mistrust, thus encouraging
P. I will say more on these later when I show how so-called confidence-
building measures aimed at narrowing the gap may frustrate bridging the
gap and so undermine the possibility of trust.
But first, let me say more on trust as bridging the gap. Why shouldn’t we
rely on prudential reasons alone, narrowing the gap to the greatest extent
possible, perhaps even closing it all together? Such a reliance on self-interest
is, I believe, insufficient. It is immensely difficult to undo the inequalities
thereby dissolving the problem of trust. That is to say, closing the gap is, for
all intents and purposes, unrealistic. As long as we are merely narrowing the
gap, there will obviously still be a gap to bridge. It is therefore crucial to
inquire how trust may be facilitated after all. Even if intervention were
possible to such an extent as to close the gap completely, cooperation would
most likely be very unstable. For to the extent that it is naked interests that
guarantee continued cooperation, to that same extent the reversing into
conflict due to a change in interests will be all the more prompt.13 Even if
I am mistaken and closing the gap may be a realistic and stable option,
undeniably there will be cases where it will be neither. Then at least in these
instances (which I suspect to be the majority of cases), closing the gap will
be unrealistic or unstable, and bridging the gap will be necessary to get
cooperation off the ground.
The most important reason not to rely exclusively on prudential reasons,
as I argue in Section 4, is that narrowing the gap as far as possible, but
failing to close it entirely, often has the counterproductive effect of eroding
trust. That is to say, confidence-building measures, having the power to
13
Recall Hardin’s example from The Brothers Karamazov (Hardin 2002: 1–3).
Dictionary: NOSD
216 Daniel Attas
make the gap a lot narrower than it initially was, may nevertheless extinguish
important moral resources that are available to bridge it.
Non-prudential reasons to reciprocate trust (and therefore to trust)
follow from the analysis of the concept of trust. Three reasons may be
discerned, corresponding to the three components of the concept. In
presenting each of these reasons I shall first introduce a minor thought
experiment to tease out our moral and psychological intuitions regarding
trustworthiness, that is, from R’s point of view, with respect to certain
features of the trust situation. This intuitive response will also indicate our
preliminary inclination to act trustworthily. This thought gets further
support from psychological experiments and I will refer to some quite
conclusive evidence from the empirical literature on how people behave in
laboratory settings (even though the explanation for their behavior remains
controversial). Finally, I will explain how these features of the situation
provide a reason and, therefore, an expectation to act trustworthily, and as
such facilitate trust.
2.1 Empathy
Let me begin with a thought experiment. Consider the two trust
situations depicted below (see Figure 8.2). Imagine yourself in the
position of R, that is to say the person who after P initiates the cooperative
move, now deliberates whether to reciprocate or to defect. The numbers
represent the gains or losses to each party: their utilities, the fulfillment
of their interests, or the value in the most general sense that they attach
to each outcome. You are then in a position to choose between payoffs
(0,20) or (10,10) in Empathy (1) and between (-5,20) or (10,10) in
Empathy (2). The first thing to notice about the two situations is that
they are identical with respect to your own payoffs. In both situations
your choice is between 10 if you reciprocate, and 20 if you defect. The
self-interested choice would therefore be defection. This is precisely the
trust gap we need to bridge. The second thing to notice is that the only
difference between the two situations is P’s payoff in case you choose to
defect. In Empathy (1) it is 0, in Empathy (2) it is -5. Now, the question
you are to ask yourself is the following: in which of these two situations
are you more likely to reciprocate?
Many people feel quite strongly, as I do, that they are more likely to
reciprocate in Empathy (2). It seems that the greater loss to P in case of
defection in Empathy (2) is somehow relevant. In any case it is nothing
about our own self-interest that moves us since there is absolutely no
difference in these terms between the two situations, so nothing in terms of
our own interest can account for the perceived difference in motivation.
Dictionary: NOSD
Terms of Trust 217
10, 10 10, 10
5, 5 0, 20 5, 5 –5, 20
Empathy (1) Empathy (2)
Fig. 8.2 Empathy
Decades of experiments in behavioral economics appear to corroborate

this intuitive response. There is a growing consistent body of evidence of
altruism in experiments (Dawes and Thaler 1988; Fehr and Fischbacher 2003;
Andreoni et al. 2007). What I refer to as empathy (being motivated by the
interests of another) is sometimes called sympathy or simply altruism in this
literature. In single-trial public goods experiments, for example, rather than
strong free-riding (i.e. zero contribution), subjects contribute 40–60
percent of their stake to the public good. In multiple-trial experiments that
contribution declines drastically, but never falls all the way to zero, as the
selfish rational model predicts. To be sure, there is much controversy over
the explanation of these findings, ranging from the suggestion that subjects
misunderstand the payoffs, to the idea that acting on a norm of cooperation
affords one an evolutionary advantage. Both of these amount to reason-
bypassing explanations: the former ascribes to the subjects a failure to
appreciate the full significance of the strategic situation they are in; on the
latter view subjects are “wired” to act against their perceived self-interest in
a way that would provide them (or their society) with advantages they could
not foresee. In between these, some economists have stipulated the idea that
individuals derive utility from the well-being of others (sometimes called
pure altruism). This allows them to stick to the simple maximizing model of
rationality and to infer a corrected set of payoffs. But there is also the
possibility (which I have not encountered in this literature) that the loss of
another provides a reason for me to act without affecting, in any (non-
tautological) way, my utilities.
Whatever the explanation, let me put forward the following hypothesis
based on the thought experiment and on a suggestive body of experimental
research: the more P is to lose from mistrust, the more R is likely to
reciprocate. This is what I shall call empathy. More precisely, the more
severe P’s losses from mistrust appear to R, the more R is likely to reciprocate.
It then follows that the stronger P’s belief that R perceives his losses as
severe, the more inclined P is to trust R. Thus, cooperation is facilitated
when P can be certain that R fully appreciates P’s losses in mistrust. Not
only must R be aware of the facts of the matter, but she must also grasp the
significance of the loss to P. This can help explain why, in societies that are
Dictionary: NOSD
218 Daniel Attas
more homogeneous in terms of values, trust will be more widespread. If P

and R share a conception of the good (at least in general contours), R will
be fully appreciative of the significance of the loss to P in case she defects.
This can also explain why discussion among subjects prior to experiments
may raise the level of cooperation. Such preliminary discussion can make
the parties more aware of the losses to the trusting party and of their
significance.14
I suggested that an explanation for some non-self-interested behavior
could be that the losses that defection inflicts on P may be a reason for R to
reciprocate. Robert Goodin argues that a central moral duty we have is
protecting those that are particularly vulnerable to us:
Failing to take positive action to prevent harm from befalling someone who is
particularly vulnerable to your actions and choices is morally akin to a bodyguard
sleeping on the job (Goodin 1986: 111).
If, following Goodin, we recognize this kind of general duty to protect the
vulnerable, and that means those who are vulnerable to us in particular,
then we can see how the trust situation engenders a moral reason to
reciprocate. In trusting, P provides a moral reason for R to reciprocate
trustworthily, in so far as P puts himself in a vulnerable position with respect
to R. Thus we see how the first condition of trust (vulnerability) creates a
reason that did not exist before trust to reciprocate so as not to harm those
who are particularly vulnerable to us.15 And so in providing such a reason
for R, P also reasonably expects R to reciprocate.
2.2 Fairness
Consider next the two trust situations depicted in Figure 8.3. As in the
previous thought experiment, imagine yourself in the position of R,
deliberating whether to reciprocate or to defect. You are now in a position
to choose between payoffs (0,20) or (20,10) in Fairness (1) and between
(0,20) or (10,10) in Fairness (2). As before, the two situations are identical
with respect to your own payoffs. In both situations your choice is between
14
In such ideal lab experiments this may be simply by making concrete the existence
of an individual likely to lose or to feel insulted by defection. See Orbell at al. (1988) and
Sally (1995), for experimental evidence that discussion among subjects raises the level of
cooperation.
15
Perhaps not all vulnerabilities provide reason not to harm. Karen Jones (2012: 72)
suggests that in the case of share trading, for example, people can be vulnerable to our
actions and yet they do not expect us to respond to that vulnerability. Such cases may be
defined by regulated competition, where the possibility of one’s vulnerability being taken
advantage of is a condition one accepts when embarking on such activities. Anyway, these
are not the kind of cases discussed here.
Dictionary: NOSD
Terms of Trust 219
20, 10 10, 10
5, 5 0, 20 5, 5 0, 20
Fairness (1) Fairness (2)
Fig. 8.3 Fairness
10 if you reciprocate, and 20 if you defect. So again, the trust gap is displayed
in the fact that the self-interested choice for you would be defection. This
time the important thing to notice is that the only difference between the
two situations is P’s payoff in case you choose to reciprocate. In Fairness (1)
it is 20, in Fairness (2) it is 10. Now, the question you are to ask yourself is
the following: in which of these two situations would you be more inclined
to reciprocate?
My sense is that I would be more likely to reciprocate in Fairness (2) and
that this reflects a common response. For people like me, it seems that the
less equal distribution of benefits in case of cooperation in Fairness (1)
somehow puts them off. In any case it is nothing about their self-interest
that moves or discourages them since there is absolutely no difference in
these terms between the two situations. So here, too, nothing in terms of
their own interest can account for the perceived difference in motivation.
Sensitivity to fairness is also corroborated by experimental evidence. An
abundance of ultimatum game studies conducted over more than three
decades have demonstrated that agents are willing to forgo personal gains in
order to punish what they consider to be an unfair offer (Guth et al. 1982).
It is frequently claimed that people are concerned not only with their
own monetary gain, but also, and sometimes to a greater extent, with
the difference between their share and that of other agents (Loewenstein
et al. 1989). An aversion to inequality in the distribution of payoffs, or
“inequity aversion” as it is often termed in the literature, was shown to be
present in trust situations too. A review of past trust game results suggests
that trustees tend not to reciprocate trust when doing so would result in an
unequal distribution of payoffs. In deciding whether and how much to send
back, trustees take into account the amount kept by the “trustor,” and
choose to send an amount that would result in payoff equality. The smaller
the amount the trustor chooses to keep to herself, the more the trustee
would be willing to send back (Ciriolo 2007; Xiao and Bicchieri 2010).
Whatever the explanation of such behavior, let me put forward the
following hypothesis based on the thought experiment and on a suggestive
body of experimental research. First, if P perceives the distribution of
advantages to be unfair towards him, that is to say, he receives less than his
Dictionary: NOSD
220 Daniel Attas
fair share of the benefits of cooperation, he will be less inclined to trust.

Although the gains of cooperation may still be rewarding and attractive, the
feeling of being taken advantage of may make him reluctant to cooperate.
(As it stands, this is not quite a case of distrust, but simply one of non-
cooperation. I will soon show how this comes to bear more directly on the
issue of trust through the notion of “reciprocity.”) Second, if the cooperative
outcome is unfair towards R, this is bound to affect the likelihood of R
behaving in a trustworthy manner. In the tension between self-interest and
value-driven decision to reciprocate trust, self-interest is more likely to gain
the upper hand if the cooperative outcome is considered unfair. This is what
I shall call fairness.
More precisely, the less fair the distribution of advantages of cooperation
appear to R, the less R is likely to reciprocate. It then follows that if P
considers the outcome as unfair towards R, or if P believes that R will
perceive the outcome as unfair, this will lead P to believe that R will not
behave trustworthily, and he will be less likely to trust R. Thus, cooperation
is facilitated when P can be certain that R is satisfied that the advantages of
cooperation are fairly distributed. This can again help explain why in
societies that are more homogenous in terms of values trust will be more
widespread. If P and R share a conception of fairness, P can act on his own
conception of the fairness of the outcome, on the reasonable presumption
that R thinks the same. This can also explain why discussion among subjects
prior to experiments may raise the level of cooperation. Such preliminary
discussion can make the parties more aware of how each of them views the
cooperative outcome in terms of fairness.16
could be that the fairness of the cooperative outcome to which P contributed
may be a reason for R to reciprocate. H. L. A. Hart argues that doing our
fair share to the maintenance of a joint enterprise is a special duty we have
towards those who have done their share:
When a number of persons conduct any joint enterprise according to rules and thus
restrict their liberty, those who have submitted to these restrictions when required
have a right to a similar submission from those who have benefited by their
submission (Hart 1955: 185, my italics).
I believe that implicit in the italicized words in Hart’s formulation is an
appeal to impartiality on the basis of which benefits and burdens of the
joint enterprise are distributed. After all, such a “principle of fairness” could
16
It is plausible also to expect the opposite: the fairer the valid trust situation is
towards the trustee, the fairer it will appear to him, and the more the truster believes that
it will seem fair to the trustee, the more probable it is that trust will prevail.
Dictionary: NOSD
Terms of Trust 221
hardly be thought valid in an exploitative or oppressive enterprise. A version

of this special duty of fairness can show how trust situations engender a
moral reason to reciprocate. Once a first party has put himself at risk as a
means towards a joint enterprise that benefits both parties in a fair way, the
other party is under a moral duty to forgo the gains of mistrust, which she
could not have otherwise had, and to reciprocate.17 Moreover, in trusting, P
provides a moral reason for R to reciprocate trustworthily, in so far as P acts
for the purpose of some gain that will be fairly shared out between them.
Thus we see how the second condition of trust (benefit) creates a reason that
did not exist before trust, to reciprocate so as to uphold a beneficial
arrangement. And so in providing such a reason for R, P also reasonably
expects R to reciprocate.
2.3 Reciprocity
Lastly, consider the pair of situations depicted in Figure 8.4. Once more you
are to imagine yourself in the position of R, deliberating whether to
reciprocate or to defect. This time, the two situations are identical with
respect to the payoffs to both parties. In both situations your choice is
between (10,10) if you reciprocate, and (0,20) if you defect. So again, the
trust gap is displayed in the fact that the self-interested choice for you would
be defection. The important thing to notice is that the only difference
between the two situations is the labels attached to P’s choices. In Reciprocity
(1) P chooses between trust and distrust, in Reciprocity (2) the choice is
between gamble and play-safe. The question you are to ask yourself is the
same as before: in which of these two situations would you be more inclined
to reciprocate?
Whenever I present this question, people feel quite strongly that they are
more likely to reciprocate in Reciprocity (1). It seems that the meaning
P attaches to his choice, as a case of trust rather than a gamble, somehow
Trust Gamble
10, 10 10, 10
Play-safe
Distrust
5, 5 0, 20 5, 5 0, 20
Reciprocity (1) Reciprocity (2)
Fig. 8.4 Reciprocity
17
This is Hobbes’s third law of nature: “That men performe their Covenants made”
(Hobbes 1991: 100).
Dictionary: NOSD
222 Daniel Attas
moves them. In any case since there is absolutely no difference between the
two situations in terms of their self-interest, nothing relating to their own
interest can account for the perceived difference in motivation. Nor could it
be anything about P’s interests or about the distribution of benefits, since
the two situations are identical in these terms too.
Experimental findings are a little less conclusive here, perhaps partly
because the issue hasn’t been directly and clearly addressed.18 Nevertheless
some of the literature is instructive. To begin with, it is often assumed that
“the trustor is confronted with a risky choice when considering whether a
counterpart is trustworthy, in a manner similar to gambling or making a risky
investment” (Eckel and Wilson 2004: 463). Yet the results of a series of lab
experiments show that trust decisions and financial gambles, for example,
are not thought of as similar. Another study, attempting to determine
responsiveness to perceived acts of trust, proceeded to investigate the reasons
behind participants’ reciprocity decisions (Pillutla et al. 2003). Findings
show that participants tended to reciprocate if they took the risk taken to
indicate high levels of trust rather than, for instance, stupidity. Since high
risk may be perceived as an indicator either of stupidity or as a signal of
trust, this study proposes:
trustors who can effectively convey the logic behind their risk-taking might be able
to offset some of their risks. Trustors who send smaller amounts might also benefit
by framing their choices as acts of trust, increasing feelings of obligation, and hence,
reciprocity. More generally, trustors need to convey their viewpoint to recipients so
that they can reach a shared understanding of their actions, their relationship, and
the potential for mutual benefits (Pillutla et al. 2003: 454, my italics).
In other words, the likelihood of reciprocation depends on the risk-taking
being appreciated as an expression of trust, rather than based on some other
explanation.
Now, I would like to propose the following hypothesis based on the
thought experiment and on a limited yet suggestive body of experimental
research: First, R would not reciprocate trustworthily unless she believes
that P acted out of trust. Indeed, R could only act trustworthily in response
to a perceived act of trust. This is what I shall call reciprocity. Trustworthiness
is triggered by an initial act of trust. Even when making the cooperative
choice, whatever R may be doing when acting without attributing trust
to P’s act could not be regarded as displaying trustworthiness. Moreover,
R would be less inclined to reciprocate, the less she could be certain that
18
It is, I think, interesting that the experimental decision is framed as a “loan.” As
Eckel and Wilson justify it: “to provide context for the decision. The responses of subjects
to other abstract games made us suspect that subjects might not think of this as a situation
involving trust, per se” (2004: 451).
Dictionary: NOSD
Terms of Trust 223
P acted out of trust. Perhaps P merely took a calculated risk or gamble, as

the label suggests? Perhaps it was an act based on foolishness, an error of
judgment, or a clumsy attempt at manipulation? Or perhaps P was coerced
into acting as he did? In all such cases P’s act could not be regarded as
an act of trust, and hence could not prompt the trustworthy response
from R. Whatever R does on the basis of understandings such as these is
now to be interpreted as a strategic response to P’s calculated or irresponsible
initial act.
Of course, acts in the real world don’t come with labels attached. We take
cues from the context and try to interpret the act as best we can. It is
interesting to note here how the fairness of the cooperative outcome can
affect context appreciation, that is to say, the interpretation of P’s act as one
of trust. If it is unfair towards P, then it is less likely that P would initiate the
process, and his act would therefore most probably be interpreted as a
cognitive failure (foolishness, short-sightedness, negligence, and so on) or
coerced. More precisely, if it is perceived by R as unfair towards P, or even
more accurately, if R believes that it is perceived as unfair by P, then R
would be less likely to interpret P’s act as trusting, and thus responding
trustworthily would not be a reason to reciprocate.
Somewhat more speculatively, if the outcome is perceived by R to be
unfair towards R, then R would be less likely to reciprocate—not merely
because R would feel offended by the terms of the arrangement—but
because R would now have reason to interpret P’s act as an error of judgment
or an attempt at manipulation rather than an act of trust. Therefore, if the
unfairness of the arrangement toward R is known to P, P should not expect
R to reciprocate trustworthily and should not trust R. To the extent that P
is uncertain with respect to the fairness of the arrangement as perceived by
R, P will have less reason to trust. If P nevertheless initiates the process that
leads to the cooperative outcome it would plausibly be taken to be on
grounds other than trust. R cannot then see herself as bound by trust.
Here we can see again how more homogeneous societies or preliminary
discussion can facilitate trust. If P and R share a conception of fairness, and
know that they share such a conception, and that both judge the outcome
fair, the initial act is more likely to be interpreted as one of trust, thus
making possible the trustworthy response. But even if P and R differ on
their views of fairness, trust may still be facilitated if they manage to
communicate to each other the genuineness of their view of the terms of the
cooperative arrangement as fair to all parties. There will then be less reason
to doubt the genuineness of the trust and consequently of the trustworthy
response to which it may lead.
may be that P’s expectation that R will reciprocate, which he communicates
Dictionary: NOSD
224 Daniel Attas
to R, may provide a reason for R to reciprocate. It is natural to assume a

moral duty to respond trustworthily, but perhaps I can say a bit more to
explain the kind of duty that it is and how it comes into being. It is similar
in certain respects to gratitude or a duty of reciprocation (Feinberg 1966:
139) but also to a duty to live up to the reasonable expectations which one
has brought about (or at least made no attempt to dissolve) and on which
others relied. Such a duty is formalized, for example, in Scanlon’s Principle
of Loss Prevention:
If one has intentionally or negligently led someone to expect that one is going to
follow a certain course of action, X, and one has good reason to believe that that
person will suffer significant loss as a result of this expectation if one does not follow
X then one must take reasonable steps to prevent that loss (Scanlon 2000: 300f ).
Scanlon wants to go beyond this principle to analyze promises: seeing the

promisor as someone who (as good as) utters the words “trust me!” thus
creating the mutual knowledge of an intention to act and creating an
expectation that will allow reliance on such an intention. Similarly, I take it,
though somewhat less demanding, the utterance “I trust you!” and the non-
response to it create the mutual knowledge of reliance and intention to act
on the expectation such a reliance assumes. In the absence of special
justification such expectations create an obligation “to follow a certain
course of action,” that is to say, to live up to one’s promise, or to respond
trustworthily.
The moral duty to respond trustworthily is not quite that of keeping
promises. It is perhaps more similar to a duty of hospitality. This is the duty
to treat well the stranger in your midst. It is not merely a duty based on
need, such as a duty of beneficence or Good Samaritanism. Firstly, it is not
universal in the sense that it is not owed to anyone in need, rather it is a
duty I owe to my guest in particular. Secondly, I owe my guest much more
than needs satisfaction—I should make the stranger feel welcome. Thirdly,
my duty is triggered by the action of the guest, not by circumstances in
which he just happens to find himself. The guest is someone who voluntarily
and purposefully enters under my auspices, into my protection or patronage.
Vulnerable as he is, he comes in, relying on the good will of the host with
the expectation not to be taken advantage of. Putting himself in a position
of dependence and vulnerability is, in a sense, an act of graciousness. For in
so doing he expresses his good view of the host and his reliance on the
protective and considerate treatment he expects. The host repays this
graciousness with hospitality. I owe my guest the decency of living up to the
good opinion he has of me that he expresses in entering my abode. Moreover,
and here the duty of hospitality is further distanced from gratitude, as host
I am bound by a duty to fulfill a reasonable expectation which I led my
Dictionary: NOSD
Terms of Trust 225
guest to believe, or (at least) said or done nothing so as to not lead him to
believe. Thus, the duty of trustworthiness is one kind of duty of reciprocation,
similar in many respects to the duty of hospitality. As in the case of
hospitality, the duty of trustworthiness is triggered by the act of trust that
expresses a good opinion of the trustee and an expectation that she will be
moved by that good opinion. Moreover, it is an expectation that she (at
least) did nothing to counter.
Recognizing this kind of duty of trustworthiness, we can see how the trust
situation engenders a moral reason to reciprocate. In trusting, P provides a
moral reason for R to reciprocate trustworthily, in so far as P acts with the
expectation that R reciprocates and, absent any protest by R, expects that
R will be responsive to such an expectation. Thus we see how the third
condition of trust (reciprocity) creates a reason that did not exist before
trust, to reciprocate. And so in providing such a reason for R, P also
reasonably expects R to reciprocate.
To recap: in trusting someone, one provides her with reasons to respond
trustworthily, reasons correlative to three conditions of trust. Firstly, in
putting himself in a vulnerable position with respect to R, P provides
R with a special duty not to harm those who are vulnerable to her in
particular. Secondly, P puts himself in a vulnerable position for the purpose
of some cooperative outcome from which they can both gain fairly, thus
providing a moral duty for R to contribute her part to the arrangement.
Lastly, P does this with the expectation that R reciprocates, thus creating in
her a duty, akin to a duty of hospitality, to live up to such an expectation.
I am assuming here that moral reasons, as such, motivate. But even if this
is not necessarily the case, I would want to argue that responding to such
moral reasons may be the best explanation of the empirically observed
behavior. Although here I will be satisfied with merely suggesting that such
explanation is at least as plausible as any other.
3. The Possibility of Trust
In trusting we overcome the gap not by narrowing it down (or closing it)
but by providing the other party reasons to reciprocate trustworthily. These
are not self-regarding reasons, appealing to the other’s prudential motives.
They are, in the broadest sense, moral, thereby bridging the trust gap,
grounding P’s expectations that R will reciprocate. I mentioned earlier that
these expectations are a special kind of belief. They are not mere predictions.
To be sure, there is a sense in which P expects that R will reciprocate. This is
a purely empirical sense based on some purported facts of moral psychology.
But in trusting, P also expects of R to reciprocate in a normative sense.
Dictionary: NOSD
226 Daniel Attas
If R fails to reciprocate, if P’s expectations do not materialize, then P may

be understandably shocked: both in the sense of being genuinely surprised
that things haven’t turned out as he (empirically) predicted, and in the sense
of feeling moral indignation at R for not having (normatively) responded as
expected of her.
So the expectation is a prediction on the assumption that one has
provided reasons to reciprocate. In trusting one creates three (mutually
reinforcing) prima facie moral reasons to cooperate. First, in putting myself
in a vulnerable position, in exposing my vulnerability, I set in motion the
general obligation not to take advantage of the vulnerable. Second, in some
instances of trusting, I do my part in initiating and maintaining a cooperative
arrangement from which we are both expected to gain. I thus generate a
duty of fairness in you to do your part in a scheme from which you benefit
when others do their part. Third, by trusting you, as it were, resolving to act
on the suspension of doubt in your motives, my expectation, which you did
nothing to counter, that you will indeed reciprocate produces a moral
requirement for you to do so.
In so far as trust provides these three prima facie moral reasons it also
institutes parallel psychological motivations to reciprocate. Moreover each
of these reasons and motivations is susceptible to degrees. The more
vulnerable I am, the stronger your obligation not to take advantage, and
also the stronger your motivation to avoid doing so. The fairer the terms of
the cooperative arrangement, the stronger the obligation as well as the
motivation to do your part. Finally, the clearer it is that I am trusting you,
the stronger your duty to reciprocate, and so too is the motivation. We
might say that P expects in the predictive sense that R do B because he
expects of her in the normative sense to be moved by his expectation that
she will do B.
The possibility of trust is thus premised on two assumptions. First,
trust is acting on the presumption of reasonableness of the other, that is,
on the presumption that R is responsive to reasons. P need neither assume
that R is merely instrumentally rational, nor that she is emotionally
attuned to P’s trusting behavior. Moreover P need not act on starry-eyed
faith in the good and benevolent nature of humankind or of R in
particular. Second, P assumes a full appreciation of the circumstances of
trust: that R is aware, and knows that P knows that she is aware of the
costs of mistrust, of the fairness of cooperation, and of the fact of trust.
That is to say, P assumes that R realizes that he has provided her with
reasons to reciprocate. Thus there are three components to P’s belief
that R will respond trustworthily: P believes (i) that R will do B if she can;
(ii) that R will do B because P did A; and (iii) that R will believe that P
did A believing (i), (ii), and (iii).
Dictionary: NOSD
Terms of Trust 227
The attempt to force the analysis of trust into a non-moral rationalist

mold is not only false, it leads to a notion of calculated interest that leaves
no room for the kind of phenomenon, with its moral overtones, that we
typically recognize as trust. Conversely, the fallback position, denying any
role for reason, viewing trust as an emotion or a kind of faith, leaves no
hope for measures we might take to facilitate trust. The proposed analysis is
reason-based, but not on the narrow instrumental rationality view. Moral
reasons may motivate us, but their activation depends on our understanding
of the circumstances we are in. What, then, are the practical measures we
may take, and what kinds of measures should we beware of when facilitating
trust? To these questions I now turn.
4. Policy Implications
Before I begin drawing some practical conclusions, a caveat is in order. The

analysis in this chapter proceeded on the basis of some formal and abstract
scenarios, as do philosophical discussions and most experimental studies.
But, of course, real life situations are a lot messier: parties in realistic
scenarios are no strangers to each other. They have a history of past dealings,
emotions engendered by that history, an entrenched opinion of their
counterparts and what they are capable of, an open account enormously
difficult to simply ignore. A large part of the problem is that they don’t
know whether the other party really wants cooperation in good faith or is
spiteful, ideological, or otherwise motivated in ways that drastically change
the incentive structure for the worse. Still, I maintain, the preceding analysis
as well as the implications drawn below portray a genuine aspect of trust in
such situations that ought to be reckoned with, even while remaining
mindful of the complicated realities.
It is not unusual for third parties in conflict situations, acting as brokers
or moderators, to propose various kinds of interventions designed to
facilitate cooperation. In international conflicts, super powers, a multi-state
alliance, or international bodies may offer a plethora of confidence-building
measures aimed at narrowing the gap. Such measures, in and of themselves,
are welcome interventions in so far as they assist the parties in reaching a
cooperative outcome.19 However, as I shall argue, such well-meaning
19
An example of this kind of thought from Voice of America, April 8, 2013 <http://
www.voanews.com/content/kerry-in-israel-palestinian-territories-on-middle-east-
diplomacy-trip/1636826.html>: “Robert Wexler, president of the Washington-based
Center for Middle East Peace, said Secretary Kerry’s peace strategy is not limited to
diplomatic contacts with Israelis and Palestinians.
Dictionary: NOSD
228 Daniel Attas
interventions may have regretful effects with respect to bridging the gap.
These collateral results may be so bad as to frustrate any cooperation that
may have otherwise been reached. In other words, confidence-building
measures must be carefully planned to avoid the counterproductive erosion
of trust.
Let me enumerate the various sources of the gap, specify the kind of
confidence-building measures designed to reduce its effect, and explain how
and which aspect of trust it may erode. First, the risk of mistrust, that is to
say the difference between P’s status quo and his situation in case R defects.
One way to narrow this gap and reduce the risk is by providing guarantees
to P, such that in case of mistrust P will be compensated by a third party,
making his loss minimal, perhaps even zero. For example, P’s concession in
withdrawing from certain territories may be accompanied by the provision
of advanced defensive technologies. In itself, this may be harmless enough.
But we should be alert to the fact that reducing P’s losses in case of mistrust
will ipso facto attenuate whatever empathy may trigger R’s duty and motive
not to take advantage of P’s vulnerability, for P in making the first move
may no longer seem so vulnerable. Moreover, if the plea for these
technologies has been advertised by P and the refusal to provide them has
been loudly publicized, the gains to P from such acquisition may appear to
R so great that R will tend to interpret P’s move as merely a calculated
strategic step designed mainly for the purpose of acquiring these advanced
technologies rather than bona fide trust. Hence the provision of guarantees
designed to minimize P’s risk of mistrust may have the primary trust-
eroding effect of diminishing empathy and a secondary effect of encouraging
an interpretation of P’s move as strategic rather than one of trust, thus
eroding reciprocity.
A second source of the gap may be the insufficient gains of cooperation
to P. That is to say, how beneficial to P the prospects of cooperation are
compared to the status quo, and the likelihood that these expected gains can
outbalance the risk of mistrust. There are two primary means to address this
gap in prudential terms: by providing incentives to P, increasing the value of
the cooperative outcome; and by imposing sanctions on P, further reducing
the value of the status quo. For example, in return for P’s withdrawal from
“Secretary Kerry is working closely with many of the foreign ministers of the Arab
League, talking about how they might—not amend the Arab peace initiative, but
implement in a way so as to create greater incentives for Israel to move forward. These
could, in certain ways, provide even greater confidence-building measures to both the
Israelis and Palestinians than some of the things the parties themselves could do,” he said.
“Wexler, a former U.S. congressman, said there are numerous positive steps that Arab
nations could take, such as enabling Israeli passenger planes to fly over their territory,
opening trade links with Israel, and accepting invitations to meet with Israeli officials.”
Dictionary: NOSD
Terms of Trust 229
some territories, P may receive privileged trading arrangements, erasure of

debt, international recognition of sovereignty over some other contested
territories, or any other benefit from a third party, thus boosting the value of
peace to P.20 Or, to counter P’s hesitation to withdraw, P may be threatened
by sanctions such as curbing diplomatic support, reducing or ending
financial aid, an embargo on arms and ammunition, general divestment and
boycotting of export, and so on. Incentives and sanctions such as these may
be helpful in so far as these measures increase the gains of cooperation,
narrow the gap, and provide greater prudential reasons for P to cooperate.
However, we should be wary of such measures considering their possible
effect on trust. If benefits, in the form of incentives, are cast uniquely at P
to such an extent that the balance of gains in the cooperative outcome seem
unfair to R, then R may feel less obliged by fairness to contribute her part
to the cooperative arrangement. Moreover, if R assumes that P too detects
the unfairness towards R, she may attribute to P a non-trusting motive.
After all P should not expect R to reciprocate for the sake of an unfair
arrangement. Therefore, P’s initiating act would be interpreted as
manipulative or foolish at best—not the kind of act that one would expect
to trigger trustworthiness in return. With respect to the other kind of
intervention, if the threats of sanctions on P are considerable, P may feel
coerced into the initiating act. It would not, therefore be an act of trust, but
a prudential move designed to eschew the penalties attached to refusal. Or
at least, this is what it may seem like to R, who would now feel less obliged
to reciprocate trustworthily.
A third source of the gap may be R’s temptation to defect. That is to say,
how beneficial to R are the prospects of defecting compared to the value of
cooperation. Here, too, there are two primary means to address this gap in
prudential terms: by providing incentives to R, increasing the value of the
cooperative outcome; and by imposing sanctions on R, reducing the value of
defection. The incentives and sanctions applied here are similar in nature to
those applied towards P to increase the gains of cooperation. The kind of
problem they may pose for trust is symmetrical to the problem arising from
incentives for and sanctions on P. If R benefits disproportionately due to
such incentives, the balance of gains in the cooperative outcome may seem
unfair to P, so much so that he will not be inclined to initiate cooperation
in the first place. Even if he does, and R perceives the unfairness towards P,
she may attribute to P a non-trusting motive such as foolishness, negligence,
20
In the latest round of Israel–Palestinian Authority negotiations brokered by the US
and led by Secretary of State John Kerry, it was reported that the US was offering to
release Jonathan Pollard, a former US Navy official convicted of espionage for Israel
almost thirty years ago.
Dictionary: NOSD
230 Daniel Attas
or manipulation. Again, not the kind of act that one could expect to trigger
trustworthiness in return. Regarding sanctions, if the threats of sanctions on
R are considerable, R may feel coerced into reciprocating. Or at least, R may
suspect that P believes that R would be compelled to reciprocate. It would
not, therefore, be an act of trust on the part of P, but a prudential move
relying on the sanctions placed on R rather than on R’s trustworthiness. If
this is how it is perceived by R, she would feel less obliged to reciprocate
trustworthily.
So we see how these confidence-building measures (guarantees, incentives,
and sanctions), designed to narrow the trust gap and facilitate cooperation,
may unwittingly demolish the foundations for bridging the gap. Table 8.1
summarizes this.
Table 8.1 Confidence-building measures and their effect on trust

Gap source Confidence-building Possible effect on trust
measure
Risk of mistrust Guarantees Diminish empathy (and
undermine reciprocity)
Gains of cooperation Incentives for P Generate unfairness (and
Sanctions on P Undermine reciprocity
Temptation Incentives for R Generate unfairness (and
Sanctions on R Undermine reciprocity
As I argued earlier in this chapter, reliance on prudential motives alone is

unlikely to succeed in closing the gap entirely, and moreover, it is likely to
produce an unstable solution. I have now shown how reliance on such
measures may damage the foundations of trust, which in many cases will be
vital to get the process towards cooperation to take off. It is essential that a
process of conflict resolution involving a series of discrete and progressive
steps be supported by trust-facilitating measures.21 Beyond the appeal to the
prudential interests of the parties, perhaps even more germane to the success
of the process, is the developing of a mutual understanding and appreciation
of the respective values—their conceptions of fairness and of the good—the
parties hold. The foregoing discussion may allow us to formulate some very
general prescriptive maxims.
To be distinguished from confidence-building measures.

21
Dictionary: NOSD
Terms of Trust 231
(1) Do not focus exclusively on narrowing the gap.

Bolstering prudential motives towards cooperative outcomes such as
peace are welcome interventions. However, relying on such motives alone is
likely, in many cases, to fail to close the gap all the way. And where it will
not fail, the solution will be unstable, relying, as it must, on interests that
may change at an instant. Therefore, along with any measures taken to
narrow the gap, further efforts to bridge the gap must be made, facilitating
trust rather than mere confidence.
(2) Manufacture “common knowledge.”
Successful trust depends to a significant extent on perceptions. First, the
obligation not to take advantage of the vulnerable can only be effective to
the extent that the vulnerability of the trusting party is appreciated by the
reciprocator. Though it would be absurd to expose P to unnecessary risk, it is
important in this respect that P publicize, make explicit and vivid, the cost he
would bear in case of mistrust (i.e. misplaced trust). No less important,
R should make a comparable effort to communicate her full awareness of these
costs. Second, the duty of fairness can only take effect if both parties consider
the cooperative outcome as fair, and both realize that both consider it so.
Clearly, the parties should strive for an arrangement that is objectively fair
to both and not one that merely reflects their uneven bargaining power. No
less important, they should announce their acceptance of the terms of the
arrangement, in the circumstances, as fair to both sides. Finally, P should
make explicit the nature of his motive and his act as an expression of trust,
and (what comes to the same thing) his expectation that R be moved to
reciprocate by this knowledge. Furthermore, R should make it very clear that
she appreciates P’s move as an expression of trust, rather than some miscalculated
gamble, a coerced act, an attempt to manipulate, or sheer foolishness.22
(3) Take care when narrowing the gap not to thwart bridging the gap.
When employing confidence-building measures, positive in themselves,
thought must be given to how these may affect trust. First, incentives to
both parties should be offered symmetrically so that the fairness of the
arrangement is not harmed. Moreover, they should not be so high to one
party that their prudential reasons would seem to overshadow any trust-
based reasons that may otherwise be operative. Second, guarantees should
22
In recent rounds of negotiations and “concessive gestures,” Israeli officials have
made a huge effort to portray the Israeli moves as responding to American pressure, their
lack of expectation that Palestinians will respond trustworthily, and their perceived
unfairness of the expected outcome. A self-fulfilling expectation, if ever there was one.
Palestinians advertised their own matching convictions.
Dictionary: NOSD
232 Daniel Attas
not be perceived as fully extinguishing P’s vulnerability. Without a minimally

significant risk, P’s move could not be interpreted as an expression of trust.
Third, sanctions should be avoided. They introduce outright prudential
motives designed to coerce the parties into the arrangement. With their
introduction a perception of trust would be difficult to maintain.
Bibliography
Andreoni, J., Harbaugh, W. T., and Vesterlund, L. (2007). “Altruism in Experiments,”
in S. N. Durlauf and L. E. Blume (eds), New Palgrave Dictionary of Economics,
2nd ed. (Basingstoke/New York: Palgrave Macmillan).
Baier, A. (1986). “Trust and Antitrust,” Ethics 96, 231–60.
Becker, L. C. (1996). “Trust as Noncognitive Security about Motives,” Ethics 107,
43–61.
Ciriolo, E. (2007). “Inequity Aversion and Trustees’ Reciprocity in the Trust Game,”
European Journal of Political Economy 23, 1007–24.
Dasgupta, P. (1988). “Trust as a Commodity,” in D. Gambetta (ed.), Trust: Making
and Breaking Cooperative Relations (Oxford: Blackwell).
Dawes, R. M. and Thaler, R. H. (1988). “Anomalies: Cooperation,” Journal of
Economic Perspectives 2, 187–97.
Eckel, C. C. and Wilson, R. K. (2004). “Is Trust a Risky decision?,” Journal of
Economic Behavior and Organization 55, 447–65.
Fehr, E. and Fischbacher, U. (2003). “The Nature of Human Altruism,” Nature 425,
785–91.
Feinberg, J. (1966). “Duties, Rights, and Claims,” American Philosophical Quarterly
3, 137–44.
Fukuyama, Francis (1996). Trust: Human Nature and the Reconstitution of Social
Order (New York: Simon and Schuster).
Goodin, R. E. (1986). Protecting the Vulnerable: A Re-Analysis of Our Social
Responsibilities (Chicago: University of Chicago Press).
Guth, W., Schmittberger, R., and Schwarze, B. (1982). “An Experimental Analysis
of Ultimatum Bargaining,” Journal of Economic Behavior and Organization 3,
367–88.
Hardin, R. (2002). Trust and Trustworthiness (New York: Russell Sage Foundation).
Hart, H. L. A. (1955). “Are There Any Natural Rights?,” Philosophical Review 64,
175–91.
Hobbes, T. (1991). Leviathan (Cambridge: Cambridge University Press).
Jones, K. (1996). “Trust as an Affective Attitude,” Ethics 107, 4–25.
Jones, K. (2012). “Trustworthiness,” Ethics 123, 61–85.
Loewenstein, G., Thompson, L., and Bazerman, M. (1989). “Social Utility and
Decision Making in Interpersonal Contexts,” Journal of Personality and Social
Psychology 57, 426–41.
Orbell, J. M., van de Kragt, A. J., and Dawes, R. M. (1988). “Explaining discussion-
induced cooperation,” Journal of Personality and Social Psychology 54 (5), 811–19.
Dictionary: NOSD
Terms of Trust 233
Pillutla, M. M., Malhotra, D., and Murnighan, J. K. (2003). “Attributions of trust

and the calculus of reciprocity,” Journal of Experimental Social Psychology 39,
448–55.
Putnam, Robert D. (1994). Making Democracy Work: Civic Traditions in Modern
Italy (Princeton: Princeton University Press).
Sally, D. (1995). “Conversation and Cooperation in Social Dilemmas,” Rationality
and Society 7, 58–92.
Scanlon, T. M. (2000). What We Owe to Each Other (Cambridge, MA: Belknap
Xiao, E. and Bicchieri, C. (2010). “When Equality Trumps Reciprocity,” Journal of
Economic Psychology 31, 456–70.
Dictionary: NOSD
Dictionary: NOSD
Index
alienation 93, 99, 102, 108, 114–26, division of labor 4, 178, 185–206
138–9, 145 domestic 100, 123, 185–9, 204
arbitrary 4, 14, 19–20, 26, 49, 120, domination 3–4, 64, 99, 110–11, 120,
128–55, 213 128–55
aristocracy 63, 157, 167–8
autonomy 44, 50, 68, 73, 83, 98, 111, egalitarian, see equality
119–22, 146, 154; see also elitism 4, 156–7, 167–9, 176–82
self-determination enforcement 2, 35, 38–42, 48–51, 55–6,
78–80, 105, 119, 143–4, 147, 169, 188
choice 2, 9, 10–12, 17, 32, 36, 40, 44, 53, epistemic 16–17, 29, 75, 154–8, 161–4,
137, 156, 177, 187–9, 196–9, 202, 179–81
213, 218, 221–2 equality 4, 21–32, 36–8, 45, 53, 60,
hypothetical 3, 18, 71–9, 88, 114, 63–9, 73–4, 79, 86, 88, 126, 134,
135–6, 139–41 139–43, 156–69, 175, 179, 183–206,
coercion 3, 37, 41, 48, 50, 60, 63–5, 68, 213–15, 219, 233
72, 78, 80, 90, 106–7, 110–12, 120–5, expertise 149, 156–8, 162, 167–9,
131, 135, 146, 158, 163, 169–70, 179 173–82
collectivity 113, 116, 165, 139, 144;
see also self-determination, collective fairness 2–5, 53, 67, 97, 157, 192–7, 213,
community 4, 37, 41, 73, 88, 93, 98, 101, 218–23, 226, 229–31
108–10, 116, 161–2, 168, 205 feasibility 1–3, 9, 102, 118, 121–6,
competence 88, 161, 166–7, 170–2, 156–7, 169–73, 176–83, 188
180–2 force 3–4, 35–6, 39–41, 45–65, 69,
consent 3–4, 36–40, 43, 50–9, 63–7, 99–100, 116, 120–3, 128, 146,
71–98, 108, 112, 115, 121, 126, 132, 159, 171
135–45, 159, 206 freedom 3–4, 35–6, 43, 52–4, 60, 67–9,
consequentialist 58, 94, 172, 177, 184 111–35, 139–55, 159, 163
contractualism 3, 71–9, 82, 87–92, 96
contribution 36, 39, 49–51, 58–61, gender 4, 15, 144, 151, 154, 185–206
103–5, 110–12, 116, 163–6, 217, 220, goods 21, 24, 36–9, 42–4, 51–3, 57,
225, 229 60, 63, 84, 92, 100, 113, 122, 148,
cooperation 4, 36, 42, 59, 101–23, 169, 190–202, 217
171, 186, 209, 212–20, 226–33 government 3, 26, 37, 49–50, 69–111,
115–19, 122–9, 133, 139–40, 144–54,
democracy 3, 66–9, 72–3, 79, 85, 92, 157–60, 163–72, 176–84, 192, 213
96–7, 100, 109–11, 125, 135, 140–1,
152–83, 233 hierarchy 63–5, 88, 105–7, 116
deontological 3, 45–67, 165 hypothetical choice, see choice,
desert 2, 36, 43, 51, 59 hypothetical
deterrence 3, 42–62
distribution 4, 21–7, 37–8, 42–5, 61, ideal theory 1, 7, 71–4, 89
67–70, 75, 80, 86, 100–1, 113, 136, independence 64, 98–100, 110, 120–5,
140, 151, 185–205, 219–22 139, 145, 151
236 Index
individualism 56–9, 63, 90–2 political authority 2, 3, 39, 64, 80, 97,
institutions 3, 72–4, 78, 96, 100–2, 128–9, 135, 148
105–25, 140, 144, 149–51, 158, 165, political community, see community
169, 187, 195–206 political liberalism, see liberalism
instrumental 83, 87, 94–6, 113, 117, political obligation 39–41, 50–1, 56,
144–6, 157, 163–9, 172–3, 182–3, 209, 68–9, 135; see also political authority
226–7 power 3, 4, 35, 39, 61–5, 72, 84, 87,
interference 2, 36, 68, 98, 114, 128–34, 92–5, 98, 117, 130–7, 142, 147–55,
137–9, 143–50, 157 165, 168–70, 173, 183, 190
procedural 21, 26, 106, 109, 118, 172–3
justice 1–5, 9–32, 37–9, 57, 68–9, promise 40, 130, 147, 224
86–9, 97, 100–1, 113, 118–26, 149, property 26, 42–3, 48, 61–3, 70, 105,
152–4, 158–61, 169–70, 181, 184, 119, 123, 171
189–205 protection 26, 36, 51–8, 85, 100, 113,
justification 2, 37, 40, 53–4, 57–8, 78, 130, 144, 177–8, 224
90, 131, 136, 168, 181, 189–95, 198, public policy 4, 156, 174–5,
202, 222 180, 204
public reason 69, 134, 153–4, 188, 192,
knowledge 4, 11, 27, 50, 57, 60–2, 71, 203–5
75–6, 83, 103–5, 115, 119, 135,
156–60, 167, 173–5, 178–9, 194, Rawls 1, 5, 19, 21–2, 32, 37, 41, 60, 69,
224, 230–1 72–3, 97, 101, 110–11, 126–9, 148,
154, 157, 166, 175, 179, 188, 191–2,
labor 4, 61–2, 106, 131, 178, 185–206 196, 205
laws 2, 50, 72, 99, 105, 109, 128–30, reasonable 3, 4, 74, 89, 106–9, 112, 115,
133–5, 139–47, 151, 156–60, 168–75 122–4, 138, 141, 149, 161, 171–82,
legislature 105, 109, 131, 139–44, 187–93, 196, 209, 220, 224–6
147–51, 160, 163, 173–7 reciprocity 4, 188, 192, 202, 206,
legitimacy 2–4, 35–7, 41, 62, 69–97, 209–33
100–2, 123–8, 136, 158, 173, 179–81, republicanism 3–4, 63–4, 69, 81, 97,
187–95, 203 126–36, 139–55
liberalism 4, 36–7, 69–73, 78–9, 85, 89, resources 11, 23, 60, 176, 199–201, 214
92, 96–7, 100, 125–31, 144, 148, respect 56, 71, 74, 77–8, 94, 105, 108,
153–7, 175, 179–82, 187–96, 203–6 115–22, 148, 159–64, 168, 178,
libertarianism 36–7, 47–9, 62, 69, 81, 86 183–4, 196
liberty 4, 21, 35, 38, 53, 69, 110, 126–39, rights 3, 23, 53, 62, 68–70, 79–81, 85,
142–57, 162, 165, 176, 192, 220 88, 96–102, 105, 110–13, 118–19,
122–4, 133, 153, 160–6, 170–82, 232
markets 186–7, 199 right to rule 2–3, 72, 158–9, 167, 174
methodology 10–12, 18, 22, 27–31, 144 rule of law 3, 4, 56, 69, 128–55
Mill 98, 126, 129, 169–70, 177, 184
self-determination 3, 98–127
natural duties 39–41, 49–50, 59, 70, collective 3–4, 74–8, 81, 84, 89–92,
119, 124 99, 101–5, 108–12, 117, 187
neutrality 4, 188–99, 202–3 self-governing 107–8, 139
non-arbitrary, see arbitrary self-rule 81, 100–2, 110, 139
social justice 23, 30, 158–61, 203–5
opportunities 23, 43–5, 51–6, 59–62, sovereignty 3, 76, 81–96, 101–3, 109,
66–7, 87, 108–10, 131, 166–9, 187, 126, 229
190–2, 195–8, 201–3 subordination 38–40, 63–8, 124
Index 237
sustain 42, 46–7, 52, 55, 105, 113, 117, trust 4, 60, 65, 113, 153,
122–3, 128, 163–7, 175, 179–82, 209–33
186–8, 198 trustworthy 210–11, 220–3
taxation 35, 49, 58–62, 70, 105, 112, voluntarism 3, 71–3, 78–81, 90–6,
134, 165 144, 147
threats 35–6, 41–2, 48, 52, 55–7, 60–6,
99, 103, 107, 116, 120, 124–5, 131–2, work 4, 114–16, 151, 185–92, 195–200,
151–2, 159, 164, 179, 186, 211, 229–30 204–6, 211

(Oxford Political Philosophy) Sobel, David - Vallentyne, Peter - Wall, Steven-Oxford Studies in Political Philosophy. Volume 2-Oxford University Press (2016)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(Oxford Political Philosophy) Sobel, David - Vallentyne, Peter - Wall, Steven-Oxford Studies in Political Philosophy. Volume 2-Oxford University Press (2016)

Uploaded by

Copyright:

Available Formats

OUP CORRECTED PROOF – FINAL, 03/02/16, SPi

OXFORD STUDIES IN POLITICAL PHILOSOPHY

Part I: Ideal Theory

Part II: The Moral Assessment of States

Part III: Issues in Social Relations

Richard Arneson is Professor of Philosophy at the University of

strongest form of a right to rule is that the state is permissible, just,

law is compatible with individual liberty (and non-domination), because it

Does it matter whether anything is just or unjust? Or does it matter only

Just and Juster 11

to relative justice or injustice. A rich partitionism would include both a

1. CATEGORICAL AND METHODOLOGICAL

Suppose that, while there is no appeal to any partition, we have good

meaning to be assigned to judgments such as that slavery is profoundly

Just and Juster 13

2. THE SCALE OF JUSTICE

Just and Juster 15

3. A THEORY OF JUSTER: THE LIMITS OF ORDINAL

I want to emphasize the limits of comparativism, and I begin with the

points on a line. Some of the limitations in ordinal comparativism would be

4. JUSTICE IS NOT LIKE UTILITY

If comparativism leaves behind so much of our thought about degrees and

Just and Juster 17

would be chosen by some hypothetical chooser, such as an impartial spectator

5.â•‡ IS “JUST” LIKE “TALL?”

Could comparativism accept that slavery is unjust after all? Suppose we

Just and Juster 19

fraction of the buildings that it is taller than, or combinations of such

6. THE VARIETY OF PARTITIONS

I am mostly concentrating on the idea of a partition between just and unjust

Just and Juster 21

a toggle or binary partition, the case in which the standard or property is

7. WHAT JUSTICE MIGHT BE LIKE (DOESN’T

One might be tempted toward comparativism by the following thought,

Just and Juster 23

8. THE CASE OF EQUALITY

One thing we are looking for is perfect arithmetical equality, if it should be

Just and Juster 25

9. SPARE PARTITIONISM

But, it might be objected, surely there are obvious comparisons of inequality.

Just and Juster 27

(Sen 2011, p. 104). Importantly, though, these are never shown to be derived

10. RICHNESS FIRST

but the vague proposition that “x has ‘more’ of the perfection-standard

Just and Juster 29

definition, be non-comparativist. And as we have seen, there is no basis

11. COMPARATIVISM AND CONSERVATISM

I turn, finally, to a second aim of Sen’s advocacy of a theory of “juster.” In

This is different from practical conservatism:

Methodological conservatism is a position about the merits of certain kinds

Just and Juster 31

of theory. The partition standard itself would be of no theoretical interest on

Political Rule and Its Discontents 37

economies, that are predicated on precisely that truism.) Instead, most

2. FROM OBLIGATION TO ENFORCEMENT

Political Rule and Its Discontents 39

of the complaint-answering conditions proposed in the last section: namely,

candidate be answered, either by anyone’s lights, or at least by the lights of

Political Rule and Its Discontents 41

we observed earlier, there will be cases in which we have no moral duty

3. DETERRENTS: THE DISTRIBUTIVE COMPLAINT

What then is enforcement? It divides into three categories, which raise

Political Rule and Its Discontents 43

Part I: Ideal Theory

Part II: The Moral Assessment of States

Part III: Issues in Social Relations

a) The dominator is someone who consents to rule on the basis of a