A Game-Theoretic Interpretation of Sun Tzu's The Art of War

Author(s): Emerson M. S. Niou and Peter C. Ordeshook

Source: Journal of Peace Research , May, 1994, Vol. 31, No. 2 (May, 1994), pp. 161-174
Published by: Sage Publications, Ltd.

Stable URL: https://www.jstor.org/stable/425030

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Sage Publications, Ltd. is collaborating with JSTOR to digitize, preserve and extend access to
Journal of Peace Research

This content downloaded from

3.104.43.49 on Thu, 21 Sep 2023 04:08:59 +00:00
All use subject to https://about.jstor.org/terms
 ? Journal of Peace Research, vol. 31, no. 2, 1994, pp. 161-174

A Game-Theoretic Interpretation of Sun Tzu's

The Art of War"
EMERSON M. S. NIOU PETER C. ORDESHOOK
Department of Political Science, Division of Social Sciences,
Duke University California Institute of Technology

Over 25 hundred years ago the Chinese scholar Sun Tzu, in The Art of War, attempted to codify the
general strategic character of conflict and, in the process, offer practical advice about how to win
military conflicts. His advice is credited with having greatly influenced both Japanese military and
business practices, as well as Mao Tse-Tung's approach to conflict and revolution. The question,
however, is whether or to what extent Sun Tzu anticipated the implications of the contemporary theory
of conflict - game theory. The thesis of this essay is that he can be credited with having anticipated the
concepts of dominant, minmax, and mixed strategies, but that he failed to intuit the full implications of
the notion of equilibrium strategies. Thus, while he offers a partial resolution of 'he-thinks-that-I-think'
regresses, his advice remains vulnerable to a more complete strategic analysis. In judging Sun Tzu's
contribution to our understanding of strategy, however, we should keep in mind that resolving circular
reasoning in some circumstances requires the use of advanced principles of probability theory and
mathematics, and so we should not be surprised to learn that Sun Tzu's treatment of information is
incomplete. Indeed, we should marvel at the fact that he understood intuitively as much as he did.

1. Introduction general way. Its age, however, is less im-

The formulation of general strategic prin-portant to us than the fact that it was written
ciples - whether applied to war, parlor at a time of prolonged conflict within an
games such as Go, or politics - has long emerging China - in an era in which the
fascinated scholars. For some, such as the leaders of competing kingdoms possessed
Chinese strategist Sun Tzu, this fascination considerable experience in the conduct of
is motivated by the necessity for formulating diplomacy, strategic maneuver, and in the
immediate practical advice, while for others art of war. As such, then, we should pre-
the motivation is merely by intellectual curi- sume that The Art of War codifies the
osity. Regardless of motivation, however, insights of an era skilled at strategy and tac-
the study of strategic principles is of interest tics. The better we understand it, the better
because it grapples with fundamental facts we understand not only the era in which Sun
of human existence - first, people's fates are Tzu wrote, but also the essentials of conflict
interdependent; second, this interdepen- today.
dence is characterized generally by conflict- But just as we might suppose that Sun Tzu
ing goals; and, finally as a consequence of offers insight into the past and the present,
the first two facts, war is not accidental but we cannot suppose that our understanding
is the purposeful extension of a state's of strategy has not progressed in 25 cen-
policies and must be studied in a rational turies. New modes of analysis, including the
way. development of decision theory and the
Sun Tzu's The Art of War, written more application of mathematics, have all entered
than 25 hundred years ago, is, of course, our the domain of strategic analysis to general-
first written record of the attempt to under- ize and refine our thinking. Thus, for
stand strategy and conflict in a coherent and scholars interested in understanding Sun
Tzu's particular contribution, we must
address the issue of how best to interpret
* This research was supported by NSF grant SES-
9223185 to Duke University and NSF grant SES- and analyze his writings in the context of
8922262 to the California Institute of Technology. these advances, since it is in this way that we

This content downloaded from

3.104.43.49 on Thu, 21 Sep 2023 04:08:59 +00:00
All use subject to https://about.jstor.org/terms
162 Emerson M. S. Niou & Peter C. Ordeshook

maximize The Art of War's contemporary of Sun Tzu's writings fail to take account
relevance. This is the issue with which this fully of what we know generally about stra-
article deals. Rather than begin our dis- tegic choice. We begin Section 1 of this
cussion with Sun Tzu himself, we begin with essay with a discussion of 'pre-game' de-
the theory of strategic behavior developed cision-making - of situations in which there
during the latter part of this century called is only one decision-maker. Section 2 takes
game theory. Briefly, game theory, which us to the core of game theory and describes
we can view as either a branch of mathemat- how interdependent choice situations differ
ics or of political science and economics fundamentally from simple decision prob-
(Ordeshook, 1986), seeks to isolate general, lems. In Section 3 we explore more carefully
abstract principles of decision-making when Sun Tzu's writings as they apply to a particu-
the outcomes of people's choices depend on lar class of games in which decision-makers
what others decide and when everyone is implement their actions sequentially, as in
aware of their mutual interdependence. parlor games like Chess and Go. Section 4
Thus, while we might interpret much of Thelooks at games that better model the stra-
Art of War as modeling and analyzing a par- tegic situation that confronts decision-
ticular type of game - interactions of puremakers in battle - simultaneous move games
conflict - game theory is concerned with theand games of imperfect information, and
general issue of interdependent decision- introduces the concept of a Nash equilib-
making, including the possibility that people rium as a general solution for such games.
may choose to cooperate so as to achieve Section 5 considers those situations that do
mutually beneficial ends. Hence, game not have simple Nash equilibria, and which
theory's applications include not only stra- cause us to introduce the idea of a mixed
tegic military planning, but also an analysis strategy. Here we argue that mixed strat-
of the decisions confronting the heads of egies - strategies that leave something to
business firms as they compete for profits or chance - are not mere mathematical cur-
market share, of political candidates who iosities, but are in fact a central feature of
must compete in order to win elections, of Sun Tzu's strategic principles. Section 6
members of committees who compete to studies the importance of information ma-
form alternative coalitions, and of nation- nipulation in war, and it is here that we find
states who compete to secure advantageous far less compelling reasons to suppose that
positions in alliances. Sun Tzu anticipated fully the analytic struc-
Game theory and The Art of War, then, ture characterizing contemporary game
each bring something different to our under- theory.
standing of strategy. Game theory offers
generality and mathematical precision and it
2. Individual Decision-making
allows us to ascertain the logical coherence
In order to appreciate the lessons and per-
of our ideas about strategic interaction; Sun
spectives of game theory as well as of Sun
Tzu provides a specific application of
Tzu's insights, it is necessary first to consider
general principles, and demonstrates the art
a situation in which there is a single decision-
of rendering logical and abstract reasoning
maker whose actions we are trying to under-
practical. The plan of this essay, then, is to
stand, and who must choose one action from
review the essential components of game
some set of alternative actions. The usual
theory in such a way that we can explore the
decision-theoretic representation of such
extent to which Sun Tzu's writings are con-
situations requires a specification of the
sistent with or illustrate parts of that theory.
following components:
Because it is unreasonable to suppose,
however, that someone writing more than (1) A list of the decision-maker's alterna-
25 centuries ago could have anticipated all tive actions, where this list is exhaustive
the nuances of strategic interaction that in the sense that at least one action in
formal mathematical reasoning reveals to us the list must be chosen by the decision-
today, we also want to learn which aspects maker in question and where the actions

This content downloaded from

3.104.43.49 on Thu, 21 Sep 2023 04:08:59 +00:00
All use subject to https://about.jstor.org/terms
 A Game-Theoretic Interpretation 163

in the list are exclusive in the sense that

Sun Tzu does not ignore the goals of those
one and only one action can be taken. who must actually implement plans in war-
(2) A list of possible outcomes (conse- fare - soldiers:
quences of actions taken) that is also The reason troops slay the enemy is because they
exhaustive and exclusive. are enraged . . . [and] they take booty from the
(3) A specification of the relationship be- enemy because they desire wealth. ... Therefore,
tween actions and outcomes - what out- when in chariot fighting more than ten chariots are
captured, reward those who take the first (II, 16-
come prevails after a specific act is
18).
chosen.
In the simplest decision-making environ-
Appropriately, Chapter 1 of Sun Tzu's ment, decision-making under certainty, we
text begins with a specification of those ele- assume that each action leads to a well-
ments of a strategic situation that mediate defined, specific outcome. In this instance,
between outcomes and actions - domestic we need only know the decision-maker's
politics (moral influence), weather, terrain, preference order over outcomes, since we
and doctrine (organizational efficiencycan of suppose that the action leading to the
most preferred outcome will be chosen.
the state and armies). The remaining factor,
command (quality of leadership), is a However,
de- a more general environment, de-
cision variable pertaining to actions and cision-making under risk, allows for the
strategies. possibility that we or the decision-maker in
An additional important component of question are uncertain as to what outcome
any decision-making environment is a speci-prevails after a specific action is taken.
fication of the goals (preferences over out- In simple decision theory we assume that
comes) of the participants. Of course, the the consequences of action are contingent
goals of the king and an Army's leaders are on the 'choices' of nature, where by nature
obvious - to win. Indeed, winning is a we mean an entity that operates without
necessity if the state is to survive:' 'War is a purpose - an entity that cannot be said to
matter of vital importance to the state; the operate in the pursuit of some goal. Because
province of life or death; the road to sur- nature has no purpose - because it is neither
vival or ruin' (I, 1), and 'Victory is the main purposefully malevolent nor benevolent -
object of war' (II, 3). Refining goals further, we suppose that it takes its form outside the
Sun Tzu is concerned also that too myopic a influence of our actions. Thus, we merely
view of victory can weaken a state. Thus, suppose that we can associate probabilities
'What is essential in war is victory, not pro- with each of nature's alternative choices.
longed operations' (II, 21). One example of this situation is a farmer's
More specifically, the first few chapters of decision about what crops to plant. The
The Art of War contain a number of pass- seeds a farmer might sow are the farmer's
ages pertaining to protracted conflicts, alternatives; yields are outcomes; prefer-
which we can quite directly interpret to ences are dictated by the relative profit-
mean that one ought to avoid conflicts that ability of outcomes; and nature's choices
deplete one's resources. 'For to win one might correspond to weather patterns for
hundred victories in one hundred battles is the year. In this instance, we would not nor-
not the acme of skill. To subdue the enemy mally view nature as either a benevolent or
without fighting is the acme of skill' (III, 3). malevolent creature with goals; instead, we
Thus, one's goal should not merely be that would merely associate probabilities with
of winning a battle, but to win at minimal alternative weather patterns (based, pre-
cost so as not to deplete one's resources for sumably, on the historical record).
future conflicts. In addition, one should use With this conceptualization, the conse-
tactics that maximize the gains from victory. quence of a specific action is a lottery over
'Generally in war the best policy is to take outcomes, and one question that concerns
the state intact; to ruin it is inferior to this' decision theorists is how people evaluate
(III, 1). these lotteries - how preferences over

This content downloaded from

3.104.43.49 on Thu, 21 Sep 2023 04:08:59 +00:00
All use subject to https://about.jstor.org/terms
164 Emerson M. S. Niou & Peter C. Ordeshook

specific outcomes determine preferences cision-making problems with nature as the

over lotteries over these outcomes. Gener-
primary adversary. It is evident that Sun
ally, game theorists assumed that we Tzu canrealized that a complete specification of
the strategic structure includes the moves
assign a value to each outcome - a numeri-
available to nature, the strategies of the de-
cal representation of preference - such that
cision-maker in question, and the strategies
we can compute the expected value of each
available to all antagonists, 'Know the
action and predict that a decision-maker will
enemy, know yourself; your victory will
choose the action that yields the greatest
expected value. never be endangered. Know the ground,
To illustrate so that we can later add some know the weather; your victory will then be
strategic complexity, consider Figure 1,total' (X, 26). Thus, to explore matters
further we must turn to game theory and the
analysis of interdependent choice.
Figure 1. Decision-making Under Risk

Nature 3. The Special Relevance of Game Theory

The preceding discussion of simple decision-
P 1 ~p
making merely serves as a point of depar-
A ~ ~4 2 ture, and, in fact, Sun Tzu offers an early
admonition that reveals that he considered
B 1 3
such a representation and any discussion of
alternative choices and goals as preliminary:
'All warfare is based on deception' (I, 17).
which assumes that a decision-maker, We cannot deceive nature, because by defi-
person 1, must choose between A and nition
B, nature is an unthinking anonymous
that nature has two alternatives which it entity. Thus, all warfare entails the strategic
chooses with probabilities p and 1 -interaction
p of two or more people, and the
next issue we must address is the distinction
respectively, and that the values assigned by
the decision-maker to specific outcomes between are decision-making against nature and
as shown. In this instance, then, the decision-making against another person.
expected value of alternative A, E(A), To see that this distinction is important,
equals 4p + 2(1 - p) = 2 +- 2p; and the suppose we replace nature in Figure 1 with a
expected value of B, E(B), equals p + 3(1 second
- decision-maker, person 2, who must
p) = 3 - 2p. Thus the decision-maker also choose between two alternatives, say C
chooses A if E(A) > E(B), or equivalently, and D. Thus, the consequences of person l's
if 2 + 2p > 3 - 2p which requires that p decisions > depend on what 2 does, and vice
1/4. Hence, if p = 1/4, then the decision- versa. Suppose, moreover, that this person
maker is indifferent between A and B; and evaluates the outcomes differently from 1 -
if p < 1/4, then the decision-maker prefersspecifically,
B suppose as is most likely the
to A. case in military conflict situations that 2 has
On occasion, Sun Tzu offers advice that precisely the opposite preferences of 1.
we might interpret as corresponding to thisFigure 2 portrays this new situation, where
simple structure. For example: the first number in each cell corresponds as
before to l's preferences, and where the sec-
There are five methods of attacking with fire. .. ond number corresponds to 2's preferences.
There are suitable times and appropriate days on
which to raise fires. 'Times' means when the
weather is scorching hot; 'days' means when Figure
the 2. Two-Person Constant-Sum Game
moon is in Sagittarius, Alpharatz, I, or Chen
constellations, for these are days of rising winds C D
(XII, 1,4, 5).
A 4,1 2,3
But there are only a few instances in The Art B 1,4 3,2
of War in which he focuses on specific de-

This content downloaded from

3.104.43.49 on Thu, 21 Sep 2023 04:08:59 +00:00
All use subject to https://about.jstor.org/terms
 A Game-Theoretic Interpretation 165

There are now two ways in which we can But, once again, if 1 anticipates this reason-
characterize the scenario that confronts ing himself . . . and so on and so forth.
persons 1 and 2: (1) perfect informationSuch
- circular reasoning does not arise in
one person must choose first and the simple
other decision-making scenarios, since by
assumption
chooses second, after learning the choice of nature does not reason. Thus,
the situation changes drastically once we
the first person; and (2) imperfect infor-
admit the possibility that 'nature' is not an
mation - both persons choose either simul-
anonymous entity, but is a person capable of
taneously or in ignorance of what each other
will do. With perfect information, case the(1),
same strategic thought as our initial de-
the situation can be analyzed straightfor- cision-maker. And this change requires the
wardly. If, for example, person 1 chooses development of new tools for thinking about
first, then he knows that if he chooses A, rational individual action.
person 2 will respond with D, whereas if he Implicit in our argument about circular
chooses B, person 2 will respond with C. reasoning, however, is the presumption that
Person 1, then, ought to choose A because not only is each decision-maker aware of
this alternative maximizes his minimum this situation, but each is aware that the
gain. other is aware, and so on. Game theorists
Matters achieve considerably greater refer to this assumption as the assumption of
complexity with imperfect information, casecommon knowledge. In the context of Sun
Tzu's analysis, such common knowledge will
(2), and it is here that we see the profound
arise if both sides to a conflict have read The
difference between Figures 1 and 2. Specifi-
cally, we can now imagine the following Art of War (or have an equivalent source of
thought process for person 1 as he contem- advice), and if both sides are aware that the
plates his options: other side has access to this volume. In
assessing the game theoretic credentials of
I think that I should choose A, because it offers meThe Art of War, then, one question we must
my best choice and provides a better guarantee than ask is whether and to what extent Sun Tzu's
B (2 versus 1). But then . . . if person 2 reasons as I
advice accommodates alternative solutions
do, he will infer that I will choose A, in which case
he will choose D, in which case I should choose B
to the dilemma of circular reasoning that
(since I prefer a payoff of 3 to a payoff of 2). But common knowledge admits.
then . . . if person 2 reasons as I do further, he will In order to make this evaluation, we must
infer my decision to switch to B in response to his first discuss the general components of a
choice of D, in which case he will conclude that C is
game. Briefly, these components are much
his better choice; in which case I ought to choose A.
But then, again, if he anticipates my reasoning, he those of the decision problem we outlined
will conclude that I will choose A, in which case he previously, except that now we must make
will respond with D . . . and so on. allowance for the fact that there are two or
more decision-makers. We also must make
Of course, circular reasoning will also char- allowances for the possibility of complex
acterize person 2's thinking. So suppose that strategic interaction in which people interact
person 1, exasperated and perplexed, con- over long intervals of time. To illustrate
cludes that person 2 will choose between C these ideas as we present them, we encour-
and D with equal probability. In this in- age the reader to keep a parlor game such as
stance, the expected value from A equals chess in mind. A description of any game,
4/2 + 2/2 = 3, whereas the expected value then, necessarily includes the following:
from B equals 3/2 + 1/2 = 2. Person 1, then,
might decide to choose A. But again, if (1) A list of relevant decision-makers or
person 1 believes that 2 will anticipate l's players. In chess, there are but two
tentative speculation that 2 will choose ran- players - white and black - whereas in
domly, I should also anticipate that 2 can card games there can be many more
infer l's choice of A, in which case 2 will not players.
choose randomly at all and will instead (2) A description of the strategic situation
choose D, because it is a best response to A. in terms of who moves at what time, and

This content downloaded from

3.104.43.49 on Thu, 21 Sep 2023 04:08:59 +00:00
All use subject to https://about.jstor.org/terms
166 Emerson M. S. Niou & Peter C. Ordeshook

in what order. In chess, white moves if he does not block them completely [then] I may
first, then black, etc. Chess also allows do so (X, 5).

long sequences of moves. In other It is important to appreciate the centrality

games, however, players may have to of this concept of a strategy to game theory.
move simultaneously, and they may If we have modeled the situation well, we
only allow short sequences, including can anticipate every possible contingency
only a single choice for each player. that a player can confront, including those
(3) A specification of the alternative actions contingencies established by nature. Thus, a
that each person can take at every strategy should allow a player to adjust his
opportunity that person has to act. In choices in accordance with things learned as
chess, White's first move allows ten
the game proceeds. And because a strategy
choices - each of eight pawns, plus thereby allows for shifting tactics depending
either knight. This list of alternatives on all possible contingencies, we can think
will expand or decrease as the game of the outcome of a game as being fully
proceeds, as other pieces are freed or determined after all players choose their
eliminated.
strategies (up to the uncertainties created by
(4) A specification of what each person nature), but before those strategies are
knows about the previous choices of actually implemented. Once strategies are
other players. In chess, a person knows
chosen events unfold according to plans, as
all the earlier moves of the opponent. In
the full character of each player's strategy is
other games, such as poker, these revealed over time.
moves may be hidden from view; and in Unsurprisingly, Sun Tzu seems to have
other circumstances, players may have
fully appreciated this idea. In addition to his
to make some choices simultaneously. early emphasis of correctly assessing situ-
(5) A description of each player's goals - an ation, Sun Tzu argues that:
evaluation of all of the final outcomes
that the game allows. In chess, as in . . . what is of supreme importance in war is to
attack the enemy's strategy (III, 4).
war, these outcomes include victory and
defeat, with the presumption that all . . . determine the enemy's plans and you will know
which strategy will be successful and which will not
players prefer victory.
(VI, 20).
The final idea we must introduce is that of a And as water shapes its flow in accordance with the
strategy - a plan of action for how to play ground, so an army manages its victory in accord-
the game. This plan - one for each player - ance with the situation of the enemy. And as water
has no constant form, there are in war no constant
specifies the alternative action a player
conditions. Thus, one able to gain the victory by
should choose for each and every con-
modifying his tactics in accordance with the enemy
tingency that that player can encounter as situation may be said to be divine (VI, 28-30).
the game unfolds. Strategies, then, consist
of contingent actions, and take the form: 'If and as to consequences,
my opponent does . . . then I will respond Know the enemy and know yourself; in a hundred
with . . .; but if my opponent chooses . . .. battles you will never be in peril. When you are
then respond with . . .'. In chess, a strategy ignorant of the enemy but know yourself, your
for white consists of an opening move, a chances of winning or losing are equal. If ignorant
both of your enemy and of yourself, you are certain
response to each of black's potential re-
in every battle to be in peril (III, 31-33).
sponses to the opening move, and so on.
Thus a victorious army wins its victories before
In the following passage, Sun Tzu is seeking battle; an army destined to defeat fights in
clearly seeking to spell out a complete strat- the hope of winning (IV, 14).
egy - a sequence of choices contingent on
what an opponent chooses: From the above passages, Sun Tzu's general
intent is clear - to analyze the diversity of
If I first occupy constricted ground [then] I must
interdependent choice situations in warfare
block the passes and await the enemy. [But] If the
enemy first occupies such ground and blocks the and to deduce efficient strategies - plans of
defiles [then] I should not follow (attack) him;action
[But] that lead to victory, broadly defined.

This content downloaded from

3.104.43.49 on Thu, 21 Sep 2023 04:08:59 +00:00
All use subject to https://about.jstor.org/terms
 A Game-Theoretic Interpretation 167

4. Solving Sequential Games Looking at Figure 2 directly, however, we

The issue, then, is how to identify good can see that if 1 chooses A, then 2 will
strategies so that we can resolve the circular choose D, whereas if 1 chooses B, then 2
reasoning that arises from games like the will choose C. Clearly, then, 1 should
one we offer in Figure 2. At this point we choose A and 2 should respond with D.
must return to the distinction offered earlier Thus, with sequential decisions, the out-
between two types of situations: (1) situ- come (2, 3) will prevail. In other words, we
ations in which the players make their can straightforwardly deduce that person 1
choices in sequence - one after the other -will choose the strategy 'A' and that person
as in parlor games like chess, go, and tic-tac-2 will choose the strategy 'Choose D if 1
toe; and (2) situations in which players must chooses A, but choose C if 1 chooses B'.
formulate strategies and make choices Game theory tells us, then, that in prin-
simultaneously, or, equivalently, in which ciple games such as chess can be 'solved'.
choices must be made in ignorance of the Practical considerations, of course, render it
opponent's actions. In this section we impossible to map out all moves in such
consider the first type of situation, gamesgames,
of which keeps us playing them as tests
perfect information. of skill and experience. On the other hand,
Consider a game like chess in which one a simpler game such as tic-tac-toe, because
player moves first, then the other, then the it too can be solved, is interesting only to
first again, then the second again, and so on. children who have not yet comprehended its
Suppose we can map out all actions in strategic structure. The skill and experience
sequence so that all possibilities are of playing chess well, though, is much like
covered. Some sequences will end quickly, the skill at playing any complex game -
as when one person makes a series of skilled using our experience and skills to simplify a
moves against an ineffective opponent. complex strategic structure so that we can
However, it is a well-known result of game learn strategies that ought to be avoided,
theory that, regardless of the complexity of and so that we can identify when our oppo-
the situation, in principal the game's out- nent is using a good or a poor strategy. Skill
come will be determinate and we will be and experience also help us reduce a game's
able to specify an unambiguously best strat- complexity so that, although its simplified
egy (plan of action for playing the game) for form may not match that of tic-tac-toe, its
each person. general principles can be understood and
This result is general. For example, recall optimal contingent plans formulated.
that if the game in Figure 2 is played sequen- And it is evident that the formulation of
tially with person 1 choosing first, then, as optimal contingent plans characterizes Sun
we have previously argued, the eventual Tzu's intent. For example. 'The doctrine of
outcome is (2, 3). Restated in terms of strat- war is to follow the enemy situation in order
egies, person 1 confronts only a single con- to decide on battle' (XI, 60).
tingency - the necessity for choosing first. We should consider one final issue with
Thus, he has only two strategies, A and B. respect to games in which the players move
Person 2, on the other hand, confronts two sequentially - namely, whether there is any
contingencies - person l's two choices - and advantage to moving first or second. Of
thereby he has four strategies: course, whether and what type of advan-
tages accrue to the first mover in a game in
sl: Choose C regardless of what person 1 which the players move in sequence
chooses. depends on the specific structure of the situ-
s2: Choose C if 1 chooses A, but choose ation under consideration. For example,
D if 1 chooses B. Sun Pin, writing one hundred years after
s3: Choose D if 1 chooses A, but choose Sun Tzu, recounts the now famous story of
C if 1 chooses B. the horse race in which the opponent pos-
s4: Choose D regardless of what person 1 sessed three horses generally superior to
chooses. T'ien Chi, the commander and chief of Ch'i.

This content downloaded from

3.104.43.49 on Thu, 21 Sep 2023 04:08:59 +00:00
All use subject to https://about.jstor.org/terms
168 Emerson M. S. Niou & Peter C. Ordeshook

Sun Pin's advice was simple: pair the worst This apparent confusion, however, is
horse against the opponent's best, the best resolved if we keep in mind the difference
against the opponent's second best, and the between strategy and action. In the first
second best against the opponent's worst. In cited passage, the positions of the players
this way, two out of three races can be won. have already been determined - the circum-
Clearly, moving second by having the stances of the game are fixed - and from that
option of determining the pairings of the point it is best to reveal one's strategy sec-
horses confers the advantage on T'ien Chi. ond. In contrast, in the second cited pass-
In order to interpret this example prop- age, the field of battle has yet to be deter-
erly, however, we must keep in mind the mined, and Sun Tzu is in fact arguing that it
distinction between actions and strategies. is better to be the one who dictates which
Strategies are plans - rules for selecting game is to be played or, equivalently, which
actions as contingencies arise - whereas player is to be assigned which position in the
actions are merely pieces of the plan. Sun game. In this second passage, then, Sun Tzu
Pin's example is especially simple because is referring to the first move in the game.
action (the assignment of horses) is equival- Thus, there is no confusion, and we can con-
ent to a strategy. In chess, on the other clude that Sun Tzu does in fact appreciate
hand, as well as in the complex maneuvers the advantages of choosing one's strategy
of war, the equivalence between actions andsecond.
strategy is lost.
With respect now to the issue of advan-
tages; for the class of games we are con- 5. Games With Imperfect Information
sidering here - games of pure conflict - the Prewar preparations - evaluating one's
advantage belongs to the player who reveals domestic power, recruiting skilled com-
his strategy last. But this does not necess- manders, training troops, and choosing
arily mean that the advantage belongs to the whether or not to engage in war - proceed
player who makes a second choice in a sequentially so that one's character, as well
sequential game. In chess, for example, as as that of an enemy, is revealed as events
in tic-tac-toe, the advantage belongs to unfold. Tactics are chosen by different
whoever moves first, which is different from rules. The success of battle tactics depends
revealing one's strategy. Now consider Sun on contingencies, including the opponent's
Tzu's view of advantages. In one instance he preparations and tactics, that often become
argues that rather than engage the enemy as apparent only after it is too late to condition
the first move, certain terrain dictates forc- on them, and decisions must be made with
ing the enemy to move first: imperfect information. In some instances,
this imperfect information arises because
Ground equally disadvantageous for both the
choices are made simultaneously by all an-
enemy and ourselves to enter is indecisive. The
nature of this ground is such that although the tagonists, whereas at other times choices are
enemy holds out a bait I do not go forth but entice not simultaneous but information is imper-
him by marching off. When I have drawn out half fect because choices are hidden from view.
his force, I can strike him advantageously (X, 4).
Regardless of its source, the game theorist's
But at another point Sun Tzu appears to approach to imperfect information is to
argue for the advantage of the first move: identify classes of games for which this fact
matters little. Referring to Figure 3, notice
Generally, he who occupies the field of battle first
and awaits his enemy at ease; he who comes later to
the scene and rushes into the fight is weary. And
therefore those skilled in war bring the enemy to theFigure 3. Two-Person Zero-Sum Game with Dominant
field of battle and are not brought there by him (VI,Strategy
1,2). c D
Ground which both we and the enemy can traverse
A 4,1 1 3,2
with equal ease is called accessible. In such ground,
he who first takes high sunny positions convenient B 2,3 1,4
to his supply routes can fight advantageously (X, 2).

This content downloaded from

3.104.43.49 on Thu, 21 Sep 2023 04:08:59 +00:00
All use subject to https://about.jstor.org/terms
 A Game-Theoretic Interpretation 169

that each player has a dominant choice - A Sequential choice, then, is sufficient to
for person 1 and D for person 2 - where by generate dominant strategies (but not
dominant we mean an action that is better necessary), and although, as we argue later,
than all others, regardless of what other Sun Tzu seems to have understood the
persons choose. In Figure 3, A is better than consequences of moving from a game of
B for person 1 regardless of what 2 chooses, simultaneous choice to one with sequential
whereas D is better than C for 2, regardless choices, much of his strategic analysis con-
of what 1 chooses. When decision-makers sists of identifying dominant and dominated
have dominant choices the analysis of ulti- strategies. First, with respect to dominated
mate decisions and outcomes avoids the cir- strategies - strategies that ought to be
cular reasoning that we applied to Figure 2.avoided in any context:
We should emphasize that the possibility
You should not encamp in low lying ground. ...
of a dominant choice is not altogether unre- You should not linger in desolate ground. . . . There
lated to the issue of simultaneous versus are some roads not to follow; some troops not to
sequential choice. Indeed, a game with strike; some cities not to assault; and some ground
sequential choices such as chess also oc- which should not be contested (VIII, 1-7).

casions dominant strategies, and thus we ... do not ascend to attack.... When an advancing
can view the concept of dominance as a gen- enemy crosses water do not meet him at the water's
edge.... Do not take positions downstream (IX, 2-
eralization of previous analysis. To see what 6).
we mean by this, recall that if the game in
Figure 2 is played sequentially with person 1 And with respect to dominant strategies,
choosing first and person 2 choosing second,
In enclosed ground resourcefulness is required. In
person 1 has two strategies, A and B, death ground, fight (VIII, 5, 6).
whereas person 2 has four strategies which
Encamp on high ground facing the sunny side; Fight
we denoted sl, s2, s3, and s4. Consider now downhill.... After crossing a river you must move
the game portrayed in Figure 4, which de- some distance away from it. It is advantageous to
allow half his force to cross and then strike. ...
Cross salt marshes speedily (IX, 1-8).
Figure 4. Normal Form of a Sequential Move Game \\7 ha.,^ ;intr^/4.,/^r^/ tU:i, nratirsn of a Hr1nai
VY IIdVC IILlUULUCU L111i IIJLIUII uI a uuiiii-

sl s2 s3 s4 nant strategy not only because it helps us

A
interpret parts of The Art of War, bu
1 ,1 2,3 2,3 because the existence of such strategies
B 1,4 3,2 1,4 3,2 helps us avoid circular reasoning. Specifi-
cally, notice the special feature of the
characterized by the joint choice of (A
in Figure 3 - once 'at' such a cell, neith
scribes t}
ie outcome that prevails for each of player has any incentive
the eight : possible joint choices of strategies change his decision. Indeed, it
by perso ns 1 and 2. Notice in particular that terminates c
although persn 1doe no hav a omiantteristic of (A, D) that termina
person 1 does not have a dominan reasoning. The existence of do
strategy, s3 is dominant for person 2 - it is
never we wrse and is sometimes better than si egies, however, is not essent
n A cA. rse and
notl? S nlnifrrmliy
1t ic sometinfrmes better
hottsMr than
thsn c/ sl ouion o a game. Consider
allU 3 UllU 1C 13 UIlIIly VII 1tilll3<
Thus, person 2 should choose s3. But, since
person 1 is also assumed to be cognizant of
Figure 5. Constant-Sum Game with a Nash Equilibrium
this game, person 1 should be able to infer
that 2 will use s3, in which case person 1 D E F

should choose A. This reasoning, of course, A

leads to precisely the same outcome we
deduced earlier - person 1 chooses A, 2 B

chooses D in response, and the outcome (2, C

3) prevails.

This content downloaded from

3.104.43.49 on Thu, 21 Sep 2023 04:08:59 +00:00
All use subject to https://about.jstor.org/terms
170 Emerson M. S. Niou & Peter C. Ordeshook

the cell corresponding to the joint choice of person assumes that his opponent is as intel-
(B, E). Notice first that neither player hasligent
a as he is and is capable of anticipating
dominant choice - for example, A is best forhis thoughts. In this event, the decision-
1 if 2 chooses D, B is best if 2 chooses E, and maker whose actions we are studying should
C is best if 2 chooses F. Nevertheless, once 'assume the worst' - should assume that
at (B, E) neither player has a unilateral regardless of what action he takes, his oppo-
incentive to defect to some other choice. nent will take best advantage of him. Bar-
Thus, (B, E) terminates circular reasoning ring the assumption that one's opponent is
as well - in other words, the existence of somehow less capable than oneself - always
dominant choices is sufficient to terminate a dangerous supposition and likely to lead to
such reasoning, but they are not necessary. unpleasant surprises - a person should then
Because it terminates circular reasoning, choose the strategy that maximizes one's
cells such as (B, E) - called Nash equilib- minimum gain, or equivalently, minimizes
rium after the theorist who proved a numberone's maximum loss (called a minmax strat-
of important results about them for differentegy).
classes of games - are profoundly important Applying this argument to the game in
as an idea about ultimate choices. Conse- Figure 5, notice that the minimum gain
quently, game theorists have devoted con- person 1 associates with A is 1, from B it is
5, and
siderable efforts analyzing the properties offrom C it is 2. Thus, person 1 should
equilibria, refining Nash's original formu-
choose B. Similarly, the minimum gain to
lation, applying those refinements to person 2 from D is 1, for E it is 4, and from
stations of far greater complexity than the F it is 0. Thus, person 2 should choose E.
simple games we describe here, and testing More interestingly, notice now that this
ideas in empirical contexts. If there is a cri- reasoning leads to the joint choice of (B, E),
tique that can be directed against Sun Tzu's which is the game's Nash equilibrium.
analysis of war, though, it is that he seems to Hence, in the case of games of pure conflict,
pay too little heed to the necessity of resolv- a prudent strategy leads to actions that are
ing the dilemma of circular reasoning, and consistent with the strategic imperatives
as a consequence he fails to infer this con- proscribed by game theory.
cept of an equilibrium as a solution to inter- Although we cannot find any direct
dependent decision-making. Put differently, reference in Sun Tzu to the notion of an
he fails to take full account of the possibility equilibrium as a means of terminating circu-
that, in addition to the king he is advising, lar strategic reasoning, it is not unreason-
enemy kings have also read The Art of War, able to suppose that he nevertheless grasped
and the fact of this common knowledge is the essence of a strategy designed to mini-
itself common knowledge. In that event, mize one's losses, but which would never-
only the notion of an equilibrium can be theless take advantage of an unskilled
used to formulate strategic plans and to opponent.
resolve circular reasoning.
Anciently the skillful warriors first made themselves
Despite this criticism, however, we can invincible and awaited the enemy's moment of vul-
find within game theory itself a reason for nerability (IV, 1).
supposing that this failure does not necess- Therefore the skilled commander takes up a pos-
arily negate the value of Sun Tzu's advice ition in which he cannot be defeated and misses no
and analysis. Specifically, one fact about opportunity to master his enemy (IV, 13).
Nash equilibria that applies to the types of Invincibility depends on one's self; the enemy's vul-
games that especially concerned Sun Tzu - nerability on him. It follows that those skilled in war
games of pure conflict in which one person's can make themselves invincible but cannot cause an
enemy to be certainly vulnerable. Therefore it is
gain is another person's loss (called zero-
said that one may know how to win, but cannot
sum or constant-sum games) - is that de- necessarily do so (IV, 2, 3, 4).
cision-makers are assured of achieving an
equilibrium if they abide by a simple rule of Sun Tzu, then, takes us at least part of the
thumb in choosing their actions. Suppose a way towards the notion of an equilibrium in

This content downloaded from

3.104.43.49 on Thu, 21 Sep 2023 04:08:59 +00:00
All use subject to https://about.jstor.org/terms
 A Game-Theoretic Interpretation 171

that if both sides to a conflict heed the to one or the other with certainty. That is,
advice of adopting strategies that minimize person 1 would be willing to stay with q and
one's potential losses against a skilled oppo- 1 - q if and only if the expected value he
nent but that are also viable against a less associates with A equals the expected value
skilled player, then an equilibrium prevails.he associates with B, as those expected
Thus, as long as the game under consider- values are determined by person 2's strat-
ation is constant-sum or zero-sum, and as egy. For the game in Figure 2, this equality
long as an equilibrium exists in simple strat- requires that
egies, Sun Tzu provides us with the requisite
4p + 1(1 - p) = p + 3(1 - p)
tools for achieving an equilibrium and for
taking advantage of an opponent who fails which we can solve, and conclude that p =
to act accordingly. 1/4. The same argument applies, of course,
to person 1- 1 should be willing to give A
and B some weight if and only if 2 is indiffer-
6. Mixed Strategies ent between C and D, which requires that,
One consequence of the preceding dis-
q + 4(1 - q) = 3q + 2(1 - q),
cussion is that whether or not a game is
characterized by perfect or imperfect infor- which we can solve to give q = 1/2.
mation matters little if there is a dominant The conclusion we reach here, then, is
choice or a well-defined equilibrium for a that if person 1 chooses randomly between
game. Players can 'solve' the game and A and B, and if person 2 chooses C with
arrive at determinate outcomes. However, probability 1/4 and D with probability 3/4,
military conflicts typically have a differentthen neither person has any incentive to
strategic character. Specifically, they gener-shift unilaterally to any other lottery
ally do not have an equilibrium in simple (including degenerate lotteries in which A
strategies. Thus, we must be careful about or B and C or D are chosen with certainty).
our arguments for the relevance of the Thus, for the situation portrayed in Figure
notion of an equilibrium, because if games 2, there exists a mixed strategy Nash equi-
do not always have equilibria then this non- librium.
existence precludes the possibility of scien- What gives this notion of a mixed strategy
tific generality. equilibrium special relevance is the import-
With this in mind, let us return to the ant theorem proved by Von Neumann &
game in Figure 2, which we used to illustrate Morgenstern (1944), which establishes that
circular reasoning and which, at first glance, every n-person game in which each decision-
does not appear to possess an equilibrium. maker has a finite number of choices has at
Recall our suggestion that person 1 might least one equilibrium in either mixed or
decide, out of exasperation, that person 2 pure strategies.2 Thus, the potential scien-
will make a random choice. We tentatively tific generality of the Nash equilibrium con-
rejected this idea because such an assump- cept is established.
tion did not avoid the circular reasoning we However, notice that a mixed strategy (1)
sought to avoid. However, notice that we minimizes one's vulnerability to an equally
assumed that person 2 used a particular lot- strategic opponent, and (2) takes advantage
tery, 50-50, and we did not check whether of an opponent who errs. Moreover, the use
all such lotteries led us in a cycle. of a random device in particular ensures that
So suppose more generally that person 2 one's tactics do not fall into a pattern that
chooses between alternatives C and D with an opponent can detect. Admitting mixed
probabilities p and 1 - p, and suppose also strategies into one's arsenal of choices has
that person 1 chooses between A and B with the consequence of taking a finite number of
probabilities q and 1 - q. Notice now that if pure strategies and rendering one's choices
person 1, given 2's probabilities of p and 1 - infinite in number. Moreover, the special
p, is not indifferent between the lotteries character of a mixed strategy is that even
provided by A and B, then he should switch after pure choices are revealed, an oppo-

This content downloaded from

3.104.43.49 on Thu, 21 Sep 2023 04:08:59 +00:00
All use subject to https://about.jstor.org/terms
172 Emerson M. S. Niou & Peter C. Ordeshook

nent cannot be certain that a choice that part of Sun Tzu's analysis into a game
appears inferior in the short term is not theory
part mold, since game theory assumes
of a grander, more all-encompassing plan - that all players are aware of all strategies
they preclude the possibility that an enemy available to an opponent. Nevertheless,
can infer a future choice with certainty having interpreted ch'i and cheng as evi-
based on previous actions. dence that Sun Tzu appreciated the role of
We cannot say whether Sun Tzu ever ex- mixed strategies, we can offer one criticism
plicitly advocated the use of random devicesof his analysis. Recall that in the previous
to disguise strategic intent, nor do the com-section we also cited evidence that Sun Tzu
mentators on his writings offer clarification. understood the role of minmax pure strat-
Nevertheless, we can offer a possible in- egies. The difficulty, however, is that if a
terpretation of his distinction between game possesses an equilibrium only in
normal (cheng) and extraordinary forces mixed strategies, minmax pure strategies
(ch'i) that renders the notion of a mixed cannot yield an equilibrium and, thus, they
strategy a centerpiece of his analysis: 'In cannot terminate cyclical reasoning. And,
battle there are only the normal and extra- unfortunately, what we cannot discover in
ordinary forces, but their combinations are The Art of War are any clear guidelines for
limitless; none can comprehend them all' ascertaining when a strategist should choose
(V, 8-11). minmax strategies and when he should
In Chinese, the character Ch'i means abide by mixed strategies. In this sense,
uncommon, unusual, or atypical. A strategy then, Sun Tzu's analysis is incomplete.
is ch'i if you think your opponent does not
anticipate it. Consequently, no strategy is
always ch'i or always cheng. A cheng can be 7. Secret Agents
a ch'i, and vice versa. It is whatever you Sun Tzu ends his text with a chapter on
think your opponent thinks about your secret agents and with the admonition that:
strategy. So the concept of ch'i and cheng
already reflects the strategic thinking be- . . . only the enlightened sovereign and the worthy
general who are able to use the most intelligent
tween the antagonists.3 If we interpret ch'i
people as agents are certain to achieve great things.
as a strategy that your opponent thinks you Secret operations are essential in war; upon them
are unlikely to use and cheng as a strategy the army relies to make its every move (XIII, 23).
that you are more likely to use, then to
make sure that any strategy you use will be The emphasis Sun Tzu places on secret
regarded as ch'i by your opponent, you must agents is understandable owing to the enor-
randomize your choices to keep the enemy mous strategic advantage to be gained from
guessing. By interpreting ch'i and cheng in knowing an opponent's strategy before-
this way, we can give fuller meaning to Sun hand. In effect, the role of the secret agent
Tzu's admonition that: is to allow a decision-maker to condition his
actions on a richer information base and to
It is according to the enemy's shapes that I lay the
render moves sequential rather than simul-
plans for victory, but the multitude does not com-
prehend this. Although everyone can see the taneous. To see the advantages of this
outward aspects, none understands the way in change, consider, for example, the game in
which I have created victory. Therefore, when I Figure 2. If choices are made simultaneously
have won a victory, I do not repeat my tactics but
or, equivalently, if no player in the game has
respond to circumstances in an infinite variety of
ways (VI, 25, 26).
perfect information, and if each person uses
his equilibrium mixed strategy, then the
Admittedly, there is an alternative way of expected payoff to each person is 2.5. On
interpreting ch'i and cheng: a strategy is the other hand, as we have already seen, if
regarded by your opponent as ch'i if your person 2 moves after 1, but knows person
opponent has no idea whether it is possible l's choice beforehand, then the final payoffs
or feasible. If we accept this interpretation, are 2 to person 1 and 3 to person 2. More-
then we cannot so easily (if at all) set this over, if person 1 is unaware of 2's better

This content downloaded from

3.104.43.49 on Thu, 21 Sep 2023 04:08:59 +00:00
All use subject to https://about.jstor.org/terms
 A Game-Theoretic Interpretation 173

information and continues to employ a nipulations are possible, the analysis of

mixed strategy, then 2's expected payoffstrategy becomes doubly complex.
rises higher to 3.5, and person l's drops to Earlier we interpreted Sun Tzu's admoni-
1.5: 'The enemy must not know where I tion that 'all warfare is based on deception'
intend to give battle. For if he does not as implying that game theory as opposed to
know where I intend to give battle he must simple decision theory is required to under-
prepare in a great many places' (VI, 14). stand his analysis. Certainly, mixed strat-
The role of secret agents, therefore, is egies are one form of deception. However,
clear - to allow a decision-maker to con- it is also evident that deception meant more
dition on his opponent's choice and toto pro-
Sun Tzu than merely randomizing one's
ceed through the play of the game as if it
choices:
were a sequential game:

Now the reason the enlightened prince and the
general conquer the enemy whenever they move
and their achievements surpass those of ordinary
men is foreknowledge. What is called 'foreknowl-
edge' cannot be elicited from spirits, nor from gods,
nor by analogy with past events, nor from calcu-
lations. It must be obtained from men who know
the enemy situation (XIII, 3, 4).
. . . when capable, feign incapacity; when active,
wise
inactivity. When near, make it appear that you are
far away; when far away, that you are near. Offer
the enemy bait to lure him; feign disorder and strike
him (I, 18-20).
Pretend inferiority and encourage his [the enemy
general's] arrogance (I, 23).

For the particular class of games with

which Sun Tzu is concerned, choosing one's Once we accept this view of deception, how-
strategy with foreknowledge of an oppo- ever, a variety of new questions arise, such
nent's choices is always advantageous - as: Is it possible to deceive an opponent
regardless of whether the opponent is aware when that opponent is aware of one's intent
of one's foreknowledge or not. The issue and opportunities? Can players deceive
with which Sun Tzu seems to have the each other simultaneously when each knows
greatest difficulty, however, is the possi- that the other is trying to deceive, and when
bility that both sides will use agents to feed each knows that the other knows?
each other false information. He advises In judging Sun Tzu's contribution to our
that: understanding of strategy, however, we
should keep in mind that although game
It is essential to seek out enemy agents who have theory is a field that has undergone nearly
come to conduct espionage against you and to bribe
fifty years of development, with hundreds of
them to serve you. Give them instructions and care
for them. Thus double agents are recruited and used researchers participating in the process, we
(XIII, 17). have only recently learned how to treat the
manipulation of information as part of a
But what of the possibility that double player's strategic arsenal (e.g. Kreps, 1990;
agents are triple agents, and so forth? Once
Myerson, 1991). Understanding the stra-
this question is admitted, we return again to tegic manipulation of information is es-
the dilemma of circular reasoning that game pecially difficult because it deals with the
theory tries to resolve. Now, however, the potential for altering an opponent's view of
game is characterized not only by imperfect the game that is being played when that
information but also by incomplete infor- opponent knows it is in your interest thus to
mation, and the players must be concerned manipulate. Resolving circular reasoning in
with what others believe about them - about this circumstance requires the use of ad-
their capabilities and their intent - as a func- vanced principals of probability theory and
tion of their choices. Thus, choices must be mathematics, and so we should not be sur-
selected not only to manipulate outcomes prised to learn that Sun Tzu's treatment
directly, but also to manipulate them in- of information is incomplete. Indeed, we
directly through the manipulation of beliefs. should marvel at the fact that he understood
And when all players know that such ma- intuitively as much as he did.

This content downloaded from

3.104.43.49 on Thu, 21 Sep 2023 04:08:59 +00:00
All use subject to https://about.jstor.org/terms
174 Emerson M. S. Niou & Peter C. Ordeshook

NOTES REFERENCES
1. Throughout this essay our quotations of Sun Tzu's Chang, Cheng-Tse, 1984, Sun Pin Pin Fa Chao Li [
The Art of War are taken from Samuel B. Griffith's Reorganization and Rearrangement of The Art of
(1963) translation. The notation following each War by Sun Pin]. Beijing: Chong-Hwa Press.
quote refers to chapter and paragraph as indexed by Cleary, Thomas, 1988. The Art of War: Sun Tzu (tran
Griffith. To ensure that Griffith's translation is con- lation). Boston, MA: Shambhala Publications.
sistent with other more contemporary interpret- Griffith, Samuel, B., 1963. Sun Tzu: The Art of W
ations, we consulted Shih I Chia Chu Sun Tzu (translation). London: Oxford University Press.
(Eleven Schools of Thought on Sun Tzu), inter- Kreps, David M., 1990. A Course in Microeconomic
preted by Kuo Hwa Zuo (1978) and The Art of War Theory. Princeton, NJ: Princeton University Press.
translated by Thomas Cleary (1988). Kuo, Hwa Zuo, 1978. Shih I Chia Chu Sun Tzu [Eleven
2. By pure strategies we mean the strategies formed by Schools of Thought on Sun Tzul. Shanghai: Shanghai
a direct analysis of the game's structure, whereas Ancient Book Publishing House.
mixed strategies correspond to lotteries over these Myerson, Roger B., 1991. Game Theory: Analysis of
pure strategies. Conflict. Cambridge, MA: Harvard University Press.
3. Notice that the concept of ch'i and cheng was elabor-Ordeshook, Peter C., 1986. Game Theory and Political
ated in The Art of War written by Sun Pin a hundred Theory. Cambridge, MA: Cambridge University
years later. Our interpretation of this concept Press.
follows from Sun Pin. von Neumann, John & Oskar Morgenstern, 1944.
Theory of Games and Economic Behavior. Prince-
ton, NJ: Princeton University Press.

EMERSON M. S. NIOU, b. 1958, PhD in Political Science (University of Texas at Austin, 1987);
Associate Professor of Political Science at Duke University. He specializes in formal theory, inter-
national relations, political economy, and Chinese politics. He is the coauthor of The Balance of
Power (Cambridge University Press, 1989). His most recent research focuses on the formation and
dissolution of alliances in anarchic international systems.

PETER C. ORDESHOOK, b. 1942, PhD in Political Science (University of Rochester, 1969);

Professor of Political Science at the California Institute of Technology. He is the author of Game
Theory and Political Theory and A Primer in Political Theory, coauthor of Positive Political Theory
(with William Riker) and The Balance of Power (with Emerson M. S. Niou and Gregory Rose). His
primary research interest at present is the study and design of electoral, constitutional, and inter-
national institutions, with special emphasis on those things that facilitate cooperation and coordi-
nation in complex environments.

This content downloaded from

3.104.43.49 on Thu, 21 Sep 2023 04:08:59 +00:00
All use subject to https://about.jstor.org/terms