You are on page 1of 222

Complexity in World Politics

Concepts and Methods of a New Paradigm

Edited by

Neil E. Harrison


SUNY series in Global Politics James N. Rosenau, editor

Concepts and Methods of a New Paradigm

Edited by Neil E. Harrison

State University of New York Press

recording. 194 Washington Avenue. JZ1305. Neil E. NY 12210-2384 Production by Judith Block Marketing by Michael Campochiaro Library of Congress Cataloging-in-Publication Data Complexity in world politics : concepts and methods of a new paradigm / edited by Neil E. Harrison. Complexity (Philosophy) I. mechanical. cm.Published by State University of New York Press. Series. 3. p. — (SUNY series in global politics) Includes bibliographical references and index. For information. International relations—Philosophy. International relations—Methodology. paper) 10 9 8 7 6 5 4 3 2 1 . electrostatic. Suite 305. Harrison. 2.1'01—dc22 2005024118 ISBN-13: 978-0-7914-6807-4 (hardcover : alk.. Albany © 2006 State University of New York All rights reserved Printed in the United States of America No part of this book may be used or reproduced in any manner whatsoever without written permission.C657 2006 327. or otherwise without the prior permission in writing of the publisher. paper) 1. No part of this book may be stored in a retrieval system or transmitted in any form or by any means including electronic. ISBN 0-7914-6807-0 (hardcover : alk. address State University of New York Press. Albany. magnetic tape. photocopying. 1949– II.

Clemens. Rosenau 1 2 25 3 43 4 73 5 95 6 121 7 137 8 143 v . David Singer Complexity and Conflict Resolution Dennis J.Contents 1 Thinking About the World We Make Neil E. Jr. Sandole Understanding and Coping with Ethnic Conflict and Development Issues in Post-Soviet Eurasia Walter C. Harrison Complexity Is More Than Systems Theory Neil E. Hoffmann Agent-Based Models in the Study of Ethnic Norms and Violence Ravi Bhavnani Alternative Uses of Simulation Robert Axelrod Signifying Nothing? What Complex Systems Theory Can and Cannot Tell Us about Global Politics David C. D. Earnest and James N. Beyond Regime Theory: Complex Adaptation and the Ozone Depletion Regime Matthew J. Harrison with J.

Harrison 165 10 183 Contributors List of Titles. SUNY series in Global Politics Index 197 201 205 .vi CONTENTS 9 When Worlds Collide: Reflections on the Credible Uses of Agent-Based Models in International and Global Studies Desmond Saunders-Newton Complex Systems and the Practice of World Politics Neil E.

Some believe that international theorists think too small and fail to synthesize relevant insights from a range of disciplines (Buzan and Little 2001). we still know so little about international relations and world politics that events continue to surprise us. Out of the blue. Booth. A year earlier. and postmodern scholars reject the ahistorical. In this book my colleagues and I describe. The abrupt end of the much-studied Cold War was widely unanticipated. civic strife in Venezuela influences the price of oil. and demonstrate the benefits of. as were the consequences of the collapse of communism in Europe. others criticize the emphasis on positivist methods (Smith. The defining characteristics of four decades of international politics were erased in a few short years. and the needs of AIDS patients in South Africa challenge international agreements on intellectual property. terrorists attacked within the United States one sunny September morning. George and Campbell 1990). a paradigm of system emergence from complex 1 . The 1997 Asian finance crisis rattled the US and European stock markets. It argues that the reality of world politics is more complex than dreamt of in current theories. Current theories of world politics assume that the social world is appropriately modeled as a simple system. Harrison Despite nearly a hundred years of theorizing. scholars and practitioners alike are constantly surprised by international and global political events. There is no agreement on the cause of this failure. After so much ink has been spilt. rationalist foundations of most international theory (Der Derian and Shapiro 1989. This book takes a different tack. in the space of a few months the global economy lurched from rapid expansion to recession and flirted with deflation. this book proposes that it should instead be viewed as a complex system. but the globalization of economic and social life has continued.CHAPTER 1 Thinking About the World We Make Neil E. and Zalewski 1996).

and the Sun. the definition of the boundary is a convenience used to assist human analysis. and theories founded in this approach will generate radically different solutions to policy problems.”1 Like realism. or predictive theories— formal statements that generate empirically testable hypotheses—based in complexity ideas and concepts are “complex systems theories. A pond is only arbitrarily separated from its shoreline. it validates novel research methods. HARRISON agent interactions that we call “complexity”. complexity views politics as emerging from interactions among interdependent but individual agents within evolving institutional formations. explanatory. Realism assumes that essential human characteristics drive political behavior within fixed structures. So world politics is a more or less self-organizing complex system in which macroproperties emerge from microinteractions. as is an animal or a country. its economy (where the distinction between gross national and gross domestic product is important). Usually. In the final section of this chapter I outline the rest of the book. I show how complexity concepts can be used in theories of world politics.2 NEIL E. An atom is a system. or its state or government. This and the next chapter outline a taxonomy of the central ideas and concepts of a complexity paradigm of world politics from which useful theories or models of complex world politics may be constructed. as when scientists define for study an individual ecosystem. I compare basic concepts of simple and complex systems and thereby frame a complexity paradigm. An automobile may be complicated. In the next section. permitting reasonable prediction of future system states. A system is simple if the units and their relations are relatively fixed. the air. or ideological orientation about the essence of political reality that organizes theorizing about and empirically investigating events in world politics. The existence of workshop manuals further illustrates the simplicity of the system: they identify all potential problems and explain how to . but it is a simple system. complexity is a thought pattern. a definition of “country” may be in terms of its recognized sovereign territory. The study of complexity in systems is “complexity science” and descriptive. FROM SIMPLE TO COMPLEX A system is a portion of the universe within a defined boundary. This ontological shift from simple to complex systems opens new paths to knowledge and understanding yet incorporates much current knowledge. and the actions of all the parts are centrally coordinated toward a collective outcome. Similarly. set of beliefs. its terrain and ecosystems. outside of which lies an environment. Following that. Each of the parts has a specific role in the system.

THINKING ABOUT THE WORLD WE MAKE 3 remedy them. One measure of complexity is the length of the shortest possible message that fully describes the system (Gell-Mann 1994. Thus. they are commonly called “agents. not only are the units diverse but each has a range of freedom of choice denied to parts in a mechanical system.1. only one unit need be described in detail. of the range of their available choices.2 But if the units also have behavioral discretion. define the characteristics of each part. They also illustrate problems and solutions. The two primary differences between complex and simple systems are diversity and decentralization. 30–38). As table 1. system description is shorter. but there is centralized coordination of their operations through mechanical or electronic management systems. In an automobile there are many diverse parts constructed for very specialized roles. Centralization of decision-making simplifies complicated systems. Because units in a complex system have discretion in their choice of behavior. system description requires description of the units (perhaps by class).1 shows. and of the rules of behavior that each will follow in making their individual choices.” Decentralized decision-making increases complexity. and the range of relations between them in exhaustive detail. Description of a jaguar in the jungle is longer than of a quark (a unit within an atom). If all the units of a system are identical. CHARACTERISTICS OF SIMPLE AND COMPLEX SYSTEMS Simple Systems Few agents Few interactions Centralized decision-making Decomposable Closed system Static Tend to equilibrium Few feedback loops Predictable outcomes Examples: Pendulum Bicycle Engine Boyle’s law Gravitational system Complex Systems Many agents Many interactions Decentralized decision-making Irreducible Open system Dynamic Dissipative Many feedback loops Surprising outcomes Examples: Immune systems Genes Molecules in air Ecosystems Markets . Modern automobiles have sophisticated management systems that use miniature TABLE 1. a living system is complex in many ways that an automobile is not. heterogeneity among the units increases description length. In living systems.

and positive feedback speeds them up. Unfortunately. It is nothing more than its parts and their defined relationships. The laboratory is designed to close the system under study. Although these systems respond almost instantaneously to multiple indicators. that a system is simple rather than complex simplifies analysis. usually implicit. Boyle’s law that pressure and temperature are inversely related can be demonstrated to be true only within a closed cylinder within a controlled environment. a living system perpetually changes. social systems are always open. Only Victor Frankenstein has yet been able to deconstruct and reconstruct a human and breathe life into it. Bacteria have fewer choices of behavior than ants. Disassembling a living system. If the system is simple. can destroy the system or. does not exchange energy with them. Simple systems usually are static and tend to equilibrium. complex systems are always dynamic and they are dissipative. it can be decomposed into its parts. Negative feedback slows down processes. And we are “dissipative structures” because we have to draw energy from our environment in the form of oxygen. The automobile is a static system that remains in equilibrium if no energy (for example. Humans age and die. Mammal societies are more complex than anthills or bacterial infections. a dynamic process of constant change in the cells within our bodies and the relationships between them. These centralized management systems prohibit freedom of choice in the units. decision-making is decentralized. Even in simple systems. The automobile can be disassembled and reconstructed and work just as well as it did before. or removing any part of it. making the system more complex. It declines only marginally by interaction with its environment (to that extent. make it much less than it was previously. there is only a single programmed response to any change in system condition. at least. The desire to simplify analysis also leads to the common assumption that the system under study is closed to other systems. food and water merely to stay alive (Prigogine and Stengers 1984). which are. In living systems. more regimented and less “free” than animals. As . a simple system can remain largely unchanged for long periods. and is not affected by them. HARRISON computers to govern feedback cycles and responses to environmental changes. it is an open system).4 NEIL E. As the “degrees of freedom” of choice for individual members in a system increase. effects can feedback on their causes. and units can choose their actions. gasoline and human control) is added to the system. the range of individual behaviors increases. This is most clearly illustrated by the “arrow of time” (Prigogine 1997). The desire among social scientists for closed systems reflects their common admiration for the analytical control of the laboratory sciences. in turn. and wishing them closed often makes assumption of closure unreasonable. In contrast. Without an input of energy. The common assumption. The thermostat is the classic example of a simple system with a negative feedback loop.

For example. Complex systems are unpredictable. prediction as a path-dependent extrapolation of historical processes runs the risk of nonlinear change. In complex systems. The collapse of the Anasazi civilization in the American Southwest has been explained by the interaction of social rules and environmental changes (Axtell et al. we should expect that “complexity will be pervasive in the world around us” (76) both natural and social. satiation. . Kremer. but they may be simulated with interacting rules for agent behavior. behavior can change from punishment/reward contact with the environment. The homeostatic behavior of animals reflects feedback from activity (hunger. The location of water temples in Bali can be simulated with a few rules of kinship and farming practice (Lansing. the range of possible system paths for a complex system widens dramatically. the circuit opens and the furnace shuts down. For example. When the air is returned to its set point.THINKING ABOUT THE WORLD WE MAKE 5 the air cools below the set-point temperature. and the prevalence of positive feedback loops inject further uncertainty into the system under study. sleep). hunt. 241–43).” That hope is partially fulfilled by simulations of social systems with agent-based models in which systems are modeled from the interactive behavior of essential agents. atmospheric scientists hypothesize that positive feedback loops caused Venus’s swirling toxic mists and 900-degree surface temperatures (Schneider 1989). and most complex systems are potentially nonlinear. an electrical circuit closes to turn on the furnace and blow hot air into the room. as described throughout this book. Complex systems usually have multiple feedback loops. Environmental selection operates on the individual agent as a form of feedback. In comparison to an automobile. 2002). 76). the flocking behavior of birds looks random and disorganized but can be modeled with three rules (Waldrop 1992. and Smuts 1998). By its nature. the game of checkers seems uncomplicated. nonlinearity is unpredictable and difficult to represent mathematically. Beyond the very short term. yet the outcome of their interaction can simulate complex systems in which agent behavior appears random and system order seems accidental. Positive feedback loops strengthen the cause and the subsequent effect in an ever increasing cycle that can lead to nonlinear transitions and system collapse. Some scientists fear that climate change on Earth could also progress with a nonlinear shift in the system (Ocean Studies Board et al. But it also “gives hope that we can find simple rule-governed models of that complexity. Decentralized decision-making and diversity among agents permits a wide range of agent actions and openness to changes in environmental conditions (the state of another complex system). Complex systems may not be predictable. These rules may be few and simple. Because complexity emerges from the simple rules of checkers. Yet it “provides an almost inexhaustible variety of settings (board configurations)” (Holland 1998. 2001).

For example. COMPLEX SYSTEMS IN WORLD POLITICS Intuitively. the properties of gases can be reduced to the mathematically describable motion of their atoms or molecules. the greater is the probability that that system is of that genus. while some simple systems appear dynamic. 65–67.g. 95–159).” Newton described a universe formed out of particles that were all made from the same material and whose movements in absolute space and time were governed by forces that followed unchanging and universal laws. world politics theories are “reposed in deep Newtonian slumber. while neoclassical economics remains the dominant explanation of economic phenomena. 47). at least temporarily. These laws could be expressed exactly through mathematics (Capra 1982. HARRISON The characteristics summarized in table 1. Thus.6 NEIL E. and property rights (Locke 1980. aided by mathematics. so human individuals would settle down in a society in a ‘state of nature. was the method for prizing open the watch case to see the workings inside (Hollis and Smith 1990. No single descriptor defines either simple or complex systems. 123–27.’” Natural laws included freedom. simple or complex. equality. The next section shows how complexity concepts can be used to construct a complex systems taxonomy of world politics (that is further elaborated in chapter 2). However. it is “an economic science after the model . and “natural laws” governed spontaneous human society: “As the atoms in a gas would establish a balanced state.. Science. in equilibrium. the more descriptors of one system genus that can be attributed to a specific system. For example. Locke and other early political and social theorists enthusiastically emulated Newton and attempted “to reduce the patterns observed in society to the behavior of its individuals” (Capra 1982. The shadow of Newton’s universe continues to obfuscate knowledge in the social sciences. simple systems may have many and diverse parts and complex systems (e.1 and described in this section are most commonly associated with each genus of system. Ruggie 1993). Kymlicka 1990. 69). of bacteria) may have homogeneous units. Thus. and complex systems can be. the social world seems complex in the sense described here. the image is of a universe constructed like a perfect mechanical watch. but current theories of world politics model it as a simple system. As Ruggie (1993) comments. complexity is an accumulation of the characteristics of complex systems. A fixed human nature was presumed to determine human behavior. For example.

agency is limited to only economic interests and programmed responses to external stimuli. In such a model. subjective self-interest. The basic unit of any social group is the individual. For example. people. Stanley Jevons—as ‘the mechanics of utility and selfinterest’” (Georgescu-Roegen 1975. Economic actors are assumed to be rational in their pursuit of undefined. Groups may be local or national. concentrating on the state as the unit of analysis causes an analytically convenient but arbitrary separation of international and domestic politics. The result is several significant simplifications of reality. and any social group. borrowed from neoclassical economics. While it is now historically located within international society. the human body is a system. In biological terms. Social and political institutions emerge from the interaction of individual humans and human groups. Out of the interactions among this mélange of groups and individuals emerges the set of institutions. A rational-choice approach. including the state. Constructivist theories—the most recent incarnation of liberalism—posit that state interests and identities are intersubjectively malleable at the margin through interaction with other states. the state remains the unit of analysis. Essentially identical units—interests and identities are assumed to be exogenously formed— are driven by “natural laws” to behave predictably in response to exogenously determined conditions. and practices that scholars call the “state. emphasis in the original).THINKING ABOUT THE WORLD WE MAKE 7 of mechanics—in the words of W. socially. Thus. Their behavior is assumed to be an objectively rational response to external forces such as the level of supply and demand of goods and services. created by the interaction of the units. and may be more or less centrally organized. Emergence A complex system is commonly described as more than the sum of its parts. universal explanations of relations between states. not change” privileges structure over agency (Smith 2004). If supply exceeds demand and prices fall. they may be looseknit coalitions or adhesive groups of fervent followers. and the theoretical focus on “explaining constancies. That is. I start with the state to better illustrate the primary concepts of a complexity taxonomy of world politics.” . each human is an essential unit within several systems. economic actors will increase their purchases. is used in an attempt to generate ahistorical. properties of the system are emergent. as in rational-choice theories. is an emergent system. Recent debates about agency and structure do not hide the similarly mechanistic paradigm that still drives orthodox theories of world politics.

Jacobson. Korten 1995. The assumption of closure thereby permits historical theorizing and supports the widespread belief among scholars that general laws can be found. Although the state is evidently an open system.3 The state also is open to other states and. 1996 among many others). In Holland’s (1995) terminology. Putnam 1988). the state is usually modeled as a unit with exogenous identity and objective interests. and Putnam 1993. Thus. But that is no reason to model them as closed systems. Despite occasional attempts to bring in domestic politics (Evans. as constructivism argues. unions. State behavior then results from the interaction of internal model and external . Gilpin 1996 and Strange 1994. open dynamic systems are inherently unpredictable (Doran 1999). Meta-agents The state is both an emergent system and a unit within the international system of states. it is open to technological. and Conklin 1996. For example. they are “meta-agents” whose “internal models” (discussed below) emerge from the interaction of domestic agents. under closure (Patomäki and Wight 2000). defined as a political system. because “constant conjunctions (empirical regularities) in general only obtain under experimentally controlled conditions”—that is. For example. Small changes that can initiate a radical system shift may come from a change in environmental conditions. simplifying causal analysis and hypothesis generation and testing. The nonlinearity of open systems prevents the theorist from mapping specific causes to observed effects. This would be impossible if social systems were modeled as open. and nongovernmental organizations (NGOs) cross boundaries and operate in several jurisdictions simultaneously (Goddard. on political economy. This greatly reduces the range of possible causal explanations for any perceived social event. cultural. and economic systems that influence political choices and processes (Skolnikoff 1993.8 NEIL E. from interactions among its constituent agents. qualitative change and emergence” (232) and “outcomes might be the result of many different causes and the same cause might lead to different outcomes” (229). HARRISON Open Systems The state is not a closed system: it is open to other natural and social systems. PasséSmith. Some social systems are both within and outside the state. or from inside. theories of world politics conventionally assume that all systems are closed to their environment much as optimal natural science experiments are controlled and isolated from unwanted external influences. is influenced by interactions with them. Open systems are “susceptible to external influences and internal. Keohane and Nye 1971). major corporations.

but the state still is assumed to be a unitary actor whose identity and interests change primarily as a result of interaction with other states (Wendt 1994). 16). but there is no assumption that the outcomes of an agent’s choices will be individually or collectively rational or will match agent intent. not unlike the concepts elaborated in Putnam’s (1988) two-level game. 6) to explain the foundations of rational-choice theory. orthodox international relations (IR) theory usually takes the state as the primary unit of interest. 186–92). This is not Simon’s “substantive conception of rationality” quoted and approved by Keohane (1988): “‘behavior that can be adjudged objectively to be optimally adapted to the situation. Rationality is subjective—within the agent—rather than objective. Constructivism and other cognitive theories treat states as subjects. Each agent can have unique desires and unique beliefs about how to achieve them. Beliefs and desires must be free of internal contradictions. behaviors of individual agents are “optimally adapted” to their situation only accidentally. change their internal models and. The concept of meta-agents can be used in any issue area in which agents and actions at more than one level of aggregation are involved. complexity does not make the simplifying jump to an assumption of objective rationality. an action is rational if it is the best way for an actor to satisfy his or her desire based on beliefs that are optimal given the available evidence and as much information as possible.THINKING ABOUT THE WORLD WE MAKE 9 reality. by assuming diversity among agents. but the potential for change as a result of domestic political discourse is usually disregarded (Hasenclever. and feedback is available on whether internal system processes and state behavior “fit” within the environment. Agents who learn from such an experience. the essential unit of any social system. Individual agents’ behaviors are responsive and purposeful but not objectively rational. their behavior. Finally. The extent to which states also may self-consciously change their interests and identity is debated. The alignment of behavior with desires and beliefs indicates agent rationality. though in a more fluid and dynamic relationship. This is substantially the same description of rationality used by Green and Shapiro (1994. . thus. while recognizing in passing the potential influence of substate and nonstate actors. and Rittberger 1997.” Because agents cannot predict the effects of their actions in complex systems. However. they will act inappropriately. given the desire. fail to achieve their goals. According to Elster (1986. Mayer.4 If their beliefs are out of synch with reality. Internal Models Each human agent. has an internal model of his or her desires and beliefs about how to achieve those desires in the world. In contrast. actions must be the intended result of beliefs and desires. and may be punished.

International norms influence behavior through the internal process of internal model formation. the state may be not be able to move its internal model—particularly in terms of its (causal) beliefs about what is possible—to accord with the reality of the international system. a sunflower. The “organization of a living system is the set of relations among its components that characterize the system as belonging to a particular class (such as bacterium. If all states are adaptive complex systems. like neorealism and world systems. a cat. Because states are themselves systems. one component of which is the desire to participate in a society of nation-states. In chapter 5.10 NEIL E. structure is a fixed or only slowly changing determinant of agent behavior. common to all living systems. As Putnam (1988) has suggested. The distinction is in the details: in conventional systems theories. or a human brain)” (Maturana and Varela 1980. Internal models drive agent behavior. the similarity is more perceived that real. . 91) describes the international system. as shown in more detail in chapter 2. simulation of agent behavior now is possible. like economic markets. Agents that consistently act in ways that are selected by their social environment as suboptimal face eradication. are individualist in origin.” and its structure is “formed by the coaction of their units” and “emerge[s] from the coexistence of states. However. . In complex systems. For example. The concept of internal models potentially extends the ideational content of world politics theories.” The structure of a complex system is the actual relations among actual physical components: “[I]n . . then the international system emerges from coevolution. Dynamic Systems Superficially. it is “formed by the coaction of self-regarding units. but those models may change when tested in a selective environment. To describe the organization. the process of matching internal model to external reality is one of trial and error. whichever the nature of their components. As Waltz (1979. Hoffmann investigates how states changed their beliefs during negotiations over regulation of ozone depleting substances and how the internal model of the United States adapted to these changes. it is only necessary to describe the relationships and not the components. International-political systems. 18). HARRISON Constructivism broadens “the array of ideational factors that affect international outcomes” and introduces “logically prior constitutive rules alongside regulative rules” (Ruggie 1998). structure is dynamic but “organization” is fixed. and unintended. self-organization is “a general pattern of organization. spontaneously generated.” However. while at the same time making analysis of agent motives more difficult. complexity appears to have some affinity with other world politics systems theories.

the system’s structure is the physical embodiments of its organization. It is equally untenable. Thus. Because “rules. In complex systems. As the momentary embodiment of prior agent interactions. the conditions were unique to each period. The common (usually implicit) assumption that the international system is homeostatic is a stronger version of the orthodox presumption that events in different spatiotemporal locations may be compared. organizational forms. Complexity science makes no such assertions: it does not assume or judge the fitness or efficiency of emergent institutional arrangements. Each state of balance. like a human standing still through tensions between opposing muscles. . Tudor England understood the need to change alliances to continually balance power in Europe. a point on a path of change. Causation The uncertainty of complex social systems calls into question conventional world politics assumptions about causation. structure has a social role but no purpose. is a fleeting event within a specific set of conditions. the international system returns to a balance between many forces. Despite its “balance of power” bromide. 959). The dynamic European system has found several momentary points of balance between myriad forces. “history is pathdependent in the sense that the character of current institutions depends not only on current conditions but also on the historical path of institutional development” (March and Olsen 1998.THINKING ABOUT THE WORLD WE MAKE 11 other words. Institutions and rules are the consequence of history but may not fit agents’ purposes. Simple dynamic systems find a point of equilibrium that is “sustained by micro-mechanisms operating in finely attuned and compensating ways” (Elster 1983. predictable from that environment” (958). structure is not fixed but a fleeting embodiment—in social systems manifested by institutions—of the deep organization within apparent chaos. Though power was balanced in Europe before World War I and in the Cold War.” While organization is static—a cat cannot become a dog—structure is dynamic. . Complex social systems are never homeostatic: in both markets and world politics the frequent and temporary equilibrium points are always distinct phenomena. Realism presumes that just as the neoclassical market continually returns to an equilibrium between demand and supply. surviving institutions are seen as uniquely fit to the environment. thus. classical realism is really about the processes of systemic change from dynamic forces. In functional social theories like constructivism and neoliberalism. norms. and institutions that exist are the inexorable products of an efficient history . complex system structure changes dynamically. Conventional world politics . 31–32). identities.



theories presume that causation is proximate and proportionate. Like most of social science, they have adopted Hume’s rules for causal explanations (Hume 1975).5 These rules require that the cause can be shown to precede the effect, that cause and effect are contiguous (there was no intermediate event), and that there is a “necessary connection” between events such that this cause can be shown to always precede this effect under consistent conditions. For at least four reasons, these rules are not appropriate causal explanations in complex social systems. First, they only apply in closed systems in which conditions can be controlled. But if social systems are open, it is unlikely that conditions will remain constant or be comparable between different states of affairs. In an open system, a cause may have different effects at different times due to changed conditions. Therefore, it is not surprising that no general laws of world politics have been found. Second, social systems are so complex that parsimonious theories that attempt to isolate single (or few) causes for observed effects may dangerously oversimplify models. In complex social systems, the events noted at the start of this chapter (among others) are surprising only when we expect to find a singular cause. Understood as the emergent consequence of multiple interacting prior events, such events are less astounding. The events of September 11, 2001, may be the result of all of the explanations commonly offered: failures of collection, coordination, and distribution of intelligence; a clash of cultures; hatred by fanatics; and so on. But each of these “causes” were themselves caused by multiple prior events. Osama Bin Laden is the product of his family, Islam, the Saudi culture, and personal experience defending Afghanistan against the Soviets. The clash of cultures (or civilizations: Huntington 1993) is as much a consequence of U.S. actions as of Muslim choices. Intelligence failures resulted, in part, from decisions that restricted human intelligence gathering, decisions made by successive US governments after several high-profile misadventures in the 1970s. Thus, September 11 could have emerged from a plethora of choices and events across the globe over decades, not as an inevitable consequence of any of them but as a path-dependent phenomenon. And if it was not path dependent, it was a symptom of a nonlinear system shift that cannot be predicted or explained. In neither case is conventional thinking about causation useful. Third, the immediate cause of an effect may, as part of a higher-order Markov chain, itself be the effect of an earlier, and possibly more important, cause. If an earlier cause is more important than later ones in the chain, this implies action at a distance in space/time that both Newton and Hume reject (Elster 1983, 26–30). Fourth, causation may be simultaneous as in open, emergent systems where the interaction of parts of the system constitute the system. In addition to these limits to the normal rules of causal explanation, assumptions of the proportionality of cause and effect are often erroneous. As dis-



cussed above, in open, emergent systems, small perturbations in the system may have very large effects, making identification of the connection between cause and effect nearly impossible and explanation problematic. Was the fact of Kaiser Wilhelm’s withered arm or his relationship with his English nanny a sufficient cause of World War I (Röhl 1998)? In a complex system, many factors symbiotically cause an effect. Theorists should look to the evolution of the system, not to individual events, for causes of observed effects. Patomäki and Wight (2000, 230) argue that “ontologically, the social world can only be understood as a processual flow that is intrinsically open and subject to multiple and at times contradictory causal processes.” Unintentionally, this is a fair exposition of complex systems. Social phenomena only occur because agents act within an existing and real context that is “not reducible to the discourses and/or experiences of the agents,” as constructivists argue. As Maturana and Varela (1980, 98) wrote: “[O]ur problem is the living organism and therefore our interest will not be in properties of components, but in processes and relations between processes realized through components.” In social systems, processes are not as automatic as they are in insects and bacteria. Humans and social groups are conscious and self-aware entities (that is, their internal models are more elaborate and complex) who, therefore, may act strategically toward some goal within their perception of their environment. PLAN OF THE BOOK Most social sciences have begun to embrace complex systems concepts. Ideas from thermodynamics coupled with a concern for economic systems’ environmental effects (Georgescu-Roegen 1975, 1971) led to the development of ecological economics that specifically models the economy as an open system (Barbier, Burgess, and Folke, 1994; Krishnan, Harris, and Goodwin 1995; Costanza 1991; Daly 1991). Brian Arthur and others have identified the presence and effect of feedback loops in economic systems (van Staveren, 1999; Arthur, Durlauf, and Lane 1997; Arthur 1990; Arthur 1989; Anderson, Arrow, and Pines 1988; Romer 1986). Complex systems approaches have attracted sociological interest (Luhmann 1998, 1990; Eve, Horsfall, and Lee 1997; Knapp 1999; Hanneman 1988; Collins, Hanneman and Mordt 1995) and touched public administration and organization studies (Griffin 2002; Stacey 2001; Marion 1999; Elliott and Kiel 1999). Even political science is not immune (Richards 2000; Axelrod 1997; Jervis 1997; Cilliers 1998; Cederman 1997; Cederman and Gleditsch 2004), though efforts are disparate and inchoate. This book is designed to drive forward the complexity research agenda as a viable alternative to orthodox theories of world politics by establishing the central concepts and ideas



needed for the development and empirical assessment of complex systems theories of issue-areas in world politics. The next nine chapters further develop the concepts outlined in this chapter and illustrate their application to several world politics issue areas. Chapter 2 begins to sketch out a taxonomy of complexity by comparing complex systems concepts to those developed more than three decades ago for a general systems taxonomy. Systems theories that were relatively popular in the early days of the Cold War have, in recent years, have fallen into disrepute as overly “grand” in purpose. Harrison, with Singer’s aid, compares and contrasts conceptual descriptions between general systems and complex systems taxonomies. Several concepts are common to the two approaches, but this chapter also identifies the important differences between the two taxonomies. Complexity is not a warmedover version of general systems theory but builds on its ideas to generate theories that better explain issue-areas in world politics. As this is a new approach to understanding world politics/IR, this book does not attempt to illustrate its application to the whole range of possible issues. The next four chapters show how complexity can generate new insights and hypotheses when applied to selected issue areas. They are arranged from the least to most technical in their use and application of complexity concepts. Because this book is an introduction to complexity in IR that is intended to initiate research rather than to develop applications adapted to all issue-areas of international relations, these chapters are only exemplars of the application of the complexity paradigm. None formally models their case but they all describe how their hypotheses might be further elaborated or empirically tested. In chapter 3, Dennis Sandole argues that complexity creates opportunities to integrate and synthesize apparently opposing worldviews. He reconsiders theories of identity-based conflict in the post-9/11 world and proposes a theoretical framework to demonstrate that traditionally competing Realpolitik and Idealpolitik (conflict resolution) approaches can coexist. Not only can they co exist, but more robust guides to identify conflict and formulate policy responses can be constructed by integrating both approaches into a single framework. In chapter 4, Walt Clemens attacks a knotty puzzle that has emerged from the collapse of the Soviet empire: why have some ex-Soviet states fared far better than others? Natural resources, education, and ethnic homogeneity do not explain why the Baltic states and Slovenia are joining the European Union, while oil-rich and more-homogeneous states are embroiled in factional fighting or war, or have stagnated in neo-Stalinism. Using complexity concepts, Clemens proposes an innovative explanation of why some newly freed states appear to have failed while others are joining the EU.



Drawing on complex adaptive systems theories (a version of complexity that uses more life science concepts), Clemens notes that some states were “fitter” than others and so better able to exploit opportunities that opened for them after the collapse of the Soviet empire. Seeking the sources of that fitness, he finds that long-standing, religiously inspired institutions in the Protestant countries developed internal models in the population that reduced ethnic tensions and increased acceptance of democratic virtues. He also shows that his marker for fitness correlates with measures of development and describes how to empirically test his hypothesis. Matt Hoffmann looks at the coevolution of states’ internal models in chapter 5. He considers two puzzles in the formation of the international regime designed to protect the stratospheric ozone layer: why did the norm governing participation change and why did the United States accept this new norm? Hoffmann shows that rational explanations are deficient and that complex systems concepts can help us to unravel both puzzles. From a complexity perspective, evolution of the universality norm is a simple story of complex adaptation. As some Southern Hemisphere states’ internal models changed to universal participation, the flux in the system eventually led other states to adapt to the new international norm. Hoffmann shows that when the United States reconsidered its internal model (with some pressure from domestic groups), it recognized it would have to accept the universality norm and negotiate in good faith with the South to achieve its goal of an effective treaty. He concludes with suggestions for theoretical, empirical, and methodological development of these ideas. In recent years, genocide within a country has become an international issue. The stimulus to this international interest in domestic interracial relations was the terrifying genocidal violence in Rwanda in 1994 that killed possibly as many as eight hundred thousand people. In addition to the moral implications, since Rwanda it is now clear that genocide in one country has serious consequences for its neighbors, making it a legitimate concern for the international community (“The Road Out of Hell,” Economist, March 27, 2004, 25–27). In chapter 6, Ravi Bhavnani shows how complexity concepts can help us to understand why the speed and magnitude of the killing was so much greater than in all previous ethnic attacks in that country. Conventional explanations of the scale of the Rwandan violence are inadequate. Bhavnani shows how bottom-up simulations can generate new hypotheses about the spread of ethnic violence. Building on evidence from the field and reasonable assumptions about relationships between extreme and moderate Hutus, he describes a simulation of how the killing rampage took hold so quickly and led to murder even within families.

if agents are assumed to be adaptive. David Earnest and Jim Rosenau in chapter 8 question whether political systems are complex systems. to be objectively rational actors. these are neither insurmountable nor critical problems.” in Simulating Social Phenomena. Inductive methods are needed to find patterns in. Like deduction. and show how multiple ABMs may be used to improve forecasting and decision-making. simple is better: like thought experiments simulations can deepen understanding of fundamental processes. complex systems thinking and ABMs already are being put to use in the service of policymakers to generate and assess multiple policy options. and argue that simulation of political systems begs the questions it attempts to answer: who are the actors and who has authority? They reject complexity as a theory. While Axelrod describes simulation as a third way. opinion surveys and macroeconomic data. Axelrod argues that simulation is best thought of as a new way of doing social science. edited by Rosario Conte. discuss potential problems with constructing complex systems theory. For them. it looks for patterns. Axelrod argues that. but it cannot prove theorems. 21–40. Saunders-Newton argues that complex systems thinking and computational methods that emphasize agent-level phenomena are part of a new transdiscipline that allows analysts to rigorously consider increasingly complex . while there are epistemological problems with ABMs. thought experiments are “much ado about nothing. Rainer Hegselmann. Chapter 9 is an indirect response to Earnest and Rosenau’s critiques.16 NEIL E. and implicitly they reject more limited applications of complex systems theory. deductive methodology suffices. In the social sciences. from the top down. In Desmond Saunders-Newton’s opinion. as commonly understood. and sometimes in international interactions. which builds systems from the bottom up rather than. as in conventional theories. the most common form of simulation is agent-based modeling (ABM). 1997). Chapter 7 is a reprint—used with permission—of part of Robert Axelrod’s chapter entitled “Advancing the Art of Simulation in the Social Sciences. but it uses data generated from defined rules rather than the real world. HARRISON The next three chapters explain the empirical validity of simulations.” They acknowledge that complexity is an attractive paradigm but argue that more development is required before it may generate viable theories of world politics. Earnest and Rosenau argue that it is no way: it lacks both the empirical appeal of induction and the disconfirmative value of deduction. As scholars debate the fine points of ontology and epistemology. because it fails the standard of theory in positivist epistemology and offers no alternate epistemology. and Pietro Terna (Berlin: Springer-Verlag. simulation starts with explicit assumptions. for example. as with deductive methods. Simulation is the third way—the only way. If social agents are assumed. in social science simulations. Like induction.

.” and “schema” are used interchangeably. It then assesses the validity of simulations and computational models and the epistemological questions they raise. Smith (2004) comments that world politics/IR theorists err in thinking of the state as solely political. The terms “world politics” and “international relations” are used interchangeably throughout this book. complex systems theories for issue-areas can be specified and models for specific problems generated. 232–33) opine that the “key error” of much international theorizing is “to treat levels of the state and the international system as related as agents to structures” instead of as “layers” within world politics. Finally. values. because political systems are complex—and becoming more complex—a new epistemology and new methods are needed to understand them. Common knowledge also shortens description.THINKING ABOUT THE WORLD WE MAKE 17 phenomena in an interdisciplinary way. from complexity concepts and ideas. he argues that efficacy or usefulness is more important than the quality of model isomorphism and method. Whether the state is modeled as political interacting with other subsystems of society or as a political unit of a social system among economic and cultural agents depends on the question being addressed. it shows that recognition of complexity in politics suggests radically new policies for addressing international problems and pursuing national interests. Patomäki and Wight (2000. The terms “complex system” and “complex adaptive system” are often used interchangeably. In this book. 4. 3. I use the term “paradigm” in the sense of a set of assumptions. He then describes how several computational social science models (including ABMs). concepts.” “mental model. the concepts described here principally derive from the latter. “Bicycle” conveys to most people a clear image of the system. the terms “internal model. NOTES 1. and in that sense it is quite comparable to “worldview” (Hughes 2000). integrated with computer-assisted reasoning methods. are being used in the PreConflict Management Tools Program being tested at the National Defense University for its ability to assess social vulnerability. and practices that comprise a view of reality. Imagine how much more complex would be a description to a Martian who is completely unfamiliar with a bicycle or any of the common parts used in its assembly. It also shows that. Yet. I use “world politics” to better reflect the multilevel structure of the political world to which complex systems thinking is so well adapted. The concluding chapter draws some general lessons from the four cases and shows how they illustrate important complexity concepts. 2. In addressing the epistemological issues surrounding computational methodologies.

1:19–39. “Competing Technologies. 1988. Santa Fe Institute Studies in the Sciences of Complexity. eds. Jason Harburger. Complexity and Postmodernism: Understanding Complex Systems.18 NEIL E. Princeton. New York: Simon & Schuster. W. 1997.” Proceedings of the National Academy of Science of the United States of America. George J. Axtell. Kenneth J. causation cannot be “real” (Patomäki and Wight 2000). Durlauf. REFERENCES Anderson. Increasing Returns. Cilliers. Robert L. and Kristian Skrede Gleditsch. 1990. Buzan. London: Earthscan for the Beijer Institute. Lars-Erik. Paradise Lost: The Ecological Economics of Biodiversity. NJ: Princeton University Press. Swedlund. supp. 1997. and Miles Parker. Alan C. 1994. MA: Addison-Wesley. Cederman. New York: Routledge. Gumerman. 2002. 2 (February): 92–99. Joanne C. CA: Addison-Wesley. 5. Shubha Chakravarty. no. “Why International Relations Has Failed. Arthur. Lane. 1989. 3 (September): 603–29. W. Dean. Reading. Arthur. Edward B. The Economy as an Evolving Complex System. Ross Hammond. Princeton.” Millennium: Journal of International Studies 30. Jeffrey S.. Barry. “Conquest and Regime Change: An Evolutionary Model of the Spread of Democracy and Peace. Emergent Actors in World Politics: How States and Nations Develop and Dissolve. 1998. and Richard Little. Philip W. 1997. no.” International Studies Quarterly 48. All we ever see is the conjunction of events. Axelrod. Redwood City. Epstein. The Turning Point. Capra. HARRISON 5. Brian. Cederman. Paul. The Complexity of Cooperation: Agent-Based Models of Competition and Cooperation. Robert M. Lars-Erik.” Economic Journal 99 (March 1): 116–31. 1982. But without an explanatory force linking cause and effect.. “Population Growth and Collapse in a Multiagent Model of the Kayenta Anasazi in Long House Valley.. and Lock-in by Historical Events. ———. eds. Brian. and Carl Folke. 10:7275–79. The Economy as an Evolving Complex System II. NJ: Princeton University Press. Burgess. As a skeptic. 2001. 3. “Positive Feedbacks in the Economy. no. 2004. Joshua M. Hume also argued that causation is a human construct. Barbier. vol. no. and David A. and David Pines. Fritjof. Arrow. . Jon Parker. 99. Steven N.” Scientific American 262. and we impute a causal relationship.

Evans. eds. Jon. “The Nature of Political Economy. Elster. 1999. Costanza. New York: W. R. and Robert D. 1997. New York: New York University Press. Elliott. 1991.. ed. and Theories. The Quark and the Jaguar: Adventures in the Simple and the Complex. Nonlinear Dynamics. and David Campbell. MA: Lexington Books. “Discovering Theory Dynamics by Simulation.THINKING ABOUT THE WORLD WE MAKE 19 Collins. eds. Der Derian. Thousand Oaks. no. Charles F. Mordt. Doran. 1986. 1990. Euel. Berkeley: University of California Press. and G.. International/Intertextual Relations: Postmodern Readings of World Politics. Rosenau and William C. Harold K. and Michael J. and L. Murray. Studies in International Political Economy. 1989. Gell-Mann. Georgescu-Roegen. Jim. Eve. “Why Forecasts Fail: The Limits and Potential of Forecasting in International Relations and Economics. Raymond A. and Mary E. James N.” Southern Economic Journal 41. Complexity and Public Policy. eds. 1999. Daly. Cambridge: Cambridge University Press. Douglas Kiel. 1996. “Patterns of Dissent and the Celebration of Difference: Critical Social Theory and International Relations. Hanneman. no. DoubleEdged Diplomacy: Institutional Bargaining and Domestic Politics. Introduction to Rational Choice. Robert G. CA: Sage. Lee. Putnam. . The Entropy Law and the Economic Process. ed. no. “Energy and Economic Myths. Explaining Technical Change. Models. 1971. 1983. Krasner and Miles Kahler. DC: Island Press. 3 (September): 269–93. 3 (January): 347–81. ed. NY: Nova Science Publishers. Complexity and Sociology: Myths.” Sociological Methodology 25: 1–46. 1994.” International Studies Quarterly 34. 1–33.” International Studies Review 1. Cambridge. Potter. James. Steady-State Economics. N.” In International Political Economy: State-Market Relations in the Changing Global Order. 1993. Stephen D. Jon Elster. Freeman and Company. Robert. Commack. Chaos. 25. 1975. Peter B. ed.. 2 (Summer): 11–41. Jacobson. Shapiro. Ecological Economics: The Science and Management of Sustainability. MA: Harvard University Press. Lexington. Gilpin. Herman E. 2nd ed. New York: Columbia University Press. eds. ———. 1991. Issues in World Politics. Washington. George. H. Sara Horsfall. R. ———. 1995. ed.

and Steve Smith. Boulder. “Variants on Six Models of the International System. Hasenclever. Passé-Smith. and Ian Shapiro. MA: Perseus Books. Explaining and Understanding International Relations. 1997. Cambridge: Cambridge University Press. Princeton. Conklin. Princeton. Hollis. Enquiries Concerning Human Understanding and Concerning the Principles of Morals. 1998. International Political Economy: State-Market Relations in the Changing Global Order. C. Robert. Hughes. NJ: Prentice-Hall. 1988. Holland. and Volker Rittberger. Nidditch. 3rd ed.. CO: Lynne Rienner. Jervis.” In International Politics and Foreign Policy. Toronto: Free Press. Conklin.. Computer-Assisted Theory Building. Griffin. Martin. MA: Perseus Books. 2002. Hume. Rosenau. Keohane. Perception and Misperception in International Politics. Donald P. Hanneman. 1988. “International Institutions: Two Approaches. 1995. H. CA: Sage. 3:22–49. NJ: Princeton University Press. and John G. Theories of International Regimes. Huntington. John T. Oxford: Clarendon Press. Newbury Park. and John G. New Haven.. The Emergence of Leadership: Linking Self-Organization and Ethics. 9–24. Peter Mayer. 1993. Barry B. “The Clash of Civilizations?” Foreign Affairs 72. Continuity and Change in World Politics: The Clash of Perspectives. 4 (December): 379–96. Green. Andreas. Passé-Smith. R. Samuel P. no. with revisions and notes by P. Englewood Cliffs. Roe Goddard. ed. Pathologies of Rational Choice Theory: A Critique of Applications in Political Science. Emergence from Chaos to Order. eds. ———. 1997. HARRISON C. Robert O. CT: Yale University Press. Reading. John H. Oxford: Clarendon Press.20 NEIL E. 1996. Morton A. 1994. System Effects: Complexity in Political and Social Life. CO: Lynne Rienner. 1969. Cambridge. New York: Routledge. ———. 1975. 3rd. David. 1990.” International Studies Quarterly 32. Douglas. Kaplan. NJ: Princeton University Press. 1976. Boulder. Hidden Order: How Adaptation Builds Complexity. Goddard. Roe. . James N. no. John T. 2000. ed.

2001. John. Knapp. 1998. MA: Harvard University Press. Kremer. Heikki. Peter. Prigogine. 1995. Autopoiesis and Cognition. “SystemDependent Selection. Maturana. P. DC: Island Press. and Barbara B. Locke. L.” Journal of Theoretical Biology 192:377–91. Olsen.. New York: Free Press. Korten. CA: Stanford University Press. 4 (Autumn): 943–69. Will. 1998.THINKING ABOUT THE WORLD WE MAKE 21 Keohane. Contemporary Political Philosophy: An Introduction. Polar Research Board. Harris. 1998. Russ. Ocean Studies Board. Cambridge. “The Institutional Dynamics of International Political Orders. 1994. . and the New Laws of Nature.” International Studies Quarterly 44. Dordrecht. Ed. Second Treatise of Government. Thousand Oaks. Smuts. The End of Certainty: Time. Stanford. and Colin Wight. and Joseph S. Luhmann. A Survey of Ecological Economics. Kiel. and J. The Edge of Organization: Chaos and Complexity Theories of Formal Social Systems. New York: Columbia University Press. Patomäki. 1999. Reidel. and Neva R. Washington. DC: National Academy of Sciences. C. Managing Chaos and Complexity in Government. West Hartford. When Corporations Rule the World. and Francisco Varela. 2 (June): 213–37. Kymlicka. Lansing. no. Transnational Relations and World Politics. ———. Rajaram. 1995. March. no. Chaos. B. Krishnan. Ilya. Macpherson. 1990. “Evolution. David C. San Francisco: Jossey-Bass. Niklas. J. Nye.. 2000. eds. Ecological Feedback and the Emergence of Functional Structure in Ecosystems. Robert O. Indianapolis: Hackett. Holland: D. Washington. “After Postpositivism? The Promises of Critical Realism. Abrupt Climate Change: Inevitable Surprise. J. In collaboration with Isabelle Stengers. Marion. Complex Systems and the Dialectic. eds. G. 1971. Stephen. CT: Kumarian Press and Berrett Koehler. Essays on Self-Reference. Goodwin. 1980. and Board on Atmospheric Sciences and Climate. 1980. 1990. Oxford: Clarendon Press. 1999.” Journal of World-Systems Research 5:74–103. CA: Sage. James N. Humberto.” International Organization 52. Jonathan M. Observations on Modernity. 1997. Douglas.

1994. “Territoriality and Beyond: Problematizing Modernity in International Relations. 2nd ed. no. Stacey. 1986. 3 (Summer): 427–60. and the Evolution of International Politics. Sprout. Paul. 1978. Susan. 1859–1888. Constructing World Polity: Essays on International Institutionalization. Cambridge: Cambridge University Press. International Theory: Positivism & Beyond. Schneider. London: Routledge. 2001. Steve. 3 (September): 499–515. 1984. States and Markets. Ruggie. NJ: Princeton University Press for the Council on Foreign Relations. Putnam.” International Studies Quarterly 48. Order Out of Chaos: Man’s New Dialogue with Nature.22 NEIL E. “Singing Our World into Existence: International Relations Theory and September 11. New York: New American Library. no. Ken Booth. Cambridge: Cambridge University Press. 1996. Smith. Complex Responsive Processes in Organizations: Learning and Knowledge Creation. 1 (Winter): 139–74. Ilya. Cambridge: Cambridge University Press. Strange. 1998. “Increasing Returns and Long Run Growth. and Margaret Sprout. 1988. 5: 1002–35. ———. 1993. Princeton. Ralph D. “Diplomacy and Domestic Politics: The Logic of TwoLevel Games. 2000. Ann Arbor: University of Michigan Press. no. Richards. 1993.” International Organization 42. Diana. The Context of Environmental Politics: Unfinished Business for America’s Third Century. Röhl. John C. Global Warming: Are We Entering the Greenhouse Century? New York: Vintage Books. Smith.” Journal of Political Economy 94. The Retreat of the State: The Diffusion of Power in the World Economy. and Marysia Zalewski. Romer. 1998. The Elusive Transformation: Science. London: Frances Pinter. 1996. Technology. Steve. Eugene B. Toward a Politics of the Planet Earth. . ———. no. Skolnikoff. Lexington: University Press of Kentucky. 2004. Political Complexity: Nonlinear Models of Complexity. London: Routledge. ———. Robert D. Harold. eds. G. New York: Van Nostrand Reinhold. 1989. HARRISON Prigogine. and Isabelle Stengers. John Gerard. 1971.” International Organization 47. Young Wilhelm: The Kaiser’s Early Life. Stephen H.

Waldrop. 1979. 1999. Wendt. “Collective Identity Formation and the International State. 1 (March): 141–67. Irene.” American Political Science Review 88:384–96. “Chaos Theory and Institutional Economics: Metaphor or Model?” Journal of Economic Issues 33. 1994. Alexander. Reading. Complexity: The Emerging Science at the Edge of Order and Chaos. Mitchell. Theory of International Politics. Waltz. no.THINKING ABOUT THE WORLD WE MAKE 23 van Staveren. 1992. M. MA: AddisonWesley. New York: Simon & Schuster. Kenneth. .

This page intentionally left blank. .

This chapter shows how the important concepts of general systems taxonomy compare with concepts in a complex systems taxonomy. which itself built upon current knowledge from orthodox theories (such as it was) and corrected the epistemological and methodological defects found in most general systems theories. In 1971 Singer proposed a general systems taxonomy that potentially supported both systemic explanation and more limited theorizing. predictive or explanatory power. We anticipate that the complex systems taxonomy can generate radically new hypotheses about world politics and develop innovative ways of testing them and thereby increase knowledge and improve policy-making. Harrison with J.CHAPTER 2 Complexity Is More Than Systems Theory Neil E. it must specify two things quite clearly: the basic constructs by which the relevant domain is to be described and the definitional relationships between and among these constructs” (Singer 1970. complexity modifies and expands on Singer’s general systems taxonomy. Despite correcting 25 . we argue that complexity’s modifications to general systems taxonomy create a more flexible taxonomy that incorporates current knowledge and integrates theories that hitherto have been considered incommensurate views of world politics. 3). David Singer The previous chapter emphasized the differences between conventional theories of world politics and international relations and complexity. As a framework for theorizing about world politics and international relations. CONCEPTS IN COMMON A taxonomy “has no descriptive. since it contains only definition propositions. But to serve as the basis for building models and theories. In addition.

but that at the same time the many entities comprising those social coalitions known as nations may also serve as useful objects of analysis. Commonly. hierarchical systems of interacting agents. In both taxonomies the international system is modeled as comprising multiple. and government agencies are only components in . and none is likely to display rational behavior.” But the state is not a solid body. 17). DAVID SINGER faults of general systems theories and a well-argued explanation of the benefits of a general systems approach. the international system is only “real” as an environment within which it operates. Most textbooks recognize several levels of analysis from individuals to the international system in which to seek explanations of global or international events.” Both the general systems and complex systems taxonomies explicitly model the ontological layers in world politics as interrelated systems. Nested Systems Textbooks and scholars agree that “there are different ontological layers in the world” (Patomäki and Wight 2000. each embraced by those at the next higher level of analysis and embracing those at all lower levels. Singer argued (1971. state. for example. it is a “coalition of all social entities at the individual. society. It follows from this that any system or set of systems at one level of analysis constitutes the environment of all the entities existing at any lower level” (Singer 1971. international theory has been largely confined to competing. flaw lies in the general tendency to focus on only one level of analysis. as described in the next section. In the general systems taxonomy the world political system is modeled as “a hierarchy of nested sets of subsystems.26 NEIL E. Singer (1970) comments. The common focus on a single level of analysis blinds theorists to influential processes operating at other levels of analysis. For each state. in pursuit of a false simplicity. rather than treat the interactions that occur across the several relevant levels. and secondary levels” (1971. Yet. at least five are identified: system. that the “perhaps fatal . and individual. 17) that the nation-state remains a “useful object of analysis. singular levels of analysis. primary. Yet. and the systems are open to their environments but can be distinguished for theoretical purposes. complexity also changes. 232). government. This section shows that some of the concepts from the general systems approach appear only slightly modified in the complex systems taxonomy. . 12). develops. and adds to general systems concepts in ways that increase its potential utility for building effective theories of world politics and international relations. HARRISON WITH J. . in the last two decades Singer’s taxonomy has attracted little theoretical interest. Each system can contain multiple feedback loops and may be susceptible to path dependencies.

or aggregations of people”—that is.g. beyond crude power measures. Thus.” largely ignoring the entities that “participate and experience them. it permits more effective separation of subsystems (e. scholars in this tradition usually explicitly posit that systems “will show rather similar patterns and processes as well as a fair degree of structural isomorphism. Not Actions Singer (1971. In social systems. it more clearly distinguishes the system from its environment (is the social system part of the political or vice versa?). In the former school.” The latter school models systems “around individuals. 7–11) identifies two schools of general systems theories: system-ofaction and system-of-entities. as described below. political from economic) where this would be theoretically more useful.1 . institutions are the environmental selection rules that govern punishment and reward for agent actions.. Agents. systems are identified around “actions. groups. complexity and general systems approaches diverge in their conception of cause-and-effect relations across levels of analysis. Similarly. path-dependent consequences of prior agent interactions through earlier patterns of institutions.” He then argues persuasively that systems must be conceived of in terms of the characteristics of their constituent entities rather than in terms of agent actions. scholars often focus on the actions and abstract away from agents. and Waltz (1979) elaborates a model of the international system that he compares to a market and in which. regardless of the characteristics of the states. the characteristics of the actors are of no interest. associations. complexity views the state as a “meta-agent” (Holland 1995) that forms its variable internal model out of the ongoing interactions of social aggregations within its domestic political processes. agent-based models assume that agents choose actions that are consistent with their individual desires and their beliefs about how to satisfy those desires. However. In contrast to the “actions” school. Kaplan’s (1957) model describes a system of actions and interactions between states. complexity assumes that the characteristics of social entities generate agent actions and participate in constructing system structure. or role. He shows that this approach is methodologically more tractable. in international relations and world politics theories. interaction. Complexity posits that internal models cause agent actions and the pattern of agent behavior reflects the interaction of the agent’s internal model with environmental constraints. Institutions are the dynamic.COMPLEXITY IS MORE THAN SYSTEMS THEORY 27 the aggregation known as the state. behavior. relationship. For example. As in Singer’s general systems taxonomy. social entities. Yet. and it clarifies levels of analysis (at what level are individual actions of decision-makers?). Historical-materialist theories focus on the structural forces that dictate state behavior.

13). and the benefits of participation in the ideological mainstream preserve the distribution of power and inhibit changes in the historical patterns that transform inevitable conflicts into costly rivalries. and the diminishing time lag between event and report. especially in poorer countries. Openness is not a measure of complexity.). the public and military desire the psychological comfort of discernible superiority. National elites use rhetoric for domestic political consumption that can incite potential enemies. Yet. However. but. it is an empirical question as to how open each is (1971. technology has .28 NEIL E.” Complexity similarly treats all social systems as open. as Singer has commented. such open communities often are both somewhat closed to their social environment and much simpler than more complex. in his 1971 monograph Singer simply states that all social systems are open—that is.2 If so. Selfrestraint within political elites and the media has diminished with the increase in the number of competing media companies. media amplify internation conflicts.d. While every model is necessarily an “ersatz” reality. assuming closure to simplify analysis and reduce empirical effort is not a reasonable parsimony but a gross distortion of reality. HARRISON WITH J. “clearly no such thing as a completely closed social system” (13). There is. Some scholars approvingly describe rural communities’ openness to their “natural” environment. Technologically induced immediacy reduces opportunities for editorial restraint. 165). and corrective mechanisms within the international system have atrophied. their geographical coverage. their boundaries are “permeable to information and energy from [their] environment”—and that no social system can realistically be treated as closed. but complex systems usually are open. Similarly. Feedback Conflict recurs in the international system because the conflicting incentives and temptations within nations and the lack of effective constraints between nations support positive feedbacks to conflict: “[I]ntra-national and inter-national events all impinge on one another in a cyclical and ongoing process within which the self-aggravating propensities frequently exceed the self-correcting ones by an unacceptably large amount” (Singer 1970. DAVID SINGER Open Systems Political systems are conventionally assumed to be closed and homeostatic: disturbances are temporary and the system tends to return to equilibrium. modern societies that are more open to other social systems but less open to ecological systems (Harrison n. it must retain a recognizable link to the portion of reality it purports to represent or it will generate inaccurate “knowledge. he writes.

But to state that the Cold War evolved out of prior history is not to claim that it was an inevitable effect of historical causes. Path Dependence and Randomness Path dependence is the idea that system development from time t to t 1 is not wholly random and can only fall within limits created by the prior state of the system. determined by the prior state of the system. Tainter 1988). Deutsch 1953). Haas 1964.g. and the international system may change its structure without becoming another system. Armed conflict is an incentive to modernize both equipment and tactics (Smith 1985). complexity posits that the interaction of multiple independent. Earlier examples come from the literature on the formation of the European Community. in complex systems the arrow of time is not reversible (Prigogine 1997). volitional agents allows positive feedback loops to develop that can drive the system to “flip” to a new state (Levin 1999).. First. It was not a discrete system and cannot be separated from its history. and that to its prior state all the way back to its nascence. Similarly. The Cold War was a period in the evolution of the international system that was in part caused by all of history that preceded it. Constructivists would point to positive feedbacks operating in the formation of norms that underpin international environmental treaties (as Hoffmann describes in chapter 5). it is better for them if the last war was more recent and a success. Like ecosystems. in which functional links or communications links were thought to increase trust and lead to eventual integration between countries (Mitrany 1966. and raw data crowd information and reduce the quality of its assessment. social systems evolve continuously. the . Studies show that in conflict prior experience matters in at least two ways. societies have often collapsed from runaway internal feedback loops (Mäler 2000. The choices of multiple discretionary agents (from individuals to states) inject randomness into outcomes. Living systems are considered path dependent: the current system state is related to and is. If generals are always fighting the last war. The Montreal Protocol is often cited as a precedent for the 1992 signing of a global treaty to mitigate emission of gases believed to fuel climate change. in part. conflict stimulates innovation in search of increased military capability. Drawing on models of ecological systems. Thus. Revolution might be an example of a flip in a social system. A major defeat may eliminate the state or so emasculate it that future aggression is militarily impractical (e.COMPLEXITY IS MORE THAN SYSTEMS THEORY 29 reduced the time available for consideration of alternatives by decision-makers. But positive feedback also can be beneficial.

Research has failed to find statistically significant linkages between war experience and later conflict choices (Singer and Cusack 1981). however. historical experience is perceived through the ever-changing lens of the present.” This does not mean that each conflict must be treated as a discrete event. Because state behavior emerges from domestic interactions. Rationality Rationality assumptions are used as a convenient simplification in both orthodox theories and general systems approaches. The Correlates of War (COW) project is a major effort to overcome a common cause of the evident failure to understand and better control internation conflict. causing an apparent US aversion to armed conflict. the project seeks patterns and commonalities among conflicts and avoids the historical fallacy that defines each conflict as a discrete and separable event. but it does show that the feedback mechanisms within states are significantly more complex than is commonly believed. current conditions and institutions. and the variable distribution of power between politically influential groups. By collecting data for more than a century of interstate and civil wars. defeat in World War I supported German aggression in the 1930s. Thus. may encourage a desire to regain position and respect. States have memories that influence future perceptions and choices. Failure. If the experience of conflict influences later conflict choices. influence state internal models. There is a randomness to the influences of memory and history that is not captured by simple theories. At this level. HARRISON WITH J. Second. Success in conflict tends to bolster militarily adventurous groups. It is conventional wisdom. and recent failure may cause caution. a complex systems theory of national security potentially allows for both path dependency through experience (state memory and capability) and randomness. for example. feedback mechanisms within the state are the likely link. Through the concept of emergence (discussed further below). which itself emerges from the past. rational- .30 NEIL E. there is institutional memory: how decision-makers and the public perceive the benefits and costs of conflict. But current social conditions and power relations are themselves historical artifacts. Analysis of the COW data showed that none of the usual hypotheses about learning from war experience is supported: “[T]he probability of the major powers getting into war is statistically independent of when and with what effects they experienced their prior wars. But in the latter rationality is assumed only at the lowest level of aggregation: the individual human. For example. DAVID SINGER Austro-Hungarian and Ottoman Empires after World War I). that a “Vietnam syndrome” influenced decades of decision-makers.

an inability to consider all options. . liberalism. only the individual can be rational (again. to be a positive relationship between high rationality in the process and the desirability of the outcome” (6–7). positive feedback in elites and media. or functional for the perpetrators” (Singer 1990. First.COMPLEXITY IS MORE THAN SYSTEMS THEORY 31 ity is consonance of behavior with desires and beliefs about how to achieve those desires within the perceived environment (primarily the social world). Conversely. and rational process can. 18). While it is claimed to be rational for both parties to defect in the prisoner’s dilemma.3 This greatly simplifies analysis but is inherently misleading. In orthodox theories. Even if individuals and groups only minimally respond to their private interests. Singer draws three conclusions. with some frequency. virtually assuring that deviation from rational choice and the implied prudent pursuit of collective interests will remain the norm” (Singer 1990. Social system behaviors are neither rational nor irrational. the rationality of processes must be judged in relation to specific social aggregations. and protecting the individual credibility of leaders are among the factors that make social decision processes extrarational. both suffer individually suboptimal outcomes. But positive feedbacks can magnify the preferences. It can mean that the outcome “was (or might turn out to be) desirable. but the bloody pages of international history remind us that it is hardly sufficient” (Singer and Cusack 1981. sunk costs (“they shall not have died in vain”). and constructivism theorize at the level of the international system and assume that states and other important agents in that system are rational actors. even though there tends . rationality is more often assigned to higher levels of aggregation. For social aggregations rationality has several meanings. thorough. like pictures of prisoners tortured by US military police. and states rarely learn from past wars: “[K]nowledge may be necessary for rational human intervention. reality is too complex to call behavior rational if agents pursue outcomes that coincide with their individual preferences. President Bush may have wanted to install democracy in the Middle East through Iraq. But successful or functional outcomes may emanate from thoroughly irrational processes. culminate in disaster. but outcomes are always unpredictable and small causes. Singer’s terminology is exact: they are extrarational. Second. In the language of complexity. in process only) and all behaviors of social aggregations emerge from social interactions: “the ‘invisible hand’ of parochial sub-system interests is ubiquitous. . Rationality also can mean that the decision process was rational. 6). which may be rational for them. successful. Third. may derail the most laudable policies. Realism. rationality in social aggregations can only describe the processes they follow. not the outcomes of those processes. “the most careful. . 417). the effect is “extrarational” for the aggregation. social rewards to conformity. Establishment and celebration of military organization.

but the process by which it is constructed may be assessed for its rationality. . DAVID SINGER Related to the orthodox presumption that social aggregations are rational is the belief. emergence was rejected by most general systems scholars as “unnecessary and scientifically misleading” (Singer 1971. construct an internal model of its environment that guides its behavior.” Social systems collapse when they no longer serve the needs of their constituent agents and the costs of belonging exceed the benefits though authority permits formal institutions (such as government agencies) to preserve themselves beyond the limits of social acceptance (Tainter 1988). social systems are not “inherently supposed to perform and survive. Hoffmann explains how the US internal model of the ozone issue changed over time. the internal model is not assumed to be objectively rational and. like an individual human agent. they have no purpose beyond the intentions and preferences of the subsystems.32 NEIL E. complexity diverges from general systems theory by accepting that a social aggregation like a nation-state may have preferences and interests. expect. domestic interest groups) interact in pursuit of their preferences. While aggregations may not be rational. This section highlights the important differences between general systems theory and complexity. HARRISON WITH J. The effects are “neither structural or behavioral. and . prefer. 13).” . sometimes implicit. Emergence Although the “interaction of individual properties (both within and among single humans) may produce emergent effects” (Singer 1968). therefore. Unlike realism. As the subsystems (for example. emphasis in original).5 DIFFERENCES BETWEEN THE GENERAL AND THE COMPLEX While the previous section discussed the similarities between several concepts used in the general systems taxonomy and concepts in complexity. . But if the nation-state is conceived as a meta-agent.4 As Singer comments (1971. also insists that any social entity can behave (Singer 1971. If systems are emergent. hope. 19. perceive. the system emerges (legislation is crafted and enacted). relatively unchanging. the latter is more than a warmed-over version of the general systems approach. or seek to do so. 18). In chapter 5 below. that systems have a purpose or function in a teleological sense. responding to changes in international negotiations and domestic political interactions. Several complexity concepts are additional to general-systems concepts and others are modifications of concepts used in the earlier approach. Singer’s general systems taxonomy specifically “denies that any social entity other than a human being can think. it may.

Fortunately. limited as it was by incomplete data sets and computing power. He rejected the “bromide” that “a social system is more than the sum of its parts” because the cultural properties of large social systems can be better described in a “strictly aggregative fashion. the emergent effects of agent interactions are both dynamic and important and cannot be captured by observation of structure. and by the end of the 1960s it seemed that he was right. or individual psychology. or of any other phenomena beyond its own component units and the relationships and interactions among them.COMPLEXITY IS MORE THAN SYSTEMS THEORY 33 and if they are cultural they can be observed as individual psychological properties. with new research techniques. Gell-Mann had developed a theory of quarks. Complexity also relates system culture to individual psychological properties but models culture as emerging from the interplay of diverse agent internal models within institutional strictures. “everything seemed to be coming up quarks.6 Social scientists who are not computer specialists can now program ABMs that model “mystical” . He was always skeptical that such partially charged subatomic particles would ever be found. no evidence had been found for their existence (Riordan 2004). one of the predictions of Einstein’s general theory of relativity was tested by astronomical observation for the first time in 2004. The modeling for the Club of Rome project Limits to Growth was a massive effort on mainframe computers by specialist programmers at the Massachusetts Institute of Technology (Meadows et al. 1972). by observing the distribution and configuration of individual psychological properties” (Singer 1968.” In complexity a social system is more than merely the aggregation of its parts: the system is modeled as emerging from the relationships and interactions between member agents. It would be unscientific to reject a theoretically useful concept merely because accepted scientific methodologies cannot record the phenomenon. emergence is real. developments of new methods permit new thinking and empirical testing of novel hypotheses. as elsewhere in the science. Similarly. Not only are the predictions of that project now largely discredited. But by 1973. behavior. is considered crude. When Singer wrote his critique of emergence. Developments in computer modeling now permit simulation of emergent systems and avoid “metaphysical pursuits” that attempt to isolate and measure emergence as a definable property of a complex system. 144). but also the design of the model. it was not practical to model complex systems. In contrast to Singer (1968). Singer also argued that “a system is nothing more than the sum of its parts and the relationships and interactions among them” (Singer 1971.” It took most of the decade for the theory of quarks to become generally accepted. Although the emergent properties of a system cannot be captured by studying the system’s parts. and it was later recognized with a Noble Prize. 19) and that a system is “not composed of [external] systems. With George Zweig. nearly ninety years after the theory was first proposed.

Atmospheric scientists use computer simulations of climatic history to explain historical and to project future climate change. Experimentation The “natural” sciences—most notably physics. mechanics. statistical manipulation (as in COW). object programming. Within a system. as discussed further in chapters 7 through 9. Indeed. as well as more speculative simulations designed to generate as much as to test hypotheses. In this way social research replicates the level of control of the classic laboratory experiment. HARRISON WITH J. In the social sciences. The development of desktop computers as powerful as supercomputers of twenty years ago. Inputs may range from the “purely speculative to the thoroughly grounded. but can it support construction of theories of cause-effect interactions across levels of analysis? We believe that complexity provides a conceptual framework for theories that can accommodate both causes working from below and from above the system under study. DAVID SINGER emergent properties as an integral part of the whole system. the concept of emergence is central to ABM construction. . 3) proposed that “the historical experiment is a perfectly legitimate mode of research” that may offer advantages over laboratory or field experiments. As discussed earlier. Cause and Effects Complex-systems concepts encourage theory that covers multiple levels of analysis. Their laboratory experimentation allows repeatable. Field experiments sometimes are possible.34 NEIL E. and chemistry—are still the measure of “scientific” for most social scientists. control over factors may be increased with comparative case studies. Singer (1977. permitting causal inference. the magnitude and variation of every input is fully controlled by the researcher” (Singer 1977). a technique rarely available to the social scientist. and social scientists can now experiment as never before. and simulations. and concepts such as neural networks now permits (at least in principle) such historical experiments. and an increased ability to ascertain covariation. controlled manipulation of isolated potential causal factors. Increased control over principal factors in laboratory experiments allows more accurate observation and measurement. emergence connects causes at lower level of analysis with effects at higher levels of aggregation. a state as meta-agent may form its internal model from the interaction of domestic constituencies and . In the “all-machine simulation .” Computer simulations can test out myriad ideas against history until a good fit is found. but they permit less control than in the laboratory. .

perceptions of environment influence agents’ internal models. perhaps energy companies or conservative Christian groups) or outside . Yet Jervis (1976). In complexity. Selection means that agents adapt or are eliminated. Agents “learn” when they change their desires or their beliefs about how to achieve their desires. among others. Its prosecution through armed attacks on Afghanistan and Iraq and formation of the Department of Homeland Security was the result of negotiation among decision-makers’ causal beliefs. The chief executive is usually accorded more influence on state policy than other participants. Small groups within the state (especially those closely associated to decisionmakers. But in the process of negotiating policy. it constrains what is possible and “selects” behaviors that are most appropriate within current institutional arrangements. coadaptation implies dynamic recursive adaptive responses between multiple agents.” at a minimum by being prevented from moving toward satisfaction of its desires. For example. The “butterfly effect” of less institutionally gifted individuals also may emerge up through levels of aggregation to influence the state internal model. First. especially in terms of causal beliefs of what is possible. internal models may change. And there may be interaction among both processes. If misperception leads to maladapted behavior. environment affects system behavior in two ways. Beyond the emergence of behavior from internal interactions is the greater problem of theorizing links between causes at higher levels with effects at lower levels. Second. has detailed the many ways in which individual decision-makers’ perceptions of the state’s environment are formed and how they influence state interests and behaviors. Agent learning is the cognitive adjustment that increases behavioral survivability in a selective environment. and his or her internal model will tend to dominate. The post-9/11 War on Terror is a realignment of preferences in decision-makers’ internal models. the agent will be “punished. by the perceptions held by individual decision-makers of the conditions in the system. even in part. often implicit assumptions. explanation from the state level requires several. For example. the most important of which is that state decisions are not influenced. The question is: how does environment affect system behaviors? For example. the ability of an individual to shape state policy depends on the institutional arrangements that regulate the individual’s influence on state behavior and the acceptance of his or her internal model by others in the decision processes.COMPLEXITY IS MORE THAN SYSTEMS THEORY 35 the interplay among participants in decision-making groups within the executive. how does the state of the international system influence state policies. Institutional arrangements in the environment create selection processes that act on system behavior. or how do national policies determine individual behavior? Singer (1961) argued that explanation of behavior at each level of analysis was problematic.

DAVID SINGER (al Qaida or Abu Ghraib prison guards) also may influence the behavior of states through their impact on the internal models of decision-makers. constructs must be sufficiently abstract to embrace “concepts that are substantially identical” (1971. Singer sets out six criteria for a good taxonomy that are equally applicable here. in world politics it is the emergence of political events (for example. for example). so much the better. HARRISON WITH J. as through information dissemination. peacekeeping missions. a good taxonomy indicates what is not known and needs to be learned. it should have theoretical relevance to the phenomena we hope to account for. as discussed below. Second. In complexity the dependent variable or outcome is the behavior of social systems. To achieve this. while many options remain open. this and the previous chapter show that this paradigm can include much that is known about world politics and. Fishing expeditions are permissible. especially if parsimony would prevent many plausible models: “Parsimony is a virtue only when well advanced toward the verification stage of the discipline and may often be a liability when we are still in the discovery stage” (6). By linking concepts from conventional theories to complexity.36 NEIL E. state-building activities. it may integrate views that are usually considered incompatible. allowing for idiosyncrasies of different fields. policies and agent actions) and institutions. Recognizing the many weaknesses and gaps in our knowledge. a taxonomy of world politics should be liberal and allow a range of eclectic approaches to many different phenomena. Complexity entails only . International organizations also may influence state choices in several ways. or as the locus of norm construction and treaty negotiations. One way to achieve sufficient permissiveness of testable hypotheses “is to develop a minimum number of classes of variables so that. First. The anticipated predictor variables are rules. a taxonomy need not be parsimonious. Historical experimentation with agent-based models is a flexible method adaptable across issue areas and levels of analysis. And if it can be relevant to theory for a broad range of phenomena. But. rules of due process (in the UN Security Council. Fourth. they also should include current knowledge within the field. 5). knowledge should be transferable between empirical domains. both internal and institutional. A BETTER TAXONOMY In his A General Systems Taxonomy for Political Science (1971). the taxonomy also remains conceptually clean and manageable” (6). where possible. Third. But it may be hoped that regularities will be found among social systems at all levels of aggregation and in all issue-areas of world politics. respectively within and between agents.

The expectation is that there are regularities in behavior that may be explained by universal causal “laws. complex systems theories may be more parsimonious than competing orthodox theories. In Singer’s general-systems taxonomy. This volume is intended to begin this work. “inside” approach that views the social world as constructed of rules and meaning through human interaction. Here the goal . whole systems are only the sum of their parts and could. environment. with concepts such as emergence. Behavior is then assumed to be explicable using methods borrowed from the natural sciences. which may be external or internal. be disaggregated and comprehended by analyzing the parts and their relations. one useful approach is to distinguish them by their “view” of reality.COMPLEXITY IS MORE THAN SYSTEMS THEORY 37 a few broad concepts that may be adapted to theory goals. Hollis and Smith 1990). Each agent tries to pick an intelligent course through multiple social engagements in which other agents bring their individual characteristics to their social roles (Hollis and Smith 1990. if necessary[. 3) and decision-makers are replaceable and only represent their position in the system with little personal volition. coverage of all levels of aggregation is needed: the taxonomy must “be able to deal with several levels of analysis and . The external or “scientific” approach assumes that the social world. the interface between these levels [must] not be a source of slippage and confusion” (6). We have argued that. So.” “Behavior is generated by a system of forces or a structure” (Hollis and Smith 1990. and perhaps most important. Sixth. Encompassing and Improving Orthodoxy Certainly the first. Orthodox theories may be classified in several ways. The complexity paradigm should support theories that fully or partially integrate these two apparently incommensurate views. and the natural world in which it exists. semantic clarity is essential: it is “preferable to select words that do convey generally accepted meaning and then.] specify the restricted or expanded definition intended” (6). test of a new taxonomy is its ability to open new research agendas by better integrating existing theories and knowledge and thereby explaining some of what was previously inexplicable. . constructs must be operational. Fifth. 6). is an environment. . Chapters 3 through 6 of this volume demonstrate how complex systems concepts can be operationalized in various issues and at different levels of analysis. independent of human agents and potentially predictable (S. Smith 1994. Orthodox theories like constructivism pursue an alternative. and internal model. Finally. complexity is eminently and uniquely able to satisfy this demand. therefore.

NOTES This chapter was written by Neil Harrison based on David Singer’s selection from among his published works of those that anticipate complexity. . 197–98). HARRISON WITH J. 1. Only theoretical development and empirical and experimental application will demonstrate if world politics theories based in complexity can overcome the incompatibility of the two views of orthodoxy and open useful new research agendas in issue-areas. therefore.38 NEIL E. As discussed. It also is difficult to see “how the system changes its structure in a closed system without a change in the units and purely functional explanations are bound to be suspect.” The inside approach is rejected by some as interpretive and. The central concept of emergence marks complexity as favoring the inside view. Explanation from the outside. 400). seeing it as intentional and meaningful behaviour” (S. The agent/structure problem is a manifestation of levels of analysis that turns on “the ‘reality’ of systems or on the need to feature them in explanations” (Hollis and Smith 1990. however firm the shove [from structures]” (Hollis and Smith 1990. And the two levels cannot not be combined to achieve an “overarching theory which explained how system-level and unit-level factors interacted to produce state behaviour” (100). and his comments and advice on earlier drafts. inherently unscientific. The two approaches usually are assumed to be incompatible. however scientific. Complex systems theories potentially integrate outside and inside orthodox views. ABMs can simulate behaviors of whole systems from the inside without consciously interpreting behavior and with no presumption of motives or meaning at any level by using randomized internal models and rules of interaction. ABMs also may be constructed to simulate historical reality. is incomplete without consideration of the units: “The anarchical character of the international system . 198). If the system is real and must be analyzed as a whole. Smith 1994. . examining “human action from within. unless they include a causal contribution from the units. This formulation echoes the recursive interaction of biological entity with its environment that is well accepted in biology and discussed in detail in Levins and Lewontin 1985. strongly suggests that the units affect the shape of relations. DAVID SINGER is to understand behavior. yet its experimental methods are potentially as scientific as those of the revered “hard” sciences. it must be shown that “wholes are more than their parts and that science is capable of establishing such a proposition” (198). . The following chapters begin that task. and the means are often hermeneutical.

Climate Change Policy. 3. Haas. New York: St. NJ: Princeton University Press. ———. Boulder: Westview.S. 1990.COMPLEXITY IS MORE THAN SYSTEMS THEORY 39 2. policies on climate change. this would seem a fair assessment. Beyond the Nation-State. Hidden Order: How Adaptation Builds Complexity. Except for “lost” tribes in places like Amazonia or New Guinea. 5. Oxford: Clarendon Press. Martin. Jervis. Nationalism and Social Community. 1969. Social Complexity. John H. Barry B. ed.S. Palma 1978 and Gunder Frank 1969). Hughes. Karl W. and the models are more accessible to social scientists not trained in the arcane tricks of effective computer programming. Explaining and Understanding International Relations. Gunder Frank. Princeton. “From the Inside Out: Domestic Influences on U. For example.” Draft manuscript. 4. is probably impossible in the modern world. 2000. Harris. and Steve Smith. . Harrison. 1964. 6. Cambridge. Also see Harrison 2000. Holland. and the Environment. Stanford. which shows how domestic politics influenced U. Latin America: Underdevelopment or Revolution. which comes with a computer model and relevant data on a CD. 1976. Neil E. see Hughes 1996. Autarky. 1996. Perception and Misperception in International Politics. social aggregations of classes and states are assumed to rationally pursue their interests. In critical theories derived from Marxian analysis. Hollis. 1953. André. International Futures: Choices in the Creation of a New World Order.d. Robert. Global Environment Policy. MA: MIT. 89–109. Ernst B.” In U. N. for example. New York: Modern Reader. MA: Perseus Books. 2nd ed. CA: Stanford University Press. Paul G. Martin’s.S. once valued by “developmentistas” (see. REFERENCES Deutsch. The equating of policy-making with sausage production reflects the inherently extrarational outcome of the policy process. “Complex Systems. 1995. Reading. Even programming of linear-world models is much more sophisticated.

“Escalation and Control in International Conflict: A Simple Feedback Model. Knorr and S. and the New Laws of Nature. New York: General Learning. and Colin Wight. 1961. David.” In From National Development . 1997. ———.” General Systems 15:163–73. no. Singer. ed. Karl-Göran.. “The Level of Analysis Problem in International Relations.physicstoday. NJ: Princeton University Press. New York: Wiley.40 NEIL E. Prigogine. ———. Mitrany. 77–92. San Francisco. ———.” World Development 6:881–924. Ecological Resources and Their Management: A Study of Complex Dynamic Systems. In collaboration with Isabelle Stengers. “The Invisible Hand. “Man and World Politics: The Psycho-Cultural Interface. 2000. 1985.” Journal of Social Issues 24. Verba. “Periodicity. 1971. HARRISON WITH J. no.” International Studies Quarterly 44. and Thomas Cusack. “Dependency: A Formal Theory of Underdevelopment for the Analysis of Concrete Situations of Underdevelopment.” In The International System: Theoretical Essays. Princeton. and Richard Lewontin. DAVID SINGER Kaplan. 1990. Singer. Chaos. Levins. The Dialectical Biologist. Reading. Fragile Dominion: Complexity and the Commons. New York: Free Press. Levin. 1966. MA: Harvard University Press. MA: Perseus Books. System and Process in International Politics. Heikki. no. K.html on March 22. 1968. “After Postpositivism? The Promises of Critical Realism. The End of Certainty: Time. 1970.” Paper presented at the Annual Convention of the American Political Science Association. “Science Fashions and Science Facts. 1957. 2004. 2000. Chicago: Quadrangle. 1977. A General Systems Taxonomy for Political Science. and Steermanship in International War. A Working Peace System.” European Economic Review 44 (2000): 645–65. Richard. “Development. 1978. Simon A. “The Historical Experiment as a Research Strategy in the Study of Politics. Palma. 1999. ———. 1 (Fall): 1–22. September 2 (June): 213–37. Gabriel. Mäler. Cambridge. Michael. Inexorability.” Viewed at http://www. David. Ilya. J. Extra-Rational Considerations and Decisional Failure. Riordan. Patomäki.” Social Science History 2. Morton A. 1981. ———. J. David. 3:127–56.

Tainter. Waltz.” International Studies Review 4. Theory of International Politics. Cambridge: Cambridge University Press. or Rereading. 1994. MA: MIT. Sterling-Folker. 2:395–405. Jennifer. Russett. Cambridge: Cambridge University Press.COMPLEXITY IS MORE THAN SYSTEMS THEORY 41 to Global Community: Essays in Honor of Karl W. Epistemology and Emancipation. “Rearranging the Deckchairs on the Ship Called Modernity: Rosenberg. Joseph A. Steve. Deutsch. 1979. 1985. Smith. 1 (Spring): 73–97. Boston: George Allen & Unwin. Kenneth. no. 1988. Cambridge. Merritt Roe. . MA: AddisonWesley. 2002. ed. The Collapse of Complex Societies. Reconstructing. no. Merritt and Bruce M.” Millennium: Journal of International Studies 23. 404–22. Richard L. “Realism and the Constructivist Challenge: Rejecting. Military Enterprises and Technological Change. Smith. The Capitalist World Economy. Immanuel. 1979. Wallerstein. Reading.

This page intentionally left blank. .


Complexity and Conflict Resolution
Dennis J. D. Sandole

The events of September 11, 2001, undermined much conventional analysis in world politics and international relations (IR). Much as the fall of the Berlin Wall was not anticipated by IR scholars, the terrorist attacks in New York and Washington did not neatly fit within conventional explanations of international conflict. In this paper, therefore, I attempt to (1) respond theoretically and pragmatically to the events and aftermath of September 11, 2001; (2) deal with Realpolitik (and one of its concomitants, ethnocentrism) and conflict resolution as traditionally contending, but potentially complementary, approaches to dealing with threats to order and security at the domestic and international levels; and (3) provide a theoretical and pragmatic basis for further research, theory building, and practice in domestic and world affairs, with a view to dealing effectively with both the deep-rooted causes and the very clear symptoms of the “new” terrorism and related identity-based conflicts fueled by the ending of the Cold War. To make sense of the attacks and to illustrate the potential complementarity of Realpolitik and conflict resolution approaches, I draw on several complexity concepts, including emergence; nonlinear, “catastrophic” responses to initial conditions; and synergistic coexistence of traditionally competing frameworks and ideas. REALPOLITIK Realpolitik is the traditional power paradigm governing efforts to manage the uncertainty and disorder inherent in “Hobbesian space.” At its most virulent extreme, it is expressed as dictatorship domestically and as imperialism internationally, with all the attendant manifestations of structural, cultural, and




physical violence—including genocide—implied by the defense and perpetuation of a preferred status quo at the expense of those who do not benefit from it (see Galtung 1969, 1996). Realpolitik has a long lineage, going back in recorded history to at least 416 BC, the midpoint of the Peloponnesian War, when Athens attempted to negotiate control over the neutral island state of Melos, a situation chronicled eloquently by Thucydides: [S]ince you know as well as we do that right, as the world goes, is only in question between equals in power, while the strong do what they can and the weak suffer what they must . . . the contest not being an equal one . . . but a question of self-preservation and of not resisting those who are far stronger than you are . . . of men we know, that by a necessary law of their nature they rule wherever they can. And it is not as if we were the first to make this law, or to act upon it when made: we found it existing before us . . . all we do is make use of it, knowing that you and everybody else having the same power as we have, would do the same as we do . . . it is certain that those who do not yield to their equals, who keep terms with their superiors, and are moderate towards their inferiors, on the whole succeed best. (1951, 331–36) This is clearly an old story, which has been repeated thousands of times up to the present day, with Hans Morgenthau (1973, 4) being one of the more “recent” successors to Thucydides and reminding us all about the “laws” that govern human behavior to Realpolitik effect. He says, • Political realism believes that politics, like society in general, is governed by objective laws that have their roots in human nature. Human nature, in which the laws of politics have their roots, has not changed since the classical philosophies of China, India, and Greece endeavored to discover these laws.

In other words, for Morgenthau and other realists, human nature—which makes “statesmen think and act in terms of interest defined as power” (1973, 5)— has not changed since Thucydides made his observations in 416 BC. Hence, the “key concept of interest defined as power is a objective category which is universally valid” (8). In the modern Westphalian world, power as interest is usually reserved for the protection of the nation-state, but it has also been used in defense of the tribe and the ethnic group.



Ethnocentrism Ethnocentrism is a natural corollary of Realpolitik: power is used by the privileged to maintain themselves and their groups at the expense of others. According to William Graham Sumner (1906), who coined the term, ethnocentrism is the technical name for this view of things in which one’s own group is the center of everything, and all others are scaled and rated with reference to it. . . . Each group nourishes its own pride and vanity, boasts itself superior, exalts its own divinities, and looks with contempt on outsiders. . . . the most important fact is that ethnocentrism leads a people to exaggerate and intensify everything in their own folkways which is peculiar and which differentiates them from others. It therefore strengthens the folkways. (quoted in LeVine and Campbell 1972, 8). Sumner also generalized “that all groups show this syndrome.” In other words, according to him and extensive research carried out by Henri Tajfel (1978, 1981) and others, ethnocentrism—following Thucydides’ and Morgenthau’s characterizations of Realpolitik—is the universal tendency for humans to divide humankind into two groups: “them” and “us.” The criteria for doing so are not fixed and can be based on, among other things, nationality, ethnicity, religion, race, class, region, or gender—criteria for which Realpolitik can mobilize defenses. Accordingly, ethnocentrism enhances intragroup community, especially under threat from out-groups (see Simmel 1955; Coser 1956), and in-group ethnocentrism works against intergroup community. Indeed, it is safe to say that, especially within a Realpolitik frame, ethnocentrism makes for a zero-sum relationship between peace at the intragroup level and war at the intergroup level. Again, according to Sumner (1906): The insiders in a we-group are in a relation of peace, order, law, government, and industry, to each other. Their relation to all outsiders, or othersgroups, is one of war and plunder, except so far as agreements have modified it. . . . The relation of comradeship and peace in the we-group and that of hostility and war towards others-groups are correlative to each other. The exigencies of war with outsiders are what make peace inside, lest internal discord should weaken the we-group for war. . . .Thus war and peace have reacted on each other and developed each other, one within the group, the other in the group relation. (quoted in LeVine and Campbell 1972, 7–8, emphasis added)



So pervasive is the tendency to subdivide humanity in this way, even under minimal intergroup differences, that fascinating experiments have been conducted with children, with the resulting “them-us” hostility between the contrived groups bordering on the remarkable. In their famous “Robbers Cave” experiments, for example, Sherif and Sherif (1953) were able to stimulate the development of hostile relationships between two groups of boys who had originally been friendly members of one and the same group. And in the famous (or to some, “infamous”) Blue Eyes/Brown Eyes exercises conducted by Jane Elliott, originally with her fourth-grade pupils in Riceville, Iowa, shortly after Rev. Dr. Martin Luther King had been assassinated: Elliott told her children that brown-eyed people were superior to blueeyed, due to the amount of the color-causing chemical, melanin, in their blood. She said that blue-eyed people were stupid and lazy and not to be trusted. To ensure that the eye color differentiation could be made quickly, Elliott passed out strips of cloth that fastened at the neck as collars. The brown eyes gleefully affixed the cloth-made shackles on their blue-eyed counterparts. Elliott withdrew her blue-eyed students’ basic classroom rights, such as drinking directly from the water fountain or taking a second helping at lunch. Brown-eyed kids, on the other hand, received preferential treatment. In addition to being permitted to boss around the blues, the browns were given an extended recess. Elliott recalls, “It was just horrifying how quickly they became what I told them they were.” Within 30 minutes, a blue-eyed girl named Carol had regressed from a “brilliant, self-confident carefree, excited little girl to a frightened, timid, uncertain little almost-person.” On the flip side, the brown-eyed children excelled under their newfound superiority. Elliott had seven students with dyslexia in her class that year [1968] and four of them had brown eyes. On the day that the browns were “on top,” those four brown-eyed boys with dyslexia read words that Elliott “knew they couldn’t read” and spelled words that she “knew they couldn’t spell.” (Kral 2000, 2, emphasis added)1 Elliott conducted the original exercise “to demonstrate to her fourth-grade students how harmful the myth of White superiority is and what, as a result of this myth, it meant to be Black in America” (1). Since then, she has appeared on television, e.g., The Tonight Show, Oprah Winfrey Show, and PBS’s documentary Frontline (see PBS 1985) in the United States. She has also gone on to conduct these exercises for adults, including police officers, around the world, letting them “experience” for themselves that

into membership in in-groups and out-groups. some commentators have been using the “clash” or “jihad” to characterize the polarizing global relationship between Judaic-Christian and Islamic “civilizations”: probably the ultimate global expression of “us-them” hostility in the history of humankind. liberals associated with flexible. therefore. Racism is not genetical. political. 2001. in some simple additive sense. Sources of Ethnocentrism Members of the conflict/conflict resolution community tend to be humanists. I have argued that biological factors play a role in human behavior—as part of a complex constellation of social. Related to the Realpolitik-ethnocentrism nexus is the seductive totality and simplicity of the clash of civilizations idea of Samuel Huntington (1993. Wilson (1979).” but of both interacting in complex ways. (quoted in Coronel 1996. political. which is associated with the Idealpolitik paradigm (see Sandole 1999a. tends to ignore—on ideological. it seems reasonable to conclude that “nature” has invested Homo sapiens with this . of “nature” and “nurture. we can learn not to be racist. Since September 11. emotional and practical grounds—the role of biology in conflict. optimistic views of human nature. among others. I have also argued (1999a. and Joseph Montville (1988) that our brains seem to be preprogrammed to bifurcate everything. Elliott has demonstrated with a simple experiment how mental models are socially constructed and can adapt to interactions with others. They tend to agree with Albert Bandura (1973) and others that whatever humans do in conflict situations is a function of learning: change what they learn and change their behavior! This view. would certainly not be in the interests of the United States or anyone else. and earlier of Benjamin Barber’s (1992) jihad. especially if accompanied by nuclear weapons. 180–85) that it is not simply a question of “nature” or “nurture. It has everything to do with power. 1996). emphasis added) In the language of complexity. such that each may be affected by the other. self-fulfillingly. 2. especially violent conflict behavior. and other factors—and adherents to that view can now “come out of the closet” without fear of being ostracized as purveyors of Nazi eugenic philosophies and programs. 110–13). more fact than fiction: a development that.COMPLEXITY AND CONFLICT RESOLUTION 47 We learn to be racist. including fellow human beings. John Pfeiffer (1984).” or indeed. economic. Edward O.2 Given the observations of. especially those with authority. This idea may become. Elsewhere (1990).

. poore. Paul Shaw and Yuwa Wong (1989) argue that the motivations that predispose human beings toward defense of their in-groups are part of “human nature”. that is. (Hobbes 1950. and (2) “the enhanced reproduction and survival of close relatives who share the same genes by common descent (a kinship component)” (Shaw and Wong 1989. brutish. Here we have the crux of the matter concerning ethnocentrism for evolutionary psychologists: “[P]roviding an ultimate. emphasis added) underpinning “we-them” distinctions. emphasis added). but the biological predisposition to bifurcate fellow members of the species into “them” and “us” nevertheless seems to be there. 103.” certain “facts on the ground. . 27). unless some kind . and such a warre as is of every man against every man . They have learned to maximize inclusive fitness—through ethnocentrism. the “seeds of warfare” lie in ultimate (in contrast to proximate) causes—inclusive fitness and kin selection. Genetic relatedness would be greatest with members of one’s lineage and one’s own kin or nucleus ethnic group” (Shaw and Wong 1989. This amounts to a “sociobiology of ethnocentrism” (44–45. including those as framed in the “clash of civilizations. nasty.” Perhaps the ultimate example of complexity in human affairs is that Humans have outfoxed themselves. in that condition which is called Warre. . . In this regard. . D. culture. learning.” and how he or she responds to them. ready to interact with culture to create certain “histories. Inclusive fitness has two parts: (1) “increased personal survival and increased personal reproduction (classical Darwinian fitness)”.3 Kin selection “implies that assistance.” where men live without a common Power to keep them all in awe.48 DENNIS J. including the genocidal ethnic cleansing that has returned to Europe in the wake of the ending of the Cold War. evolutionary rationale for cooperation and civility among genetically related individuals also provides an ultimate rationale for anticipating origins of reduced cooperation among less related individuals” (41. . . R. and short. out-group enmity. 104) Clearly.” that then become the bases of violent conflict spirals. favors or altruism would be directed at individuals who were genetically related enough to give the common gene pool greater survival advantages. and other aspects of “nurture” can impact significantly whom an agent defines as “threatening. nationalism and patriotism—to the extent that they have created the means to destroy the very inclusive fitness they seek to foster and protect. SANDOLE particular kind of “hard wiring” to protect us from one another in Hobbes’s infamous “state of nature. 26). where the life of man [is] solitary. .

and other isms are “normal. (197) CAN “NURTURE” INFLUENCE THE “NATURE” OF ETHNOCENTRISM? Concerned members of the international community could join with Jane Elliott and start to teach children in the schools. should be—and can be—changed! This would be quite a challenge to bring into any level of classroom. unfair. discuss. chap.”) They would brainstorm where those feelings come from. brought genocide back to Europe. there is no reason to believe that Homo sapiens will escape nuclear devastation. 2) counsel: to “separate the people from the problem.” that is. and on changing the mental models through which individuals comprehend the world around them and in terms of which they choose their behaviors. some of those definitions may be counter to our survival. however. where pupils and students are actively encouraged. part of our “human nature. self-fulfillingly counterproductive and dangerous and. and so on. oftentimes dysfunctional expressions of our biological predisposition to bifurcate people into friend and foe. the consequences of those feelings. or turned the United States into the most violent country in the industrialized world (see Sandole 1999a. difficulties in doing so. the culturally defined targets of those feelings are not part of our nature: they may be wrong. therefore. not that racism. by conflict resolution–trained facilitators. we are sort of stuck with it. if not extinction. either as members of in-groups or as an entire species. . This is a complex tall order: the feelings that we experience have a “natural” base. for example. therefore. originally meant to have survival value. anti-Semitism. (They would thereby make it exceedingly difficult to do what Roger Fisher and William Ury (1983. Given that the predisposition is part of our “wiring. . 4). Indeed. examples throughout the country and the world where those kinds of feelings have translated into violent conflict situations. Recent examples of its violence include the Washington. how to work on changing those feelings. but it is a necessary one if we are to make a dent on the levels of violence that have. or interact with members of certain groups.” but that they are learned.COMPLEXITY AND CONFLICT RESOLUTION 49 of action is forthcoming . Imagine classrooms at all levels.” However. as implied. to brainstorm the kinds of emotions they experience when they think about. We are not. DC–area . Hence. motivated nineteen young Arab Muslim males with box cutters to turn passenger-filled aircraft into cruise missiles against the World Trade Center and the Pentagon. up to university level. they are. stuck with the culturally/experientially determined referents of that predisposition. it would be in our best interests to work on changing those definitions.

[is] now virtually equal to the murder rate for teenagers. and as Stepp (2002. But few of us have been educated that way: we receive our degrees usually in only one discipline. Bush’s strident declaration that “you are either with us or the terrorists” has radicalized Muslims all over the world. we are returning to the dangerous simplicity of a bipolar world. elevating the level of analysis to a more global version of us-them hostilities. everything is connected to everything else (see Waldrop 1992). 2002) and the murder of three professors at the University of Arizona by a failing student (October 28. “jihadic” character. where. infants . in practice. and therefore we as analysts may also be more a part of the problem than we are of the solution.” But what about the adults. Sandole 1999a). conflict resolution is intuitively similar to complexity and provides a conceptual basis for capturing the complexity of complex con- . One of its major assumptions is that.if not interdisciplinary. 2002). given the Bush administration’s continuing war rhetoric to keep the patriotic fervor flowing beyond the ebbing impact of military successes in Afghanistan and Iraq. President George W. tended to become more a part of the problem than of the solution: more and more it has been revealed to be a significant source of self-stimulating and self-perpetuating conflict systems (see Vasquez 1993. who feel threatened and victimized by governmental security services as well as by purveyors of hate crimes (see Pierre 2002a.S. the Realpolitik-ethnocentrism nexus and the intragroup peace versus intergroup war dynamic are taking on a more global.50 DENNIS J. Conflict Resolution As an interdisciplinary field. 2002b). and the worsening insurgency in Iraq. but at the expense of the security of many American Arabs and Muslims. In other words. It has also made many Americans feel a closer sense of community. D. A3) points out. civilizational. . any attempt at problem solving must be at least multi. some of whom may be killing their kids (A17)? To what extent can nurture close the nature-nurture gap of ethnocentrism for them. . especially in the post-9/11 world? The War on Terror is currently being waged within a Realpolitik framework. Hence. Accordingly. it has. “The homicide rate for U. according to a new analysis of government data that revealed a surprising demographic milestone. SANDOLE sniper incident (October 2–24. although Realpolitik has been conceived as a rational approach to the defense of individual and national interests. among other things. Complex Systems Complexity offers insights in this regard. again.

the consequences may be catastrophically or otherwise radically different. as a result of apparently slight changes in current information. 1960a) work on the dynamics of an arms race is another source of ideas in orthodox conflict studies that show similarities with those of complexity: in a dyadic relationship. thereby keeping alive the possibility of prediction. plus the constraints of each on further arms spending (limiting factors).” which focuses on basic human needs (see Sandole 1993. 117–20. 2001. and underlying grievances. 24. For instance. to prevent the planes crashing into their intended targets. (1962. Also. complexity had been around awhile before it was conceptualized as such (see Saperstein 1995). Nevertheless. and other systems. and this gives a stability to the system in the large that it may not have in the small. Sadole 1999a. 110–13. depending upon each actor’s sensitivities to the other’s arms levels (mutual fears). However. to be effective in the long run. COMPLEX SYSTEMS AND CONFLICT STUDIES Given small differences in the start-up conditions of biological. On the other hand. if not all. 137–40). 293–305). for example. economic. they also have a good deal of stability. there could be a stable balance of power with regard to “rate[s] of rearmament or .4 To paraphrase Heisenberg’s uncertainty principle from quantum mechanics (see Nagel 1961. The distributed decision-making in complex systems and their consequent dynamism and tendency toward nonlinearity make them unpredictable: though patterns may hold for a period of time. often of dramatic change. Realpolitik must always be included in a larger frame.COMPLEXITY AND CONFLICT RESOLUTION 51 flicts. emphasis added) Lewis F. physical. their sensitivity makes them liable to change out of all proportion to any stimulus. Kenneth Boulding remarked that in conflict analysis and resolution. These goals are susceptible of change. in such cases neither analysts nor policymakers would be able to predict a system’s behavior with unlimited precision. Richardson’s (1939. Idealpolitik. Part of its appeal is that it does not replace Realpolitik as such. September 11. and something I call “nonMarxist radical thought. where it coexists and coevolves with. a metaparadigm. Marxism. because on occasion we need the military to prevent or stop atrocious acts of violence such as the genocidal conflicts of recent times in Rwanda and Bosnia. innovations in thought. there would also be discernible patterns underlying chaos. As with many. Human beings are moved not only by immediate pressures but by distant goals that are contemplated in the imagination. we could have used our police or military as armed marshals on board the four hijacked aircraft on Tuesday.

98) refers to as a schismogenic “regenerative causal circuit or vicious circle”—is implicit in Realpolitik: the use of a “measured” amount of force. where attempts to deal with them could backfire. 2. at least up to the Dayton Peace Accords of October– December 1995. . 129–31). within the unstable condition. “for relatively modest variations in . But we should be fair: the danger of paralysis derives from the possibility that conflicts-as-process could. with the exception of the NATO bombing campaign that. . . or at least nothing of major significance: witness Bosnia and Herzegovina. There could also be radical shifts between stable and unstable systems (in either direction). could backfire. 152). beyond some threshold. The danger in this. unpredictably. or there could be an unstable equilibrium in which either complete disarmament or a runaway arms race is possible. we can talk of entropic conflicts: conflicts that approach entropy. assumptions” regarding mutual fears. SANDOLE disarmament”. even as part of an Idealpolitik strategy to achieve negative peace as a necessary (but not sufficient) condition of positive peace. is that conflict researchers may be paralyzed into recommending nothing and policy-makers paralyzed into doing nothing. limiting factors. Robert Axelrod’s (1984) fascinating theory of cooperation has not been applied to Bosnia. also see Boulding 1962. Rapoport 1960. led to the Dayton Peace Agreement. or predict the course of any particular conflict-as-process.5 COMPLEX SYSTEMS AND CONFLICT RESOLUTION: MANAGING ENTROPIC CONFLICT SYSTEMS IN THE POST–COLD WAR WORLD It is clear that conflict researchers and policy-makers cannot predict with certainty what kinds of conflicts-as-process will emerge from various kinds of conflictsas-startup conditions (see Sandole 1999a. chap. Or. conflicts can assume an entropic character (as Iraq may already have)—what Gregory Bateson (1973. leading to destruction of the conflict systems themselves. there could be radical shifts from complete disarmament to a runaway arms race (or vice versa). In such cases. In other words. chap. conflicts-as-process could escalate into self-stimulating/self-perpetuating spirals. in part. and because of very small shifts in existing conditions. unpredictably. escalate out of control (a continuing risk in Iraq). of course. The danger that. This may explain why. 1. or progressive disorder. and Saperstein 1995). making matters worse.52 DENNIS J. resulting from “small shift[s] in the position of the initial point” of armament expenditures at “time zero” (152. D. and grievances (Nicholson 1989.

that is. 4). In the absence of the “provocability” of the Bosnian Slavic Muslims. plus the “nonprovocability” of the international community.COMPLEXITY AND CONFLICT RESOLUTION 53 Axelrod has argued that in all situations involving the prisoner’s dilemma—a classic game of Realpolitik analysis and prescription that applies to the interpersonal as well as international levels—the best way to act is in terms of the TIT FOR TAT strategy: TIT FOR TAT’s robust success [in prisoner’s-dilemma situations] is due to being nice.” which was prosecuted by the Serbs against the major victims of the wars in former Yugoslavia. chap. however. Serbian military successes. Its retaliation discourages the other side from persisting whenever defection is tried. Its niceness means that it is never the first to defect. And its clarity makes its behavioral pattern easy to recognize. it is easy to perceive that the best way of dealing with TIT FOR TAT is to cooperate with it. Subsequently. “ethnic cleansing. also see 54) For TIT FOR TAT to work. and once recognized. Extending Axelrod’s theory to the wars in former Yugoslavia during 1991–95 leads to the following scenario: 1. and this property prevents it from getting into unnecessary trouble. Bosnian Slavic Muslims. forgiving. that is. the international community was “forgiving” toward the Serbs to avoid stimulating new or exacerbating ongoing violent conflict spirals. Its forgiveness helps restore mutual cooperation. emphasis added. 2. 3. it “requires that the players have a large enough chance of meeting again and that they do not discount the significance of their next meeting too greatly” (174). Slovenian and especially Croatian declarations of independence from the Yugoslav Federation in June 1991 resurrected Serbian fears (especially among Serbs living in Croatia) of Croatian defection from the stable TIT FOR TAT equilibrium that had existed up to that point. “the future must have a sufficiently large shadow”. the international community was effectively shamed into becoming “provocable” and retaliating against the Serbian “defection” from the previously stable TIT FOR TAT equilibrium. and clear. see Axelrod 1984. provocable. (176. although in a very restrained way (as in the live-and-let-live system of trench warfare during World War I. . stimulated the development and exacerbation of a violent. asymmetrical conflict-as-process.

. While Serbs celebrated the first “anniversary” of the fall of the UN “protected safe area” of Srebrenica. and reiterated their goal of keeping the territory “ethnically pure. 1996 (see Hedges 1996a).54 DENNIS J. Axelrod’s theory is certainly appropriate for Realpolitik-defined realities. . in the future. But even with Dayton. Paul Stern and Daniel Druckman (1994/1995. Serbs marked their victory . a humiliating retreat and one that was greeted with jubilation in the self-styled Republic of Srpska” (Hedges 1996b). 114) view Axelrod’s theory as an example of the strongest evidence of the “hegemonic position of realism” in US international relations thinking and practice. during the summer of 1996 demands by the international community and threats of sanctions were followed by vague promises by the Bosnian Serbs to comply. Until Dayton. freedom of expression. according to many western diplomats. perhaps because of paralysis associated with the unpredictability of the consequences of even minor adjustments in complex systems capable of generating entropic conflict processes. become appropriately “provocable. involvement in the Yugoslavian imbroglio was considerable uncertainty as to the wider ramifications of the gathering storm” (emphasis added). and their failure to do so was followed by a breakdown on sanctions. As Michael Lund (1996. the “provocability” of the international community remained an issue. because it effectively legitimates cooperation within the Realpolitik paradigm. 111) puts it: “From 1990 into 1992. a major obstacle to European and U. and municipal elections a year later. with the exception of the NATO bombing campaign leading up to Dayton Axelrod’s theory remained basically untried and untested in former Yugoslavia.” In terms of this analysis. The international community embarked on a “train-and-equip” program for a joint Bosnian Muslim–Croat army (see Pomfret 1996a) so that Bosnian Muslims in particular could. “It was. even though the Organization for Security and Cooperation in Europe (OSCE) declared that national elections could nevertheless take place on September 14. it may be remembered. then it is clear that.”6 Taken together with the observation that the conditions specified by Dayton for “free and fair elections” in Bosnia—freedom of movement. war crimes investigators were sorting through the remains of men and boys captured and shot after the Muslim enclave fell. such as the wars in former . D. and freedom of association—had not been met. SANDOLE 4. for instance. nearly one year following the cessation of hostilities in Bosnia. the international community had not been sufficiently “provocable” in Bosnia and Herzegovina. freedom of press. This only emboldened the Serbs.S. the “provocability” of the international community (more so than of the Bosnian Muslims) was the issue.

Idealpolitik. only two—provocability and clarity—reflect Realpolitik as such. such knowledge can most frequently be attained only through conflict. but of using other tools as well when those conditions have changed (also see Fisher 1993. paradoxical as it may seem.” But of the four elements of TIT FOR TAT. In Bosnia and Herzegovina. struggle may be an important way to avoid conditions of disequilibrium by modifying the basis for power relations. 115–17) and by my own “4 2 framework. and non-Marxist radical definitions of reality. that “We will defend our frontiers biologically” (Dobbs 1996a. 110–13). 259–60. 8). 292–94).” in which “conflict may be an important balancing mechanism” designed to achieve the very equilibrium that may be absent to begin with: Conflict consists in a test of power between antagonistic parties. a Bosnian Serb official in the city of Brcko characterized the Dayton Peace Agreement’s call for the return of refugees as “a clear attempt to change the biological structure of the city. Marxist. for instance. Indeed. through learning. the development of cooperation out of the “coevolutionary dance of competition and cooperation” (see Waldrop 1992. there must be. admittedly.” which combines Realpolitik. emphasis added) . 136) refers to as a “trial by ordeal. Accommodation between them is possible only if each is aware of the relative strength of both parties. thereby implying a continuation of the doctrine and practice of “ethnic cleansing. Consequently. Anthony Lewis (1996. This “complex” orientation shares with Fisher and Keashly’s (1991) “contingency model” the prescription of using what is necessary under one set of conditions. ethnic cleansing had eroded). Unless a stable balance exists.” He went on to assert. TIT FOR TAT’s other two elements—niceness and forgiveness—locate it in a more “complex” constellation of options. However. But for TIT FOR TAT to be ultimately successful. 262–65. chap. plus cooperative and competitive means for dealing with conflict (see Sandole 1999a. Fisher 1997. emphasis added). TIT FOR TAT is a response to “complexity”: it can encourage. (137. in addition to a “sufficiently large shadow” of the future (which.COMPLEXITY AND CONFLICT RESOLUTION 55 Yugoslavia. TIT FOR TAT is an example of a process of agent interactions in which each agent learns to coevolve with others. since other mechanisms for testing the respective strengths of antagonists seem to be unavailable. with Muslim refugees in mind. very much like that suggested by Stern and Druckman’s own “contours of a new paradigm” (1994/1995. stability in the sense of Richardson’s (1939. the parties may engage in what Lewis Coser (1956.” Also. 11) had concluded that “The only thing that ever moved the Bosnian Serbs to more than empty promises during the war [there] was force. 1960a) “balance of power”—another Realpolitik aspect!—between the “coevolving” parties in their respective capabilities to inflict pain on each other.

It is by this process that out-group enmity and ethnocentrism have been reinforced and carried over from nucleus ethnic group to band. to tribe. . a problem with “balance of power.” as there is with Realpolitik in general. .” especially in the relative absence of such on the part of the international community. intervention.56 DENNIS J. proximate. in the shorter run. struggles to maintain balances of power [have given] rise to more complex societal units which [have] continued the legacy of intergroup warfare. And the same nationalist parties that incited the conflict had been .” as manifested in former Yugoslavia. conflict. The US-led effort to arm the Bosnian Muslims was designed to “make possible a reassessment of relative power and thus serve as a balancing mechanism which helps to maintain and consolidate societies” (137) and to provide a material basis for increased Muslim “provocability. 47) imply. also see USIP 1997). to nation-state. Motivated by resource competition. encounter the difficulty that the assessment of the actual power relations between the contenders can hardly be made before their relative power has been established through struggle” (135–36). emphasis added.g. 1996). “trials by ordeal” to determine “relative strength. the official originally in charge of the US program. Indeed. in positive peace fashion. the weapons “would be used for Bosnia’s defense and would contribute to stability in the region. (45) Applying complexity concepts. and the functions of government relied on Western representatives with sweeping powers. . .S. groups and their expansion figure as forces of selection in our theory. however. mediation or arbitration). therefore. the balances a “natural” consequence of social processes (see Galtung 1969. SANDOLE Apropos less lethal forms of conflict handling (e. to chiefdom. environmental cause [of war]”: Since failure to maintain a balance of power could have resulted in extinction. It depended on a North Atlantic Treaty Organization (NATO) force of twelve thousand foreign troops.. involves more than stable balances associated with negative peace of an enforced temporary respite from violent conflict. Bosnia was unable to function as a sovereign state. according to James Pardew. . exploring the dynamic of deep-rooted processes and conditions that make. D. it also involves building upon and transcending these and. There is. Coser tells us that such “Efforts . The purpose of the trainand-equip program [therefore] is to prevent war by creating a military balance in Bosnia” (Pomfret 1996b. As Shaw and Wong (1989. and warfare. Seven years after the U. are associated with “groups as forces of selection [that] represent an emergent. thereby establishing a stable balance of power and ensuring that TIT FOR TAT succeeds in Bosnia without further international intervention.

complexity in general—still remain to be fully applied to Bosnia. John Mearsheimer’s (1990a. regularity. as “extreme” Realpolitik practitioners implement more and more antidemocratic and threat. But the irony here is that. chap. but also— with provocability still an issue—that the negative peace is not a stable one. Realpolitik philosophers. therefore—and with it. Hence. Kenneth Waltz’s (1964) earlier defense of a “bipolar” international system as inherently more conducive to stability than a multipolar system. tend to find democracy too chaotic. therefore. THE “COMPLEXITY” OF COMPLEX SYSTEMS As implied thus far.000 personnel. the NATO force was replaced by a British-led EU peace keeping force of some 7. 1990b) lamenting of the end of the Cold War and its simplicity. President George W. unpredictability. of order and security—in their domestic and international environments through the simplistic bifurcation of the species into “them” and “us” and by the threatened or actual use of force against “them” whenever circumstances within the Realpolitik/ethnocentric frame call for such.or force-based measures in pursuit of order and security. TIT FOR TAT. generating “security dilemmas” (Herz 1950) and the eventual collapse of their own systems. their efforts tend to become more and more counterproductive and self-defeating (see Burton 1972. and practitioners tend to respond to the disorder. 6). At the extreme “right-wing” end of the Idealpolitik-Realpolitik continuum. and insecurity inherent in “Hobbesian space” by advocating and/or pursuing the enhancement of predictability. therefore. we find authoritarians who have a low threshold for uncertainty and insecurity and who. If such policymakers are alive and well at the end of the day and respond to the “cognitive dissonance” (Festinger 1962) generated by their failed Realpolitik-based policies and expectations with a “paradigm shift” (Kuhn 1970) to Idealpolitik-based norms and polices. theorists. and stability (the “PRS needs. Bush’s strident declaration to the entire world that “[y]ou are either with us or the terrorists!” This desire for simplicity in the face of real complexity is the foundation of most conventional perspectives on conflict and its resolution.” see Sandole 1984)—and. then we might have a situation as we did following the termination of World War II in Europe. and following the terrorist attacks of September 11. when the erstwhile mortal enemies Germany and France established the basis for what has become the .COMPLEXITY AND CONFLICT RESOLUTION 57 reelected (WP 2002). 2001. The problem with the Dayton Peace Agreement for Bosnia is not only that the physical and emotional reconstruction of the country (positive peace) has lagged behind the enforced prevention of violence (negative peace). In December 2004.

” But there is irony here as well: even Idealpolitik may contain the seeds of its own destruction. therefore. Once the “bite-and-counterbite” dynamic of terrorism versus counterterrorism reaches some critical threshold. successful experiments in peacebuilding (still ongoing!) in pursuit of “positive peace. for “drama. and Koch 1961. motivated by fundamental ideology more than by a political agenda. As Paul Sites (1973) and Kenneth Boulding (1962). instability. D.” lest frustration/aggression-based cycles of violence degenerate further into the “new” terrorism (see Sandole 2002)! One of the interesting aspects of the “new” terrorism—where terrorists are quite prepared to die in the execution of their acts to inflict catastrophic damage and destruction on their symbolic and human targets—is that the terrorists.” This is the crux of complexity: the “need” to nudge systems at the “edge of chaos” so that neither chaos nor order prevails at the zero-sum expense of the other. and Brody 1968). for example. disorder and insecurity in the West and those supported by the West. former Yugoslavia was “able” to implode genocidally in the 1990s (an incredibly negative event) after years of intergroup stability and relative prosperity (a positive setting). Hence.58 DENNIS J. Hence. Given that complex systems can shift “catastrophically” from one end of the continuum to the other with apparently little effort. the self-fulfilling confirmation of Huntington’s (1993. Holsti. their ultimate intention and effect are to generate maximum unpredictability. North. psychologically. Frankel’s feat is an example of the potential for diversity in human mental models. whatever else we as humans may “need” to develop effectively—physically. because “too much stability” may lead to boredom and atrophy. another consequence is each side’s overperception of and overreaction to the actions of “the Other” (see Zinnes. 1996) otherwise contentious “clash of civilizations” . among others. one implication here for the architects of globalization is the need to creatively influence the balance between order (“McWorld”) and disorder (“Jihad”). Quite a challenge. SANDOLE European Union: one of the most illustrative. North. Consequently. and. especially when people are stressed by threats at home and abroad. those in the developing world who have traditionally borne the brunt of colonialism and imperialism have a chance to close the gap between the “haves” and the “havenots. emotionally. Yugoslavia’s flip to violent conflict may reflect complex system sensitivity to initial conditions and small events. have argued. so that. One of the major lessons of complexity. and socially—we also have a need for stimulation. Viktor Frankel (1985) was able to survive the horrors and brutalities of a Nazi concentration camp (an incredibly negative setting) by discovering “meaning” in his adversity (a remarkably positive occurrence). Similarly. are not deterred by traditional Realpolitik threats or the actual use of force. is to never take anything for granted for too long on either end of the Idealpolitik/voluntary order– Realpolitik/force-based order continuum. therefore.

rush to war against. and pillar 3.S. manifest conflict processes (MCPs). 2003) as one such framework for identifying and integrating factors associated with traditionally competing frameworks. all further enhanced by the U.1. chap. Basically. and apparently long-term occupation of. 6.COMPLEXITY AND CONFLICT RESOLUTION 59 or of Barber’s (1992) “jihad. Iraq. I have used this framework as a basis for developing a “new European peace and security system” (NEPSS) potentially relevant to preventing “future Yugoslavias” (see Sandole 1999a. complex-systems concepts have given new meaning to a possible antidote: frameworks that can potentially integrate most if not all disciplines in an effort to explain and to facilitate dealing with the foci of any one of them. the three-pillar framework comprises pillar 1. 1999a. 7 and below).” with weapons of mass destruction in the bargain. another developing country in the Arab/Muslim world. Three-Pillar Framework While complexity may have generated paralysis over Bosnia. I have developed the “Three-Pillar” comprehensive mapping of conflict and conflict resolution (see Sandole 1998a. chap. THREE-PILLAR COMPREHENSIVE MAPPING OF CONFLICT AND CONFLICT RESOLUTION Pillar 2 Conflict Causes and Conditions Individual Societal International Global/Ecological Pillar 1 Conflict (Latent [Pre-MCP]) MCP/AMCP Parties Issues Objectives Means Conflict-handling Orientations Conflict Environments Pillar 3 Conflict Intervention Third-Party Objectives Conflict Prevention Conflict Management Conflict Settlement Conflict Resolution Conflict Transformation Third-Party Approaches Competitive and/or Cooperative Processes Negative and/or Positive Peace Orientations Track 1 and/or Multitrack Actors and Processes . or aggressive manifest conflict processes (AMCPs)—while pillar 2 deals with conflict causes and conditions. conflict (third-party) intervention (see table 3. TABLE 3. conflict—latent conflict (pre-MCP).1).

1999b) and that. a military officer.” The New European Peace and Security System (NEPSS) I have used the three-pillar framework as a basis for designing the NEPSS: an intervention into post–Cold War Europe that just might be relevant to preventing “future Yugoslavias” (see Sandole 1998b. resolution. a humanitarian aid worker. management. the long-term objectives they hope to achieve by waging conflict over certain issues. societal. and “track 1” (official. “negative peace” and/or “positive peace” orientations. Complexity is especially relevant here. The working hypothesis of the three-pillar framework is that to design and implement an effective intervention into any particular conflict “space” under pillar 1. the issues about which they are in conflict. 1999a. For example. a journalist. chap. Reflecting the complexity perspective. Similarly. as it provides the essential conceptual basis for combining into a coherent whole traditionally competing frameworks and ideas. can be accommodated within the three-pillar framework. and the conflict “spaces” within which their conflict is occurring. plus the means for achieving any of these objectives: competitive (confrontational) and/or cooperative (collaborative) processes.60 DENNIS J. a psychologist. Pillar 2 comprises four levels of explanation—individual. D. if we were to ask an anthropologist. the three-pillar framework maps the conflict. we would also get different responses. unofficial. the means they are employing. an international relations specialist. a diplomat. also may be relevant to conflict interventions outside Europe. appropriately adapted. governmental) and/or “multitrack” (nongovernmental. and potential interventions at multiple levels. Finally. a citizen activist. This is precisely what I have attempted to do with “NEPSS. its causes. and global/ecological—that capture potential causes and conditions of the conflict occurring in the “conflict spaces” of pillar 1. a historian. and it is not possible to anticipate the effects of interventions. Complex problems are characteristics of whole systems. pillar 3 deals with third-party objectives such as violent conflict prevention. we have the parties. and a religious leader about how to deal with Yugoslav-type conflicts. their preferred conflict-handling orientations. and a sociologist for their views on why former Yugoslavia imploded into a genocidal frenzy during the 1990s. international. settlement. however. and/or transformation. 7. a political scientist. So. . and other) actors and processes. they do not have simple solutions. SANDOLE Under pillar 1 (the “Middle Kingdom”). we would likely get radically different responses. if we were to ask a businessperson. a potential third party under pillar 3 will have to “capture the complexity” of the conflict as represented by all four levels of potential “drivers” under pillar 2. All these. an economist.

. Quite simply. Under the vertical. at its December 2002 summit in Copenhagen. the EU issued invitations to Cyprus. Hungary. facilitating a genuine paradigm shift from Realpolitik. private citizen. NATO represents an example of political and military aspects of a reframed. training. and Slovenia. research. Prescriptively. Latvia. and media—corresponding to each level (see Diamond and McDonald 1996). Peter Wallensteen [1998] and Walter Kemp . positive-sum common security. societal. NEPSS also employs the basic structure of the OSCE as a conceptual and operational framework for enhancing the complementarity and synergy of all mechanisms working together on common problems. religious. the European Union (EU) an example of economic and environmental aspects.g. subregional. business. more comprehensive sense of security. NEPSS makes use of existing international organizations in Europe such as OSCE. the Czech Republic. at its November 2002 summit in Prague. and Poland as members. More importantly. Descriptively. the European Union (EU). Poland. Slovakia. and the North Atlantic Treaty Organization (NATO). NATO. there has been a need for something else to deal with the conflicts of the post–Cold War world. with track 1–9 actors and processes—governmental/official. assuming an early warning system to activate the preventive diplomacy envisaged by Michael Lund (1996) and others (e. Lithuania. funding. Hence. Slovakia. This is where the prescriptive element enters the picture. zerosum national security to Idealpolitik. all these organizations are basically interstate in nature. Estonia. while the problems posed by conflicts in former Yugoslavia and elsewhere are essentially intrastate in nature. For example. Subsequently. and education. Romania. Within this framework.” And. and the Council of Europe (CoE) an example of humanitarian and human rights aspects of comprehensive security. which had already taken in the Czech Republic. developments that are actually occurring as well as those that could or should occur. and Slovenia—all of which became members by March 2004. Malta. activist. we would have a mapping of Europe in terms of the local. Estonia. the Council of Europe (CoE). regional. nongovernmental/professional. NEPSS is characterized by integrated systems of conflict resolution networks. with vertical and horizontal components. Lithuania. post–Cold War institutions. issued invitations to seven other former members of the communist world—Bulgaria. each of these heretofore Cold War institutions has been reaching out to its former enemies. and global levels of analysis. The idea here is that “all conflicts are local. these developments are nothing short of revolutionary. Latvia. all of which became members by May 2004.COMPLEXITY AND CONFLICT RESOLUTION 61 NEPSS comprises descriptive and prescriptive elements—that is. But revolutionary though these developments are. Hungary. inviting them to become members or join together in constituting new.

and the escalating development of a “clash of civilizations” or “jihad” between the Judaic-Christian and Islamic worlds against the background of easily available weapons of mass destruction—are not only tragic but. with one or both sides trying to “preempt” the other. for example. Should the vertical dimension fail to prevent “the house from catching on fire. conflicts developing at any local level could be responded to by a synergistic combination of track 1–9 resources at that level—plus. The primary message of a complexity approach to conflict analysis and resolution is that there may be a need for a policeman to pull the attacker off a victim—for example. therefore. Even the 1999 NATO air war against Serbia over Kosovo—albeit clearly for the humanitarian purpose of preventing further genocidal ethnic cleansing of Kosovar Albanians—falls more into the category of the narrow use of Realpolitik force basically within a Realpolitik (instead of an Idealpolitik) framework. all Saudis. the destruction of Grozny and killings of tens of thousands of Chechen civilians in the Russian Federation). via the law of unintended consequences. all Palestinians. that all Afghans. for NATO to stop genocide. the possibility of a nuclear war between India and Pakistan over Kashmir. the U. the “new” terrorism. societal. While some recent developments in Europe are suggestive of progressive reinforcement of NEPSS’s descriptive character and the “vertical” dimension of its prescriptive character—such as the emergence from the November 1999 OSCE Summit in Istanbul of the Charter for European Security. all Arabs. But this is not the same as declaring or insinuating.” then there may be a need for the horizontal dimension to become operational.62 DENNIS J. These and other developments—the relentless Israeli-Palestinian carnage.. but basically within an Idealpolitik framework. potentially very destabilizing. to achieve negative peace (suppression of the fire) but only as a necessary (not sufficient) condition for achieving positive peace: the elimination of the (pillar 2) underlying causes and conditions. to the extent necessary and possible. in part.” and then proceeding to eliminate (or be perceived to be eliminating) the entire population and its culture as a way to deal with the “terrorist” problem. all Wahhabis. Paradoxically. this problem is being. or all Muslims are “terrorists. D. 1999b)—other developments are suggestive of the sole narrow use of Realpolitik force (e.–Iraq war. This would involve the judicious use of Realpolitik force. created self-fulfillingly by this perspective and corresponding behavior! Within the terms of the argument posed here. SANDOLE [2001]). a framework that also allows for and encourages conflict resolution (dealing with the underlying causes of the fire . inclusive of the Platform for Co-operative Security (see OSCE Istanbul 1999a. regional. subregional. all Chechens. and global levels as well.S.g. Realpolitik force must always take place within a bigger picture.

China.g.g. paving the way for a new definition of “the enemy” as any and all assaults to the global commons: a truly complex approach to a set of complex problems at the “edge of chaos.” we can leave it to the children: the next generation of decision-makers. as well as (violent) conflict prevention (preventing the house from catching on fire in the first place). If peace is not positive as well as negative—if it does not ultimately deal with the underlying “conflicts-as-startup conditions”—then “conflict-as-process” will never be far from the surface. global terrorism itself just might provide the motivation for the international community to come together. always available to come back to haunt us time and time again! This is the ultimate message and categorical imperative of a complexity approach to conflict analysis and resolution: not only to think and act outside the Realpolitik-only box. have failed to rise to the status of William James’s (1989) “moral equivalent of war. Palestine. This would be a truly superordinate undertaking that could galvanize the international community into developing a culture of global problem-solving that transcends traditional ethnocentrism and a reliance on Realpolitik-only perspectives and measures. forcefully putting out the fire). among other compelling elements of the global problematique (e. “biological”) reasons. but to combine it synergistically with other. Russia. AIDS). In the meantime. North Korea. 2005). ways of knowing and acting.. Madrid train bombings (March 11..” Among the conceptual tools that could facilitate movement in this constructive (albeit ambitious) direction is the Three-Pillar Framework (3PF) or . and a growing gap between haves and have-nots. but only in collaboration with others (see Sherif 1967). But for some inexplicable (perhaps. ecological degradation. CONCLUSION It would take extremely enlightened leadership—in the United States. preventing the spread of the fire). 2004). Europe. and London transit bombings (July 7. exponential population growth. Saudi Arabia)—to pursue “positive” as well as “negative peace” in coordinated response to assaults to the “global commons”: superordinate goals that no one state can achieve on its own. especially after the Bali bombings (October 12.” Perhaps. among others (e. 2002). conflict management (if initial conflict prevention fails. usually competing. in part. and Japan.COMPLEXITY AND CONFLICT RESOLUTION 63 at hand) and conflict transformation (dealing with the long-term relationships among the survivors of the fire). and not just to “root out” terrorism (Realpolitik). Israel. but to deal with its root causes as well (Idealpolitik). 2002). “if we have time. and conflict settlement (if management fails. Moscow Chechen hostage crisis (October 23–26. Iraq. however.

5. Analysts working together with policy-makers could use either or both to capture the complexity of complex conflict situations. a flip in the system to something quite different—as when a forest becomes a desert. small changes in initial conditions may lead to nonlinear system changes. furthermore. In this way. To a very large extent. 1. 2003. emphasis in the original). D. Sensitivity to initial conditions is a characteristic of complex systems that can also be found in some simple systems. those interventions may actually be exacerbating the causes of 9/11-type terrorism. Such counterproductivity is the price that policy-makers—and the rest of us— might continue to pay for rejecting or otherwise avoiding conceptual tools that transcend symptoms and capture the complexity of complex conflicts.7 NOTES An earlier version of this paper was presented at the 44th Annual Convention of the International Studies Association (ISA).” 2. .64 DENNIS J. SANDOLE 3PF-generated new European peace and security system (NEPSS) discussed briefly in this chapter. the US-led invasions and occupations of Afghanistan and Iraq (especially the latter) seem to be addressing only the symptoms of the conflicts that have torn these Muslim countries apart. This section reflects and builds upon parts of chapter 8 (especially pp. “Inclusive fitness thus equals an individual’s Darwinian (egoistic) fitness augmented by an allowance for the effect that the individual can have on the reproductive success of those who share identical genes by common descent” (Shaw and Wong 1989. February 25–March 1. 4. see Cowley (2003) and Oldham (2003). Also see Rosenthal and Jacobson’s (1968) classic study “Pygmalion in the Classroom. 3. 26–27. Panel on “Global Complexity: Agent-Based Models in Global and International Studies. however. Oregon. For other discussions in this regard.” Portland. The author gratefully acknowledges comments and suggestions made by Neil Harrison and Patrick James. But it does not follow that a system in which small changes in initial conditions “cause” large system changes is necessarily complex. 193–201) of Sandole (1999a). In complex systems. they could deal with relationships that have gone wrong and the underlying causes and conditions driving negative developments in those relationships as well as the symptoms (indicators) of those negative relationships.

1997. 1–2. England: Paladin. no. Englewood Cliffs. 1972. England: Manchester University Press. 3 (March): 53–65. I have applied the 3PF to an analysis of the causes of 9/11type terrorism (see Sandole 2002) and the 3PF-generated NEPSS to a design for an EU intervention into post-NATO Bosnia (see Sandole 2004). New York: Basic Books. Aggression: A Social Learning Analysis. Axelrod. 1997. Bateson. Burton. Psychiatry. 2003.” Contrast. The Evolution of Cooperation. 288. ———. “Serbs Mark a ‘Victory. 1973.” Atlantic Monthly 269. 1996. Barber.” Scientific American.html. whose ancestors adopted the Islamic faith of Turkish invaders] in Republika Srpska. The Complexity of Cooperation: Agent-Based Models of Competition and Collaboration. no. Albert-Laszlo.magenta. Manchester. 1996.” said General Milenko Zivanovic. NJ: Prentice-Hall. ———. emphasis added). Kenneth E. Bandura. Mercita. and Epistemology. International Herald Bertalanffy. General System Theory. Boulding. Evolution. Ludwig von. “There is no place for Turks [the derogatory term the Serbs use for Bosnian Muslims. http://www. NJ: Princeton University Press. 12 July: 10. Steps to An Ecology of Mind: Collected Essays in Anthropology. the regional commander. Harmondsworth. Conflict and Defense: A General Theory. “Interview with Jane Elliott. . 1992. REFERENCES AP. who led the final assault on Srebrenica” (AP 1996.) 7. New York: Harper and Row. Violence Explained.’ Victims Are Exhumed. and Eric Bonabeau. Cambridge: Cambridge University Press. Coronel. Gregory. Princeton. 1973. Robert. Benjamin R. John W.” Associated Press. Thus far. “Scale-free Networks.COMPLEXITY AND CONFLICT RESOLUTION 65 6. McWorld. 28 March. 1973. 5 (May): 60. “Jihad Vs. 1962. Barabasi. (Also see Honig and Both 1996 and Rohde 1997. Frogmore. World Society. Albert. England: Penguin. 1984.

CT: Kumarian Press.66 DENNIS J. no. London: Penguin Books.S. Festinger. ———. “The Story of Brcko Is the Story of Bosnia. Hedges. Man’s Search for Meaning. West Hartford. 1956. Cowley.” International Herald Tribune. Syracuse. A Theory of Cognitive Dissonance. Lewis A. 1 (February): 29–42. Fisher. 1962. 1996a. 1985. Roger.. and John W. Our Fears (Anxiety and Your Brain: How Living With Fear Affects the Mind and Body). and Loraleigh Keashly. NY: Syracuse University Press. Multi-Track Diplomacy: A Systems Approach to Peace. Peace and Peace Research. 26 June. 1987. 42–52. and William Ury. London: Sage Publications. Chris. 1996. Geoffrey. NATO Officials Say. “Bosnian Vote in the Fall Wins a Wary Approval. 6. CA: Stanford University Press. 1996a. 1996b. Frankl. 2003. James. Viktor E. Diamond. New York: Free Press. McDonald. “The Potential for Peacebuilding: Forging a Bridge from Peacekeeping to Peacemaking. 1993. Chaos: Making a New Science. 3 (July): 247–66. “Our Bodies. SANDOLE Coser. Revised and updated. D. Ronald J. 24 February. Peace By Peaceful Means: Peace and Conflict.” International Herald Tribune. “Violence.” Peace & Change 18. ———. 28 June. 1991. Fisher.” Journal of Peace Research 28. 1969. Stanford. 1996. Interactive Conflict Resolution. Oslo (PRIO).” Journal of Peace Research 6.” Washington Post. London: Collier-Macmillan. no. “U. Galtung. The Functions of Social Conflict. Jr. Fisher. for the International Peace Research Institute.” Newsweek. 3:167–91. A25. ———. Getting to Yes: Negotiating Agreement Without Giving In. Ready to Participate in New Force in Bosnia: Gore Signals Willingness. Michael. Leon. 1983. “The Potential Complementarity of Mediation and Consultation within a Contingency Model of Third Party Intervention. Gleick. Ronald J. 2. New York: Penguin Books. Dobbs. 14 November. New York: Washington Square Press. 3rd ed. Louise. . 1997. Johan. Development and Civilization. no.

1996. 1996. James. publ. 1990a. Homewood. New York: Simon & Schuster. Institute of Peace Press. and R. Chicago: University of Chicago Press.” Horizon. August 1910. Leviathan.” International Herald Tribune.” International Herald Tribune. New York: John Wiley. (Orig. 2000. 1989. New York: Free Press. Lewis. Dent and Sons. 2 (January): 157–80.. 9 July. http://www. Michael S. ed. Preventing Violent Conflicts: A Strategy for Preventive Diplomacy. “West’s About-Face on Karadzic: Familiar and Sad. A. William. Walter A.” International Security 15. Lund. Dutton. . David Singer.) Holsti. 1 (Summer): 5–56. The Clash of Civilizations and the Remaking of World Order. New York: E. Hobbes. Jan Willem. 1–5. Robert A. Kuhn. “Back to the Future: Instability in Europe After the Cold War. Mearsheimer. 1997. 1968. ———. 2 July. “Perception and Action in the 1914 Crises. IL: Dorsey Press. “Karadzic and Co. 1972. Huntington. no.) Jervis. Brigitta. 11. 2nd ed. and Donald T. 463–68. “The Eyes of Jane Elliott—Blue Eyes Brown Eyes Exercise. New York: Penguin Books.. John J.S. NJ: Princeton University Press. 2001. 1993. 3 (Summer): 22–49. Kral. Thomas S. O.” In International War: An Anthology and Study Guide. Ethnic Attitudes and Group Behavior. The Structure of Scientific Revolutions.” In Quantitative International Politics: Insights and David Singer. 10. 1. Brody. P. “Idealist Internationalism and the Security Dilemma. M. R. Quiet Diplomacy in Action: The OSCE High Commissioner on National Minorities.. C. Washington. London: Collier-Macmillan. 1996b. no. Melvin Small and J. ed. John H. (Orig. 1996. DC: U. Srebrenica: Record of a War Crime. 1950. 1970. R. Samuel P. System Effects:Complexity in Political and Social Life. no. 1950. in McClure’s Magazine. Kemp. Anthony. The Hague: Kluwer Law International. in 1651. Honig. Robert. London: J. publ. Thomas. Ethnocentrism: Theories of Conflict. 2nd ed. Princeton.COMPLEXITY AND CONFLICT RESOLUTION 67 ———. “The Moral Equivalent of War. Herz. J. Have to Be Tried. and Norbert Both. “The Clash of Civilizations?” Foreign Affairs 72. LeVine. Campbell.horizonmag.asp. ed.” World Politics 5. North. 1996.

D. 5th ed. New York: Alfred Knopf. 1973. 1961. “Victims of Hate. “Human Nature: The Universe Inside Your Skull. Cambridge Studies in International Relations. PBS. but Repeated Use May Cause Lasting Harm. Formal Theories in International Relations.” Science Digest 92:92. Starts Delivery of Heavy Weapons to Bosnia’s Muslim-Croat Forces. 3. Nagel. 1984. 14 September. ——— 309. August. Northvale. The Dreams of Reason: The Computer and the Rise of the Sciences and Complexity. Heinz R. Politics Among Nations: The Struggle for Power and Peace. John. 2002b. F1. Peters). New York: Simon & Schuster. Montville. 35–50. “Fear and Anxiety Permeate Arab Enclave Near Detroit: Muslim Americans Feel They Are Targets in War on Terror. Morgenthau. 5. 1990b.” Atlantic Monthly. John. The Structure of Science: Problems in the Logic of Scientific Explanation. A3. NJ: Jason Aronson. F5. “A Class Divided. OSCE Istanbul Summit Declaration. Pagels. 1988. 1999a. SANDOLE ———.” International Herald Tribune.” Washington Post. 18–19 November ’99: Charter for European Security. Pierre. http://www. Cambridge: Cambridge University Press. But Much Hardship. Michael. 8 July.” Washington Post. ———. 2003.68 DENNIS J. ———. Hans J. “U. 4 August. 1988. Nicholson. London: Routledge and Kegan Paul. 18 February. “Why We Will Soon Miss the Cold War.html. Oldham. .” Washington Post. OSCE Istanbul. OSCE Istanbul Summit.” Frontline (PBS documentary produced by W. 1985. A10. 1996b. 18–19 November ’99. Pamela. A44. Foreword to Vamik Volkan’s The Need to Have Enemies and Allies: From Clinical Practice to International Relationships. 1999b.” Washington Post. “Croats and Muslims Edge Toward a Joint Army. Pfeiffer. Ernest. by Vamik Volkan. Pomfret. “A Flight of Mind: The Act of Dissociation Can Protect Children Emotionally from Trauma. 22 November. Joseph V. Robert E. 1989. 1996a. Istanbul: Organization for Security and Cooperation in Europe. A1. 2002a. Now Feeling Forgotten: Family of Man Killed After 9/11 Finds Little Charity. Istanbul: Organization for Security and Cooperation in Europe.

Chicago: Quadrangle Books. 1960a. T. New York: Holt. 5 May. Manchester. Michael Banks. Pittsburgh: Boxwood Press. 2002. New York: St. Ann Arbor: University of Michigan Press. 1998a. Sanders. We Can’t Think Straight. and Debates. 1968. Irene. and Intrusions.” Washington Post. 1990. Dennis J. New York: Farrar.” In Conflict in World Society: A New Perspective on International Relations. 1997. Martin’s. ———. England: Manchester University Press. 1939. Straus and Giroux. D. New York: St. B2. Monograph Supplements 23. Endgame: The Betrayal and Fall of Srebrenica. Anatol. Rosenau. Richardson. Sandole. NJ: Princeton University Press. Chicago: Quadrangle Books. “Generalized Foreign Politics: A Study in Group Psychology. Burton. John W.” In Conflict Management and Problem Solving: Interpersonal to International Applications. Statistics of Deadly Quarrels. London: Macmillan. “A Comprehensive Mapping of Conflict and Conflict Resolution: A Three Pillar Approach. “To Fight Terror. Inroads. 2 (December): 1–30. Sandole and Hugo van der Merwe. ed. ———. Dennis J. ed. David. Sandole and Ingrid Sandole-Staroste. ———. no. England: Wheatsheaf. ———. and Metaphors in Conflict and Conflict Resolution: Coherence or Confusion?” In Conflict Resolution Theory and Practice: Integration and Application. and Lenore Jacobson. Theories. Rosenthal. Martin’s. Brighton. “Conflict Management: Elements of Generic Theory and Practice. Princeton. 1984. Games. Lewis F. 1992. 1993. D. Robert. no. Europe’s Worst Massacre Since World War II. “The Biological Basis of Needs in World Society: The Ultimate Micro-Macro Nexus. ed. Rohde. 7 (June): 1–91. 1987.COMPLEXITY AND CONFLICT RESOLUTION 69 Rapoport. ———. . Rinehart and Winston. Post-Modernism and the Social Sciences: Insights. Pauline Marie. Martin’s Press. Arms and Insecurity.” Peace and Conflict Studies 5. New York: St. London: Frances Pinter. Pygmalion in the Classroom. New York: New York University Press. 1960b. Dennis J. ———.” British Journal of Psychology.” In Conflict: Human Needs Theory. 1960. “Paradigms. “The Subjectivity of Theories and Actions in World Society. Pittsburgh: Boxwood Press. Fights. D. ed.

” In Conflict: From Analysis to Action. Stepp. London: Pinter/Cassell (Continuum International). “Building Peace in Post-NATO Bosnia: A Recommended Action Plan. London: Continuum International. no. Causes. ———.. and Carolyn W.” In Terrorism: Concepts. 2003. Frederic Labarre. Groups in Harmony and Tension. Felberbauer. Kurt H. ———. “War and Chaos.” Ethnic Conflict Research Digest (University of Ulster) 2. Paul. Stern. “A Peace and Security System for Post-Cold War Europe: Preventing Future Yugoslavias. and Daniel Druckman. 1999a. Genetic Seeds of Warfare: Evolution. Trans. Conflict and the Web of Group-Affiliations. Experts Say. Desmond Dinan. Paul C. Paul. Predrag Jurekovic. Sherif. . Simmel. and Larissa Fast. SANDOLE ———. “Infants Now Murdered As Often As Teens: Actual Rate May Be Higher. Alvin M. Boulder. Sherif. 1967. and Patriotism. and Ernst M. 2002. New York: Free Press. Sites. D.S. 1998b. Georg. 1973. Boston: Unwin Hyman. ———. New York: Harper and Row. Scott Moore. Group Conflict and Cooperation: Their Social Psychology. ———. Muzafer. www. R. and Conflict Resolution.gv. 1955. ed.” In From Peace Making to Self Sustaining Peace—International Presence in South East Europe. Sandra Cheldelin. no.70 DENNIS J. CO: Lynne Rienner. 2002. 1953. “Has the ‘Earthquake’ of 1989 Toppled International Relations Theory?” Peace Psychology Review 1. Shaw. New York: Dunellen. Nationalism.” American Scientist 83. Defense Threat Reduction Agency. and Yuwa Wong. “The Causes of Terrorism. Fort Belvoir. no.php?id=160.bmlv. London: CollierMacmillan. Capturing the Complexity of Conflict: Dealing with Violent Ethnic Conflicts of the Post-Cold War Era. “A Typology. London: Routledge and Kegan Paul. ed. 1994/1995. A17. ———. 1989. 2004. Laura Sessions. 2 (Autumn/Winter): 109–22. Control: The Basis of Social Order. 1995. 10 December. Saperstein. R. ed. Wolff and Reinhard Bendix. ed.” Washington Post. 6 (November–December): 548–57. Vienna: National Defence Academy. Sherif. Muzafer. “A Design for Peace and Security in Post-Cold War Europe. VA: U. Daniel Druckman. 1 (February): 41–42.” In Encyclopedia of the European Union. 1999b.

Tajfel. Rosenau. Differentiation between Social Groups: Studies in the Social Psychology of Intergroup Relations. M. Cambridge: Cambridge University Press. ed. 1992. Complexity: The Emerging Science at the Edge of Order and Chaos. Uppsala. Uppsala University. “The Stability of a Bipolar World. Catastrophe Theory: Selected Papers. 1972–1977. by Roger Lewin. ed.” Special Report of the Dayton Implementation Working Group.” Daedalus 93 (Summer): 881–909. New York: Simon & Schuster. New York: Free Press. E. Catastrophe Theory. Zeeman. Mitchell Waldrop. The unabridged Crawley translation with an introduction by John H.” Editorial. ———. Edward O.S Institute of Peace. Reading. Vasquez. Koch. 1997. North and H. New York: Academic Press. Waldrop. 27. 1978. 1993. New York: Modern Library. by M. Mitchell. 2002. WP. September. Kenneth N. John A. E. Jr. Washington Post. William Graham. and the Outbreak of War. C. London: Collier-Macmillan. Henri. Washington. Curt. . Threat. New York: E. 1992. A. Sweden: Department of Peace and Conflict Research. 10. Finley. D. and Monte Davis. and Complexity: The Emerging Science at the Edge of Order and Chaos. Wilson. “The Order of Things. 1964. Thucydides. Report no. New York: Bantam Books. 1977. 1951. Peter. Zinnes. London: Cambridge University Press. The War Puzzle. “After the War. 51 (20 December): 1. 1978. On Human Nature. 1961. Washington Post Book World 22. Cambridge Studies in International Relations.” In International Politics and Foreign Policy: A Reader in Research and Theory. Folkways. A24. The Peloponnesian War. 12 November. James N. Wallensteen. Suplee. MA: Addison-Wesley. New York: Ginn. 1979. “Capability. C. 48. 1981. Jr. 1906. no. Woodcock. R. 1998. ed. P. Dutton. Waltz. USIP. “Dayton Implementation: The Train and Equip Program. DC: U.COMPLEXITY AND CONFLICT RESOLUTION 71 Sumner. Alexander..” Review of Complexity: Life at the Edge of Chaos. Human Groups and Social Categories: Studies in Social Psychology. Preventing Violent Conflicts: Past Record and Future Challenges.

This page intentionally left blank. .

a major concern of complex systems theories. These concepts do not contradict explanations centered on the success or failure of movement toward democratization (Snyder 2000). complex systems theories have much less utility for projecting alternative long-term futures or prescribing international strategy. But for several reasons discussed in chapter 10 and elsewhere in this volume. complexity science integrates concepts from many fields to produce a new slant on evolution. The analysis here suggests that ideas and concepts from complexity can enhance our ability to describe and explain the past and present. economic. Jr. Complexity starts with a wider lens than democratization but includes it. subsumes political. Still. 73 . and cultural strengths. The precise role played by each strength in shaping societal fitness becomes an important but secondary question. ideas and concepts from complexity can enlarge our vision and complement other approaches to social science. Clemens. If complexity fulfills this goal. The contributions of complexity to this understanding are evaluated in this paper. Generated by scholars from various disciplines. This chapter contends that basic complexity concepts do much to explain the movement toward or away from resolution of ethnic problems in newly independent states.CHAPTER 4 Understanding and Coping with Ethnic Conflict and Development Issues in Post-Soviet Eurasia Walter C. it should also help us to understand ethnic and other problems in post-Soviet Eurasia and other troubled regions. The concept of societal fitness.1 Its exponents seek a general theory able to explain many different types of phenomena—social as well as biological and physical. but rather enrich them and offer linkages to other fields of knowledge.

CAS defines fitness as the ability to cope with complexity. However. criticality.3 No organism evolves alone. and society coevolves with others and with their shared environment. To survive challenges and make the most of opportunity. Creative and constructive responses to complex challenges. the Internet. Fitness. Complexity theorists endeavor to explain the process of complex adaptation within complex systems—whether they be ecosystems. agent-based systems. As discussed in the introduction to this volume. All the hypotheses discussed here are pitched at the macrolevel: they focus on emergent properties of state and society. CLEMENS. fitness. Three of these—emergence. . nor in gases whose molecules move at random. ESSENTIALS OF COMPLEX ADAPTIVE SYSTEMS THEORY Nonlinearity and complexity are hallmarks of human social networks. A change in any one actor or environment can alter the environment of multiple actors and challenge their fitness. A full assessment of the past. however. Fitness is found in the middle ranges of this spectrum between rigid order and chaos—not in a crystal.74 WALTER C. Coevolution. where every atom resides in an ordered hierarchy. the harder it is to anticipate how change in one element will affect others (the “butterfly effect”). or political systems. let me summarize the essence of complexity and then apply it to explain divergent policy outcomes in the former Communist states of Eastern Europe and the USSR. present. species. a fit organism can process information about and deal with many variables. Having registered these caveats.2 The version of complexity used here—derived from the interpretation of complex adaptive systems (or CAS) developed by Stuart Kauffman and others at the Santa Fe Institute—is anchored in eight basic concepts. and you lose fitness. and self-organization—were described in chapter 1. are more likely to be found close to the edge of chaos than toward the other end of the spectrum. and punctuated equilibrium that are particularly relevant to the discussion in this chapter. and future of any social system would have to analyze the key individuals and groups who shape it. The theory posits that all life forms exist on a spectrum ranging from instability (chaos) to ultrastability (ordered hierarchy). The more variables shape a system. JR. This section considers in more depth the ideas related to coevolution. Move too far toward either pole. or on the international system as an emergent phenomenon of the interactions of states. this chapter does not specifically address the ultimate actor—individuals. complexity produces bottom-up theories and models. Every individual. often the decisive factors in tipping the balance of forces one way or the other.

may collapse in an avalanche. The concept of punctuated equilibrium underscores that evolution is often marked by surges of speciation and avalanches of extinction (Gould 2002). however. Thanks to mutation and self-organization. The fitness of the United States hovers close to the edge of chaos. they must adapt or disappear. Coevolution of units within a complex system can be mapped as a rugged landscape in which the relative fitness of each organism is shown as a peak rising or falling as a consequence of coevolution. DIFFERENCES ACROSS EURASIA: VARIATIONS THAT NEED EXPLANATION The huge area to which we shall try to apply CAS is Eastern Europe and the former USSR. we identify four large domains that took shape in Eurasia after the breakup of the USSR in 1991—each distinguished by the way it dealt with ethnic and development issues. When their environment changes. The sandpile metaphor. Because the fitness of countries is an emergent property. endure with little change for a long time.4 Species often develop quickly. Self-organized Criticality. As in an arms race. This means that their fitness is a function of the interaction of individuals within the social and political parameters of the society. Punctuated Equilibrium. the peaks of a predator and its prey may gain or decline according to changes in their offensive and defensive capabilities. In zone A was a . countries that are more decentralized are expected to be more adaptable and. is not universally accepted and is not essential to complexity theory. the fitness peak of the prey will drop. fitter. while that of Singapore teeters on the brink of rigidity. Fitness Landscapes.ETHNIC CONFLICT AND DEVELOPMENT 75 The behavior of social systems emerges from the interactions of their members. If individuals among the prey population acquire characteristics that reduce their vulnerability. therefore. Balanced between order and chaos. If attackers acquire more lethal weapons. if one more grain of sand is added. Adapting Snyder’s (2000) analysis. their peaks will rise. This fragile equilibrium is called self-organized criticality. a fit being is like a sandpile that. Scientists in many fields noticed in the 1990s that critical events occur more often—both earlier and later than forecast by the model of a bellshaped curve. How long the system is stable and endures is difficult to predict—especially in politics. and then die out suddenly—not gradually. it is not possible to predict with precision how countries will react to changes in the environment. Up to a point. members of the species find their niche and hang on to it.

Kyrgystan had a free press for a time. Hungary. In the former Soviet republics of Central Asia. and perhaps even Serbia might follow suit. a region virtually frozen in time—with little ethnic conflict and stagnant economic life (except in countries where the promise of carbon fuels brought injections of outside capital). Where to place the other states not clearly in one of these three zones? By the early twenty-first century Slovakia had clearly moved into zone A. most of former Yugoslavia.7 President Aleksandr Lukashenko tried to russify Belarus and negotiate its union with the Russian Federation. Each of these countries could readily drop into zone B or C. but ethnic differences were not at issue. and a late rising star. Zone A consists of societies and states that have experienced almost no ethnic violence and have made strong progress toward democratic institutions and economic development through market economics. in zone B. Zone B comprises societies that became embroiled in severe ethnic fighting in the 1990s—Chechnya. But the scales teetered. partial democratization probably aggravated ethnic tensions. Azerbaijan. Georgia. where dictators suppressed ethnic or other challenges to their rule. but Lukashenko repressed them with little overt violence. His opponents sought to establish and maintain a clear Belarusian identity.5 Of former Soviet republics. but could still become embroiled in . as Snyder says. we may distinguish a hybrid zone D where major countries—Russia and Ukraine—shared some but not all characteristics of the other regions. erstwhile Communist leaders became dictators claiming to be both nationalist and democratic. Latvia. Serbia made major strides toward real democracy and peace with Montenegro in 2001–2. In each case. but this ingredient of a true democracy disappeared in the mid-1990s. Each showed a very low capacity for coping with ethnic differences and the problems of establishing a viable economy and a stable democracy. Romania. only Estonia. and the erstwhile Soviet republics of Armenia. and Lithuania belong in zone A (Clemens 2001). because nationalist firebrands could mobilize votes against them. in zone C.76 WALTER C. “democracy” made it harder for Armenia’s leaders to negotiate any kind of compromise with Azerbaijan over Nagorno-Karabakh. CLEMENS. a shatterbelt of ethnic conflict and material regress. Thus. Finally. Slovakia. set of countries that benefited from ethnic calm and enjoyed gradual economic and political development. In the 1990s Tajikistan experienced much fighting between political rivals. Croatia. the exemplar is Slovenia. and Moldova. From the former Yugoslavia. Thus.6 Zone C refers to Central Asia and Belarus. From erstwhile Soviet allies in Eastern Europe. Montenegro. JR. There were signs that Bulgaria. the leaders are the Czech Republic. Poland.

by accession to or distance from the European Union. Like Russia. President Viktor A. however. many of the country’s old problems had reemerged. By year’s end. Moscow recognized Tatarstan’s “sovereignty” within the Russian Federation (Rossiskaia Federatsiia. By the early twenty-first century neither Russia nor Ukraine had achieved a real democracy or a strong market economy. As this chapter is designed to illustrate the uses of CAS for understanding ethnic conflict and development in post-Soviet Eurasia. except for Russia’s wars against Chechnya (1994–96 and again after 1999).ETHNIC CONFLICT AND DEVELOPMENT 77 more ethnic warfare with Kosovars or the Hungarian-speakers of Vojvodina. APPLYING COMPLEX ADAPTIVE SYSTEMS THEORY TO EXPLAIN PAST AND PRESENT FITNESS Adopting the language of CAS. for example. influenced. however. and Estonia in the 1990s demonstrated a high level of . the precise allocation of these countries to group B or group C is not crucial. But neither suffered from outright ethnic violence. Neither had suffered much ethnic violence in the previous fifteen years. By 2005 Bulgaria and Romania were becoming difficult to classify. but neither came close to qualifying for membership in the EU. this chapter argues that countries such as Slovenia. Transparency International placed Russia and Ukraine among the world’s most corrupt countries in the late twentieth and early twenty-first centuries. the Czech Republic. Kyiv avoided war with Russian irredentists in the Crimea and with Moscow over its claims to ships and naval facilities in Sevastopol. on the hope that they would contribute to George W. where rossiskaia is more inclusive than the term russkaia. Russia’s ethnic nationalism was qualified by civic nationalism. There were signs early in the century that the Russian Federation’s Duma and President Putin might require that any would-be Russian citizen be fluent in Russian. Their location on the A-to-D spectrum may well change. but each had a low HDI ranking compared to. Slovenia or Slovakia. Thus. Ukraine’s “orange revolution” in 2005 promised fundamental changes. albeit with new faces. The two largest Slavic states emerging from the USSR comprised the hybrid zone D. say. Yushchenko’s administration followed a Western orientation even as it labored to overcome the misgivings of diffident Russian-speakers in Ukraine. Each had been admitted to NATO. Ukraine failed to use effectively its vast natural resources and highly educated work force (D’Anieri 1999). as “British” takes in more diversity than “English”). Ukraine achieved a kind of civic nationalism incorporating native Russian and Ukrainian speakers. Bush’s War on Terrorism.

n 175 Freedom Index 2003 Honesty Rank 2003. n 133 GDI Rank 2003.a.1.TABLE 4.a. THE HIGHEST-RANKING COUNTRIES IN “HUMAN DEVELOPMENT” AND OTHER VALUES FROM VARIOUS CIVILIZATIONS Country 3 2 6 5 11 22 25 25 28 29 30 31 34 38 n.a. 92 33 43 53 37 HDI Rank 2003. Republic of Brunei Darussalam Argentina Estonia Cuba Belarus Malaysia 1 6 7 8 9 22 24 25 28 29 30 31 34 41 52 53 58 . n 175 Economic Freedom Rank 2004. 68 (MF) 6 (F) 155 (RE) 151 (RE) 72 (MU) 8 17 18 11 21 21 50 27 5 29 50 n. 48 53 Free Free Free Free Free Free Free Free Partially Free Free Free Not free Partly free Free Not free Not free Partly free 28 (MF) 19 (MF) 6 (F) 18 (MF) 35 (MF) 33 (MF) 56 (MF) 22(F) 2(F) 62 (MF) 52 (MF) n. n 153 Cultural Tradition Protestant Catholic Protestant Catholic Japanese Israeli Orthodox Orthodox Mixed Catholic Asian Muslim Catholic Protestant Catholic Orthodox Muslim (continued) Norway Belgium United States Canada Japan Israel Greece Cyprus Singapore Slovenia Korea.

TABLE (continued) Country 61 82 83 91 103 107 mostly free. MF Sources: “HDI” and “GDI” are from U. F free. Development . n 175 Freedom Index 2003 Free Free Not free Partly free Free Not free 40 (MF) 89 (MU) 127 (MU) 99 (MU) 119 (MU) 72 (MU) 70 n. Human Development Report 2003 (New York: Oxford University Press. 43 n. MU mostly unfree. RE n 153 Honesty Rank 2003.heritage.1.a. “Honesty Rank” is from Transparency International at www. n 133 Cultural Tradition Buddhist African Chinese Muslim-Hindu Hindu-Muslim African Thailand Cape Verde China Indonesia India Swaziland 74 103 104 112 127 133 Code: For economic freedom.transparency. HDI Rank 2003. 83 n. “Freedom Index” is from Freedom House at www. 2003). tables 1 and 22. GDI Rank 2003.N. “Economic Freedom” is from Heritage Foundation at www.a. n 175 Economic Freedom Rank 2004.

Children and young adults. Ethnic minorities were very small in Belarus. and D experienced great difficulty in dealing with ethnic minorities. Countries in zones B. was never an independent state before 1992. Thus. by contrast. few countries showed much interest in or had much prospect of joining NATO or the EU in the foreseeable future. Most countries in zones B. language.1. Estonia and Latvia had only two decades of independence between the two world wars. C. Armenia. Success in one domain helped them cope with problems in others. Economic advances in Estonia. Parts of Ukraine and Russia have better soil as well as much richer mineral deposits than any country in zone A. CLEMENS. make it easier for Tallinn to provide welfare benefits for Russian-speakers residing in Estonia but who were not citizens. Still. Azerbaijan. As we see in table 4. and Belarus. by contrast. Indeed. Moldova. the city councils in Riga as well as Tallinn were sometimes dominated by coalitions of old leftists and “unity” parties devoted to the interests of Russian-speakers. consolidate democracy. Estonian and Latvian leaders espoused a kind of ethnic nationalism tempered by civic moderation. Still. fitness. and D faced simpler ethnic challenges than in many zone-A countries. because they were more homogeneous. the South Caucasus. . and civic tests. Romania. and nourish creativity. Societies in zone A achieved high levels of fitness on many fronts after the demise of the Soviet empire. and Russia possess energy resources far superior to those in any lands in zone A. By contrast. Estonia even permitted noncitizens to vote in local elections. Ethnic peace made it easier to raise living standards. and D displayed low levels of overall fitness even though many possessed assets lacking in zone A. and in most of Central Asia (except for Kazakstan). of course.80 WALTER C. Each country in zone A joined both NATO and the European Union. JR. learned Estonian or Latvian more readily than most of their elders. C. They instituted a naturalization process that required aspiring citizens to pass residency. and Russia have evolved from states and cultures dating back more than a thousand years. for example. they scored much higher on the UN Human Development Index and in Freedom House ratings for political and civil liberty than did comparable peers such as Serbia. By the early twenty-first century—more than a decade since independence—few of either country’s Slavic speakers had acquired a working knowledge of the official state language. Ukraine. Slovenia. In zones C and D. About four-fifths of the Russian Federation’s population was Russian but most other groups in the federation spoke Russian. Kazakstan. Georgia. the governments in zones B. Estonia and Latvia in the 1990s faced minorities of Slavic speakers that made up more than one-third of the resident population. ethnic tensions produced no deaths in the Baltic. C. A million or so Chechens occupied only a dot on the federation’s periphery.

President V. from the bottom up. As in Soviet times. So great was the plunder that by 2004 there were more billionaires in Moscow than in New York. and D attempted to direct economic and cultural life as well as politics from the top down. produces innovation and ways to meet needs and exploit opportunities. they repressed newspapers and news media that contradicted the official line. each state focused on joining Western Europe and NATO—not on cooperating for shared ends with its immediate neighbors. Instead. but this is a very limited facet of coevolution. Kazakstan “coevolves” with foreign oil drillers. its overall fitness will suffer. Emergence Nowhere in the formerly Communist lands did there emerge strong patterns of cooperation. Most countries close to Western Europe have coevolved with the West more quickly and thoroughly than those that are more distant. Central Asian states proved unable. It entails also a market economy and a social system that. Privatization in Russia and most other countries in zones B. but its Communist rulers sought autarky. it was more like “every state for itself”—indeed. . Albania abuts Greece.” Even in zone A. Coevolution This concept explains several features of post-Soviet Eurasia. “every national and subnational group for itself. and D permitted privileged insiders to seize public resources at low cost. C. the Czech Republic is more “First World” than is Kyrgyzstan. The Commonwealth of Independent States had many accords registered on paper but never executed. C. even to find ways to stop the shrinkage of the Aral Sea—an environmental disaster that affects the whole region. after as well as before independence. Thus. Putin was designated acting president by his predecessor before a snap election that confirmed the appointment—bolstered by a then popular war against ethnic aliens. Thus. Rivalries persisted in the Caucasus even though both Georgia and Armenia needed the energy that Azerbaijan could provide and for which it needed buyers. Subgroups meant to either to resist or to strengthen the commonwealth also achieved little.ETHNIC CONFLICT AND DEVELOPMENT 81 Self-Organization Self-organization takes in more than democratic politics. But if a country shuts itself off or is otherwise isolated from global trends. V. but the government’s orientation toward Moscow serves to minimize productive exchanges with the West. The centralized regimes in zones B. Belarus abuts Poland and Lithuania. In many respects Kazakstan and other Central Asia states resemble Communist Albania—cut off from the West by government fiat.

Punctuated Equilibrium The concept of punctuated equilibrium warns us not to expect steady progress in fitness. Long plateaus without improvement may drive some people to depart or to take drastic steps to effect change. Meaningful social change often requires a period of preparation.” Self-Organized Criticality CAS warns that societies may be less fit than they appear. Still. as if guided by an “invisible hand. it was EU and NATO demands for settled borders and ethnic peace that persuaded Hungary and Romania to patch up their differences and convinced Estonia and Latvia to renounce some border regions seized by Moscow in the 1940s. Fitness depends on the harmony of many factors.” which Kauffman’s version of CAS attributes to established ecosystems (such as coral reefs). a new or heavier burden could seriously weaken an apparently fit society. But regress . JR. Accumulating experiences may tip even middle-aged Russian-speakers toward integration with native Balts. New generations can be educated. West European unification did not emerge gradually but in sharp jumps and with some steps backward. This is not quite “order for free. CLEMENS. In Estonia and Latvia many young persons who speak Russian at home are learning the official state language. it resembles the positive results that Adam Smith expected if individuals were allowed to do what they do best. but one cannot be sure what grain of sand—what policy innovation or social change—may start an avalanche that radically changes a society and its fitness. Whatever the shortfalls of the European Union. How would Lithuanians respond if a faulty nuclear reactor shut down their energy supply or spread poison to the air and soil? Or if Russians for a prolonged time simply turned off the oil and gas flows on which Lithuania (and many post-Soviet societies) depends? Each Baltic country endured severe stresses in the 1990s. it is a triumph of cooperation compared to the beggar-thy-neighbor behaviors of ex-Communist societies. Just as an extra grain of sand may cause a sandpile to collapse. Indeed. Agent-Based Systems In zone A individual agents are free to innovate and carry on their business with a minimum of government control. The system is shaped by its members rather than by a central command.82 WALTER C.

This insight helps explain why Central Asia is frozen in time. The fundamental insight of CAS is its prediction that fitness will be found along the middle of the bell curve ranging from rigid order to random instability. Though it is difficult to show all these variables in a single peak. How long will displaced persons in Bosnia and other parts of the former Yugoslavia wait until they return to their homes? Fitness Landscapes The relative fitness of a fruit fly and a frog population may be portrayed as “peaks” that rise and fall with coevolution. if CAS is correct about the role of self-organization in fitness. lack of political and civil liberties. Can we graph changing patterns of fitness among the societies of post-Soviet Eurasia? This is not a simple task. it does no worse than most competing theories—few of which provide useful handles for predicting or shaping the future. We expect that low fitness in this domain will tend to correlate with low scores in overall human development. and why Slovenia and Estonia adapt well to their new freedoms. though high creativity is most frequently found close to the edge of chaos.8 If we focus on ethnic problems.9 PREDICTING ETHNIC VIOLENCE AND PRESCRIBING REMEDIES Theories of complex adaptive systems provide useful concepts for analyzing ethnic issues and other ingredients of societal fitness. Indeed. if only because fitness among humans is multidimensional. In this regard.ETHNIC CONFLICT AND DEVELOPMENT 83 is also possible. dislocations. and corruption. education. and material living standards. social Darwinists and ultrarealists are wrong: success in politics does not derive from raw power plus cunning. cultivate selforganization—not a system steered from on high. why the Caucasus explodes. low technological achievement. But they offer only general principles for anticipating future outcomes or prescribing constructive policies. This insight has clear policy implications: avoid the extremes of dictatorship and anarchy. Western policymakers and investors should not count on authoritarian regimes in Kazakstan or Azerbaijan to . The UN Human Development Index provides a solid starting point to measure public health. and deaths caused by ethnic unrest. To generate a healthy and innovative society. and why Russia resorts to an iron fist to overcome chaos. we would also study measures of ethnic harmony and its opposite—injury. however. a cobweb graphic could illustrate the correlations suggested here.

the healthier and fitter they will be. What if democracy terminates democracy—as happened in Germany in the 1930s? Is self-organization desirable if the majority votes against the minority. and open debate (relative to most other societies) long ago. groupthink. however. Even if the goal of self-organization seems clear. questions arise about the road to this goal. and a welfare mentality that discourage initiatives from the bottom up. maintain order forever. After the Peasants’ Revolt. For the first time in history. and D moved toward universal literacy only in the past 100 years. The stronger and more diverse the independent agents shaping the formerly Communist societies. Outsiders cannot compel internal reforms but should do what they can to nudge these societies toward greater selforganization. as in parts of Nigeria? HOW TO ACQUIRE AND NURTURE FITNESS Culture matters. and NGOs that enlarge public goods and are not dominated by government. free thought. This twin revolution helped to liberate all who experienced it (Clemens 2005). He then wrote his Short Catechism instructing people what to believe. businesses. Their attitudes as well as formal structures will determine whether Azerbaijan and Kazakstan use their petrodollars to create values for the entire community (as in Norway) or follow more closely the Saudi Arabian or Nigerian models. JR. CLEMENS. These independent agents face a difficult struggle against the moral legacies of Communism—corruption. some princes and religious leaders also urged individuals—female as well as male—to read and interpret sacred texts on their own.10 All the societies in zone A became oriented toward universal literacy. Constructive policies will cultivate creative individuals. Martin Luther. as happened in Sri Lanka and as Serbs feared would happen in a majoritarian Bosnia? And what if the majority brings in a government that imposes the laws and mores of one religion. They should not prop up local dynasties in the hope of securing privileged access to oil and gas. Jan Hus. each society in zone A acquired its sacred religious texts in the vernacular between the fifteenth and seventeenth centuries. Countries such as Azerbaijan suffer not only from top-down controls but also from a rent-seeking mentality among many well-educated persons who will eventually play major roles in business and politics. and other reformers.84 WALTER C. C. Following the leads of John Wycliffe. CAS attention to independent actors agrees with the growing conviction among political scientists that formal and informal institutions of civil society help to buffer the ravages of free markets and curb the excesses of willful governments. Luther feared that he was provoking chaos. But Luther could not stop the transformation he had .11 Many regimes in these zones still discourage or try to prevent open debate on policy and other important issues. The societies in zones B.

Catholic France and Italy had Bibles in the vernacular even before Luther’s challenge to Rome. Lepore 2002).ETHNIC CONFLICT AND DEVELOPMENT 85 Figure 4.1. Bibles in the vernacular also helped cultivate a sense of national identity (Hastings 1997.1 shows a strong correlation between high HDI scores and early publication of Bibles in the vernacular. Bibles in the vernacular were not widely published until the late nineteenth century or the twentieth century. the discovery of the New World. which published both the New and Old Testaments in the seventeenth century. . but figure 4. Date the Bible Published in Vernacular Correlated with HDI Rank unleashed. Where Orthodox Christianity prevailed. In the seventeenth century Sweden’s monarchy and state church wanted their subjects—even servant girls—to read and discuss the Bible. the Renaissance. and growing refinement of scientific methods. The synergies of literacy and individualist thinking were empowered by the printing press. Certainly many factors shape human development. (The sole exception was Romania.)12 Wide-scale literacy came to the Orthodox countries much later than in Protestant and Catholic countries or in Jewish communities.

and pushed through reforms permitting the Kremlin to appoint regional governors instead of having them directly elected by their subjects. CLEMENS.86 WALTER C. because of state secrecy and communications networks that ran vertically but not horizontally. and D still do not encourage free thought and debate. In the mid-1990s Georgians welcomed Eduard Shevardnadze back from Moscow to Tbilisi. Unlike the Christian Bible. But Communist regimes and schools discouraged freethinking. Few Bosnians. looking for ways to propel their peak(s) upward. Rather. Mikheil Saakashvili. Centralized controls channeled thought and discouraged debate. like a legendary vozhd. as a rule. they will lag their more westernized neighbors in many ways. it added to the already heavy burdens of corruption at the center. Comparatively unfit. Elections held in 2000 and 2004 suggested that most Russians still hoped that a vigorous leader. CONCLUSION In the early twenty-first century most governments in zones B. Vladimir Putin. Lacking self-organized economies and polities. Islamic societies did not encourage mass literacy or. jail the country’s richest man when he sought to shape political life. Until they do. Shevardnadze was ousted in 2003 in a popular revolt led by an American-trained lawyer. they will tend to repress dissent rather than create solutions for mutual gain. and Chinese were for many years largely in the form of paraphrase and commentary. or Central Asians have been able to read classical Arabic. Azeris. Putin proceeded to silence independent media. Democratic in form but authoritarian in substance.13 Translations of the Qur’an into Persian. Having won many votes by intensifying the war against Chechens. The Soviet dissident Andrei Sakharov lost his security clearance and was sent into internal exile. But reliance on topdown leadership did not end turmoil in Georgia. this proved difficult.14 By the 1950s Communism had brought near universal literacy to the USSR and Eastern Europe—even to Albania. counting on him to end a reign of chaos. who . JR. the language in which the Qur’an was first written is regarded by Muslims as sacred—the only truly accurate way to express God’s message. individual interpretation of sacred texts. For Arabs as well as non-Arab Muslims. they will have great difficulty dealing with ethnic issues within and across borders. Even when Communist regimes sought to foster technological innovation. Turkish. they will not possess a necessary ingredient of social fitness. In the language of CAS. many other dissidents suffered worse fates. memorization and recitation of the Qur’an have been far more important than discussion. would unite and mobilize the people for a better life. these countries— even erstwhile superpower Russia—will wander in the valleys of a fitness landscape. C.

Would closer ties with America improve fitness in former Soviet republics? For countries such as Georgia and Uzbekistan.ETHNIC CONFLICT AND DEVELOPMENT 87 promised to replace corruption and chaos with a rule of law. hyperpower might bring material gains but could also weaken self-organized fitness. each variable would have to be studied in depth. Thus. To really understand these relationships. CAS suggests each actor’s most basic need is a capacity to cope with challenges at home and abroad. and that societal fitness in a large. Where Can We Go from Here? Complexity cannot generate precise algorithms for analyzing ethnic conflict. But it does provide valuable conceptual tools for this task—principles. The graphics in this essay suggest a correlation between HDI rank and the date when the Bible was published in the vernacular of the countries that later became units of the Soviet bloc and Yugoslavia. Saakashvili had learned the rhetoric of democracy. By 2004. 57). he was promoting his own cult of personality. but—in a society that wants a strong. Lacking internal strength. models. How could we operationalize these concepts? Let us assume that HDI rank is an approximate indication of societal fitness. How much literacy was there before the printing press and Martin Luther? How did it evolve in the decades and centuries after Luther? What forces and institutions resisted or facilitated the growth of literacy? 2. Unlike neorealists who believe that relative material power—missiles and GDP—is the best guide to world politics. charismatic leader— he gravitated toward the national norm. modern society depends on universal literacy and free expression. however. When? By whom? Why? How many copies? Were they repressed (as in Russia in the 1820s)? Did they sell? When did subsequent printings take place? Trace the evolution of literacy in these countries. closer ties with the U. however. Here are a few of the tasks: 1. Identify the conditions in which the Bible was rendered in the vernacular and published in each country. . Let us assume also that the onset of universal literacy and free thought can be traced to the dates when the most sacred books of that society were published and when conditions were established in which they could be subjected to individual interpretation. a major insight of CAS is the concept of societal fitness. each people’s capacity to cope with ethnic diversity might then decline—especially if exploited by political entrepreneurs hoping to gain power and wealth from others’ differences (Singer 1999. Develop a common standard for measuring literacy.S. including ethnic diversity. metaphors.

against others. violent nationalism. CLEMENS. conditions favoring individual freedom were stronger in some regions (such as Bohemia) than in Byzantium or Russia. What indicators— in science. 7. Consider the shortcomings of the twin revolution. Ask which came first: individual freedom or the twin revolution? Long before Luther. 12.N. such as literacy. Distinguish technology from the culture where it is applied. Why did most Europeans respond with alacrity to the printing press while Islamic cultures did not? Learn from outliers: Romania (Orthodox but closer to Rome than to Russia) had the Bible relatively early but nonetheless has a low HDI ranking. politics. and economics—show independent thinking? Such indicators are more evident in a relatively open metropole such as England than in a repressed dependency such as the places we now call Estonia or Slovakia. Rich data on individual countries is available in the human development reports produced by local social scientists in many East European and former Soviet states.88 WALTER C. How should we weigh the contribution of one factor. . 3. 6. Development Programme. 11. 5. they are accessible online from the U. 8. such as growing wealth? Analyze the chicken-and-egg. Consider the policy implications: If high levels of human development may be traced to a twin revolution that began five hundred years ago. among Bohemia’s Hussites) and that of the last century or two. and war? Perhaps the twin revolutions were necessary but not sufficient for overall fitness. how can they be fostered in societies that have experienced one or both revolutions only in recent decades? Can the benefits of the twin revolution be nullified by manipulation of mass media and government power by authorities seeking to create their version of a Brave New World? 4. Distinguish the kinds of ethnic/national consciousness that existed in previous centuries (for example. Adapt the approach used to study traditionally Christian countries to those in the Muslim and other religious traditions. why has the West shown so much intolerance. the arts. Slovaks got the Bible relatively late but achieved a fairly high HDI score in the late 1990s. 9. If literacy and independent thinking conduced to human development. 10. Determine a way to prove causation rather than mere correlation. JR. Trace the growth of independent thinking.

For a skeptical view of complexity theory.” and “species. (1995). “Arborscapes: A Swarm-based Multi-agent Ecological Disturbance Model” (98-06-056). Robert Jervis (1997) examines the complex interactions of social units. and Simplicity Part I: A Tutorial” (98-04-027). 4. Complexity. international relations theory. For a more balanced appraisal.ETHNIC CONFLICT AND DEVELOPMENT 89 The argument here is that to understand present trends and develop constructive ways to deal with ethnic diversity. NOTES 1. The following interpretation of complexity theory is based largely on the work of Stuart A. 1997b) and Axelrod and Michael D. Gould may have confused “individual.” “class. For applications to management. For related work by IR specialists. Richards says that nonlinear modeling can build directly on existing economic theory. Hoffmann (1999). we must review not just decades but centuries of history—cultural. In the same vein. Wilson (1998). and Edward O. see also Charles J. On coevolution. also. Compare with James N. Nowak et al. 2. political. The utility of complexity theory is assessed by Hayward R. Cohen (1999) have used a variety of methods to resolve complex problems. see Lars-Erik Cederman (1997). The Santa Fe Institute publishes the journal Complexity and working papers such as Martin Shubik. Lumsden and Edward O. Robert M. see papers given by Michael Lipson (1996) and Matthew J. 3. Wilson (1981). 11. and Melisa Savage and Manor Askenazi. Alker and Simon Fraser (1996) and continued in Alker (1996). economic. however. In political science. The tasks in such work are complex and vast. see “Edge of Chaos” and many relevant entries in Ian Marshall and Danah Zohar (1997). Rosenau (1990). 1995. see Hendrik Spruyt (1994). For a book that blends historical analysis. alliance formation. For early work at the Santa Fe Institute. Axelrod (1997a. see John Horgan (1996). Kauffman (1993. 2000) and other scholars—from the Nobel physics laureate Murray Gell-Mann to the Nobel economics laureate Kenneth Arrow—who have interacted at the Santa Fe Institute. Many aspects of nonlinearity are examined in Diana Richards (2000). and outbreaks of war. epochs in political economy. but can be made more manageable by using the conceptual tools developed by CAS for studying societal fitness. where nonlinear models are applied to federalism. but says little about self-organization. Martin A. For an application of complexity theory by a former student of Axelrod. see Roger Lewin (1992). . and systems analysis. see Roger Lewin and Birute Regine (2000). nonlinear modeling must invent a method unique to the problem at hand—from dynamical systems to spatial voting models to timeseries analysis.” See Mark Ridley 2002. environmental regimes. But “punctuation” may result from an incomplete fossil record. “Game Theory.

(1998). Department of State—all accessible on the Internet. Huntington (2000). For a model. 10. 9. 8. Snyder concludes that the ethnic content of the Moldovan conflict was ambiguous. and the U. Harrison and Samuel P. and Aili Piano (2001) and the country reports published regularly by Human Rights Watch. For a range of viewpoints. In the 1990s Mongolia moved quickly toward democracy. 2 percent Russian.S. For discrepancies between the UNDP Human Development Report published annually in New York and country reports published by UNDP offices in Baltic and East European The Moldovan government in the early 1990s was nationalistic. even though it was poorer than most parts of the USSR and had a weak infrastructure for education and communication. Literacy rates are difficult to track and measure. see Clemens 2001. each spoken now by only a hundred or so persons. are given in www. Transparency International. 24. JR. but Russian-speakers in the breakaway “Transdniestr Republic” were driven more by nostalgia for the Soviet empire than by nationalism (Snyder 2000. But Bosnians could have read a translation into Serbo-Croat in 1875 (Swartz 2004). Dates of Bible publication in many languages. 4 percent Kazak. 13. but estimates for many formerly Communist countries are at Snyder 2000. see Maruca 2000. including those native peoples of Siberia and North America. Abutting the former USSR. 250–51). for him a completely unknown tongue. 200–202.90 WALTER C. 2 percent other) and did not clash with China despite the potential for expansionist claims by each side. 7. Citing several studies. 5. the United Nations Development Programme. Moscow’s one-time client state Mongolia is a special case. CLEMENS. 6. Some translations into Malay have been so literal that they were not intelligible without prior knowledge of Arabic. 2 percent Chinese. This survey is so detailed that it notes the very different years for Bible publications in Tartu Estonian (no longer spoken) and Tallinn Estonian and for Eastern and Western Livonian. Mariano Grondona (1996). Alexander Motyl. which also reproduces the opening lines of St. An Azeri in Moscow showed me his family Qur’an written in Arabic. . The country had few internal ethnic problems (90 percent of the population is Mongolian. scriptures. John’s Gospel in each language. 11. Amnesty International. see the essays in Lawrence E. 110–11. See the studies done for Freedom House by Adrian Karatnycky. 12. and Dominique Jacquin-Berdal et al.

Reformist.” Paper delivered at the American Political Science Association Annual Meeting. Clemens. Paths to Peace: Is Democracy the Answer? Cambridge. Miriam Fendius. and Socialism. April 23–24. 2004. and Robert Axtell. Lars-Erik. 1997. ———. Debating the Democratic Peace: An International Security Reader. Axelrod. Albany: State University of New York Press. Joshua M. 1996. MD: Rowman & Littlefield. MD: Rowman & Littlefield. Hayward R.. Doyle. NJ: Princeton University Press. Jr.. Cambridge: Cambridge University Press. Epstein. Elman. Axelrod. and Simon Fraser. and Human Development. or Secessionist Politics: Explaining Minority Behavior in Multinational States. 1999.ETHNIC CONFLICT AND DEVELOPMENT 91 REFERENCES Alker. Lanham.. 1996.” Work-in-progress at Boston University and Harvard University. Lanham. “On Historical Complexity: ‘Naturalistic’ Modeling Approaches from the Santa Fe Institute. Valerie. Ways of War and Peace: Realism. 1997b. Robert M. Robert M. Secession and Inter-Ethnic Cooperation and Conflict. 1997. ———. Michael E. Economic Interdependence in Ukrainian-Russian Relations. Bunce. ed. Hayward R. The Baltic Transformed: Complexity Theory and European Security. 2nd ed. Norton. The Complexity of Cooperation: Agent-Based Models of Competition and Cooperation. and Michael D. W. Dynamics of International Relations: Conflict and Mutual Gain in an Era of Global Interdependence. Walter C. Princeton. Paul J. New York: Free Press. Individual Freedom. Cederman. Growing Artificial Societies: Social Science from the Bottom Up. Emergent Actors in World Politics: How States and Nations Develop and Dissolve. D’Anieri. .” Paper prepared for the Workshop on Nationalism. ed.. 1997a. Liberalism. Rediscoveries and Reformulations: Humanistic Methodologies for International Studies. MA: MIT Press. Michael W. 1999. New York: W. 1996. Princeton. Cohen. MA: MIT Press. “Advancing the Art of Simulation in the Social Sciences. ———.” Paper delivered at the International Conference on Computer Simulation and the Social Sciences. Cambridge. Brown. Cornell University. Cortona. MA: MIT Press. Harnessing Complexity: Organizational Implications of a Scientific Frontier. Alker. San Francisco. 1997. 2001. NJ: Princeton University Press. 2004. Italy.. 1996. 2005 “The Culture of Democracy: Literacy. Cambridge. “Status Quo.

” Paper presented at the Annual Meeting. 1997. Stephen Jay. Robert. Washington. 1981. Lepore. A. and Birute Regine. The Structure of Evolutionary Theory.” Paper presented at the Annual Meeting. Lipson. 2002. New York: Basic Books. Grodona. Horgan. Armonk. 1999. Lewin. and International Relations: Or. Lumsden. New York: Oxford University Press. 1998. New York: A. 1992. Stuart A. New York: Oxford University Press. CLEMENS. eds. Changing the Rules by Playing the Game. Culture Matters: How Values Shape Human Progress. eds. et al. Martin’s.92 WALTER C. Sharpe. 1996. “Nonlinearity. “Constructivism and Complexity Science: Theoretical Links and Empirical Justification. and Barrett L. Roger. and Markets in East Central Europe and the Newly Independent States. Chicago. eds. 2002. and Culture: The Coevolutionary Process. New York: Cambridge University Press. Religion. Roger. Edward. MA: Harvard University Press. Hoffmann. 2001. Alexander Motyl. and Aili Piano. New York: Oxford University Press. New York: St. and Samuel P. Complexity: Life at the Edge of Chaos. Huntington. The Soul at Work: Listen. 2000. MA: Addison-Wesley. 1996. 1996. Gould. 1993. Democracy. The Construction of Nationhood: Ethnicity. Genes.. . ———. E. Dominique. The Origins of Order: Self-Organization and Selection in Evolution.. Mind. System Effects: Complexity in Political and Social Life. Nations in Transit. Cambridge. 2000. Jill. NJ: Princeton University Press. Jacquin-Berdal. Lewin. What If China Doesn’t Democratize? Implications for War and Peace. Karatnycky. Adrian. 2000. Friedman. Matthew J. At Home in the Universe: The Search for the Laws of Self-Organization and Complexity. New York: Simon & Schuster. Cambridge. International Studies Association. Jervis. NY: M. 1995. Culture in World Politics. Wilson. ———. Adrian. Let Go: Embracing Complexity Science for Business Success. 2000. The End of Science: Facing the Limits of Knowledge in the Twilight of the Scientific Age. Hastings. JR. A Is for American: Letters and Other Characters in the Newly United States. Constructivism. Reading. Knopf. 1997. Michael. Charles J. Kauffman. Buenos Aires: Planeta. El mundo en clave. New York: Freedom House. MA: Harvard University Press. 1999–2000: Civil Society.. DC. Harrison. and Edward O. eds. American Political Science Association. Princeton. Lawrence E. and Nationalism. McCormick. Investigations. John. Mariano. New York: Macmillan. Respond.

org. Explanation. 1994. New York: Morrow.ETHNIC CONFLICT AND DEVELOPMENT 93 Marshall. New Haven. “Prediction. 11. Rosenau. 1998. 2003. 2:47–59. “The Evolution Revolution. The Sovereign State and Its Competitors: An Analysis of Systems Change. 1998.worldscriptures. Knopf. Waert. Wilson. NJ: Princeton University Press.” American Political Science Review 90:512–33. United Nations Development Programme. 1999. and David R. Maruca. Spencer R. 2000. 1997. and Danah Zohar. http://www. Ian. 1918–1988. Swartz.” International Journal of Peace Studies 4. Christopher Gelpi. Jack. Turbulence in World Politics: A Theory of Change and Continuity. David L. Princeton.” Harvard Business Review 78 (July–August): 24. 2000. Russett. 1995. Who’s Afraid of Schrödinger’s Cat: All the Science Ideas You Need to Keep Up with the New Thinking. James N. Human Development Report. and Dan Reiter. “The Arithmetics of Mutual Help. 2000. W. Ridley. Mark. New York: A. . Hendrik Spruyt. no. CT: Yale University Press. NJ: Princeton University Press. “Assessing the Dyadic Nature of the Democratic Peace. Merlin. A. 2002. 1950–1985.” Work-in-progress at Boston University. Political Complexity: Nonlinear Models of Politics. Regina Fazio. Richards. 1998.” Scientific American 272:76–81. and the Soviet Exit from the Cold War. 2004. Davis. From Voting to Violence: Democratization and Nationalist Conflict. Never at War: Why Democracies Will Not Fight One Another. Singer. Snyder. Rousseau. Bruce. Consilience: The Unity of Knowledge. Princeton. Edward O.. Spruyt. 1990. New York: Oxford University Press. “The Third Leg of the Kantian Tripod for Peace: International Organizations and Militarized Disputes. Nowak. New York: W. “Competitive Fitness. Martin A. David. J. et al.” New York Times Book Review. Diana. Norton. John Oneal. March 17.. ed.” International Organization 52:441–67. “Translations of the Quran. Ann Arbor: University of Michigan Press. 1996.

This page intentionally left blank. .

and Snidal 2001). neoliberal instantiation. and Rittberger. This is not a wholesale indictment of regime theory. Meyer. Krasner 1983. The goal of regime theory is simple: to explain when rules. Hoffmann This chapter undertakes a deceivingly simple endeavor—to show how insights from complexity concepts (drawn from the complex systems taxonomy described in chapters 1 and 2) can aid our understanding of environmental regimes. and Rittberger. regime theory is mostly concerned with bargaining. 1997). I illustrate how complexity concepts can help us to move beyond regime theory with a brief examination of the ozone depletion regime. Lipson.CHAPTER 5 Beyond Regime Theory Complex Adaptation and the Ozone Depletion Regime Matthew J. Koremenos. In its traditional. 95 . Meyer. principles and norms (Krasner 1983) are likely to occur in a particular issue-area. and a good deal of time and expertise has been spent researching the ways in which such bargaining has and can take place across a number of issues (Keohane 1984. 1997. I focus on one aspect of a complexity approach that is missed by traditional (especially neoliberal) regime theory—namely. Hasenclever. it is not possible to indict the whole enterprise (Hasenclever. how actors coevolve with their political context and how adaptive actors come to understand both the environmental problems that they face and the potential solutions to those problems. Rather. How can states (for the most part) come up with a set of rules/institutions to help them achieve/exploit common interests in the absence of a hegemonic authority? This is the driving question for much of the regime theory enterprise. procedures. As there is no single regime theory.

states know what they want and can relatively easily perceive the problems they face and the rules of the game. Second. 1999) provides a regime life cycle. you have a dependent variable— regime emergence—and a number of independent variables: problem type. A complexity approach considers that regime theory as traditionally conceived freezes too much of the dynamism in environmental regimes. norms. Others focus on social practice models (Young 2002). Arthur 1994b). A complexity approach has a great deal to add to the study of regimes and to move us beyond (neoliberal) regime theory. and how they fit together is not self-evident. Rational choice is often the crucial behavioral assumption for explaining how the independent variables come to determine whether or not a regime is likely. with the laudable exception of Oran Young (1997. taken as a diverse whole. there is the challenge of mechanisms. 1997. once formed. In this chapter. there is the challenge of complexity. However. 1999). and system . extant institutions. Most work in regime theory has been dedicated to discerning factors and ascertaining if actual bargaining matches our hypotheses— especially those drawn from rational choice. in his institutional bargaining approach that parses regime dynamics into prenegotiation. and decision-making procedures that characterize regimes.1 Beyond rational choice. actors are constantly adapting to the system rules. principles. Oran Young (1994. In other words. stick and have influence (Ikenberry 2001). Pierson 2000) allow us to understand why regimes. I explore an additional aspect of complexity that moves us beyond regime theory—complex adaptation. In other words. so to speak. Models of evolutionary cooperation and bounded rationality from the complexity literature can inform sophisticated regime bargaining studies (Axelrod 1997. Hypotheses follow on when to expect regimes to form and function effectively (List and Rittberger 1992). I discuss how treating states as adaptive actors ensconced in a coevolutionary process with their context helps us to understand the ozone depletion regime. situation type. exogenous shocks. Studies of increasing returns in politics and economics (Arthur 1994a. regime theory has faced a number of challenges that have yet to be resolved. A complexity approach calls this into question and claims that the system rules and the actors are entwined in a coevolutionary relationship. Less time. and so forth. negotiation. HOFFMANN In the best social scientific tradition. has been spent on exploring the mechanisms that produce the rules. Regime theory generally conceives of the actors and their context to be relatively static and independent. What’s wrong with this? Nothing per se. and postnegotiation phases. norms. The list of independent variables is long.96 MATTHEW J. First.

Haas 1992. I demonstrate the plausibility of this contention by tracing the coevolution of the dominant state in the ozone depletion regime (the United States) and the system rules within which the United States was embedded. adapt its understanding of the system rules to fit the change?.g. Second. and I conclude with some thoughts about further empirical testing of the hypotheses presented. It is impossible to understand the ozone depletion regime without understanding the underlying requirement for universal participation that developed. I discuss the aspects of complexity used to analyze the ozone depletion regime. e. First. thus providing insight into the formation of regimes and their evolution through time. why did the United States. how did the system rules governing participation in the regime change from calling for North-only negotiations in the early 1980s to the universal negotiations of the late 1980s? The international community’s understanding of who should participate in the ozone depletion regime changed over time and influenced what issues were addressed. Rowlands 1995) has been hailed as a singular success in the realm of environmental politics. it is possible to understand actors’ perceptions of the problems and the solutions deemed possible—crucial prior information for explaining the bargaining that is the usual fodder for regime analysis. Tolba 1998. I then give a brief overview of the ozone depletion regime for background purposes. Through complexity. Yet there are puzzling aspects of the regime that have been overlooked with serious consequences for how we understand the response to ozone depletion itself as well as the “lessons learned” for regime analysis in environmental politics more generally. . Litfin 1994.BEYOND REGIME THEORY 97 rules (both the specific regime rules and the larger constitutive rules of the system) are dynamically changing through the self-organized actions and interactions of actors. I explore two aspects of this complicated regime. Benedick 1991. To demonstrate such an approach. which could have ignored this change.. The discussion of the two main puzzles is next. and what rules would be encompassed in the regime itself. I turn to the ozone depletion regime. I contend that one way to understand both of these puzzles is to treat states as complex adaptive actors and to consider that universal participation emerged through the process of complex adaptation. First. Parson 1993. Young (1999) notes that such regime foundations or “discourses” “not only provide a way of framing and addressing problems and the behavioral complexes within which they are embedded but also contain normative perspectives on the importance of the problems and appropriate ways to resolve them” (206–7). This oft-studied regime (see. Specifically. A complexity approach provides insight into how this coevolution takes place.

predict the consequences of action. and goals. The system rules. the rules are behavioral. they become incorporated. and Lane 1997). agent characteristics are not deducible from knowledge of the system. but they can also represent identities. This process links adaptive actors in a coevolutionary relationship with each other and with the larger system. Adaptive agents are always trying to “fit” with their context. and act. HOFFMANN COEVOLUTION AND COMPLEX ADAPTATION Complexity provides a process that links agents and a broader (social) system. System change results when innovation on the part of a subset of agents throws the system rules into flux and other agents then adapt their rule models and therefore their actions and interactions. In complexity. the agents are successful. Agents evaluate the results of actions and assess the ‘fitness’ of their rule models. crucially.2 while in turn shaping those same actions and interactions. Durlauf.. or kept in relation to the evaluations. 1998. identity. and behaviors are appropriate or fit.g. through the evaluation process. Similarly. into the agents’ rule models. changed. produced by agent actions and interactions. The actions that adaptive agents undertake and the interactions in which they participate reproduce or alter the larger system.4 In this way. In a complex system. the two are inextricably entwined. Adaptive agents are defined by internal rule models or schema (Holland 1995. though. Gell-Mann 1994). neither is reducible to the characteristics of the other. and behavior—while the agent’s actions feedback and affect those same system rules. These rule models represent the agent’s internal (or subjective) understanding of the world (the larger system) around them. System characteristics cannot be directly derived from knowledge about agent characteristics. Explaining the system-agent linkage is done analytically by positing the process of complex adaptation. When their internal rule models fit their context. and Arthur. the agents are not successful. Holland 1995. When some agents change their behavior. the system rules influence agents’ internal rule models through coevolutionary processes. interests.98 MATTHEW J. A new context “forces” agents to alter their rule models as the context determines what goals. They allow the agents to perceive and define their situation. this alters the system for the other agents. weakened. . When their rules do not fit. an agent’s system shapes its internal rule models—its interests. This irreducibility is characteristic of complex systems whereby the system characteristics emerge from the actions and interactions of the agents and those actions and interactions are shaped by the system (see. The system rules that define the agents’ context emerge from the actions and interactions of the agents. In most applications of adaptive agents. At the agent level this adaptation is facilitated by self-evaluation of behavior.3 Internal rule models are strengthened. e. interests. do more than constrain potential actions.

I trace the process of complex adaptation and discuss how universal participation emerged. At any time states “know” what the global response to environmental problems like ozone depletion should be. thus. this “global” problem entailed varied responses. domestic politics. reifying or changing the rules that dominate in an issue-area. and each state has an internal understanding of the system rules. though not all states will share the same understanding. 1985. 1990. THE FORMATION OF THE OZONE DEPLETION REGIME Ozone depletion was recognized as a potential problem in 1974 when two scientists put forward the hypothesis that chlorofluorocarbons (CFCs) could destroy the ozone molecules in the stratosphere that protect the Earth from UV radiation (Rowlands 1995. what impact it had on the ozone depletion regime. Copenhagen Amendments. States act on the basis of these internal rule models. chap. inherent dynamism in the internal rule model/system rules relationship—adaptive actors shape and are shaped by their political context through their actions and interactions. Montreal Protocol. The initial regime activity and negotiations that produced the Vienna Convention of 1985 consisted mainly of bargaining between the United States and the EU states. and negotiations resulted in a succession of agreements (Vienna Convention. States are embedded in a system with rules. There is. Their actions (bargaining. Some states. After a brief background description of the ozone depletion regime. Change is driven by interpretation of scientific information. International activity began in the late 1970s. 1992) that moved the state of the ozone depletion regime from a call for research through a complete phaseout of CFCs. At first the EU was opposed to reductions in ozone-depleting chemicals. and why the United States adapted to this emergent participation rule.6 Politically. notably the United States and the Nordic states. the logic of complexity entails the coevolution of actor understandings and the political context. Very simply. and social interaction with other states—the evaluation of actions and the altering of rule models. 1987.BEYOND REGIME THEORY 99 A complexity perspective thus provides a model of agent behavior— adaptive—and a feedback process that dynamically links the internal understandings that agents have of their system rules with the system rules themselves. London Amendments. for instance) shape the system rules. 2).5 The ozone depletion regime provides fertile ground for a demonstration of the empirical utility of a complexity perspective. took unilateral domestic action and began to regulate CFCs. because collectively the European states had come to dominate the .

The compromise or “Grand Bargain” (Tolba 1998) struck in London was an accelerated CFC phaseout (100 percent emission reductions by 2000) combined with a pledge by Northern states to compensate Southern states for non-CFC development paths and their accession to the ozone depletion regime. Post-Montreal negotiating sessions routinely included upwards of one hundred states. With the EU and US at loggerheads. the international community went from calls for research coming from a negotiation with mostly Northern states to a full-fledged phaseout of CFCs and a regime encompassing most of the globe. The Montreal Protocol is still hailed as an exemplar of environmental negotiating and the ability of the international community to take decisive action on an urgent environmental problem. The system rule defined . The Montreal Protocol itself laid out a compromise on binding reductions that would see CFC emissions decrease by 50 percent by the year 2000. and the urgency surrounding the issue ramped up once again. The discovery of the ozone “hole” in 1986 dramatically heightened the sense of urgency surrounding the ozone depletion problem. At the London meeting of the Parties to the Montreal Protocol bargaining was much more North-South than it was United States–European Union. except for an important call for further negotiations and a statement of guiding principles. the Vienna Convention accomplished little. and the North-South dimension of these negotiations altered the regime dynamics significantly. After the protocol was signed. First. The negotiations were larger in number (up to about sixty states from twenty-five) and the Global South took part in significant numbers for the first time. Before this time. Regime dynamics did not end with the Montreal Protocol. they forged ahead with negotiations to reduce CFC emissions. Two parts of this regime formation story demand our attention. The move to amend the Montreal Protocol and strengthen the ozone depletion regime was complicated by the now universal attendance at the ozone negotiations. The signing of the Montreal Protocol is generally conceived of as the emergence of the ozone depletion regime—when the rules fostering cooperation to solve the ozone depletion problem were put into place. and though the international community lacked scientific proof of the connection between CFCs and ozone depletion. HOFFMANN worldwide CFC market when US domestic actions served to decrease its output. the proper response consisted for the most part of Northern states alone. in 1987 we see a breakpoint change in how the international community perceived the proper response to ozone depletion. however. in five short years. These negotiations culminated in the Montreal Protocol and were again mainly comprised of US/EU bargaining. new scientific findings solidified the proof of the CFCozone depletion connection.100 MATTHEW J. Thus.

The actions of Northern and . other actors adapt to the flux in the system rule. This would not be so puzzling if the United States were not the hegemon. Why did the ozone depletion regime go global?7 Second. though profound. This rule model drove US actions in the Vienna Convention and Montreal Protocol negotiations. Understanding the emergence. Further. some actors change their rule models when they negatively evaluate the outcomes of their actions (or inaction. the altered system rules in ozone depletion would come to have an impact beyond the ozone depletion regime. like global warming. The participation system rule goes into flux. After 1987. the US rule model changed. Northern states change their rule models and begin advocating universal participation. Why did the United States change its internal rule model to fit the altered system rules? These puzzles are more than academic nitpicking. the United States. The definition of the ozone depletion problem and solutions had changed.BEYOND REGIME THEORY 101 ozone depletion as a Northern problem and deemed that solutions would be formed and adhered to by Northern states. did not have to change its perceptions of ozone depletion. and the single most important player in the ozone depletion regime. In stage 1. functioning. The altered system rules determined how the bargaining to amend the Montreal Protocol would take place. of all states. According to conventional wisdom (and conventional international relations theory). and lasting impact of the ozone depletion regime requires that we understand how ozone depletion came to require universal participation and. came to be defined. Both Northern and Southern states define ozone depletion as a problem requiring North-only participation. EXAMINING PARTICIPATION AND REGIME FORMATION 8 A complexity perspective tells a simple. story about the evolution of the participation system rule. in 1987. In stage 2. further. In stage 3. the stability of the Northonly rule erodes. the US internal rule model clearly defined ozone depletion as a North-only problem. how even the United States came to adapt to a particular set of system rules. universal participation in the ozone depletion regime has been the rule. in this case). because the transition can be described in three straightforward stages. influencing how other environmental problems. The story is simple. When they act on this new internal model and participate in the regime negotiations. the system rule and internal rule models of the states match—states’ behaviors are driven by internal models that are adapted to and reify the system rule for North-only participation. After Montreal. Southern states come to define the ozone depletion problem differently and perceive that it requires their participation as well.

HOFFMANN Southern states. I trace in greater detail the transitions in the participation system rule and in US rule models for dealing with ozone depletion through the three stages outlined above. This is quintessential complex adaptation. Paul Horwitz of the US Environmental Protection Agency (EPA) and UN Environment Program (UNEP) noted that “decisions were being made by countries that were the problem—they believed they could get a hold of the problem. and by 1987. instantiate a newly stable system rule for universal participation.102 MATTHEW J. The microprocess of agent-adaptation and internal rule model change is linked to the macroprocess of evolution in system rules. and the actions of both Northern and Southern states reified the North-only system rule.9 Stage 1: North-Only Participation In 1985 there was a stable system rule calling for North-only participation. Agents are driven by rule models and their actions (re-)create the system in which they are embedded. The story is profound because of the implications of the transitions. because they determine what gets bargained over and how such bargaining takes place. [The] group thought they owned the issue. This is not always a poor set of assumptions. Only four of the original twentyone signers of the Vienna Convention were Southern states. but it reifies a misleading understanding of politics as relatively stable. With a complexity perspective we can account for change in system rules and more fully explain the process of regime formation. The system rules can be considered the political context (or “episteme”—see Johnston 2005) within which regimes are formed. only two developing nations had ratified the convention (Benedick 1991.” He also noted that in the beginning there was “less of a stress on global participation” and that it “didn’t make sense to negotiate a global agreement.”10 A global agreement did not make sense to South- . 201–14). now operating with similar rule models again. Innovation in a subset of the population of agents disrupts the system rules and leads to adaptation across the whole system. It was obvious to both Northern and Southern states that ozone depletion was a Northern problem (Sims 1996. It blinds us to the dynamic (even if slowly changing) nature of system rules and to how system rules shape and constrain the bargaining that is the main focus of regime theory. Both Northern and Southern states had internal rule models that defined the ozone depletion problem as one requiring a Northern negotiated regime. they are the rules of the game and structure the process of regime formation. 265–69). Understanding the system rules is a key aspect of analyzing regimes. In traditional regime theory. In the following discussion. the system rules are assumed to be static and unproblematic.

and India’s CFC production remained marginal to world production. The bargaining at this stage was shaped by the participation rule. In Montreal in September 1987.BEYOND REGIME THEORY 103 ern states. there was no proof of any threat to India. They saw little change in the situation—the scientific uncertainties about ozone depletion continued. financial resources. Canada. and the main issues in contention were whether to undertake CFC emission reductions before unassailable scientific proof of the CFC–ozone depletion link was available and how to deal with technology transfer among Northern states once substitutes for CFCs were developed.”11 Southern disinterest in participating matched Northern indifference to Southern participation and would continue past the Vienna negotiations. 1987a. and the Nordic countries] and there was the EU[. Southern participation through the summer of 1987 and discussion of Southern issues were sparse. Because everyone knew that ozone depletion was a Northern problem. In the four meetings in 1986–87 the main bargaining continued to occur between the United States and the European Union. everyone also knew what issues to discuss. India’s position was broadly representative: In the post-Vienna period. did not participate in the preparatory meetings for the Montreal Conference. as Stephen Seidel of the US EPA observed. either. the provisions for development assistance remained nearly unchanged from what was included in the Vienna Convention. Essentially. and technology transfer— were not of primary concern.12 The North-only rule continued to shape the formation of the ozone depletion regime as the international community responded to the discovery of the ozone hole and the associated increased urgency of ozone depletion with negotiations toward an ozone protocol. the United States remained focused on North-only negotiations and set its sights on Japan and the Soviet Union in . There was the Toronto group [the United States. The main bargaining was undertaken between the United States and the European Union. India. therefore.] the key negotiations took place between them. it is no surprise that the issues that traditionally engage the South—economic development. with the United States pushing for deep CFC cutbacks quickly and the European Union urging slower action (see UNEP 1986. Indian policy makers continued to believe that ozone depletion was mainly the concern of the developed countries. Thus. 59) This stability around North-only participation pervaded the negotiations that culminated in the Vienna Convention of 1985. (Rajan 1997. The Vienna Convention only discussed the need to take into account the “circumstances and particular requirements of developing countries” and only vaguely discussed technological cooperation. 1987b). “[T]he developing countries didn’t want a big role.

a number of reasons are evident for why Southern states would weaken the North-only rule model and come to feel that their own participation in the ozone depletion regime was necessary. but briefly. states with less than a third from the South. . A17). In the summer and fall of 1987. a universal membership organization. 1987. Southern rule models underwent a transition. Very simply. Southern states started to participate. Southern states were not pining to be included. However. at Montreal 65 percent of the fifty-seven participants were Southern states (UNEP 1987c). Whereas previous negotiating sessions had been attended by twenty-to-thirty. In addition. While the United States was “looking for an effective agreement involving as many nations as possible” (San Diego Union-Tribune. It was simply the case that both Northern and Southern states had internal rule models that defined the ozone depletion problem as one that required North-only participation. The United States and the rest of the Northern states were not trying to convince Southern states to participate. UNEP and its executive director.104 MATTHEW J. HOFFMANN addition to the European Union (Boston Globe. Stage 2: Instability in the System Rule But the stability of the North-only participation rule was not destined to endure. September 8. Northern and Southern actions reified the North-only participation rule. Similarly. September 10. The United States still considered the ozone depletion problem to be a Northern issue. the negotiations were held under the auspices of the UNEP.13 First. Throughout the early regime formation process. it is clear that the nations that the United States wanted to involve were Northern states. why the Southern states changed their rule models is in some ways less important than the fact of their changed rule models and behavior. there were incentives to change rule models based on traditional notions of self-interest (the desire to have continued access to cheap. In-depth explanation of the transition in Southern states’ rule models is beyond the scope of this chapter. Keeping Southern states informed of the process contributed to the negative evaluation that Southern states had of the ozone depletion regime and their own North-only rule models. 1987. for this discussion. On the contrary. evident in changing Southern behavior leading up to and including the Montreal negotiations in September 1987. The system rule determined what states were present and what was bargained over—important parameters of regime formation. useful chemicals) and environmentalism (the desire to be seen as environmentally friendly). 1). Mostafa Tolba. aggressively advocated universal participation (Hoffmann 2002). Southern states were not barred from the ozone depletion regime.

the South started ‘playing’ too late to be a major factor at Montreal. If Northern states deemed the Montreal Protocol sufficient to . Most Northern negotiators thought “the terms [of the Montreal Protocol] were attractive enough to encourage other developing countries to sign onto the document” (Rowlands 1995. while further eroding the North-only system rule. what emerged was a participation rule in flux. The instability in the system rule paradoxically had both relatively minor as well as extremely significant consequences. This thinking was a vestige of the understanding of the ozone problem as one to be dealt with through North-only measures—a rapidly deteriorating vision of a global response for ozone depletion. 169–70). as most of the provisions of the MP had already been worked out (Miller 1995. The flux in the participation system rule would soon come to alter how states perceived the ozone depletion problem and the solutions necessary to solve it. as the Northern states were still operating on a North-only participation rule. As Northern states would soon discover. The change in Southern rule models and behavior altered the system rules and presented a new context to which the Northern states had to adapt. There was a great deal of inertia behind the proceedings to this point. The flux had little impact on the Montreal negotiations themselves.BEYOND REGIME THEORY 105 In participating. 78–79). blindsided the United States and other Northern states. This action. With Northern states still following the Northonly rule and Southern states following a new internal model that called for their own participation. the substance of the discussions at Montreal remained focused for the most part on Northern issues of CFC reductions and substitutes. On the other hand. the transition in Southern rule models was significant in that it changed the post-Montreal negotiating landscape entirely and drove the adaptation of American and other Northern states’ rule models. Yet Northern adaptation to the flux in the participation system rule was not inevitable. Stage 3: US Adaptation and the Emergence of Universal Participation Southern states were not satisfied with a protocol that was negotiated essentially without them. a regime negotiated by states driven by North-only participation rule models would not be effective in a context where the North-only system rule had eroded. and they signaled this dissatisfaction by declining to sign the Montreal Protocol. Southern states altered the stability of the system rule calling for North-only participation. In addition. and Southern participation at Montreal did not greatly alter the work that had been accomplished over the previous two years. While Southern issues were raised.

the international community came to realize that the Montreal Protocol was not enough to solve the problem (see. the enthusiasm for the protocol waned. March 19. certain Southern states (Brazil. 706). China. and Southern production to this point was primarily in joint ventures with Northern companies. In 1988. This was not an outlandish assumption. the United States was very pleased with the substantive results of the protocol. Environmental Nongovernmental Organizations (NGO) representatives were even blunter. the urgency surrounding the ozone depletion problem dramatically increased. Initial US evaluations of the protocol were enthusiastic. with the North responsible for devising solutions. but it continued with a different focus. especially the 50 percent reduction in some CFCs. and perhaps Indonesia) had large enough domestic markets to create a viable CFC industry. In addition. However. Benedick 1991. Congressional Quarterly Weekly Report. 1988. the ignored or forgotten potential of the South to produce/consume CFCs and thus contribute to the problem became critical to Northern states with a changing perception of the ozone depletion problem.. A24. HOFFMANN solve the ozone depletion problem. 1988. David Doniger of the Natural Resources Defense Council (NRDC) argued that “There . The United States negotiated the Montreal Protocol operating under a rule model (and political context) that told it that Northern states had control of the problem. and the imminence of a rule model change was not evident in the United States in the months immediately following its signing. immediately after the Montreal Protocol was signed. as the science became more certain and the reality of the lack of Southern signers became more evident. and the EPA began calling for a reassessment and accelerated cutbacks (Mills 1988. Crucially.g. However. 1987. unilateral action. However. CFC technology is relatively simple and was widely available. Even as the United States prepared to ratify the protocol. then Southern participation may never have been an issue of concern.14 In 1988–89. The South produced and consumed relatively small amounts of CFCs. however. the importance attached to the Southern potential and the expectation that the South would comply with the Montreal Protocol changed significantly in 1988–89. The regime formation and/or strengthening process thus continued in the late 1980s. the US Senate prepared to consider further. The South claimed a voice and demanded to be a part of the decision-making process that promised to alter development paths. October 28. In late 1987. Washington Post. Southern states no longer assumed that ozone was a North-only problem. this potential to contribute to the problem was as evident in 1986 as it was in 1988. September 27. A3. India. San Francisco Chronicle. With scientific proof of the CFC–ozone depletion link in hand and more information on the potentially catastrophic effects of ozone depletion. e. 370).106 MATTHEW J.

It is hard to overestimate the importance of industry’s role in pushing universal participation in ozone depletion. The North-only rule had driven US behavior in the negotiations of a landmark agreement. and the NRDC prepared to force the EPA to take action through the courts once again (Global Climate Change Digest 1988). hoping to delay or defeat unilateral measures (Carnevale 1988). India’s spokesperson praised the “polluter pays” principle and “made known the Third World’s doubts about the industrialized countries’ political will to come up with the required financial aid and technology transfers for CFC technology” (International Environment Reporter 1989b. “[T]he treatment of developing countries. 129). Though industry did not create this vision of a global response. With the participation rule weakened by negative evaluation. It was at this conference that the South became vocal. The work ahead was clear-cut. Participating in earnest in the negotiating process for the first time in large numbers. the United States adopted a universal participation rule. The United States adapted to the new system rule and began advocating universal participation. As Richard Smith of the US State Department reported: “At Helsinki [the first meeting of the parties to the Montreal Protocol]. 169). they would have to accede to an international . but the Montreal Protocol was a landmark only as a bare beginning in the fight against ozone depletion. A Chinese official protested that Southern nations “resented the rich ‘telling them what to do and not to do’” (International Environment Reporter 1989b. It was clear that if the Northern states were to achieve Southern cooperation. the South began pushing for developmental assistance at a 1989 London conference on the ozone depletion problem attended by 123 states (UNEP 1989.15 These evaluations of the Montreal Protocol significantly weakened the US internal rule model calling for a North-only response to ozone depletion. became a central concern” (Litfin 1994. As Litfin observed.BEYOND REGIME THEORY 107 is virtually no chance that the current protocol will be sufficient to solve the problem” (International Environment Reporter 1988a. it was clear that many developing countries want to participate but are understandably concerned about the potential costs to their economies” (US House of Representatives 1989. The new consensus on the system rule calling for universal participation made a significant difference in how the bargaining proceeded. Even industry joined the calls for further international action. 106). which hitherto had been considered a minor issue. paragraph 11). 111). “Some nations will not find it easy to forego the use of CFCs in their quest for industrialization” (International Environment Reporter 1989a. once it was in place industry played a crucial role in fostering the evolution of US rule models toward universal participation. 169). Daniel Arap Moi of Kenya voiced the generic Southern concern when he stated. Representatives from India and China were more direct. 77).

1990. The United States was loath to submit to Southern demands. they agreed to do this with new and additional funds administered by a new institution. Transition through Complex Adaptation Universal participation was not a natural or inevitable rule for the ozone depletion regime.108 MATTHEW J. Regime theory misses these dynamics and is thus hampered from understanding the bargaining that takes place in regime formation and beyond. however. because the new institution was to be jointly controlled. rather than solely administered by donor states. the South (crucially India and China) agreed to the Montreal Protocol. This was a huge victory for the South. and the fight against ozone depletion became a truly global affair. The United States and other Northern states agreed to pay the full incremental costs of the transition away from CFCs incurred by the South. What did change were the internal rule models of the Southern states—how they perceived and acted toward the ozone depletion problem. This was the bargain that the South. why should we care about the emergence of universal participation and the US adaptation to this new rule? The second is . and the impasse threatened to scuttle the ozone negotiations. a deal had been struck (UNEP 1990). was driving. Nothing about ozone depletion or the interests of the major actors radically changed between 1986 and 1988. The South made it clear that they would not participate without a fund. HOFFMANN fund and technology transfer mechanisms—they would have to make side payments to the South. A21). A1). This change set off a coevolutionary transition in both the internal rule models of the Northern states and the system rule for participation. BEYOND REGIME THEORY? Three questions remain. and the United States. in particular. and by June 29. but the bigger accomplishment of the London negotiations was the establishment of the Multilateral Fund and technology transfer mechanisms. At London. and more importantly. With the funding provision in place. held on to its opposition to a new fund. now fully engaged in the negotiating process. Just prior to the London meeting of the parties to the Montreal Protocol. The London Amendment to the Montreal Protocol contained the ultimate phaseout by 2000 desired by all parties (and most nonparties). compromise ruled the day. First. advocating instead the use of the World Bank and other existing institutions (Weisskopf 1990a. the United States relented and agreed to an independent fund with the caveat that any ozone fund set up would set no precedent for other environmental issues (Weisskopf 1990b.

It shapes what issues are discussed and even considered. . Examining and explaining the emergence of universal participation in the ozone depletion regime aids our explanations of the formation of regimes beyond ozone depletion. Regime theory. Ikenberry 2001) and. Similarly. other issues were also seen as requiring universal participation. through increasing returns (Pierson 2000). The universal participation rule that emerged in the ozone depletion regime ushered in an era of global responses to environmental problems. system rules tend to lock in (Arthur 1994a. tend to be relatively stable. Perhaps states were making objectively rational choices in response to the incentives and constraints they faced in the ozone depletion issue.BEYOND REGIME THEORY 109 more fundamental to this chapter: why do we need complexity concepts to address the puzzles of universal participation? Finally. Once this happened in ozone depletion. Universal participation was therefore the obvious choice: involve the South in negotiations and actions toward a solution. because the system rule for participation has an enormous influence over the substance of regime negotiations themselves. how can the complexity-based claims about regime transformation put forward here be rigorously assessed? Why Study the Transformation of System Rules? Substantively. Southern issues of development and technology transfer were virtually ignored. the story of the emergence of universal participation is crucial. which too often takes the number and identity of the actors for granted. Before Montreal. Do We Need Complexity Concepts to Comprehend the Emergence of Universal Participation? It is not immediately clear that complexity is the only perspective that can explain the emergence of universal participation. the Southern states’ decision could have been a rational reaction to the imminent regulation of an important class of chemicals. After Montreal. and a commitment to North-only participation did not facilitate the accomplishment of this goal. For example. the altered system rule for participation ushered in a transformed set of negotiations. especially climate change (Hoffmann 2005). The United States wanted to solve the ozone depletion problem. the transition in US thinking could be characterized as a rational updating of strategies/beliefs in light of new scientific information or new understanding of economic interests. Northern issues of CFC reductions and replacement technologies dominated negotiating agendas. to the detriment of its ability to explain regime outcomes. cannot capture this dynamic transition. Further. North-South issues came to dominate the agenda.

the United States committed to universal participation because that requirement had come to pervade the system. the United States could not hope to actively or effectively participate. instrumental acceptance of universal participation would not diminish the influence of the nascent system rule for universal participation (see Risse. however. Multiple actors with widely varying motivations all arrived at the same conclusion: the ozone depletion regime required universal participation. to catalyze their participation. but those reasons were constant in the 1985–88 period when their behavior changed. Without the appropriate understanding of the foundations of the ozone depletion regime. In fact. In addition. This was a coevolutionary process whereby the system rules. It was taken as natural for the United States and other Northern states to pursue universal participation and extend the ozone agreement to the Southern states to eliminate future damage to the ozone layer. especially given the hegemonic position of the United States both in terms of traditional notions of power and power in the ozone depletion issue itself. and Sikkink 1999 for a similar argument about norm acceptance). The US understanding no longer fit with the prevailing system rule. First. First. Second. the US “instrumental” acceptance can also be seen as adapting to a new context. No options beyond universal participation were considered. and it is certainly not clear that the trade restrictions in the Montreal Protocol would not have eventually brought the South on board anyway. Instead. Ropp.110 MATTHEW J. It took a transition in how they perceived the ozone depletion problem. and so US understandings had to change (Bernstein 2001). Crucially. HOFFMANN However. this was the thinking at the EPA in the immediate aftermath of the Montreal Protocol negotia- . Southern states certainly had good (rational) reasons to participate. altered by the participating Southern states. contained a new understanding of the ozone depletion problem.16 It is not clear that the Southern states had as much bargaining leverage as they are usually credited with. Either choice may have been a more effective way to meet its goals. In hindsight. the United States could have stayed the course of the Montreal Protocol provisions or could have advocated limited Southern participation. rather than pursuing universal participation. close scrutiny reveals that a rational updating explanation falls short. the specific direction of the change in US definitions of ozone depletion relied on the existence of a system rule requiring universal participation. not a rational calculation. though plausible alternatives are imaginable in hindsight. The United States did not calculate from a number of potential options that committing to universal participation was the way to maximize its utility. the United States and Europe could have remained committed to a North-only negotiated regime and pursued a coercive strategy to force Southern states to accept the ozone depletion regime.

internal rule . and Brazil—increase their domestic consumption of CFCs to the levels allowed by the protocol.17 The emergence of universal participation and US adaptation to it was not rational in the traditional sense. a complexity approach provides concepts and tools useful in meeting the challenges faced in the study of global environmental regimes. because once its original definition of the ozone depletion problem was weakened. Complex adaptation. we gain greater understanding of the foundation of the ozone depletion regime. The large Southern states were the main concern. It might have been a rational choice to retain a North-only vision of the ozone depletion problem and let the trade sanctions and fear of technological obsolescence “force” the South to comply with the Montreal Protocol. and Brazil to agree to the Montreal Protocol and join the process. In addition to the growing interest of Southern states. India. In other words. thus constraining possible choices for US definitions/strategies. even if it was objectively rational to include Southern countries in the regime formation process. March 6. the ones with real leverage. Instead.BEYOND REGIME THEORY 111 tions (International Environment Reporter 1988b. Empirically Testing the Complexity Explanation? In this chapter. 226). the United States (and other Northern states) came to understand the problem as universal—it came to accept the current understanding of the problem. this does not necessitate universal participation. The United States and other Northern states could have worked to involve the large Southern states. The international community had already instantiated the system rule for universal participation. it was a result of complex adaptive processes. “if just four developing countries—China. 1). likely would have been less expensive for the United States in terms of necessary development concessions and perhaps would have been more efficient. By conceiving of states as adaptive actors and tracing the dynamics of complex systems. Indonesia. It could have been perfectly plausible to entice China. India. Neither of these potential alternatives to universal participation was considered by the United States or other states. First. avoiding the problems associated with large negotiations. CFC production on a worldwide basis would double from the 1986 base level” (quoted in the Los Angeles Times. why bother negotiating with Southern countries if the provisions of the protocol combined with the power of Northern nations would make Southern ascension to the regime inevitable? Second. the system rule was also already enshrined at UNEP. I have endeavored to demonstrate two points. Dealing with just these states. According to Irving Mintzer of the World Resources Institute. and the new participation rule was ensconced in the structure of the negotiations.

Going beyond (neoliberal) regime theory with a complexity approach for studying environmental regime emergence and transformation is not obviously necessary. Lipson. I briefly discuss directions for future and comprehensive complexity work—a research program for rigorously assessing complexity explanations of environmental regime transformation. The brief empirical discussion above hints at but does not fully demonstrate the superiority of a complexity approach over neoliberal regime theory. I chose the transformation in the ozone depletion regime from North-only participation to universal participation precisely because the two approaches differ in their expectations and explanations.112 MATTHEW J. Second. For regime theory. A complexity approach to environmental regimes provides explanations of bargaining among dynamic. These different foundations obviously provide for different empirical expectations. In this concluding section. enhancing understanding of environmental regimes. adaptive actors in a coevolving context. As noted above. Indeed. these concepts and tools can be successfully employed in empirical assessment. The question that remains is how to craft a research agenda that would provide such a demonstration.and micro/macroprocesses needed to frame our analyses. a bare beginning. HOFFMANN models. and if. as are the interests of the states facing the problems. Environmental problems and their solutions are actively created and re-created as actors coevolve with their context. The two approaches have clear and quite distinct sets of assumptions. and Snidal 2001) and why they might change. So does the complexity approach go beyond regime theory? What steps are required to test the complexity explanation? First. we would not need to consider complexity. arguing that the transformation of participation rules emerged through the coevolutionary actions and interactions of states. we can discuss expectations about participation (Koremenos. Without this anomaly. because regime theory and a complexity approach do not agree on what can change. the transformation of the regime from requiring Northonly to universal participation is an (important) anomaly for regime theory. it is at least plausible on its face that a complexity approach can provide a superior explanation of regime dynamics. system rules are assumed and exogenous to the analysis. however. This is. Direct comparison is somewhat difficult. Regime theory provides explanations of bargaining (and bargaining outcomes) among static. However. A complexity approach views system rules as malleable and endogenous to the analysis. Regime theory says such a transformation should come at the behest of powerful actors working with new information—rational updating—and the explanation is found in rational bargaining. Environmental problems and their solutions are assumed known. rational actors in a static context. and coevolution form the foundation of micro. as in .



this case, the complexity explanation can accommodate the anomaly, we have gone beyond regime theory. A rigorous test of the complexity explanation begins with a method not discussed explicitly in this chapter: formal modeling. Proponents of a complexity approach (in the social sciences and beyond) have developed a set of computer simulation tools that offer a “laboratory” traditionally denied to social science (see, e.g., Epstein and Axtell 1996; Hoffmann 2005; Axelrod 1997; Cederman 1997). These agent-based models allow proponents of a complexity paradigm to rigorously assess the logic of their arguments—putting a proposed explanation onto the computer forces one to explicitly define critical assumptions. Such simulation analysis facilitates the discovery of boundary conditions, unexpected hypotheses, and perhaps most importantly, understanding of under what conditions we can expect to see certain outcomes (expected or otherwise). Joshua Epstein (1999) claims that computer simulation experiments provide researchers with rigorously arrived-at “candidate explanations” for social phenomena. Where regime theory uses game theory to provide a rigorous foundation for its explanations of bargaining, a complexity approach uses agent-based modeling. Thus, the logic of a complexity explanation can be assessed formally and at least as rigorously as the regime theory explanation. Regarding the ozone depletion regime, I have in other places (Hoffmann 2002, 2005) reported on agent-based modeling experiments designed to explore the emergence and evolution of system rules (or norms) through the interactions of coevolving agents. These experiments demonstrated the abstract plausibility, though not the empirical validity, of the explanation for regime transformation developed in this chapter. With confidence in the logical soundness of the explanation based on complexity concepts, empirical assessment of the complexity explanation of the ozone depletion regime requires further and detailed process tracing (see Hoffmann 2005 for an effort in this direction). This is a nontrivial task, and it is no wonder that, with laudable exceptions (Brunk 2002; Cederman 2003; Jervis 1997), complexity scholars have shied away from empirical work. A full account of the emergence and influence of universal participation requires an analysis of the coevolution of multiple actors’ rule models with the participation requirements, as well as the coevolution of the actors themselves. Complexity processes are far from parsimonious, and a full comprehension of a complex system requires thick description and rich empirical detail. Specifically, rigorous testing of the explanation proposed here would entail tracing the development and adaptation of the European Union’s and Southern states’ rule models in addition to that of the United States. In addition, more attention needs to be paid to the dynamics of the rule models themselves, exploring the domestic and global political processes



through which actors define their rule models, evaluate their behavior, and alter their rule models. As discussed above, complex adaptation is an abstract model of actor behavior that needs to be fleshed out with significant empirical detail. Combining agent-based modeling experiments and empirical process tracing (other empirical methods can be used as well, depending on the research question) in a recursive process provides the most rigorous test of the explanation. The simulation experiments inform the process-tracing case studies, and the empirical work feeds back to inform further modeling. With such testing we can assess whether or not the complexity explanation asserted and initially explored above actually addresses the anomaly of universal participation, and we can demonstrate that a complexity approach takes us beyond regime theory. NOTES 1. A special issue of International Organization from 2001 (vol. 55, no. 4) is entirely dedicated to exploring rational choice mechanisms in regime emergence and design. 2. Emergence is an oft-debated and imprecise concept. For an introduction to emergence and emergent processes, see Holland 1998. 3. Agents are not always or even usually treated as unitary. When dealing with meta-agents (agents composed of other agents), subagents within the agent do evaluation. For instance, environmental groups evaluate the outcome of negotiations that the United States participates in. 4. The evaluation stage of complex adaptation is crucial. At this stage, agents alter their rule models, which is key for understanding how ideas become an ingrained part of internal rule models. In addition, however, this stage adds variation in a population of agents, because different agents may have different evaluation processes and different criteria for fitness. 5. Internal rule models can be difficult to operationalize empirically, because they are inherently unobservable. However, one advantage to studying large, corporate actors such as states is that such agents usually write down the understandings that comprise their rule models. As a proxy for the actual (unobserved) rule models of the United States, for instance, I treat the negotiating positions that the executive branch used in the various negotiations as the rule models. These negotiating positions reveal how the United States defined the problems over time as well as what it desired out of agreements and what it was willing to commit to. In order to get a full sense of the negotiating position for any particular set of negotiations, I utilized several sources in an attempt to “triangulate” and get a true picture (US Congressional documents, UN meeting reports, newspaper accounts, and interviews). The system rules are equally



difficult to directly observe. They are essentially the norms/structures or, in the broadest sense, rules that are external to the United States. I also triangulate around these system rules from several sources (UN documents, newspaper accounts, and interviews). 6. For in-depth analysis of the ozone depletion regime, see Tolba 1998; Benedick 1991; Litfin 1994; Rowlands 1995; and Hoffmann 2005. 7. The dominant way to approach this change in the literature is to analyze the ozone depletion regime before and after Southern states joined the proceedings. Unfortunately, almost no attention is paid to the transition. This is likely an artifact of the rational choice approach’s propensity to downplay history and to treat each bargaining situation as if no interactions occurred before the current negotiation. See Hoffmann 2005; and Mitchell and Keilbach 2001. 8. For an elaboration on this discussion of the formation of the ozone depletion regime, see Hoffmann 2005. 9. Participation is obviously just one of a number of important system rules. 10. Interview with Paul Horwitz, US EPA. 11. Interview with Stephen Seidel of US EPA. The Toronto Group consisted of the United States, Canada, and the Scandinavian countries. 12. Text of Vienna Convention—reprinted in Benedick 1991, 218–29. 13. For more on the Southern transition to participation, see Hoffmann 2005; Morrisette et al. 1991; and Downie 1995. 14. Interview with Paul Horwitz. 15. Competitiveness was the main concern of industry, but this nonetheless led them to advocate universal participation as the solution for the ozone depletion problem. 16 Such coercive strategies are well within the bounds of rational choice predictions—see Mitchell and Keilbach 2001. 17. UNEP report printed in US House of Representatives 1989, 1050. REFERENCES Arthur, Brian. 1994a. Increasing Returns and Path Dependence in the Economy. Ann Arbor: University of Michigan Press. ———. 1994b. “Inductive Reasoning and Bounded Rationality.” AEA Papers and Proceedings 84, no. 2:406–11.



Arthur, W. Brian, Steven N. Durlauf, and David A. Lane, eds. 1997. The Economy As an Evolving Complex System II. Reading, MA: Addison-Wesley. Axelrod, Robert. 1997. The Complexity of Cooperation: Agent-Based Models of Competition and Collaboration. Princeton, NJ: Princeton University Press. Benedick, Richard. 1991. Ozone Diplomacy: New Directions in Safeguarding the Planet. Cambridge, MA: Harvard University Press. Bernstein, Steven. 2001. The Compromise of Liberal Environmentalism. New York: Columbia University Press. Brunk, Gregory. 2002. “Why Do Societies Collapse? A Theory Based on SelfOrganized Criticality.” Journal of Theoretical Politics 14, no. 2 (April): 195–230. Carnevale, Mary Lu. 1988. “Du Pont Plans to Phase Out CFC Output.” Wall Street Journal, March 25. Cederman, Lars-Erik. 1997. Emergent Actors in World Politics. Princeton, NJ: Princeton University Press. ———. 2003. “Modeling the Size of Wars: From Billiard Balls to Sandpiles.” American Political Science Review 97, no. 1 (February): 135–50. Congressional Quarterly Weekly Report. 1987. “Members Hail Ozone Agreement.” September 19, 2283. Downie, David Leonard. 1995. “UNEP and the Montreal Protocol.” In International Organizations and Environmental Policy, ed. Robert Barlett, Priy A. Kurian, and Madhu Malik, 171–86. London: Greenwood Press. Epstein, Joshua. 1999. “Agent-based Models and Generative Social Science.” Complexity 4, no. 5:41–60. Epstein, Joshua, and Robert Axtell. 1996. Growing Artificial Societies: Social Science from the Bottom Up. Washington, DC: Brookings Institution Press. Gell-Mann, Murray. 1994. “Complex Adaptive Systems.” In Complexity: Metaphors, Models, and Reality, ed. George Cowan, David Pines, and David Melzer. New York: Addison-Wesley. Global Climate Change Digest. 1988. “NRDC Opposition to CFC Ruling.” Vol. 1, no. 3 (September), Haas, Peter. 1992. “Banning Chlorofluorocarbons: Efforts to Protect Stratospheric Ozone.” International Organization 46, no. 1:187–224. Hasenclever, Andreas, Peter Meyer, and Volker Rittberger. 1997. Theories of International Regimes. Cambridge: Cambridge University Press.

Koremenos. no. ed. Strategic Restraint. Charles Lipson. 2 (February): 111. “Nations Agree on Risks From CFC Use. “Companies Should Plan on Bigger Cuts in CFC Production. NJ: Princeton University Press. and Duncan Snidal. Andrew Hurrell and Benedict Kingsbury. “The Power of Interpretive Communities. Emergence: From Chaos to Order.” International Organization 55. 1988b. “Plans for Assessments Under Protocol Should Begin in Fall. “The Rational Design of Institutions. International Environment Reporter. G. NJ: Princeton University Press. ———. Johnston.” In Third Workshop on Agent-Based Simulation. 2002. 11. Cambridge: Cambridge University Press. no. 1988a. Cambridge: Perseus Books. 2005. and Volker Rittberger. 4 (April): 169. Keohane.” Vol. Krasner. no. Ozone Discourses. 1995. and the Rebuilding of Order After Major Wars. Karen. New York: Oxford University Press. Matthew J. 1989a. Jervis. Stephen. 3 (March): 106. “Entrepreneurs and the Emergence and Evolution of Social Norms. ed. Ghent. ———. Albany: SUNY Press. 4: 761–99. Holland. 1984. Belgium: SCS-European Publishing House.” Vol. 1989b. Hidden Order. Robert. After Hegemony: Cooperation and Discord in the World Political Economy. “Regime Theory and International Environmental Management. Barbara. no. Litfin. 1998. After Victory: Institutions. Ikenberry. 1997. Princeton. International Regimes. Ithaca. 32–37. John. 12. New York: Columbia University Press.” In The International Politics of the Environment. 2001. 1983. ———. “Over One-Hundred Nations Gather in London to Focus on Need for Tighter Limits on CFCs. 1994. Princeton. John. Robert.BEYOND REGIME THEORY 117 Hoffmann. 2001.” In Power and Global Governance. Consumption. no. 12. But Not on Speed of Additional Control Steps.” Vol. New York: Addison-Wesley. Princeton. 2005. Martin. ed. List. 1992. ———. Iain. 11. Christoph Urban. NY: Cornell University Press. Robert Duvall and Michael Barnett. Ozone Depletion and Climate Change: Constructing a Global Response.” Vol. NJ: Princeton University Press. Thomas Writes Tolba. System Effects. ———. 4 (April): 210–11. .

Cambridge. ed. The Third World in Global Environmental Politics. Joel Darmstadter. Risse. “Increasing Returns. eds. “Situation Structure and Institutional Design: Reciprocity. Pierson. Rajan. Robert Keohane. UNEP/WG. 1986. and Marc Levy. Mills. Geneva. 1999. 2000. and Patricia Keilbach. Peter Haas. Mike. no. MA: MIT Press. 2001. 3 (June): 209–23. no. L.. 1995. 1993. Holly. Mitchell. “But Some Say ‘Too Little Too Late’: Ratification of Ozone Pact Recommended. “Protecting the Ozone Layer. Boulder. Paul. Vienna. 4: 891–917.” Policy Studies Journal 24.167/2/Report_of_wg_on_its_ 2nd_session. Ronald. 2: 251–68. MA: MIT Press.” International Organization 55. 1996. and Exchange. Peter M. 27–74. Andrew Patinga. Draft Report of the Ad Hoc Working Group on the Work of its First Session. ———. 1991. 1988. CO: Lynne Reiner.” Congressional Quarterly Weekly Report. Cambridge: Cambridge University Press. 1987a. “The Unsheltering Sky: China. 1–5 December 1986. New York: Manchester University Press. . The Politics of Global Atmospheric Change. Global Environmental Diplomacy: Negotiating Environmental Agreements for the World. India and the Montreal Protocol. 1997. 370. Path Dependence. Rowlands. 1995. Stephen Ropp. Report of the Ad Hoc Working Group on the Work of its Second Session. Thomas. Cambridge.118 MATTHEW J. Morrisette. UNEP. Edward.” Global Environmental Change 1. HOFFMANN Miller. no. Parson. UNEP/WG. Mukund Govind. 1998. 2:201–14. and Kathryn Sikkink. Sims. “Prospects for a Global Greenhouse Gas Accord: Lessons from Other Agreements. Coercion. February 20.” American Political Science Review 94. Delhi: Oxford University Press.151/1. Ian. and Michael Toman. 23–27 February 1987.4/draft_rep_of_the_ first_sess. no. Tolba. Global Environmental Politics: India and the NorthSouth Politics of Global Environmental Issues. Marian A. The Power of Human Rights: International Norms and Domestic Change. Mostafa.” In Institutions for the Earth: Sources of Effective International Environmental Protection. 1973–1992. and the Study of Politics.

April 1987. 2002. 1987b. Governance in World Affairs.172/2/Report_of_wg_on_its_ 3rd_session. Ozone Layer Depletion. A1. 1990b.” Washington Post. 1990. Geneva. ———. International Governance: Protecting the Environment in a Stateless Society. “Administration Defends Resistance to Plan for Helping Third World Cut CFCs. 1990. ———. 4 September 1987. ———. 1999. Foreign Leaders Had Urged Reversal. . May 6. Michael. “US Drops Opposition to CFC Phaseout Fund.BEYOND REGIME THEORY 119 ———. Weisskopf.2/3. Ithaca. ———. 1994. US House of Representatives. June 16. UNEP/OzL. UNEP/cpp_list_of_par ticipants_Montreal. Business. Cambridge. 1997.R. Young. June 29. Ithaca. 1987c. 1989. Oran. MA: MIT Press. Report of the Second Meeting of the Parties to the Montreal Protocol on Substances That Deplete the Ozone Layer. Comm. Report of the Ad Hoc Working Group on the Work of Its Third Session. May 10. Interplay. UNEP/WG. The Institutional Dimensions of Environmental Change: Fit. Print 1989. MA: MIT Press. Montreal List of Participants. ———. Report of the Parties to the Montreal Protocol on the Work of Their First Meeting. 1989. and Scale.Pro.1/5.” Washington Post. Global Governance: Drawing Insights from the Environmental Experience. ———. Cambridge. 1989. 1990a. H. NY: Cornell University Press.Pro. ———. A21. NY: Cornell University Press. UNEP/Ozl.

.This page intentionally left blank.

. some five hundred thousand to eight hundred thousand Tutsi were killed by civilian militias and the Hutu-dominated army in a matter of weeks. Few refused to participate in the killing. killing Tutsi became the norm. 84). Northern Ireland. Guatemala.CHAPTER 6 Agent-Based Models in the Study of Ethnic Norms and Violence Ravi Bhavnani During the nightmarish April. and school teachers killed their pupils” (Gourevitch 1998. and the former Yugoslavia. this explanation does not to do justice to the level of participation—anywhere between two hundred thousand and five hundred thousand Hutu participated in the genocide (Des Forges 1999. While Rwanda’s culture has been described as one of fear and conformity. and June of 1994 following the assassination of President Juvénal Habyarimana. individuals with no prior disposition to engage in ethnic violence were turned into efficient killing machines. the result that “neighbors hacked neighbors to death in their homes. When confronted with the choice of killing ethnic others or being killed by members of their own ethnic group. 115). instructed to kill their Tutsi neighbors or face death at the hands of Hutu militias. or complicity with. As Gourevitch (1998. doctors killed patients. and colleagues hacked colleagues to death in their workplaces . . time and time again. Moderate Hutu who failed to participate in the violence were. and similar behavioral norms have motivated mass participation in. “Every Rwandan I spoke with seemed to have a favorite. 96) notes.1 The end result was that among the Hutu. May. Mamdani 2001)—or the vehemence with which Tutsi were massacred. unanswerable question. group violence in settings as diverse as Cambodia. leading events in Rwanda to be described as the “fastest killing spree of the 20th Century” (Power 2001. it was how so many Tutsi had allowed themselves to be 121 . For Nkongoli.

because of a congenital transhistorical condition—‘a culture of fear’ or of ‘deep conformity’—would require stretching one’s sense of credibility” (Mamdani 2001. the question was how so many Hutu has allowed themselves to kill. or why similar catalysts can lead to different outcomes.122 RAVI BHAVNANI killed. An agent-based model (ABM)—defined in terms of entities and dynamics at the microlevel—can be used to explore why such behavioral norms emerge in only some conflicts. and stands in marked contrast to violence in 1963 in which roughly thirteen thousand lives were lost. Likewise. regions. In contrast to explanations that point to a culture of conformity or highlight the importance of structural factors. and countries are not inherently peaceful or prone to interethnic violence. state capacity or penetration. For one thing. whose father was Hutu and whose mother and wife were Tutsi. there is little doubt that structural factors that pertain to the economy. in their hundreds and thousands. . It follows that an adequate explanation for mass participation by reluctant Hutu in the Rwandan genocide must address the associated issues of why they participated in the killing. the scale and duration of violence inevitably vary over time and across social contexts. as exemplified by the relatively localized and contained episode of violence in Rwanda in 1963. and why these norms can either promote interethnic violence or cooperation.2 Also explanations that emphasize the role of a particular factor or triggering event—such as the assassination of President Habyarimana—point to the correlation between the magnitude of the catalyst and the scale of violence. Rather than being driven by fear of Tutsi. and perhaps more. states. Yet the conventional preference for tracking structural factors—which either tend to remain constant or are replicated to some degree in most episodes of conflict—is overstated. how they were persuaded to participate. For Francois Xavier Nkurunziza.” The scale of violence was simply unprecedented. “To believe that ordinary Rwandans killed. but need to clarify why violence can erupt in the absence of such a catalyst. Hutu resistance to the killing was evident at both the individual and community level (Des Forges 1999). 200). my contention is that mass participation by reluctant Hutu in violence directed at Tutsi can be explained by the emergence of a violencepromoting norm among the Hutu community at large. I argue that a complexity theory with its simulation by agent-based modeling lends itself well to the study of ethnic norms—behavioral norms defined in ethnic terms that effectively persuade members of an ethnic group to participate in violence against nominal rivals. it was fear of fellow Hutu that drove the reluctant to participate in the genocide. In Rwanda. or international aid flows—to name but a few— have important implications for the nature and onset of violence. and the effect of widespread participation on the scale and duration of violence. Thus. Cities. prior levels of violence are inadequate predictors of future levels of violence. prevail in some ethnic groups but not in others. a Kigali lawyer. Rather.

he might denounce us later. . one may reasonably assume that individuals varied in their level of extremism and thus in the extent to which they harbored antipathy for nominal rivals or believed they posed a threat. This changed remarkably over the course of the genocide despite high expectations of behavioral conformity among the Hutu—expectations that were clearly and repeatedly broadcast by the extremist regime and interhamwe. and the subsequent emergence of a violence-promoting norm during the genocide in 1994. in extreme cases. OK. For instance. Could they have killed under duress— knowing that if they refused or even appeared reluctant. . They tell him. behavioral adaptation was more complete: Everyone was called to hunt the enemy. Say that guy comes with a stick. INTERACTION. “Hey. AND ADAPTATION Rwanda is a densely populated. I then provide a brief introduction to ABMs. I briefly describe the heterogeneity of Hutu beliefs and behavioral dispositions. .3 Given the variation in levels of Hutu extremism. 221) In other instances. colleagues. “No. . he does. relatives although individuals adapted to behavioral expectations in different ways (Des Forges 1999). the manner in which intragroup interaction increased behavioral conformity. reluctance to participate was manifest in inconsistent behavior: The most ambivalent stories of the genocide I heard from survivors were about Hutu who saved a friend or colleague in one place. They say. characterized by relatively frequent interactions (and intermarriage) between members of the two major ethnic groups. or. Casti 1997).AGENT-BASED MODELS IN THE STUDY OF ETHNIC NORMS 123 In the next sections of this chapter. . But let’s say someone is reluctant. RWANDA 1994: HETEROGENEITY. Everyone must help to kill . The end result was that large numbers of Hutu killed Tutsi who were acquaintances. complicity with the state’s genocidal agenda was initially low (Des Forges 1999). and equation-based approaches. only to go and join the killing in another. but he doesn’t kill. As a result. . get a masu. game-theoretic. I conclude the chapter with a comparison of agent-based. He must kill. and describe how ABMs may be used to study processes of norm formation and change within ethnic groups. discuss my preference for an exploratory as opposed to a consolidative modeling approach (Bankes 1994.” So. and he runs along with the rest. heterogeneous society. they would surely be killed—and saved a life when the opportunity presented itself ? (Mamdani 2001.

The threat of sanctions therefore extended the violence over a wider range of targets. These expectations were underscored in no small measure by Radio Milles Collines. and observe and punish their behavior. and the message that all Hutu should either kill Tutsi or be killed (Des Forges 1999. Whereas some Hutu may have willingly redressed their grievances against particular Tutsi. and Radio Rwanda broadcasts of the notorious ten commandments which included calls for Hutu unity in the face of the common Tutsi enemy. And the next day it’s become a game for him. When a population mixes randomly. An expla- . Prunier 1995). in turn.124 RAVI BHAVNANI at least one person. 24) Behavioral expectations were initially formed at the local (neighborhood. who bore grudges against them. If a person was let go by one party. calls for all Hutu to stop being merciful and undertake umuganda—a reference to customary work symbolizing the killing of Tutsi. observes. quickly update their behavior to conform to group practice. to capture the heterogeneity and adaptation of individual preferences. and in so doing reinforce this behavior. (Gourevitch 1998. no Hutu (moderate or otherwise) could be sure of whom they may encounter or trust. HutuTutsi) during the genocide. It is the process an ethnic group goes through in the transition from conflict to violence that determines whether an episode progresses beyond the common occurrence of low-level violence to more isolated instances of high-scale violence. escaping became more difficult for members of the targeted group—moderate Hutu and all Tutsi. the causes of conflict constitute partial explanations at best. or coveted their property. For instance. Alternatively stated. sought favors. As a result. Many Hutu were turned in by relatives or neighbors. village or town) level.” So this person who is not a killer is made to do it. RTLM. it does underscore the importance of studying violence from the bottom up—to specify behavior at the level of the individual. community. While this brief discussion does not do justice to the intricacies or complexities of the 1994 Rwandan genocide. under these conditions they were pressed to be indiscriminate. and updated as roving bands of interhamwe motivated communities that were less than zealous in their willingness to massacre the local Tutsi population. You don’t need to keep pushing him. and are inextricably linked to the process by which conflict unfolds. one key element that stands out is the random nature of encounters. and punishes whom and how often. In the undeniably complex pattern of interaction (Hutu-Hutu. he or she would be caught and killed by another. and to assess the importance of group networks that determine who interacts with. Moderates. extremists eventually have the opportunity to interact with moderates.

albeit small. revenge (“tit-for-tat” behavior). who notes that “fear could silence opposition. but it could not generate enthusiasm (for killing) . peasant or professional. does one go about explaining the participation of hundreds of thousands of Hutu? Why were similar levels of participation not evidenced in previous episodes of Hutu-Tutsi violence? The explanation I advance is rooted in the concept of an ethnic norm comprised of two basic components: agreed-upon behavior and punishment for deviation from this behavior (Axelrod 1986). as they grew in scope.4 These schemas may emphasize obedience (an inherent rationale that orders are to be followed. Each of these schemas may transform or undermine traditional constraints on violence or (in the case of cooperation) introduce constraints where none previously existed. patronage or work (a duty with a quid pro quo of material rewards). with an allowance. a norm thus defined either compels or dissuades members of a group from engaging in violence against ethnic rivals. paranoia (fear of being killed by an enemy if the enemy is not killed first). By implication. to be capable of explaining phenomena at the aggregate or group level. seek revenge for prior acts of violence. detachment (cutting off one’s feelings for an “enemy”). Individuals may follow their own convictions or beliefs. . How. In the context of interethnic rivalry. who refused to join in the . follow or go with the crowd. cultural models—tacit knowledge structures or schemas that are both widely shared by members of a social group and induce participation by a large number of individuals—may be used by “ethnic entrepreneurs” as motivating templates for interethnic violence or cooperation.AGENT-BASED MODELS IN THE STUDY OF ETHNIC NORMS 125 nation that affords due attention to process therefore needs to be specified at the level of the individual. Individual members of an ethnic group are motivated to participate in or oppose violence against nominal ethnic rivals for a variety of reasons. Where individual motives are absent or insignificant. the strength of a norm dictates how typical this behavior is of the group as a whole. or simply derive entertainment value from participation.5 Yet while group consciousness is necessary to ensure collective action. . for disobeying orders that are contrary to any logic). it often proves insufficient when the individual costs of participation are high— a point that is underscored by Mamdani. and to incorporate feedback mechanisms from the aggregate back to the individual level. or religious duty (such as a jihad or the building of a Hindu Rashtra). the massacres targeted anyone. honor (saving face). ETHNIC NORMS As baffling as the scale of violence in Rwanda is the number of those purported to have participated in the killing. then.

” The environmental entities in a model usually have their own dynamics. or simply absent—the severity of punishment directly affects the strength of the norm. and enforce each other’s behavior). Given that the existence of an ethnic norm is a matter of degree—norms may be strong. ABM are comprised of one or more types of agents. The nonagent environment can encompass any variables external to the agents that are relevant to behavior. resulting in behavioral change at the group or system level. Each agent’s behavior affects other agents as well as the nonagent environment. The profile or “state” of an agent can include various characteristics and preferences. It follows that while punishment. is specified in terms of various entities or dimensions.126 RAVI BHAVNANI melee” (2001.6 To capture the emergent properties of norm creation and change. and leave less room for individual choice. Agents may also possess adaptive mechanisms (learning or evolutionary) that lead them to change their heuristics based on their own experience. describing how they change over time independent of agent behavior. weak. as well as particular social connections (i. Norms. and social structure all influence norm formation and change. or . it is difficult to attribute the emergence of a norm to any specific factor. I turn to complexity theory and its simulation by ABM. each with an associated “state. 219). In addition. memberships. the emergence of ethnic norms depends critically upon the composition of an ethnic group (the number of those who support or oppose a particular course of action and their influence within the group) as well as the speed with which behavioral expectations are transmitted to coethnics (how frequently individuals observe. as well as a nonagent environment in which the agents are embedded. influence. on the other hand. networks) and a memory of recent interactions and events. the mobilization of ethnic rivals.e.. An environment. since norms are an emergent property of social systems and their existence depends upon complex patterns of interaction. and internalization among individuals. AGENT-BASED MODELS Generally. identities. they could represent the effects of shocks or “triggers” such as sudden economic collapse.7 In addition to individual characteristics. agents are defined by their decision-making heuristics and capabilities to act in response to inputs from other agents and from the environment. therefore. In addition to punishments. group composition. ranging from physical features such as geography or topography to things comprising states of the world like political. clearly prescribe appropriate or expected behavior. interact with. economic and social conditions. These variations can reflect natural progressions (or regressions) according to logical rules and also involve uncertainty or noise.

One then runs the program to simulate the behavior of the agents and the dynamics of the environment. what to do when they interact. and how to interact with the environment. The output from model simulations consists of both the microlevel behavior of agents and changes in the environment. the simulation can be run hundreds or thousands of times—with various tracking measures or outcome variables summarized across runs—to study the variations in and sensitivity of results. When an ABM is simulated on a computer. examples of a consolidative approach include complex multiagent models that use agent architectures based on “naturalistic decision-making. usually involves the development of “model” systems that represent “real-world” systems with easily measurable physical characteristics and components. but often the outputs are quite detailed as well. decision rules. Ideally.AGENT-BASED MODELS IN THE STUDY OF ETHNIC NORMS 127 a military invasion. is largely inappropriate for studying violent conflict. the environment may also adapt in response to agent behavior. These models often require exhaustive inputs. MODELING THE EMERGENCE OF ETHNIC NORMS Constructing an ABM to explore the dynamics of norm formation and change within an ethnic group is part art and part science—the exact proportions of which can vary greatly. as well as the emergent macrolevel structures. an exploratory model can serve as an experimental device to examine how members of an ethnic group might behave under a variety of assumptions. on the other hand.” The consolidative approach. The existence of interlaced feedback relationships—agent-agent and agent-environment interaction—leads to the nonlinear. adaptive mechanisms. For instance. while stopping short of offering precise and detailed forecasts of how they will act given a particular set of circumstances. relationships. and dynamics that result from the aggregation of this microlevel activity. which are then processed with computer programs that can run to millions of lines of code. and the agent-interaction topology all require decisions on the . this large amount of data can be transformed into a useful or manageable form. In principle. Besides following its own dynamic rules. The specification of agent attributes. ABMs are well suited for studying dynamic processes—such as emergence and spread of ethnic norms—that are sensitive to both historical contingencies and situational factors. agent behavior is generated as agents determine which other agents to interact with.8 Consolidative modeling. which is characterized by significant information uncertainties and practical barriers to experimental validation. however. path-dependent dynamics that are characteristic of complex systems. The model’s dynamics are studied by implementing the agents and the environment as a computer program. Although most often applied in settings where each component has clear physical properties.

“hypocrites” who punish others for their failure to conform to behavioral expectations but personally shirk. and Prahl 1988. Finally. agents who simply act upon their own preferences (to engage in or refrain from violence) without punishing others. and discuss how such a model could be applied to analyze the conditions under which mass participation in ethnic violence occurs. punishment often is used to bring individual behavior into conformity with group practice and who talks to. agents may have different degrees of in-group influence. McAdam and Paulsen 1993. Watts and Strogatz 1998). family members. Marwell. Tilly 1978. Oliver. intragroup punishments can vary in severity from killing coethnics. Kandori 1992. As a result. Most ethnic groups face collective action problems and these problems are likely to be more pronounced when the costs of participation or compliance are high. observes. the model may include agents who vary in terms of their out-group extremism (their disposition to engage in or oppose violence against nominal ethnic rivals) and their level of in-group tolerance (their propensity to punish coethnics for failing to adhere to their own. interacts with. As noted earlier. while leaving out much that is important (as well as much that is peripheral). agents may vary in their levels of in-group extremism (the strength of punishments administered to coethnics). one may have agents who both engage in (or refrain from) violence and punish those who fail to engage in (to refrain from) violence. or externally defined. behavioral standards). The precise mix of punishment (and reward). In addition.9 In this section. I provide a high-level description of a simple ABM to study the emergence of ethnic norms. that is. or ultimately sanctions whom is of critical importance. Such a model also makes it possible to embed agents in a set of social relationships to determine interaction patterns within the group. and relatives. and public humiliation to the destruction of personal property and the loss of one’s status within the group. as well as agents who remain neutral in the face of individual or group pressure. variation in the strength of these inducements. threats to individuals. As a result.10 Networks that connect members of an ethnic group to one another therefore become instrumental in determining the success . As a result.128 RAVI BHAVNANI part of the modeler to include essential components and mechanisms that capture the problem at hand. Opp and Gern 1993. and distribution of agents willing to apply these measures may all be designed to capture the specific mechanisms used by groups to induce collective compliance (Oliver 1984. what effectively distinguishes ethnic entrepreneurs or leaders from other agents. one is principally interested in how and how often “like-minded” individuals observe and sanction the behavior of individuals with contrasting or opposing views (Granovetter 1976. Posner and Rasmusen 1999). At the most basic level.

agents may update their behavior to conform to the expectations of other agents they encounter. resulting in more or less conformity within the group. assume some correlation between out-group extremism and in-group tolerance. past interaction. the interaction topology. under what conditions social structure affects the emergence and maintenance of norms. network structure. By explicitly specifying the structure of a social network—designed to capture patterns of interaction at the level of local communities. behavior or type of the punishing agent. Finally. the perceived punishment for deviating from group behavior. One way to capture behavioral change at the level of the individual is to specify a set of update rules that determine the degree of behavior modification based upon factors such as the individual agent’s current behavior. and punishment regimes to norm formation and change. The agent-based model outlined above generally is implemented as a computer program in which agents interact with and influence each other. the model outlined here may be used to generate hypotheses linking group composition.11 It follows that individual agents may adapt their behavior in a variety of ways depending on the nature of the encounter. or whether norms can emerge in the absence of punishment. and prevailing local or global behavior. one may choose to specify agent characteristics independently. . Such a model provides the user with a great deal of flexibility. based on factors such as spatial proximity. the information environment. or introduce a high degree of correlation across all agent characteristics. making it possible to alter the composition of an ethnic group to represent domination by extremists or pacifists. by specifying measures of aggregate behavior to capture the emergence of ethnic norms. or to adhere more closely to perceptions of local behavior. One also may “grow” such networks endogenously.AGENT-BASED MODELS IN THE STUDY OF ETHNIC NORMS 129 of efforts designed to achieve collective compliance and a variety of network types—ranging from random to small world networks in which “ethnic entrepreneurs” have a disproportionate level of influence on the group—can be specified. One may also alter individual update rules. Alternatively. strength of punishment administered. or larger spatial configurations—it becomes possible to assess the extent to which network structure hinders or promotes behavioral conformity. Spatial relationships can bias which agents are more likely to interact with other agents. or to behavioral expectations broadcast to the group as a whole. as well as in-group tolerance and in-group extremism. For instance. One may therefore seek to determine whether norms are equally likely to form in groups with similar aggregate preferences. or similar (or different) preferences. and the number of agents with a disproportionate level of influence on group members. or by highly contentious or largely apathetic groups. and the structure of social relations or networks that characterize the group. neighborhoods.

I conclude this chapter with a rationale for the use of ABM. with ABMs it is relatively easy to embed agents in both physical and social spaces in the same model. This stands in marked contrast to EBMs. with ABMs one can accommodate myriad differences in agent and environmental characteristics. ABMs are distinct in that they are constructed in a “bottom-up” manner—specified at the level of individual agents and their interactions with each other and the environment. Most EBMs instead employ a mean-field approach to describing trajectories and variances. once again for reasons of analytical tractability.130 RAVI BHAVNANI MODELING CHOICES Because other formal approaches—namely. ABMs are better suited to modeling complex adaptive systems than EBMs (Parunak et al.” Second. where space may be built into partial differential equations. as well as processes of change and adaptation within the same model. agents can move in a two-dimensional spatial topology with the specific structure of this topology designed to reflect the social context being modeled and bias agent interaction—who interacts with whom and how often. precisely because the heterogeneity and adaptivity of agents lead to sensitive. or they may capture variability over time (ordinary differential equations as used in systems dynamics). equation-based and game-theoretic models—are available to study processes of norm formation and change. For several reasons. These equations may be algebraic. path-dependent dynamics that are not adequately captured by the mean trajectory or even by a simple distribution over such trajectories. EBMs generally focus on macrolevel entities and relationships. because equations at this level are easier to handle analytically and aggregate variables are among the few observables that are consistently available. Thus. 1998). . but where concerns about tractability make it difficult explicitly to include social topologies and how they bias interactions between agents. By contrast. For example. Equation-Based Models An equation-based model (EBM) consists of a set of equations that usually express relations among observables. The focus on expected trajectories can be misleading. each exhibiting nonlinear rules of behavior and adaptive processes of various kinds. ABM vs. The evaluation of these equations produces the evolution of the observables over time. First. whereas EBMs are disposed to treat their actions as a “black box. ABMs make it possible to represent heterogeneous agents. Third. or over time and space (partial differential equations). Thus. ABMs are capable of providing insight into the behavior of individual agents.

partly out of the overriding concern for being able to derive equilibrium solutions. they still remain equilibrium-centered. GTMs typically assume that systems go to equilibria as limiting states. can capture both forms of heterogeneity within the same model. ABMs offer some distinct advantages. Scholars have attempted to capture the richness of interaction of real-world problems by increasing the number of players. It follows that it is difficult for the players in GTMs to change their defining characteristics over time. Given the nature of the social problem. GTMs share the capacity to specify interactions in terms of individual agents with particular sets of preferences and to evaluate their responses to different conditions. environmental differences can only be obtained by specifying many separate models. Game-theoretic Models Likewise. GTMs do not lend themselves well to studying agent and environmental heterogeneity. yielding certain nonlinear results. While these refinements have led to a number of interesting insights. at the most. Moreover. and implement dynamic changes—so that these model components are also heterogeneous over time. introducing the option to “exit” and adding noise. each with their own particular structure. Third. First. Meanwhile. enable greater scope for variation. however. GTMs therefore lend themselves less readily to studying dynamic processes—such as norm formation and change—that emerge over time and are sensitive to both historical contingencies and situational factors. Second. most applications of GTMs employ two or. three types of agents. unlike the agents in . permitting nonsimultaneous play. and working on finite repetition (Axelrod and Dion 1988). where the stability of outcomes is determined by institutional arrangements such as the rules of jurisdiction and amendment control and thus by the structure of the game itself. by contrast. but only with multiplayer games that quickly become intractable. and they do not focus on processes that unfold over time. “weak” or “strong” states in deterrence games. and player “types” typically tend to be fixed—for example.AGENT-BASED MODELS IN THE STUDY OF ETHNIC NORMS 131 while interacting with a variety of other agents selected as a result of spatial and social interaction topologies. In fact. This flexibility is perhaps the fundamental reason why ABMs are capable of replicating phenomena commonly exhibited by complex social systems. altering the payoff matrix. an ABM is well suited as a methodology for studying emergent processes such as norm formation and change relative to a game-theoretic model (GTM). in GTMs the characteristics of players are typically determined exogenously. It is feasible to specify larger numbers of agents with varying characteristics. the solutions generally fall within the class of structure-induced equilibria. ABM. ABM vs.



ABMs, who can alter their preferences and even their traits in response to new environmental conditions and via other processes of adaptation. Thus, ABMs enjoy an advantage in this instance to GTMs, because they address the dynamic nature of the social problem, accommodate requisite heterogeneity of agents and environments, and build in the ability of agents to change in response to the conditions, including group norms, they encounter and the experiences they undergo. CONCLUSION In a seminal article, Gould (1999) takes issue with the assumption that group interests either implicitly lead to group action or explicitly stem from group conflict and result in group violence. Ethnic groups—more often than not—are confronted with the problem of retaining the commitment of moderate members as interethnic rivalry progresses from conflict to violence, and extreme members as rivalry progresses from conflict to cooperation. To explain how groups resolve these dilemmas, the approach I advocate in this chapter explores the emergence of a behavioral norm defined in ethnic terms—a macrolevel outcome—by focusing on microlevel dynamics. ABMs, in particular, lend themselves well to the study of emergent phenomena such as norm formation and change within a group. By repeating simulations and observing trajectories of participation, one can learn about outcomes associated with different initial conditions as well as about processes of norm formation and change, which supports the notion that where the system ends is only as important as how it gets there, if not less important. ABMs also afford an intuitively satisfying representation of real-world political situations. Most models we construct “in our heads” involve individuals interacting with each other and some environment. This accessibility of ABMs facilitates construction of “what if”–type experiments that are critical for policy analysis, and yields output that may readily be translated back into policy recommendations and practice. NOTES Acknowledgments: I am grateful to Robert Axelrod, David Backer, Pradeep Chhibber, Neil Harrison, Ken Kollman, Scott Page, and Rick Riolo for their comments on previous drafts. I would also like to acknowledge generous support from the Center for the Study of Complex Systems at the University of Michigan for this work. All faults remain my own. 1. One way to understand this phenomenon is to assume that internal models are dominated by the survival instinct when individuals are faced with the threat of death.



2. Bhavnani and Backer (2000) specify a computational model to explain variation in the scale and duration of ethnic violence using data from Rwanda and Burundi. 3. Strauss (2004) argues that the motivation to participate in the violence is likely to have been heterogeneous, and that several theories are probably right though he finds strong support for the argument that intra-Hutu coercion is more likely to explain participation by “less violent” individuals, whereas fear or anger are more likely to explain participation by “more violent” individuals. 4. Cultural models can concern things other than the ethnic norm surrounding issues of violence, as certain rationales of conduct can be invoked in support of actions other than aggressiveness or moderation. Obedience, detachment, patronage, and honor all have meanings independent of issues of violence, as does arguably a jihad, rashtra, or the equivalent. Paranoia and revenge are harder to separate. For approaches that use cultural models in explaining ethnic conflict, see Das 1996, Engineer 1995, and Kakar 1996. 5. Hinton (1998) illustrates how the Khmer Rouge used indoctrination to reinforce a cultural model of detachment—“cutting off one’s feelings or heart”—toward an enemy who moments earlier might have been a friend or family member. 6. In its starkest form, punishment involves the killing of coethnics who refuse to adapt their behavior. Less-severe punishments include the destruction of personal property; threats to the individual or his or her family; public humiliation; loss of status, honor, or reputation within the group; ostracism from the group; or bodily harm. Where punishments are subtle and executed discreetly, “consciousness-raising” meetings, speeches, pronouncements, songs, slogans, or chants may be used to call for a specific action or set of actions to be taken against ethnic rivals. Such pronouncements are often couched in terms of a communal, religious, or national duty, are accompanied by calls for ethnic solidarity or unity, and invoke a “moral” obligation on the doer to perform the stated task or assignment. Pronouncements may also carry thinly veiled threats directed at more moderate (or extreme) members of the ethnic group, equating transgression with sympathy for or even identification with ethnic rivals—thereby making transgressors fair game for the very behavior they disdain. In many instances, enforcement costs may be high but positive for individuals who obtain some payoff from the enforcement of a norm. To simplify matters, one may assume that the strength of the punishment captures the enforcement cost borne by the punisher. Where enforcement costs are high, punishments are more likely to be weak. Low enforcement costs, on the other hand, are more likely to give rise to strong punishments. 7. Ancient history often provides important symbols and myths through which to interpret current events. For instance, the Balkan conflict was, at least in



part, fueled by traditional identities of the “other” that were grounded in the fifteenth century. Such symbolic histories are a part of learning—and, thus, a source of behavior—as much as other ideas and beliefs and personal daily experience. 8. Barring clear instances of ethnic norms—as in Rwanda, where punishments were widespread and in evidence, given the desired aim of mobilizing the entire Hutu population to kill the entire Tutsi population—the fact that such pernicious norms operate within ethnic groups exacerbates the difficulty of measuring them. This is compounded by the fact that individuals are often reluctant to divulge punishment for fear of further reprisal. 9. My discussion of ABM to study ethnic norms is not intended to be mechanistic and or to minimize the gravity of interethnic violence. Given that my unit of analysis is the individual, and that my primary concern is to understand and model individual participation in violence against ethnic rivals, the model is specified mainly in terms of individual characteristics, heuristics, and behavior. My framework therefore stands in marked contrast to aggregate studies of war in international relations, where individual motivation and participation are, more often than not, filtered out of the analysis and more attention is devoted to aggregate outcomes (conflict, deterrence, resolution). 10. Social movement theorists do regard networks as important for recruiting participants for protest or rebellion. Despite their prominence in this literature, social networks have received limited attention in the context of ethnic violence. For instance, Brass (1997) notes that all riot-prone towns do have—to a greater or lesser degree—informal organizational networks that serve to mobilize members. He does not, however, distinguish between different types of networks. Likewise, Varshney (2002) bases his argument on the existence of interethnic networks that promote civic engagement and reduce conflict, but does not specify the structure of these networks—how these networks may differ across contexts. 11. This also speaks to the difference between bonding (intragroup) and bridging (intergroup) social capital.

REFERENCES Axelrod, R. 1986. “An Evolutionary Approach to Norms,” American Political Science Review 80, no. 4:1095–111. Axelrod, R., and D. Dion. 1988. “The Further Evolution of Cooperation.” Science 242:1385–90. Bankes, S. 1994. “Exploratory Modeling for Policy Analysis.” Operations Research 41, no. 3:435–49.



Bhavnani, R., and D. Backer. 2000. “Localized Ethnic Conflict and Genocide in Rwanda and Burundi.” Journal of Conflict Resolution 44, no. 3:283–307. Brass, P. 1997. Theft of an Idol: Text and Context in the Representation of Collective Violence. Princeton, NJ: Princeton University Press. Casti, J. 1997. Would-Be Worlds: How Simulation Is Changing the Frontiers of Science. Hoboken, NJ: John Wiley & Sons. Das, V. 1996. Mirrors of Violence. London: Oxford University Press. Des Forges, A. 1999. Leave None to Tell the Story: Genocide in Rwanda. New York: Human Rights Watch. Elster, J. 1989. “Social Norms and Economic Theory.” Journal of Economic Perspectives 3, no.4:99–117. Engineer, A. 1995. Lifting the Veil: Communal Violence and Communal Harmony in Contemporary India. Hyderabad: Sangam Books. Gould, R. 1999. “Collective Violence and Group Solidarity: Evidence from a Feuding Society.” American Sociological Review 64, no. 3:356–80. Gourevitch, P. 1998. We Want to Inform You That Tomorrow We Will Be Killed with Our Families: Stories from Rwanda. New York: Farrar, Straus and Giroux. Granovetter, M. 1976. “Network Sampling: Some First Steps.” American Journal of Sociology 81, no. 6:1287–1303. Hinton, A. 1998. “‘Why Did You Kill?’ The Cambodian Genocide and the Dark Side of Face and Honor.” Journal of Asian Studies 57, no. 1:93–122. Kakar, S. 1996. The Colors of Violence: Cultural Identities, Religion and Conflict. Chicago: University of Chicago Press. Kandori, M. 1992. “Social Norms and Community Enforcement,” Review of Economic Studies 59, no. 1:63–80. Mamdani, M. 2001. When Victims Become Killers: Colonialism, Nativism, and the Genocide in Rwanda. Princeton, NJ: Princeton University Press. Marwell, G., P. Oliver, and R. Prahl. 1988. “Social Networks and Collective Action: A Theory of Critical Mass. III.” American Journal of Sociology 94, no. 3:502–34. McAdam, D., and R. Paulsen. 1993. “Specifying the Ties Between Social Ties and Activism.” American Journal of Sociology 99, no. 3: 640–67. Oliver, P. 1984. “Rewards and Punishments as Selective Incentives: An Apex Game.” Journal of Conflict Resolution 28, no. 1 (March): 123–48.

New York: Columbia University Press. University of California.” Nature 393:440–42. Veerhagen. 1978. Strogatz. Power. MABS’98. “Dissident Groups. Posner. EquationBased Modeling: A Case Study and Users Guide.” British Journal of Social Psychology 21 (2): 139–49. Watts. Opp. E. and R. Savit. 1993. “Collective Dynamics of Small World Networks. Ullmann-Margalit. and S. 3:296–306.. no. 2001. “Bystanders to Genocide: Why the United States Let the Rwandan Tragedy Happen. From Mobilization to Revolution. Sherif. MA: Addison-Wesley. Rasmusen. no. 2004.” Doctoral diss. “Creating and Enforcing Norms. 5:659–80.. no. G. and Spontaneous Cooperation: The East German Revolution of 1989.. H. 2001. 1936. and War in Rwanda. Parunak.. Oxford: Oxford University Press.” International Review of Law and Economics 19. and E. 1999. S. A. “Simulation of the Learning of Norms. Berkeley. K. R. New York: Harper. New Haven: Yale University Press. D. S. Prunier. R. “The Evolutionary Emergence of Norms. 1982. V. Gern. 3:369–82.” Proceedings of Workshop on Multi-Agent systems and Agent-Based Simulation.. 1998. with Special Reference to Sanctions. 1998. Riolo. 2002.136 RAVI BHAVNANI Opp.” American Sociological Review 58.” Social Science Computer Review 19. 1995. and C. “The Order of Genocide: Race. Ethnic Conflict and Civic Life: Hindus and Muslims in India. “Agent-Based Modeling vs. Varshney. Strauss. M. Reading. The Psychology of Social Norms. Personal Networks. C. Department of Political Science. . Power. The Emergence of Norms. K. 1977. Tilly. The Rwandan Crisis: History of a Genocide.” Atlantic Monthly September:84–108.

While this definition is useful. and discovery. process them by taking hypothesized mechanisms into account. Prediction. Simulation can also be used to perform certain tasks. the artificial intelligence method can be thought of as simulation of human perception. simulations of task environments can also help design new techniques. entertainment. or social interaction. and Schrage 1987. performance. This is typically the domain of artificial intelligence. An important example of the use of simulation for training is flight simulators for pilots. proof. and function optimization. To the extent that the artificial intelligence techniques mimic the way humans deal with these same tasks. These purposes include: prediction. education. Tasks to be performed include medical diagnosis. and then generate their consequences as predictions. Fox. if the goal is to predict interest rates in the economy three months into the future. speech recognition. ix). For example. Simulation is able to take complicated inputs. training. Training. 137 . Performance. To the extent that the artificial intelligence techniques exploit the special strengths of digital computers. it does not suggest the diverse purposes to which simulation can be put.CHAPTER 7 Alternative Uses of Simulation Robert Axelrod Let us begin with a definition of simulation. decision-making. simulation can be the best available technique. Many of the earliest and most successful simulation systems were designed to train people by providing a reasonably accurate and dynamic interactive representation of a given environment.” (Bratley. “Simulation means driving a model of a system with suitable inputs and observing the corresponding outputs.

1978) simulation of residential tipping provides a good example of a simple model that provides an important insight into a general process.1 For example. The main use of simulation in education is to allow the users to learn relationships and principles for themselves. the easier it may be to discover and understand the subtle effects of its hypothesized mechanisms. even highly complicated simulation models can rarely prove completely accurate. Using simulation for prediction can help validate or improve the model upon which the simulation is based. From training and entertainment it is only another small step to the use of simulation for education. The result is that very segregated neighborhoods form. the simpler the model.g. Education. in the social sciences induction is widely used in the analysis of opinion surveys and the macroeconomic data. As a scientific methodology. social scientists have been quite successful in using simulation to discover important relationships and principles from very simple models. For example. SimCity is an interactive simulation allowing the user to experiment with a hypothetical city by changing many variables.138 ROBERT AXELROD Entertainment. Schelling’s (1974. For educational purposes. But the use of simulation for the discovery of new relationships and principles is at least as important as proof or prediction. Deduction. it pays to think of it as a new way of conducting scientific research. Induction is the discovery of patterns in empirical data. and discovery. The model assumes that a family will move only if more than one-third of its immediate neighbors are of a different type (e. in particular. but social scientists are not as successful in accurately simulating the movement of workers or armies. proof. Nevertheless. it is only a small step to entertainment. Physicists have accurate simulations of the motion of electrons and planets. Simulation as a way of doing science can be contrasted with the two standard methods of induction and deduction. a simulation need not be rich enough to suggest a complete real or imaginary world. simulation’s value lies principally in prediction. race or ethnicity).. involves specifying a set of axioms and proving consequences that can be derived from . on the other hand. Proof. Indeed. From training. So are simulations of completely imaginary worlds. To appreciate the value of simulation as a research methodology. Flight simulations on personal computers are fun. A good example is the computer game SimCity. In the social sciences. Discovery. as discussed below. Conway’s Game of Life (Poundstone 1985) demonstrates that extremely complex behavior can result from very simple rules. such as tax rates and zoning policy. Simulation can be used to provide an existence proof. even though everyone is initially placed at random and everyone is somewhat tolerant. Prediction is the use that most people think of when they consider simulation as a scientific technique.

Good examples include the neoclassical economic models in which rational agents operating under powerful assumptions about the availability of information and the capability to optimize can achieve an efficient reallocation of resources among themselves through costless trading. Simulation is a way of doing thought experiments. the consequences may not be at all obvious. the dominant form of modeling is based upon the rational choice paradigm. The real advantage of the rational choice assumption is that it often allows deduction. simulation is often the only viable way to study populations of agents who are adaptive rather than fully rational. deducing the consequences is often impossible. or it may be at the population level through differential survival and reproduction of the more successful individuals. Throughout the social sciences today.2 There are some models. however. The large-scale effects of locally interacting agents are called “emergent properties” of the system. since its unrealistic assumptions undermine much of its value as a basis for advice. Game theory. The adaptation may be at the individual level through learning. Simulation is a third way of doing science. The discovery of equilibrium results in game theory using rational choice axioms is a good example of deduction. in particular. One of the main advantages of simulation is that it allows the analysis of adaptive as well as rational agents. Like deduction. it starts with a set of explicit assumptions. But unlike deduction. the consequences of adaptive processes are often very hard to deduce when there are many interacting agents following rules that have nonlinear effects.ALTERNATIVE USES OF SIMULATION 139 those assumptions. While people may try to be rational. Either way. In my view. . in which emergent properties can be formally deduced. Instead. March 1978). Thus. however. it does not prove theorems. simulation becomes necessary. Emergent properties are often surprising. and deduction can be used to find consequences of assumptions. they can rarely meet the requirement of information or foresight that rational models impose (Simon 1955. The main alternative to the assumption of rational choice is some form of adaptive behavior. simulation modeling can be used as an aid to intuition. because it can be hard to anticipate the full consequences of even simple forms of interaction. the simulated data comes from a rigorously specified set of rules rather than direct measurement of the real world. While induction can be used to find patterns in data. the reason for the dominance of the rational choice approach is not that scholars think it is realistic. Nor is game theory used solely because it offers good advice to a decisionmaker. a simulation generates data that can be analyzed inductively. While the assumptions may be simple. But when the agents use adaptive rather than optimizing strategies. is typically based upon the assumption of rational choice. Unlike typical induction.

accuracy is important and simplicity of the model is not. As pointed out earlier.” This type of simulation is characterized by the existence of many agents who interact with each other with little or no central direction. or to develop tactics for a new fighter aircraft. Instead. rather than “top-down” direction. 21–40. Likewise. The emergent properties of an agent-based model are then the result of “bottom-up” processes. A simulation of the economy aimed at predicting interest rates three months into the future needs to be as accurate as possible. 1997). When a surprising result occurs. but this raises the question: surprising to whom? . This requires adhering to the KISS principle. The point is that while the topic being investigated may be complicated. stupid. there are other uses of computer simulation in which the faithful reproduction of a particular setting is important. the goal of agent-based modeling is to enrich our understanding of fundamental processes that may appear in a variety of applications.” The KISS principle is vital because of the character of the research community.” in Simulating Social Phenomena. Simplicity is also helpful in giving other researchers a realistic chance of extending one’s model in new directions.140 ROBERT AXELROD An important type of simulation in the social sciences is “agent-based modeling. Induction as a search for patterns in data should not be confused with mathematical induction. it is very helpful to be confident that one can understand everything that went into the model. which stands for the army slogan “Keep it simple. the assumptions underlying the agent-based model should be simple. and Pietro Terna (Berlin: Springer. Rainer Hegselmann. For this purpose. “Advancing the Art of Simulation in the Social Sciences. which is a technique for proving theorems. then simplicity of the assumptions is important and realistic representation of all the details of a particular setting is not. 2. the assumptions that go into the model may need to be quite complicated. Some complexity theorists consider surprise to be part of the definition of emergence. and is used with permission. Rosario Conte. ed. Although agent-based modeling employs simulation. not in the assumptions of the model. NOTES This chapter is excerpted from Robert Axelrod. if a simulation is used to train the crew of a supertanker. it does not necessarily aim to provide an accurate representation of a particular empirical application. But if the goal is to deepen our understanding of some fundamental process. Both the researcher and the audience have limited cognitive ability. The complexity of agent-based modeling should be in the simulated results. 1.

.” Quarterly Journal of Economics 69:99–118. R.) Simon.ALTERNATIVE USES OF SIMULATION 141 REFERENCES Bratley. 1963. ed. Ambiguity and the Engineering of Choice. H. . Fox. The Recursive Universe. Robert Morris. Schelling. “Exploration and Exploitation in Organizational Learning. A Behavioral Theory of the Firm. Schrage. “Bounded Rationality. 1985. Cyert. and L. ———. A. (See especially 137–55. NJ: Prentice-Hall. 1974. 19–64 (see especially 43–54). B. W.” In The Corporate Society. March. A Guide to Simulation. “On the Ecology of Micromotives.. New York: Springer-Verlag. Poundstone. Norton. New York: Wiley. and J.” Organizational Science 2:71–87. 1978. G. March. 1955. 2nd ed. ———. 1987. W. G. Chicago: Contemporary Books. 1978.” Bell Journal of Economics 9:587–608. “A Behavioral Model of Rational Choice. New York: W. J. T. 1991. P. Englewood Cliffs. Micromotives and Macrobehavior.

This page intentionally left blank. .

furthermore. adaptability. 1995. and adaptation. scene 5 So laments Shakespeare’s tragic protagonist at the news of his wife’s death. While one may forgive the Scottish king for his pessimistic metaphor. Langton 1994. 547. 1991. Earnest James N. Kauffman 1995. learning. quoted in 143 . argues that human societies are “superorganisms” and global politics are becoming a “super-superorganism. Corning (2002). So perhaps it is no surprise that the complexity sciences—explicitly concerned with these properties of a variety of systems. a poor player That struts and frets his hour on the stage And then is heard no more. for one. Corning notes that the biologist Edward O.” Johnson’s “myth of the ant queen” (2001) explicitly postulates that human cities organize themselves much like colonies of insects do: without the centralized authority of an “ant queen. —The Tragedy of Macbeth. and unpredictability of organisms and ecological systems. Wilson argues “the humanities and social sciences shrink to specialized branches of biology” (Wilson 1975. full of sound and fury. Langton and Shimohara 1997). It is unremarkable. from physical to social—not only invoke the metaphor of life but also have postulated the idea of “artificial” life (Langton et al.” Smith and Stevens (1997) reduce social organization to the attachment behavior regulated by the brain’s cyclic production of neurotransmitters known as “opioids.CHAPTER 8 Signifying Nothing? What Complex Systems Theory Can and Cannot Tell Us about Global Politics David C.” More provocatively. Signifying nothing. “life” for most of us connotes roseate meanings: dynamism. act 5. It is a tale Told by an idiot. Rosenau Life’s but a walking shadow. that social scientists observe in human societies the dynamism. evolution. growth.

Complexity appears at first glance to be precisely the paradigm we need to understand global politics today. the simulation techniques and computer skills necessary for the application of complex systems theories are within the grasp of international relations scholars who have mastered more-conventional statistical or formal methods. there is a growing field of “biopolitics” (see Somit and Peterson 1997). ROSENAU Corning 2000.144 DAVID C. social systems have structures of authority that may be inconsistent with the definition of complex adaptive systems. 1998) and. Why? Macbeth might claim complex systems theories are tales told by an idiot. authority serves to minimize complexity. Although they may behave in complicated and confusing ways. surprising cascading events like the Eastern European revolutions of 1989 or the current crisis of multilateralism. The proliferation and influence of supra. We argue. 103). One therefore cannot use complex systems theory to model even partly centralized or hierarchical systems—precisely those types of systems that proliferate in the world of politics. though we are more optimistic about complexity’s prospects. we argue that in social systems. that by construction the simulation methods of complex systems theory cause the researcher to make assumptions about those issues that are of most interest to international relations scholars in particular and to political science in general: who the actors are and . international relations scholars who use complex systems theories—not to mention complex systems theorists who study international politics—are few and far between. by and large. and Axelrod and Cohen (1999)—differ in important ways from social and political systems. Clearly. EARNEST AND JAMES N. the seemingly chronic inability of existing theories to provide reliable predictions—all these facts understandably make many students of politics and societies sympathetic to theoretical approaches that posit instability. Jervis (1997). and change in the international system. 1995. while complexity is a meaningful metaphor. In this chapter we argue that those who study international relations have failed to use complexity as a general theory of complex systems (“complex systems theory”) because. Yet. life and its connotations of dynamism are central to an important line of contemporary thought about social systems. unpredictability. complex adaptive systems—at least as conventionally formulated by theorists like Holland (1992. This includes the many IR theorists today who accept complexity and nonlinearity as a metaphor for the inordinate intricacy of global and international politics. Furthermore. in political science. international relations theory has been slow to embrace complex systems for reasons other than the barriers to learning its methods for investigating the intricacies of global politics. Indeed. Axelrod (1997). then. the transformative effects of global information technologies. furthermore.and subnational actors. These differences are more than mere definitional or typological differences. Rather than a walking shadow in our analyses.

Otherwise. as some argue. it is silent on precisely those accidents and path dependencies that are most important to international relations theory. we take complex systems theory at its word and assume it is indeed “theory. We base our criticisms on two premises.SIGNIFYING NOTHING? 145 where authority resides. To the degree that complex system theory is embedded in a nonpositivist epistemology. At a time in our discipline’s embrace of the contingency of social agency. For this reason. complex systems theory has few if any methodological bases for staking its claim as “theory” in the positivist sense. who argue their methods combine deductive and inductive reasoning. given the knowledge claims of Axelrod (1997) and Epstein and Axtell (1996). among others. . First. to be sure. Using these computer-based models. Without agent-based models. complex systems “theory” reduces to a paradigm (“complexity”) rather than a theory—an indispensable element of theory construction. practitioners have engaged in little formal discussion of either the epistemology of complex systems theory or the standards for knowledge they set out for their work. our criticisms may be inappropriate. We believe. transmissible. however. generalizable. Our second premise relates to the first: to the degree complex systems theory makes theory-like claims. but in the spirit of encouraging practitioners of complex systems theory to debate explicitly the foundation of their knowledge claims. by extension. but a starting point in the process rather than its culmination. But to our knowledge. We offer our criticisms not to condemn the theory. known as agent-based modeling. IR scholars risk modeling dynamic processes and systems that are theoretically uninteresting. signifying nothing”—what one might call “Macbeth’s objection. There is an underlying irony here. We acknowledge that many complex systems theorists do not share our premises. it does so on the basis of its principal method. the challenge of complex systems theory is to model not merely dynamics but also the emergence of actors’ identities and of political authority itself.” Scholars who apply complex systems theory to questions of global politics need to understand both these perils as well as the promise of its methods. the inferential deficiencies of complex systems theory. We argue these standards are appropriate. and replicable results (to the degree dynamism and indeterminacy are replicable conditions). it is little surprise that few scholars are embracing theory whose methods treat as exogenous the identities of political actors and the sources of authority. that the epistemology and ontology of complex systems theory are poorly defined. Although complex systems theory embraces contingent phenomena.” For this reason we apply standards of positivist epistemology to their findings. complex systems theorists claim they have found nonobvious. the methodological shortcomings of agent-based modeling are. From the perspective of international relations theory. “sound and fury. We argue that without the simulative methods of agent-based modeling.

Individuals form their political opinions from a host of information resources that are as diffuse and decentralized as in a complex adaptive system. 1997) is but one example of a model that broadly articulates global politics as a complex adaptive system. Urry 2003. EARNEST AND JAMES N. we can use simulated social systems to investigate these questions. some types of information cascades appear to emulate the informational dynamics of complex adaptive systems. Economists long have recognized the similarity of markets and complex adaptive systems. it articulates a world of numerous “spheres of authority” or adaptive and decentralized political actors. numerous IR scholars. 1999. In a range of questions about voting. invoke complexity as a metaphor (Rosenau 1990. Earnest 2001a. chaos theory. authority in fact is as dispersed as it is in a complex adaptive system. Hughes 1997. Rosenau’s turbulence model (1990. It is tempting to believe that these intricacies and dynamism are the emergent properties of numerous. Anderson 1996. In politics. as paradigms to understand patterns of conflict between states (Saperstein 1996. When assumptions about agency are unproblematic. see also the essays in Alberts and Czerwinski 1997). ROSENAU Before we elaborate these criticisms. intricate. among whom we include ourselves. Indeed. Complex systems theory also holds promise for the investigation of interest aggregation and voting behavior. and dynamic is beyond question. Lohmann 1994). then. markets. and information dynamics. the (hypothesized) diffuse nature of exchange allows them to simulate a host of transactions and markets and to create interesting emergent phenomena like poverty and the concentration of wealth. dispersed. Some go even further and use complex systems theory or its antecedent. Axelrod 1997) and even the evolution of military organizations (Beau- . 1997. and autonomous political actors independently enacting simple local decision rules. These information processes may include riots and protest movements (Kuran 1991. GLOBAL POLITICS AS A COMPLEX ADAPTIVE SYSTEM That global politics today is mystifying. Opp and Gern 1993. It explicitly posits nonlinear relationships or cascades in politics. we believe there are inherent limits to the application of complex systems theory to a broader range of questions about global politics. This is precisely the domain of theoretical inquiry in which the simulation methods of complex systems theory hold the greatest promise. we should note there is a galaxy of social and political issues in which authority and actors’ identities are not problematic. and undoubtedly the reason that complexity appeals to IR scholars as a metaphor.146 DAVID C. 2003. But for reasons we discuss below. and it posits recursive relationships between political actors and their environments—what Rosenau calls “macro-micro linkages” (see also Smith 1997). Jervis 1997.

social researchers are most interested in emergent properties such as cooperation. Although complexity simulations seek in principle to discover patterns. they do so through one of two different ways. Because simulations of complex adaptive systems typically do not rely upon empirical data—though it seems to us that this need not be so—they may be of little help in inducing patterns in the actual politics of the world around us. we first discuss how complexity theorists use computer models. We then explore the deeper ontological assumptions complexity theorists must make about political actors and their motivations. To understand why. This massively parallel structure of the complex adaptive model usually begs for the use of computing technologies that allow either parallel processing (that is. with their attendant limitations. There are both methodological and. computers with a microprocessor for each actor—a serious practical limitation) or quasi parallel processing (software that iteratively processes instructions for each actor on a single microprocessor before taking the next “step” in time). and other social phenomena. epistemological reasons for this. Computer simulations of complex adaptive systems are. more importantly. By the definition of a complex adaptive system. Such computer-based models face a number of barriers to their acceptance among IR theorists. researchers may observe an empirical phenomenon that they hypothesize is the emergent property . warfare. Though our discipline’s cultural aversion to new methods may be one of them. students of IR and global politics have embraced the paradigm of complexity. and a ruled piece of paper to study residential segregation—the volume of calculations the researcher needs to undertake may be prohibitive. trade. we choose to focus instead on the broader question of how a researcher may use these computer simulation methods within a broader research program. its dynamics and temporal trajectories— what complexity researchers call its “emergent” properties—generally are what most interest complexity theorists in the social sciences. Two problems are immediately apparent. neither a deductive nor an inductive method. Clearly. Complexity theorists thus often rely on computers to conduct their social simulations. First. first of all. these emergent properties result from the local interactions of numerous autonomous. The Third Way The overall behaviors of a social system. to model complexity in global politics. disease transmission.SIGNIFYING NOTHING? 147 mont 1994). Yet the paradigm of complexity holds greater sway than the theory does.1 While it is in principle possible for the researcher to simulate these systems manually—a memorable example is Thomas Schelling’s (1978) use of pennies. Usually. dimes. independent agents pursuing local decision rules.

ROSENAU of a complex adaptive system. therefore. Yet complex systems theory is not a deductive theory. they face two criticisms. and that the same axioms are likely to produce different. and feedback from the environment and create and run on the computer a model of the system to see if they can “grow” the phenomenon or property (Epstein and Axtell 1996. giving rise not only to questions of model validity but to inaccurate predictions as well. Complex systems theory by construction posits. and the environment. would have no objections to the problematic assumptions we think complexity theorists make. grow out of the simulated system. deductive theorists have long argued that as a matter of epistemology. we should look at its explanatory and predictive value. their decision-rules. because they do not prove theorems. one should not reject a model or theorem on the basis of its axioms. An alternative approach to induction is to base the model’s assumptions on empirical data and then see what interesting emergent properties. deductive methods seek explicitly to prove consequences that one may logically derive from axioms. EARNEST AND JAMES N. rules. As we argue below. outcomes. To put it another way: the essence of path dependence is that while the researcher may be able to deduce a set of possible outcomes. Meadows et al.2 Of course. these assumptions can often be highly problematic and potentially tautologous. Epstein and Axtell’s “sugarscape” model (1996) typifies this approach to using a simulation for inductive purposes. therefore. The researchers then make assumptions about actors. The nature of contingency in complex adaptive systems means that numerous consequences are possible under a given set of assumptions. even divergent. the simulation methods of complex systems theory are not deductive. If successful.148 DAVID C. unexpected ways. Unlike game theory or other deductive methods. This approach still relies on the modeler’s assumption of which variables are salient. Second. complexity researchers cannot explicitly . 20). A deductive theorist. for two reasons. When complexity theorists use inductive methods to inform their models. however. Either their assumptions are empirically groundless and theoretically underdetermined. that it is difficult for the researcher to deduce consequences from his or her initial assumptions. if any. assumptions about relevant causes can drive a model in hidden. Though it was not a simulation of a complex system per se. such an approach risks oversimplification. Rather. As Miller (1998) shows in his testing of the World3 model. however. such a simulation relies upon the researcher’s assumptions about actors. he or she cannot deduce “the” (or even “the likely”) outcome(s) of a process.’s “World3” model of population growth and resource exhaustion (1974) is a good example of this inductive use of empirical data as a basis for an algorithm-based computer simulation. or their simulations produce uninteresting dynamics that at best have no referent in real world politics and at worst are indecipherable. Given our apparent consensus that global politics are intricate. and in any case may not yield any theoretically interesting emergent properties. First.

and paradoxically irrational yet calculating in their interactions with other human beings.3 It is unclear. of course. that empirical tests of computer-simulated processes can in fact test our hypotheses about actual dynamic systems. . Of course. Despite these advantages. while humans may follow simple decision rules. then. These observations about the complexity sciences are not novel. A second. but it cannot have both . Humans are both adaptive and habitual. he calls this method a “third way” of doing science.SIGNIFYING NOTHING? 149 test hypotheses about global politics. we remain yet to be convinced. authority in human societies—and even among social animals like apes or wolf packs—may be logically incompatible with the definition of a complex adaptive system. capable of both learning and misapprehension. Yet until a computer simulation can disprove a hypothesis. in which “simulated data . Complex systems theory. complex adaptive systems are little more than thought experiments on a computer—much ado about nothing. Although Elliott and Kiel (1997) advocate such a complementary approach. In this respect. for reasons we enumerate below. however. . lacks both the empirical appeal of induction and the disconfirmative value of deduction. it may be complex and adaptive. A social system may have authority present and be complex. it cannot be a theory of politics. and perhaps more important. To make sense of the intricacies of contemporary global politics. Unlike in physical complex adaptive systems. Epstein and Axtell (1996) similarly call the complexity paradigm a “generative” science: “Artificial society modeling allows us to ‘grow’ social structures in silico demonstrating that certain sets of microspecifications are sufficient to generate the macrophenomena of interest” (20. both Axelrod and Epstein and Axtell argue complexity simulations combine elements of inductive and deductive methods as an aid to the researcher’s intuition. they also may not. this conjoint simulative-empirical research design may hold greater promise for physical and biological systems than for social systems because of several distinctive features of humans. can be analyzed inductively” (Axelrod 1997. Axelrod (1997) acknowledges that the simulation of complex adaptive systems is neither inductive nor deductive. For one. researchers who use simulations of complex adaptive systems must supplement these efforts with empirical investigations. Practitioners of complex systems theory have defended their work against these charges. therefore. emphases in original). this requires scholars to embrace once again those methods—such as case studies or statistical models like ordinary least squares—which we have derided for their emphasis on stasis and linearity. 4). feature of societies is the role of authority. We recognize that complex systems theorists argue they can simulate learning and strategic behavior in their computer-based worlds. Rather. furthermore. the methods of complex systems theory alone cannot prove or disprove hypotheses about global politics—although it is a theory of process.

we rely upon heuristics and other simplifications in our everyday decisionmaking. These are software routines that each agent in the simulation follows to learn. The literature on the psychology of decision-making also suggests that humans indulge in wishful thinking and other forms of motivated bias (see also Levy 1997). because humans’ expectations are contingent in part on their identities. Again. less Delphic knowledge to inform their behavior. our decisions depend. it is unclear that this assumption holds for human agents. complexity researchers capture the adaptive behavior of agents through genetic algorithms (see Holland 1995. EARNEST AND JAMES N. or adapt to his or her environment. We suspect.150 DAVID C. In short. Through iterative feedback from the simulated social environment. but the psychology of decision-making suggests that human beings make decisions in messy ways that are difficult to capture with simple assumptions about decision rules. The Role of Expectations The three witches prophesied to Macbeth his murderous actions. or both. we focus here on modeling the evolution of decision rules of agents rather than their attributes. The modeler typically assigns to the population of agents either randomly generated or theoretically informed decision rules. the agents adjust their decision rules through mutation. Typically. While complexity theorists can reasonably assume that biological or physical agents pursue these simple decision rules. through . since centralized authority and individual autonomy. then a simulated complex adaptive social system will make assumptions about expectations and interests that are unlikely to capture this dependence. As we argue below. the agents’ decision rules may adapt through learning. Agents themselves may adapt by changing their attributes or passing advantageous characteristics to future generations. ROSENAU authority and adaptation through complex interactions of autonomous agents. these lines of thought seem to support a rejection of the assumption of rationality and the adoption of “satisficing”-type decision rules that typify agents in complex adaptive systems. however. The appeal of the rationality assumption is its simplicity. A substantial literature suggests that human beings are not perfectly rational. evolve. by definition. are mutually exclusive. by contrast. on who we are. 1998). at least in part. on their own. If so. that this simplicity itself may gloss over some important characteristics of expectations and interests in human actors. it is useful to understand the methods complexity researchers use to simulate the adaptive and learning behavior of agents. Human agents typically rely. Since students of global politics typically are interested in actors’ expectations and interests rather than in their biological characteristics. At first glance. the conceptual conundrum of authority in complex systems presents a fundamental challenge to the application of complex systems theory to questions of world politics.

are either majoritarian or entrepreneurial. Fitter agents come to dominate the system. If multiple and divergent outcomes may result from the same set of assumptions. Likewise. Wilson (1980) argues that the concentration or diffusion of both the costs and of benefits from public policies will affect the type of contestation in which individuals engage. In this respect. Axelrod (1997. An important question is. for example. for example. and to what degree from the modeler’s assumptions. Through these processes. diffuse costs and concentrated benefits yield client politics. While complex systems theory in principle embraces precisely this type of contingency in decision rules. or through reproduction. an agent’s interests and. There are two reasons to suspect that they cannot. then. it becomes exceedingly difficult to disentangle those results which emerge endogenously from those that are true by construction. determining to what degree the emergent behavior of the social system results from the dynamics of the system. or decision rules merely assumes that interests are diffuse and that politics.4 For example. expectations. in turn. James Q. however. from a randomly generated set of rules. the literature on interest groups tells us the spatial organization of interests matter: a political actor’s identity as a passive or active participant. as complex system theory posits. 1) has shown how tit-for-tat emerges as an optimal strategy in a complexity simulation of an iterated multiplayer prisoner’s dilemma. therefore. First. while less fit agents may be “selected out” or become in some sense “peripheral”— though this term belies the importance of these “lesser” agents in sustaining the system’s dynamics. The inferential problem is. majoritarian politics will result. The modeler’s . If the modeler assigns specific interests or decision rules. depends in part on how the interests of others are spread throughout a system. interest group politics will result. the researcher may find that the genetic algorithm produces a small population of “fit” strategies. Building on Olson’s seminal work (1965) on collective action problems. When both costs and benefits are concentrated. whether or not genetic algorithms can emulate the processes of how human beings make decisions. agents typically narrow the population of decision rules. when both are diffuse. he or she places the model in either the world of client politics or interest group politics.SIGNIFYING NOTHING? 151 selection. the problem arises with the a priori assumption of the spatial organization of costs and benefits. the modeler’s choice of fitness criterion makes an assumption about the concentration or diffusion of costs and benefits from the environment. therefore. both of which arise from how a political actor’s identity can influence his or her expectations. A randomly generated population of interests. Wilson’s insight for complex systems theory is that the modeler’s assumptions about interests—no matter what those assumptions may be—are not theoretically neutral. and concentrated costs but diffuse benefits create entrepreneurial politics. expectations will be shaped by the spatial organization of interests in the system as a whole. chap.

then interests and expectations are highly path dependent in ways that genetic algorithms fail to capture. If one accepts the constructivist critique that identities themselves are contingent (a notable “if” that we address later). The Emergence of Authority Politics is. therefore. no “macro” and no “micro” levels. the very contingency of one’s political identity—whether or not one has a stake in a given policy. (122) As Urry suggests. any rule for the assignment of decision rules. The exercise of authority is central to our understanding of global politics today. or identities—requires the researcher to make unfounded assumptions about the structure of politics and its influence on decisions. may end up shaping the path of the evolution of agents’ decision rules. A second problem with genetic algorithms is a classic problem of most models of human agency: they share an inability to capture the nuanced psychology of human decision-makers. our self-identification can affect our perceptions and expectations: we interpret information as supporting our desired outcome. EARNEST AND JAMES N. the random assignment of decision rules—indeed. expectations. This is because each such notion presumes that there are entities with separate and distinct essences that are then brought into external juxtaposition with its other. it has migrated downward . as much as the genetic algorithm. furthermore. In a simulated social system. The danger of a tautology is obvious. Humans do not follow simple decision rules. whether human agents can accurately perceive environmental feedback in the perfect way that the use of genetic algorithms suggests. As the literature on motivated bias suggests (see Lodge. Human psychology and decision-making are so idiosyncratic as to make the ascription of simple behavior rules highly problematic. and we incorporate discrepant or disconfirming information as supporting our predispositions or earlier decisions. and Galonsky 1999). no “societies” and no “individuals”. It is debatable. the authoritative allocation of values. and no “system world” and no “life world”. on who we are.152 DAVID C. our textbooks tell us. Indeed. whether or not one chooses to partake in political contestation—makes interests and in turn expectations accidental and path dependent. we are subject to a panoply of psychological biases and errors. Many of these biases depend. It has migrated upward to international and nongovernmental institutions and to global corporations. Urry (2003) articulates an extreme variant of this argument: [T]here is no “structure” and no “agency”. interests. Tabor. ROSENAU assumptions. particularly since so many researchers argue that authority has migrated away from the institution of the nation-state.

no matter how needy.” Of course. First. and others. will go to jail for stealing a loaf of bread. Unlike in a complex adaptive system. the hegemon. in which decision-making is perfectly diffuse (or to extend our example. Authority is perfectly decentralized. how can a social system with its attendant structures of authority be a complex system? And if a complex system can have “some” hierarchy. This is the logical antithesis of social authority. or the “leviathan. each agent decides and acts on the basis of internal rules that evolve in response to environmental feedback. the firm. in which a privileged agent makes allocative decisions for a group of other actors. one that derives its dynamism and adaptability from its precarious balance on the edge of chaos. most of us no longer assume the primacy of the nation-state. and short according to Hobbes. we need a complex adaptive system that shows how authority emerges in the first place from the interactions among autonomous agents. what is the difference between a complex system and other definitions or types of systems? What theoretical leverage does the ontology of the complex adaptive system offer. and formidably more important. This raises the question of the appropriateness of the complexity metaphor for the study of politics: are authoritative systems logically incompatible with complex adaptation? Though they did not consider the question in these terms. is centralized authority: the state. A complex adaptive social system. As we have already noted. is nasty. The “natural” response to such systems. classical thinkers clearly thought so. a market in which agents can break contracts or steal from others). terrorist cells. we need a complex adaptive system that shows how authority shifts from one authoritative actor to another and how “layers” (or “spheres. while in others authority may represent the boundaries of the system or the “rules” within which autonomous agents enact their rule models—much like economic agents pursue rule models (“get rich”) within a system (“market”) in which property rights and the enforcement of contracts are unproblematic.SIGNIFYING NOTHING? 153 to local governments. particularly if its treatment of authority is so elastic? The definition of a complex adaptive system thus seems inconsistent with our conventional understanding of what an authoritative system . But if we take the definition of a complex system at its word. Second.” “nodes.” or “attractors”) of authority may result. political authority often compels individuals to act contrary to their internal rules: the beggar. the role of authority in a complex system may be one of degree: some complex systems are characterized by little if any authority. Authority therefore is problematic. according to classical thinkers and organizational theorists alike. This feature of global politics poses a double challenge to the application of complex systems theory to international politics. the pattern of authority in a complex adaptive system is one of its distinctive features: it has none. civil society. or alternatively some balance of centralized and decentralized decision-making. brutish.

decentralized social structures rather than classic. 1997) calls them “spheres of authority”. this only begs the question of why states—and not voters. Marion (1999) calls these informal forms of cooperation “social solitons”. for the application of complex systems theory to problems of global politics. they may not rely upon coercion. ROSENAU is. IGOs. and Lohmann (1994) each explain the Eastern European revolutions of 1989 as the result of cascading information processes among leaderless individuals. Such supranational actors may in fact possess coercive powers typical of strictly defined authority. however. Both the literature on and the popular perception of the “democratic deficit” suggest. If the answer is simply that it is easier to model authority as a unitary actor.and subnational actors may not be forms of authority as we have strictly defined it. The breakdown of authority is another of the central concerns of the study of global politics that demonstrates the shortcomings of the concept of the complex adaptive system. How can these patterns of authority result from the complex interactions of autonomous actors? While it is tempting to argue we can model these global and transnational processes as interactions among states. both of the persuasive and the coercive varieties. Rosenau (1990. This process of breakdown arguably is itself the result of the complex interactions of dispersed autonomous agents. For example Kuran (1991). implement. Much of the study of global politics today is concerned with the problematic relationships between sovereign states and supra. one area where complex systems theory holds considerable promise is the investiga- . While the question of cooperation under anarchy once preoccupied IR scholars. but because they are easier to construct. then we have merely committed the same methodological error as other methods: we choose our models not because they are conceptually appropriate and theoretically useful. Some of these supra. Riots and other forms of the erosion of authority may be important phase transitions in complex adaptive social systems. for example. to maintain their influence over political actors. Persuasive authority itself may derive. these days the discipline of IR has a much more nuanced conception of authority in the international system. for example. furthermore.” Indeed. and understand. EARNEST AND JAMES N. from informal. While this may offer certain conceptual advantages. that the coercive power of supranational actors is not trivial.and subnational authorities. This question is particularly troublesome. Global politics today is rife with examples of new forms of social authority.154 DAVID C. Opp and Gern (1993). NGOs. Hobbesian centralized authority (Earnest 2001b): “leaderless” groups may derive moral legitimacy precisely because citizens view them as decentralized and spontaneous. or transnational elites—deserve ontological primacy. furthermore. and Harrison (2001) terms them “nodes of order. complexity theorists have yet to reconcile an authoritative social system with the decentralized authority of a complex adaptive system.

furthermore. One possible way to do this is to endogenize authority at the genesis of a complex adaptive system. both debtors and creditors are functionally identical.SIGNIFYING NOTHING? 155 tion of how authority cascades through political systems. and if. One might argue that creditors on the sugarscape. they may not truly simulate the emergence of the authoritative allocation of values for a population of agents. But these processes and questions require us to make some initial assumptions about authority in a complex adaptive system. when. a complex adaptive social system must have political authority in the first place. While these approaches are an important first step. in fact are qualitatively different actors or authorities. Epstein and Axtell (1996) show how markets may emerge and how actors will assume specialized roles as creditors or debtors. But for other questions these assumptions risk making any findings about authority highly problematic. Each follows local decision rules. Axelrod’s (1997) tribute model shows that aggregate collectivities may emerge from the behavior of decentralized agents in a complex adaptive system. and voting behavior—researchers can make some reasonable assumptions. For some questions of global politics—such as the study of market transactions. as the emergent property of a complex adaptive system. how it shifts from one authority “attractor” to another. Kollman. information dynamics. Likewise. Epstein and Axtell’s sugarscape model is similarly devoid of agents that are qualitatively different. Setting aside the question of whether or not this is truly a complex adaptive social system. and Page (1997) simulate “instability” or variations in the effectiveness of political institutions. Miller. this approach cannot tell us about the sources of authority. A number of complexity theorists already have tackled this challenge. the tribute model produces only quantitatively different actors. Yet. The alternative method—and in our minds more challenging but theoretically more fruitful—is to grow authority from the bottom up. or strong states in the tribute model. Cederman (1997) seeks to endogenize the processes of the constitution of states in the international system. some states in the model develop more power than others. Though these weaker . But the emergent actors are not qualitatively different: they do not make decisions for other agents in the system. Though the economy of the sugarscape creates debtors and creditors. We can assume that a single agent makes allocative decisions for all other agents and can enforce those decisions in the face of agents’ internal rule models. and where it may achieve a degree of dynamic equipoise. path dependent (and there are good reasons to suspect it is). If the emergence of authority is. Axelrod’s tribute model purports to demonstrate how collective action arises through coercion. Before we can model shifts in authority. how power creates its own authority. this exogenous approach may assume away important evolutionary dynamics. since their power or wealth deprives weaker agents of viable choices.

and dissolve with extraordinary speed. rather. and hence no choice at all. a complex adaptive system can neither simulate social authority nor describe how social authority constitutes the very actors that are its subjects. particularly chap. the argument might go. As we noted above. “protosystems” that include an infinite number of latent actors and dormant systems. or a host of other processes) or removing them (conquest. whether domestic. EARNEST AND JAMES N. the power of stronger actors presents them with Hobson’s choices. Both approach the . 1994) and others have argued. Social authority therefore includes the power to remove any given actor from society itself. Ruggie (1986. Spruyt (1994) shows that even the nation-state is a contingent social construction. rather. we go to jail. shift. After all. there are latent identities. any one of us may choose not to pay our taxes and to face the consequences from the power of the authority known as the IRS. 8) each have attempted to endogenize this type of latency in political agents. Lustick (2000) and Cederman (1997. it therefore not only allocates values but affects the aggregation of interests. Whether they be states. social authority itself can constitute political actors by adding them to society (through enfranchisement. and so on). liberation. They are. if we do not pay our taxes. erode. adapt. or global. 1993) similarly argues that differentiation among nation-states is a historically path-dependent constitutive process. This is where complex adaptive approaches to authority and the emergence of new political actors have fallen short. or values that are context and spatiotemporally dependent. it is theoretically groundless for scholars to assume that any agents are ontologically primitive in global politics. political agents themselves therefore are contingent and indeterminate. In politics. how complex systems theory’s approaches to authority fail to engage the constructivist critique of structuralism in international relations theory. attributes. therefore. with political agents entering and exiting. this analogy shows how inappropriately the tribute model or the sugarscape model treat the concept of an authoritative actor. Societies are not merely open systems. which other actors may invoke to mobilize or remove political actors. In this respect. international. It is also the force to define and shape the collectivity as well. the constitutive rules of identity—of who participates in the contestation of specific values—themselves may vary and adapt in a way that is not true of a biological or physical complex adaptive system. As Wendt (1987.156 DAVID C. Yet. In politics. voters. imprisonment. Because decision-making is by definition perfectly diffuse in a complex adaptive system. the rules of identity also may be issuespecific—the political issue may determine who the “actors” are. or organizations. ROSENAU agents technically are not compelled in their enactment of their individual decision rules. These actors and systems emerge. This argument anticipates. After all. Authority is not merely the force to make allocative decisions for a collectivity. furthermore. there are no “agents” per se.

SIGNIFYING NOTHING? 157 question of agency by endowing the agents in their complex adaptive social systems with latent identity attributes. Soon. that students of global politics remain skeptical about complex systems theory’s methods. but a few guidelines may help develop a research program in artificial authority. By assigning rules of identity exogenously—that is. most complexity simulations of politics treat these questions axiomatically. by assuming the existence of authority rather than endogenizing it as a problem—complex adaptive approaches to world politics overlook those very concepts that are of greatest interest to us as students of global politics: institutions and identities. If and . Following the example of the Turing test. dynamic. Cederman (1997). but from the researcher’s own assumptions about authority and agency at the model’s genesis.5 A complex adaptive simulation of global politics thus risks a cleverly disguised tautology: the emergence of authority and agency results not from the adaptive. but they are less developed on the question of the contingency of authority. One way to respond to our criticisms of complex systems theory is. therefore. we believe. The simulative methods of complexity thus risk obfuscating important assumptions made by complexity researchers about actors and authority in global politics today. SIGNIFYING NOTHING? “Can machines think?” the mathematician Alan Turing (1950) asked more than a half century ago. the works of Axelrod (1997). contexts may emerge in which the actors’ latent identities are activated. As their models run. These are promising approaches to the issue of the contingency of agency. and he set out to create a measure by which artificial intelligence researchers could assess their progress in creating cognizance. The resulting standard—the Turing test—specifies that scientists will have succeeded in creating artificial intelligence when a human interrogator cannot distinguish between a human respondent and a computer respondent. Lustick (2000). It is little wonder. complex systems theorists and students of global politics may face a question similar to Turing’s: how will we recognize authority in a simulated complex adaptive system? It is beyond the scope of this chapter to answer this question fully. but also an important explanatory assumption: the very existence of political actors. we propose an experimental design that incorporates human subjects into the virtual world of the simulated social system. It remains to be seen whether complexity theorists can endogenize fully these constitutive processes. THE SOUND AND THE FURY. In contrast to Lustick and Cederman. and others hold some promise in “growing” authority as the emergent property of a complex adaptive system. to ask a similar question about “artificial authority”: can a machine command compliance? Toward this end. Complexity theorists make not just procedural assumptions. nonlinear interactions of the agents.

EARNEST AND JAMES N. receive the same rewards or punishments. contribute to our theoretical understanding about contingencies of agency and authority. that the agents are ontologically primitive. considerable conceptual appeal. In politics. they are full of sound and fury. Of course. Their decisions and expectations depend in turn on who they are and their position not only in political space. at least—precisely the type of spatiotemporal path dependencies that complexity researchers are interested in understanding. but also the exogenous treatment of interests and identities that are of most theoretical interest to students of international politics today. these contingencies are—conceptually. but as yet signify very little. in short. the complex adaptive simulations of world politics that we know of are promising yet incomplete—as Macbeth might say. We hope such a research program on artificial authority will begin to address the ontological conundrum of authority in a complex system and to articulate the degree to which complex systems theory depends upon a nonpositivist epistemology. but also in time and the context of values. or voters—have expectations that are more than the simple rule models posited for complex adaptive agents.158 DAVID C. But in our view. The concept of a complex adaptive system appeals to us precisely because it embraces the intricacies. nonlinearities. actors—whether states. ROSENAU when a complexity researcher succeeds in generating authority from the bottom up in a complex adaptive social system. IGOs. when human subjects comply with the system’s authority in the same manner as the silicon agents do. The methods of complexity do not easily transcend the disciplinary boundary between the social and biological or physical worlds. Human subjects receive and transmit the same information from and to their environment as the silicon subjects do. The problem to us is not with the paradigm but with the epistemology of complexity. in fact. standard. the researcher should rewrite the program so that human subjects can participate as if they were agents within the system. We believe this is a difficult. Until the methods of complex sys- . We cannot even assume. It behooves us as students of international politics to demonstrate how complex systems theory can answer the difficult questions of authority and agency. but they are free to follow their own decision rules. and influence their neighbors and environment according to the same rules. then one may reasonably argue the system is authoritative.6 Such a program may not only begin to address our criticisms but may. though not unattainable. NGOs. because the world of politics is conceptually different than the world of physical and biological complex adaptive systems. It has. The practical necessity of relying upon computers to simulate complex adaptive social systems risks not merely oversimplification. Complex systems theory offers a paradigm that explicitly rejects the concepts of equilibrium and stasis that seem so inappropriate for our understanding of international politics. furthermore. terrorist cells. Under such conditions. and unpredictability that we observe daily in world politics.

to establish its bona fides we may use it to disconfirm hypotheses that we have already falsified with other methods.brook.northwestern. All are freely available with numerous sample simulations. See Hoffman and Johnson 1997. According to methodological falsificationism. 2. See Taber 2001. particularly NetLogo (developed at Northwestern we can derive some confidence in our knowledge when we fail to disprove them. of course. Indeed. models/ascape/ReadMe. available for a number of different operating systems ( logo/) and StarLogo (developed at MIT.html) are two other simulation software and Ascape (the Brookings Institution. but also in its failure to disconfirm hypotheses once it has demonstrated the ability to do so. it must establish its ability to test and disconfirm hypotheses. RePast (University of Chicago. complex systems theory faces two challenges. Otherwise. Miller (1998) argues that genetic algorithms may themselves be useful in probing the weaknesses of the specification of complexity models and identifying the modeler’s key assumptions.SIGNIFYING NOTHING? 159 tems theory can create artificial authority from the ground up. swarm. For example. http://repast. For example. 5. the Santa Fe Institute’s Swarm software. we fear its hour on our conceptual stage is drawing to a close. only after it has established its ability to disconfirm can complexity theory be used to probe and test new hypotheses. the utility of a method lies not only in its ability to disconfirm a hypothesis.sourceforge. Taber (2001) recalls a conference at which two modelers of fractals confessed they had “no earthly idea” how their model worked. First. 4. therefore. 3. In this sense. the assumption that actors have interests is not procedural. 24. All of these “assumptions” that agent-based modeling requires are precisely those questions of greatest theoretical interest to an important school of thought about global politics. The best-known simulation software is. Other software packages have lower learning thresholds. we cannot know whether or not the failure to disconfirm a hypothesis arises from the robustness of the hypothesis or from the methodological shortcomings of complex systems theory. This second challenge is arguably more difficult to achieve. and the assumption that these interests are related to—or unrelated to—the actors’ position in space and time is not procedural. NOTES 1. to many constructivists the assumption that a political system exists is not procedural. This is the essence of logical positivism: while we cannot prove hypotheses. http://ccl. .

2002. or alternatively that they would not have. Cohen. In other words. we cannot assume that silicon agents would necessarily have complied with authority in the absence of human agents. REFFERENCES Alberts. NJ: Princeton University Press. February 20–24. EARNEST AND JAMES N. Beaumont. and Thomas J. Czerwinski. NJ: Princeton University Press. and Future. Anderson. 1997. Montreal. ———. Chaos. and Michael D. eds. Corning. Westport. and History. 1994. Robert. War. David S. New York: Free Press. Washington. Global Politics.” Paper presented at the Forty-second Annual Convention of the International Studies Association. Peter J.. Since complex systems theory posits that agents’ behaviors are correlated with each other. 2001a. 1997. Axelrod. 1996. February 20–24.” Paper presented at the annual meeting of the Association for Politics and the Life Sciences. Chicago. Princeton. 1997.” Paper presented at the Forty-second Annual Convention of the International Studies Association. we cannot assume the independence of the behavior of silicon agents when human agents participate in the system. Axelrod. Princeton. Complexity. DC: National Defense University. and National Security. Global Politics of Power. 1999. Chicago. “The Complexity of Globalization: Meeting the Challenges of Structuration Theory. Present.’ Past. Emergent Actors in World Politics: How States and Nations Develop and Dissolve. 2000. August 11–14. It is admittedly difficult to operationalize measures that test whether or not human subjects comply with a model’s authority in the same manner as silicon subjects. .160 DAVID C. Robert. Justice and Death: An Introduction to International Relations. Roger. London: Routledge. Lars-Erik. “Synergy and the Evolution of ‘Superorganism. CT: Praeger. Cederman. ———. ROSENAU 6. Earnest. The Complexity of Cooperation: Agent-Based Models of Competition and Collaboration. David C. Harnessing Complexity: Organizational Implications of a Scientific Frontier. Peter A. “The Sociobiology of Democracy: Is Authoritarianism in Our Genes?” Politics and Life Sciences 19 (March): 103–8. “Will No One Rid Me of This Meddlesome State? Social Inequality and the New Social Contract. 2001b.

Sara Horsfall. “Nonlinear Dynamics. Hughes. New York: Oxford University Press. “Complex Adaptive Systems. no.” In Chaos. MA: Addison-Wesley. Kollman. Emergence: The Connected Lives of Ants. John H. 1 (Winter): 17–30. Complexity and Sociology: Myths. System Effects: Complexity in Political and Social Life. ed. Cambridge.. Kuran. . Kauffman. and L. Growing Artificial Societies: Social Science from the Ground Up. NJ: Princeton University Press. Barry. Stuart. Euel.” World Politics 44 (October): 7–48. Thousand Oaks. and Theories. held June 1992 in Santa Fe. NJ: Prentice-Hall. 64–78.” Prepared for the 42nd Annual Convention of the International Studies Association in Chicago. MA: Addison-Wesley. 2001. 1997. 1991. “The Value of a Complexity Metaphor for International Political Economy. Johnson. ———. Reading. Raymond A. Joshua M. Complexity. 1999. and Public Policy. Cities and Software. Lee. Jervis. Hoffman. Matthew J. Washington. 2001. “Now Out of Never: The Element of Surprise in the East European Revolutions of 1989. Upper Saddle River. Continuity and Change in World Politics: Competing Perspectives. Hidden Order: How Adaptation Builds Complexity. Boulder. 1997. CO: Westview. Timur. 1996. Langton. Epstein. Reading. Miller. and Mary E. Steven. MA: Perseus.” American Economic Review 87. Artificial Life III: Proceedings of the Workshop on Artificial Life.. DC: Brookings Institution. and Robert Axtell. Brains. Holland. Christopher G. Harrison. 1997. Cleveland. Eve.” Paper presented at the International Studies AssociationMidwest meeting.SIGNIFYING NOTHING? 161 Elliott. CA: Sage. Princeton. 1997. Models. Page. no. Douglas Kiel. Neil E. “Political Institutions and Sorting in a Tiebout Model.” Daedalus 121. Emergence: From Chaos to Order. Ken. 1997. 1992. and David Johnson. ———. New Mexico. February 20–24. ———. At Home in the Universe: The Search for the Laws of SelfOrganization and Complexity. International Futures: Choices in the Face of Uncertainty. 1994. 5 (December): 977–92. Robert. “Change and Process in a Complex World: Using Complexity Theory to Understand World Politics.. John H. and Scott E. 1998. 1995. ed. October 3–5. New York: Scribner. 1995.

Atlanta. ROSENAU ———. Mancur. Princeton. 2003. Redwood City. Lohmann.” World Politics Turbulence in World Politics: A Theory of Change and Continuity. and Aron Chase Galonsky. Charles Taber. 1 (March): 87–112. 1994. 1990. Along the Domestic-Foreign Frontier: Exploring Governance in a Turbulent World. East Germany. Langton. 1997. Christopher G. “The Political Consequences of Motivated Reasoning: Partisan Bias in Information Processing. MA: Harvard University 1. Ian. Available at http://www.” Management Science 44. John H. and International Relations. Marion.162 DAVID C. CA: Sage. Susanne. New Mexico. Langton. eds. 1999. Rational Choice.” American Sociological Review 58. Thousand Oaks. Artificial Life V: Proceedings of the Fifth International Workshop on the Synthesis and Simulation of Living Systems.” Paper presented at the Annual Meeting of the American Political Science Association. no. 1999. Levy. Opp. no. NJ: Princeton University Press. Rosenau. “Active Nonlinear Tests (ANTs) of Complex Simulation Models. Meadows. “Dissident Groups. 1997. 1993. NJ: Princeton University Press. and Christiane Gern. 1 (October): 42–101. Milton. Russ. 5 (October): 659–80. Karl-Dieter.surrey. Princeton. The Edge of Organization: Chaos and Complexity Theories of Formal Social Systems. 1998.” International Studies Quarterly 41. Cambridge. “The Dynamics of Information Cascades: The Monday Demonstrations in Leipzig.. MA: MIT Press. Personal Networks and Spontaneous Cooperation: The East German Revolution of 1989. “Agent-Based Modeling of Collective Identity: Testing Constructivist Theory. Cambridge.soc. Distant Proximities. 1989–1991. ———. Dennis L. Artificial Life II: Proceedings of the Workshop on Artificial Life Held February. Miller. 1995. no. 1974. The Logic of Collective Action: Public Goods and the Theories of Groups. 2000. 1965. Lodge. no. “Prospect Theory. 6 (June): 820–30.. James N.” Journal of Artificial Societies and Social Simulation 3.. 1 (January). Lustick. 1997. ———. Artificial Life: An Overview. MA: Wright-Allen Press. and Katsunori Shimohara. Christopher G. 1990 in Santa Fe. eds. Dynamics of Growth in a Finite World. MA: MIT Press. et al. . Olson. 1991. Jack S. CA: Addison-Wesley. September 2–5.html. Cambridge: Cambridge University Press. et al. Cambridge. no. EARNEST AND JAMES N. Cambridge.

” International Organization 47. CT: Praeger. 1978. Wilson. “Nonlinear Dynamics and the Micro-Macro Bridge. ed. Lee. Alexander E. Sara Horsfall.” In Chaos. John. Darwinism. Turing. . ed. ed. 1986.SIGNIFYING NOTHING? 163 Ruggie. Eve. “Computing Machinery and Intelligence. New York: Basic Books. Complexity and Sociology: Myths. 1997. “Continuity and Transformation in the World Polity: Toward a Neorealist Synthesis.” In Neorealism and Its Critics. 1993. Peterson. “Territoriality and Beyond: Problematizing Modernity in International Relations. 139–63. Robert O. W. Albert. 3 (Summer): 335–70. Urry. Dominance. Cambridge: Polity. Smith.. 197–214. and Computational Social Science.” Political Methodologist 10. Micromotives and Macrobehavior. 1975. John Gerard. Alan M. Douglas Kiel and Euel Elliott. Raymond A.” International Organization 41. Alvin M. Potions. Lee. L. “Of Spells. NJ: Princeton University Press. 52–63. Models. New York: W.” American Political Science Review 88. Norton. The Politics of Regulation. Charles S. Complexity and Sociology: Myths. no. Models. Global Complexity. Eve. and Theories. Cambridge. Spruyt. Stevens. “The Agent-Structure Problem in International Relations Theory. Sara Horsfall. no. ———. 2003. The Sovereign State and Its Competitors: An Analysis of System Change. no. 1980. ———.” In Chaos Theory in the Social Sciences: Foundations and Applications. “Biological Foundations of Social Interaction: Computational Explorations of Nonlinear Dynamics in Arousal-Modulation. Edward O. no. 1 (Winter): 139–74. Wendt. Wilson. Sociobiology: The New Synthesis. 1994. Thomas S. 1994.” Mind 59. and Mary E. Thousand Oaks. 1997. Keohane. 1997. Hendrik. 2001. James Q. and Steven A. New York: Columbia University Press. and Mary E.” In Chaos. 1 (Fall): 23–26. 1987. Princeton. Taber. Thomas C. Raymond A. “Collective Identity Formation and the International State. “The Prediction of Unpredictability: Applications of the New Paradigm of Chaos in Dynamical Systems to the Old Problem of the Stability of a System of Hostile Nations. and Democracy: The Biological Bases of Authoritarianism. and Gregory T. Saperstein. 1996. Thomas S. and Theories. CA: Sage. Ann Arbor: University of Michigan Press. 131–57. Thousand Oaks. Smith. no. MA: Harvard University Press. Schelling. Somit. 236 (October): 433–60. 2 (June): 384–96. Westport. CA: Sage. 1950. ed.

.This page intentionally left blank.

Moreover. the complexity paradigm offers new thinking about old problems and suggests new hypotheses and explanatory 165 . But before exploring these inquiry possibilities. Their second objection is that complex systems theory cannot capture and model the essence of political systems: the central role and importance of authority. or rather an ability to combine. and decompose theoretical views of the world not easily accomplished in the past. The first is that complex systems theory “lacks both the empirical appeal of induction and the disconfirmative value of deduction. it is prudent to bound these possibilities by considering the theoretical challenge to the credible use of computational models of agency. as alluded to in the title. more precisely stated. like those in some of the earlier chapters in this volume. agency-level computational models—will reshape the practice and use of social science/inquiry. The preceding chapter by Earnest and Rosenau questions whether there can be a complex systems theory of political systems. compare.” Therefore.CHAPTER 9 When Worlds Collide Reflections on the Credible Uses of Agent-Based Models in International and Global Studies Desmond Saunders-Newton It is quite possible that Agent-Based Models (ABMs)—or. For issues in which authority is less pervasive. it is not really a theory. it is possible to envision the realization of an algorithmic social science and the more explicit inclusion of social science theory and insights into the praxis of policy analysis and applied international/global studies. they argue. When contrasted with methodological approaches that solely focus on the aggregate levels of analysis. ABMs allow for a metaphorical collision of worlds. They raise two objections.

2 or induction. In my opinion. I do not directly challenge either of these criticisms of the usefulness of complex systems theory in explaining or predicting political systems. either via deduction. I show that ABMs can improve policymaking even if both of these criticisms are true. exchanges related to methodological adequacy will likely fall into the category of reasonable scholars may differ. Instead. Such a methodology offers the prospect . complex systems theory cannot describe the importance of authority and ABMs cannot simulate its effects. can also be raised when we consider the “status quo” methodological approaches used by scholars in this field. On the other hand. Several US government agencies.166 DESMOND SAUNDERS-NEWTON methods. Even before theorists have worked out the ontological and epistemological problems associated with using complexity to explain political systems.1 abduction. I argue that the epistemological problems associated with ABMs are worthy of note but that they are overstated. In this chapter.” but also are consistent with the innovation generated by the tools of this intellectual approach. the new science (chaos and complexity) has “placed within our grasp a set of very powerful tools—concepts to think with. Earnest and Rosenau’s objections have merit in that they raise concerns about our ability to rigorously create knowledge about our world. are already exploring the design and employment of systems using multiple ABMs to generate policy options and model potential costs and benefits of different choices. their concerns about our ability to reason from models. EFFECTIVE COMPUTATIONAL INQUIRY The introductory chapter defined agent-based modeling as the simulation of complexity in the social sciences. But Earnest and Rosenau argue that in those issue-areas in which authority is more influential.) With this in mind. Such emergent applications not only reflect a desire to bring social science knowledge to bear on “problems of the day. this chapter explores a scheme for improving our ability to use these new perspectives and methods. We can use them well or badly. retroduction. xii. As noted by Frederick Turner. This section will be followed by a more explicit explication on some of the anticipated uses of ABM methods in the realm of international security praxis and inquiry. as well as the appropriate substantive focus of the disciplines that constitute the field of study known as international relations or world politics. but they are free of many of the limitations of our traditional [methodological] armory” (Turner 1997. inclusive of the Defense Advanced Research Projects Agency (DARPA) and the Center for Technology and National Security Policy. we see the use of complex systems thinking—and the agent-based models that it supports—in an attempt to improve the choices of political leaders and reduce the risks of action in international politics.

Regardless of pedigree. Agency-based modeling reflects a methodological approach that allows for considering phenomena at levels of aggregation. A Typology of Agency-Based Methods . what these models have in common are an ability to explicitly consider. and environment. but also to pursue such inquiry in a more interdisciplinary fashion. and sociophysics. agents. more granular than that of the Westphalian nation-state. I would like to deepen this definition by asserting that agent-based modeling—or. A potential typography for envisioning agency-based methodology is shown in figure 9. and relate. more broadly considered. computational social network analysis. At least three distinct approaches potentially explain ecologies of agents and their interactions: multiagent simulation. computational methods that emphasize agency-level phenomena—reflect a maturing transdiscipline that allows analysts and inquirers not only to consider increasingly complex phenomenology in a rigorous fashion. and they vary by distinctive methodological styles and model ontologies (disciplines). or resolution.1. This capability affords an opportunity to explore theories of complex international and global social phenomenon. Several methodologies afford the analyst a means of studying the forms and dynamics of a social system. Figure 9.1.WHEN WORLDS COLLIDE 167 of representing the interactions of agents comprising a heterogeneous system of autonomous actors. interactions.

377) characterize the efficacy of ABMs in the social sciences: • The agents in an ABM can be based upon the wide breadth of actors that arise in natural and artificial systems—for example. will vary across methodological approaches. Lempert. As Bankes. no methodological approach comes cheap or free of costs. Closely related to this notion of credibility is the willingness of practitioners such as decision-makers or professional analysts to accept the value-added data and information generated by a methodology. the ability to consider intuitively resonant levels of resolution is a true benefit of the approaches. Even though “what you can know” may well differ between agency-based methodological approaches.3 Thus. ABM is an appropriate computational ontology for representing much of the knowledge and data that arise from social actors operating and interacting in an international and global context. . As a result. I address the general issue of computational epistemology and methodology as it relates to the use of computational models of agency. modeling neophytes with decision-making authority and analysts trained in traditional modeling traditions and shaped by hard-earned expertise. and Popper (2002. COMPUTATIONAL EPISTEMOLOGY AND METHODOLOGY As a modeling formalism. 1996). and Popper suggest. computational ontologies. or rather the “what is” that is instantiated as algorithm. individuals. Put another way. or institutions. one reason for the growing popularity of agent-based modeling is its flexible representation of reality: “[I]t is an appropriate ontology for representing much of the knowledge and data that is available about social actors and social systems” (377). and this gives rise to epistemological implications: the ability to “know” will vary. I consider how such concerns are more specifically related to the examination of complex international and global social phenomenon. All tools used by individuals involved in inquiry come with constraints such as theoretical assumptions or computational artifacts or a lack of ontological isomorphism. this modeling approach makes sense to important sets of constituents beyond the analytic community—for example. groups. Lempert. Put simply. The following paraphrases suggest how Bankes. does the model result in a convincing or resonant narrative of analysis or inquiry? To that end.168 DESMOND SAUNDERS-NEWTON While these methodological approaches allow us to represent the behavior of agents in the form of either individuals or institutions. they differ in terms of the mechanisms that allow for—or give rise to—interactions between the agents. Of course. Further. a prime concern as we consider the use of such a methodology is its credibility (Dewar et al.

Environmental processes and effects that are “not inherently agent based in character” can readily be reflected in hybrid computational models with significant agent-based components. these questions point to the necessity of identifying and developing criteria for performing research based on computational modeling. Whatever the shortcomings of non-ABM methods. I suggest that there are at least two important uses of models: prediction and exploration. the answer to this question when focused on non-ABM methodological approaches partly explains the many serious attempts to incorporate ABM into social science/inquiry practice. or institutions. the agent-based approach provides much greater facility for capturing the information that is available. To that end. given the nature of ABM as a computational method. While this modeling approach to representing human complex systems can be very effective. What we know about the relationships between such agents can easily fit into the agent-based mechanism. perceived or actual. and Popper 2002. what insights generated from the use of an agent-based model are valid relative to how we frame phenomena that occur in the world? Without doubt. this same question should be considered when employing other methodological approaches. I posit the following model-centric version of these questions as a springboard for considering the issue of computational epistemology: What must characterize a model in order for it to be useful in answering a scientific or policy question? Focusing on this variant of the epistemological question. such as numerical models based on systems of differential equations or symbolic logic or linear models of behavior. Lempert. how do we deepen our knowledge by performing computations? While philosophical in their nature. we are drawn to consider the nature or use of models of social phenomena. it is still necessary to consider the primary principle of epistemology: how do we credibly learn “things” about the world by using ABMs? Furthermore. 377). representation and verisimilitude are only a part of our epistemological concerns. More precisely. of the associated individuals. • • In summary. executing the resulting simulations can be used to infer the dynamic implications of the combination of knowledge and assumption that is incorporated in the model (Bankes. groups. Moreover. With respect to predictive models. and in comparison with competing modeling frameworks. we immediately are faced with challenges that have harried users of analytic methods for studying social processes .WHEN WORLDS COLLIDE 169 • The decision algorithms that these agents use can be based on knowledge and data available regarding the decision behavior.

if a model can be shown to accurately represent the world and predict outcomes. does not . This revision in perspective is important. are partially addressed by technological advances. Assuming a model accomplishes this goal. the need for perfect prediction is lessened. As asserted by Flyvbjerg. at least analytically.170 DESMOND SAUNDERS-NEWTON since the first half of the 1900s. In addition. Physics as a discipline is not viewed as any less rigorous for these debates. There continue to be questions across the disciplines of physical and biological sciences about mechanisms that define the world’s operations—for example. given that a single model capable of allowing for trusted assertions about the future states of a complex social system would reflect a “mirror world” (Gelernter 1991). more correctly. Much as a cured patient does not question the successful treatment. an isomorphic algorithmic artifact—would be a model sufficiently correct that we could peer into it and then learn about the world in which we exist. a model is predictive if its output is comparable to outcomes in the real world or actual system of interest within some well-characterized error process. Problems once viewed as intractable. and consequently supersedes the aforementioned challenges. the ability to predict the one systemic trajectory—diminishes as we come to value less the finding of the optimal responses to social problems. however. however. Some of these issues. These characteristics are admittedly something of a caricature of how social scientists practice their craft. as well as the correct frame for including important physical science principles such as path dependency and model parsimony. To that end. the numerous frameworks for reconciling the quantum world with the macroworld. epistemological issues are resolved. However. These challenges include a lack of veridicality and associated difficulty with identifying the “correct” representation of the nature and trajectory of human complex systems. Such a model—that is. the relative importance of path dependency—or. epistemological quibbles become irrelevant. By focusing on finding robust yet acceptable solutions that are valid across many possible or plausible future outcomes. Wallerstein 1996). For example. this challenge is not specific to ABM methodology. are less so as a result of advances in computational methods and reductions in the cost of computation. for a number of historical reasons. Moreover. The inability to predict. Many problems of interest to students of complex international and global social systems are not amenable to predictive modeling. part of this perception is likely driven by using the wrong metric for considering inquiry efficacy and quality. the inability to ascertain the “truth” of how the world operates is not a challenge only faced by those engaged in social inquiry. social science praxis has been viewed as being less precise and rigorous than colleagues involved in the physical sciences (Flyvbjerg 2001.

but serves as laboratory equipment. many epistemological concerns are mitigated. Lempert. complex systems are inherently unpredictable. Modeling based on computational experimentation has been called “exploratory modeling. specification. The . and scenario. agent-based and systems dynamics simulations. a model is not considered a mirror of the world. The rigor of experimental science is based on abductive and inductive logic. a computer-based model or simulation serves as a platform for performing computational experiments in which one can map the inputs for a specific case to the outputs that measure the associated systemic behavior. In fact. Given the aforementioned epistemological concerns and the failure of other methods to predict social outcomes. Thus.” In such an approach. a good model is not necessarily the one that is an isomorph of the actual system. Methodological differentiation allows us to contrast. and Popper 2002.WHEN WORLDS COLLIDE 171 and should not suggest that the ABM approach is not useful. the methodology for using computational approaches such as ABM embraces a transition from using just one single. as presumed. utilize. Moreover. The epistemology of experimental science has been considered in great detail. computations in support of inquiry or generation of “surprise” are viewed as an experiment. In such an instance. criteria other than the quality of point predictions must be used to measure the quality of theory. If. method. there is no reason to believe that such a model need be realistic at all. A very different and equally credible use of a model is in the context of a computational experiment that supports the use of models in “exploration. 379). the inability of ABMs to predict social outcomes should not be a primary measure of their efficacy. as noted by Bankes. but is rather one that can be used to perform crucial experiments that are useful in the context of an argument or problem (Bankes. Since predictive modeling arises from the praxis of theoretical science.” It is differentiated from predictive modeling by not attempting to limit the explicit uncertainty that arises from not having the one “correct” model. These alternative models can be differentiated along dimensions of theory. Theoretical differentiation suggests comparing model results that are attributable to the disciplinary frameworks underlying the models. one can use a computational model or simulation to perform experiments whose outcomes are useful in constructing credible arguments. In such a case. “correct” model of the world to the use of an ensemble of alternative models. it is biased toward deductive reasoning and measures research quality using the criteria of validity. In this inquiry framework. and is defined in terms of falsifiability and reproducibility (Popper 1979). and synthesize modeling techniques—for example. and by analogously casting computational social science problems in terms of experimental science.

the ensemble of models will likely contain more information than one single model. collaboratively and integratively (Saunders-Newton and Graddy 2001). it is easy to envision how changes in model specification. can give rise to very different results. but each provides useful insights into understanding social behaviors in isolation. Figure 9. as modified by Desmond Saunders–Newton. For the policy community. Knowledge Management inthe Intelligence Enterprise (Boston Artech house. Non-Peircian Reasoning Processes foci of these models are different (one is at the level of the agent and the other at the macrosystem level). which can be viewed as a separate model. Regardless of the basis upon which these alternative models are generated.2.172 DESMOND SAUNDERS-NEWTON Based upon Ed Waltz. 177. this speaks to coupling a given model with certain expectations about the world in which it operates. 2003). the scenario describes how a represented system may behave as a result of changes in policy or the social environment. it is possible to derive insights through the exploration of the properties of an ensemble of . As for scenario differentiation. by conducting large numbers of modeling experiments. p. With respect to specification.

Further. Moreover. and policy analysts have began to explore how ensembles of ABMs can even now be used to generate insights into complex social and political systems. The integration of these approaches is very dependent on the ability to exploit computational approaches—for example. it is not difficult to assert that the issues of interest in the realm of international security can be categorized as not only complicated. the relationship between human “wetware” and algorithmic “software” can be structured to leverage human ability to adeptly make inferences from complex patterns. one can use figure 9. and induction.2. . 1996.2 to differentiate between methods typically associated with standard hypo-deductive practice from those that make use of exploratory modeling to reflect reasoning approaches such as retroduction. a number of technologist. It is worth further emphasizing that these benefits are made possible by the availability and use of computational experiments. 1996. models and responses. the use of induction and abduction in a collaborative fashion reflects an intriguing interaction between human and machine “intelligences. This approach lies at the heart of the exploratory modeling approach (Bankes 1993. Saunders-Newton and Scott 2001). the reasoning process reflects an interesting variation on the efforts of Charles Peirce’s consideration of reasoning. Further. we move further along in the process of making use of algorithmic approaches to aid in decisionmaking and inquiry.WHEN WORLDS COLLIDE 173 alternative models that no single model computer model could reveal.In this instance. While for many these terms are often synonymous. Thus. Dewar et al. abduction. With respect to motivation. experimentation to expansively consider new hypotheses. As well as this ability to more effectively use alternative reasoning or inferencing approaches. by inferring across this ensemble as opposed to seeking insights from one model. TOWARD AN ALGORITHMIC SOCIAL SCIENCE The previous section considered how traditional critiques of computational models of social systems are not necessarily a true delimiting factor in the use of ABM. but also often complex. we can induce invariant properties and abduct robust responses to these model outcomes. the graphic illustrates the various interrelations between reasoning approaches. Before briefly describing an effort relevant to international and global studies. it is important to make explicit the motivation for such efforts and a rationale for using computational models as a key methodological pillar.” As illustrated by figure 9. scientists. and how this approach is being integrated into the policy community. In anticipation of an algorithmic social science with ABM as a prominent methodological approach.

if the problem is defined as chaotic. structured approaches to consider the problem. Figure 9. Complicated and Complex Problems they have very distinct meanings for research methodologists.4 . If such approaches are used to consider a “wicked” problem. a complicated problem can often be viewed as one with identifiable casual relationships and known uncertainties. and often viewed by practitioners and laypersons as near-chaotic.3. It is not difficult to further assert that the current international regime is easy to represent as complex. and to use a methodology that is appropriate to the problem domain. However. As illustrated in figure 9. Complexity science moves the problem from the chaotic to the complex domain. The objective then is to frame near-chaotic problems so that they are classified as complex. Unfortunately. Intelligence Innovation Division) and David Snowden (IBM). while ABMs provide a means to assess phenomena in such regimes. a complex problem is one much more consistent with emergent patterns. its supposed incomprehensible nature will often result in the unwillingness to use rigorous. or near-chaotic. the necessary simplifications often make the model and its associated results of questionable use to decision-makers.3.174 DESMOND SAUNDERS-NEWTON Mark Lazaroff (BAE Systems Advanced Information Technologies.

As a general statement.4. a computational (algorithmic) social science is the algorithmic instantiation of social science theory. As shown in figure 9. The ability to both make tough security problems less chaotic and provide a methodology more supportive of effective analytical narrative5 is a driving motivation for a number of Department of Defense near-term and longer-term research and development efforts using ABMs. models serve as a language for expressing theory. the language of algorithmic social science (computational symbolic processing) falls between natural language and mathematical formalism. Bridging these two “factions” is important as a means of exploiting the insights derived about individuals and their artifacts. such as institutions and technologies. More to the point. and maintain the rigor associated with mathematical dialects. 3). Computational models. While the efforts are fairly numerous. Thus. but also provides a bridge between the two most contentious epistemological factions that reside in the disciplines that study human processes: naturalistic inquiry and positivist or reductionist science. are a way to render theory in which a model is loosely defined as a representation of theory about real-world phenomena that serves as a bridge between theory and data. The Language of Algorithmic Social Science .4. it is able to provide insights into the deep narrative and rich detail associated with natural language. a few Figure 9. in order to deal with ever-present challenges such as violent conflict and complex humanitarian disasters. Another way of envisioning such a language is as analytic narrative. as defined by Taber and Timbone (1996.WHEN WORLDS COLLIDE 175 An algorithmic social science—an applied and theoretical transdiscipline based on the use computational models in the form of ABMs—not only affords a more appropriate consideration of a “complex” problem domain.

and identifying actions that could preclude or reduce the likelihood of such an event. This effort began in the spring of 2003 and was primarily concerned with developing a process and associated technologies for both anticipating “potential” violent conflict. System-Level Architecture of the Pre-Conflict Management Tools Program directly resonate with a substantial amount of the literature in international and global studies. The general conceptual format for this effort is shown in figure 9. One such effort is a program being managed out of the National Defense University’s Center for Technology and National Security Policy called the Pre-Conflict Management Tools (PCMT) Program. While the database—which is concerned with the near- . The PCMT approach was an attempt to efficaciously couple a self-sustaining analytic database with models of social vulnerability and authoritative networks and process so as to support activities occurring in a collaborative decision-making and analysis environment. These include efforts in the realm of preconflict and postconflict operations—that is.5. and also enhance the likelihood of postconflict success. It is further thought that information and knowledge of relevance to the avoidance of conflict can also be used to reduce conflict intensity and duration if such an event cannot be forestalled.176 DESMOND SAUNDERS-NEWTON Figure 9. actions necessary to avoid violent conflict or more effectively deal with the consequences of using military force.5.

In light of budgetary constraints. who focused on the relationship between civil war and development policy (Collier et al. categorization. the effort developed and integrated a limited number of models of social structure and dynamics in order to demonstrate an ability to use multiple models in a coordinated and structured fashion. the PCMT effort was to be composed of multiple models that allowed for the representation of individual and institutional behavior and interactions at various levels of resolution—for example. and a model of authoritative network dynamics. as well as traditional nation-state-level data—is of interest. The PCMT researchers employ computer-assisted reasoning methods in conjunction with simulation models such as ABMs to create large ensembles of plausible future scenarios. and are integrated by the use of computer-assisted reasoning methods. In addition.WHEN WORLDS COLLIDE 177 real-time collection. To that end. ABMs have proven quite helpful in the collaboration effort. The network dynamics models of authority were developed by the Institute for Physical Sciences. a qualitative computational model of internal conflict and state failure. it is necessary to consider an appropriate means of translating model output into decision-enabling information. including a single ABM as of the summer of 2005. because they provide an effective starting point for converting data into compelling information and knowledge for decision-makers spanning various US government agencies and important stakeholders from other societal sectors. These models comprise a collection of disparate modeling approaches. To craft information and knowledge so to increase its use in these decision-making environments.and regional-level data. regional-.6 It is important to note that ABMs are of importance in thinking about the issues of societal formation and fragility. and the qualitative model is a computational instantiation of the Fund for Peace’s Conflict Assessment System Tool (CAST). the initial PCMT effort made use of three models of social vulnerability and authoritative network dynamics. and national-level phenomena and system-level models. nongovernmental organizations. This particular stratagem supports a robust adaptive planning (RAP) approach to reasoning under conditions of complexity and “deep uncertainty” . The econometric model is based on the work of the World Bank’s Paul Collier. and visualization of local. the models include an econometric model of development and civil war. ABMs of local-. of greater relevance to this chapter is the aspect of the program concerned with identifying conditions that could precipitate violent conflict and with better understanding the actors and institutions capable of avoiding or ameliorating conflict. In terms of modeling classes. such as coalition partners. 2003). As originally envisioned. and multinational corporations. as well as the networks of authority capable of redressing or addressing conditions that give rise to social fragility.

Popper. the use of ABMs are viewed in a positive fashion by many persons interested in making better use of social science insights in future international security operations. particularly for the aspect of the PCMT architecture that is concerned with identifying “proponents and opponents to peace. In fact.178 DESMOND SAUNDERS-NEWTON that typically defeat analytic approaches (Lempert.” A number of organizations such as the Defense Modeling and Simulation Organization9 and the Defense Advanced Research Projects Agency are finding that ABMs are effective mediums for addressing the current analytic shortcomings in the area of effects-based operations (Saunders-Newton and Frank. 2002) as well as for considering notions such as long-term strategic assessment (Lempert. Popper. However. and their related models. a majority of all future models used in the social-vulnerability aspect of the program will be ABMs.7 It is anticipated that the RAP approach being used by the PCMT effort will provide greater insight into the vulnerabilities of societies and policies often undertaken to address these vulnerabilities. Interestingly enough. Such a capability underpins operational concepts such as “effects-based operations. Retraining analysts from formulating problems in a fashion amenable to a point prediction or optimal solution to framing problems in terms conducive to the generation of a robust solution or to computational experimentation reflects a cultural shift that will require time. it is evident that proposed solutions generated by these highly accurate models are not particularly effective in addressing the problems of greatest concern. and Bankes 2002). because they cannot have the predictive accuracy of models of fluid dynamics or classical dynamics. . However.8 CLOSING THOUGHTS In closing. the instantiation of many of the hard-earned insights of social processes in algorithmic models is aiding in their acceptance among many in this praxis community. It is fully expected that once the PCMT program moves from the technology demonstration stage into operational use. more accurate weapons do not change the conditions that give rise to terrorist or transnational criminal networks. an effort such as the PCMT program suggests that such a shift may be less than a generation away. The greater challenge may be less about ABMs than about their use. For example. and Bankes 2003).” and are viewed as credible ways of “knowing” in their own right. a number of efforts are currently exploring the use of ABMs to better understand how actions propagate forward into time as consequences and externalities. ABMs are an important component of this effort. I would be remiss if I did not note that the pervasive engineering culture may make more difficult the acceptance of social science disciplines.

and maintaining large ensembles of these societies and cultures. and would likely be impossible to realize without the use of agency-based models of culture. if we had a larger sample of planets to observe. However. societies or cultures created in-silica are now becoming increasingly possible. or rather the possible consequences of actions as they propagate into time. because they require thinking deeply about mappings between real-world and virtual-world ontologies. By instantiating. policies. are not overly different from the considerations we should make now when we attempt to use analytic tools based largely on Newtonian physics to understand social processes. Such a technical approach would be quite consistent with the earlier discussion of experimental design and computational epistemology. as well as means of considering the epistemological rigor and methodological sufficiency. Conversely. Unfortunately. the lack of pervasive faster-than-light technology and a severe shortage of observable inhabited planets make induction from a pool of planetary evidence difficult.WHEN WORLDS COLLIDE 179 One interesting approach under consideration would allow us to address an assertion often attributed to Herbert Simon: “[T]he soft sciences are actually the really hard sciences. we could likely infer generalizable insights that would aid us in thinking about actions on our own world. however. NOTES 1. It should be noted that some suggest that Peirce’s ultimate . Such considerations are the price we must pay in order to understand the answers we receive to the questions we ask. While the concept of retroduction as originally defined by Charles Sanders Peirce has multiple—seemingly contradictory—meanings. By exploiting interesting advances in the realm of computational anthropology (Gessler 2002). as well as insights garnered from the observation of persistent massively multiplayer online gaming communities. it possible to envision the creation of artificial cultures as a means of better understanding the plausible trajectory of cultures or societies being impacted by formulated policies and strategies. The challenges for developing and using such modeling artifacts are daunting. These challenges. it becomes possible to assess the impact of strategies. A possible effort under consideration is the creation of persistent artificial cultures whose emergent institutions and artifacts can be viewed over thousands of computer years to assess social and cultural stability under various stressors.” Implicit in this assertion is our general inability to make generalizable statements based on a sample size of one. growing. and actions across extremely rich/deep computational models of different groups. I suggest that it can best be understood as the process of conjecturing a new hypothesis beyond a current frame of discernment coupled with a search for evidence to affirm the new hypothesis.

means of sharing complex information or knowledge supportive of the decision-making process. no. REFERENCES Bankes. . In actuality. Future variants of the PCMT effort will move toward including additional ABMs into the modeling suite. nonlinear dynamics. In July 2003. methods beyond ABMs will allow complexity perspectives to address problems in the chaotic domain—for instance. The expectation is to use insights from this workshop as a means to improve the explanatory capabilities and level of realism of the next generation of models and simulations (Workshop 2003). As defined by Bayesian decision theorists. 6. Other recent efforts that have explored the use of ABMs include the Marine Corps Combat Development Center’s Project Albert and the Advanced Research and Development Activities’ Non-Linear Human Dynamics Program. or inference to the best explanation. 4. the prior probabilities for the uncertain parameters of the system model. 1993. 3. “Analytic narrative” is a phrase meant to suggest a rigorous. 9. Abduction. is defined as a form of inference that attempts to identify the most appropriate or plausible hypothesis for explaining a given collection of data or body of evidence. and induction under the general conceptual frame of retroduction (Chiasson 2001). deep uncertainty is the condition where the decision-maker does not know. 5.” Operations Research 41. This type of inference process is constrained by the quality of each hypothesis—individually and relative to one another—as well as the extensiveness of the search across “explanation or hypothesis” space and the net importance of drawing a conclusion (Josephson and Josephson 1996). 2. DMSO managed a workshop exploring the ability to model and simulate personality and culture. “Exploratory Modeling for Policy Analysis. 3 (May–June): 435–49. This refers to the less-than-perfect mapping between natural or realworld ontology and the ontological principles underlying the instantiation of a artificial society or culture in a computational environment. 7. abduction. and/or the value function. or multiple decision-makers cannot agree on. 8.180 DESMOND SAUNDERS-NEWTON intent was to relate deduction. yet compelling. the system model. S.

” World Bank Policy Research Report No. “Confronting Surprise. Elliott. 1991.. How It Will Happen and What It Will Mean. and Steven Popper. 2002 . 7–8). no. V. and Theories. 1931–58. Proceedings of the Center for the Study of Evolution and the Origin of Life. Gelernter. “Making Computational Social Science Effective: Epistemology. Shaping the Next One Hundred Years: New Methods for Quantitative. no. Phyllis. MA: Harvard University Press. Gillogly. Cambridge: Cambridge University Press. Gessler.” Social Science Computer Review 20. Philosophy. 2003. and Nicholas Sambanis. and Mary L. MA: Jones and Bartlett Publishers. Cambridge. Nicholas. Long-Term Policy Analysis.” In Evolution in the Computer Age. Methodology and Technology. 1996. “Breaking the Conflict Trap: Civil War and Development Policy. Peirce. Abductive Inference: Computation.. A. Dewar. Flyvbjerg. Weiss (Vols. 2002. Robert Lempert.. S. Josephson. CA: RAND Publications. ed. 2001. ed. Burks (Vols. Bankes. 2003. The Collected Papers of Charles Sanders Peirce. “Exploratory Modeling: Search through Spaces of Computational Experiments. 1994. Eve. Sara Horsfall. Anke Hoeffler. CA: Sage. eds. James. Oxford: Oxford University Press. Hartshorne and P. Chiasson. “Credible Uses of the Distributed Interactive Simulation (DIS) Environment. 26121. V. J.WHEN WORLDS COLLIDE 181 Bankes. T. NJ: World Scientific. Lee. ———. Santa Monica. Thousand Oaks. Technology. 1997. Popper. Sudbury. 353–60. Raymond A. Vye. Fogel. Robert. Hodges. Making Social Science Matter: Why Social Inquiry Fails and How It Can Succeed Again. 1–6) and A. 1996. Saunders-Newton. Mirror Worlds: Or the Day Software Put the Universe in a Shoebox. S. Complexity. Lempert. . Josephson. P. C. Collier. 2002. Bankes. and P. John R. Bankes. David. and J.” In Proceedings of Third Annual Conference on Evolutionary Programming. Amsterdam: Rodopi. Models. D. 4 (Winter): 420–40. 4 (Winter): 377–88. Peirce’s Pragmatism: The Design for Thinking. B. J. S.” Social Science Computer Review 20. L. S.. Sebald and L. Lucas. River Edge. and Susan G.” RAND Publications (MR-607-A). “Computer Models of Cultural Evolution. Cambridge: Cambridge University Press. Chaos. Marta Reyaal Querol. C. and Sociology: Myths. and S. 2001. Havard Hegre. Steven.

October. “Foreword: Chaos and Social Science. July 2003. Immanuel.. CA: RAND Publications. 1 (Spring): 47–65. “But the Computer Said! . Objective Knowledge: An Evolutionary Approach. “Effects-Based Operations: Building the Analytic Tools. . Delft.182 DESMOND SAUNDERS-NEWTON Popper. 2002. Taber. Saunders-Newton.” 2003. 2001. Palo Alto. 1997. “‘Way to Better Way’: Simulation Paradigms. Netherlands. ed. no. Saunders-Newton. 1996. Oxford: Oxford University Press. Karl R. D. CA: Sage. VA. Alexandria.” In Chaos.. Center for Technology and National Security Policy Publication. 19. Scott. Charles S. no. and Mary L. Fredrick. D. Santa Monica. Decision Inferences. Open the Social Sciences: Report of the Gulbenkian Commission on the Restructuring of the Social Sciences. Lee. “Workshop on Cultural and Personality Factors in Military Gaming. and Richard J. Graddy.” Social Science Research Review 19. Thousand Oaks. Computational Modeling.. Turner. 1979. and E. CA: Stanford University Press. Sponsored by the Defense Modeling and Simulations Office. 1996. and H. Complexity and Sociology: Myths. Wallerstein. Eve. A Typology for Using Computational Modeling Methods in Public Sector Decision-Making. Sara Horsfall. and Public Sector Enterprise Management.” Proceedings of the Eurosim Congress. Raymond A. Models and Theories. 2001. Saunders-Newton. Timpone. . .” Defense Horizon. D. and Aaron Frank..

emphasis in the original). I first show how complex systems concepts can improve how we think about and understand world politics. further elaborated the complexity concepts and ideas that may be used to construct complex systems theories of issue-areas in world politics. In a short coda. Rational choice preserves the status quo. . . In this chapter. This conveniently simple model has misled generations of scholars and policy-makers (Smith 2004). Thus. In the next section. I show how complexity could reform policy in world politics. Chapters 7. Like a cancer it changes minds and institutions until its simpleminded rationality seems utterly human: “Taking a preference for the maximization of self-interest or even utility as a given begets both a cognitive and a political reality in which individuals and political leaders alike come to view such behavior as normatively acceptable and as the standard by which government should operate. In the third section. I summarize the many benefits of the complexity 183 . . 8.C H A P T E R 10 Complex Systems and the Practice of World Politics Neil E. and 9 debated the epistemology and methods of complex systems. I consider Earnest and Rosenau’s epistemological critique of complex systems theory in chapter 8. . Harrison The study and practice of world politics has for too long been distorted by rational choice. thereby. . . Chapters 3 through 6 illustrated complexity and its benefits by applying complexity concepts and sketching complex systems theories for specific issue-areas. public policy as is becomes the public policy interest as it ought to be” (Petracca 1991. This book has proposed a better way of understanding world politics. Chapter 1 described the complexity paradigm built on an understanding of the characteristics of complex systems and shows how ideas from complexity can be adapted to world politics. Chapter 2 compared general and complex systems taxonomies and.

Constructivism sees identity as formed and changed through socialization in the international system. He treats international negotiations as the coevolution of adaptive states. He does not deny the usual explanations of the causes of the violence. Improving Current Theory Complex systems concepts can extend or elaborate current knowledge. Hoffmann shows how a complex systems theory can provide the microfoundations to international negotiations and agreements. because “agents and structures are produced or reproduced by what actors do” (Wendt 1994. though more statically. 385). 390). The ethnic hatreds. Constructivists describe identity as “grounded in the theories which actors hold about themselves and one another and which constitute the structure of the social world” (Wendt 1992. This language is close to complexity concepts of internal models. Throughout this chapter I indicate several paths for further development and application of the complexity paradigm.184 NEIL E. and allow exploration of new possibilities. Bhavnani’s analysis of the Rwanda genocide also uses complexity concepts as an adjunct to conventional theories. which fixes attention on the internal processes by which states exercise their freedom in choosing their identity and interests and thereby influence international processes and other states’ choices. Conceptualizing the state as a complex system that is an agent in the international system (a meta-agent). much as suggested. A BETTER WAY TO UNDERSTAND WORLD POLITICS This book has outlined the concepts of a complex systems taxonomy for world politics and offered four cases that demonstrate their application. Constructivism argues that interests and identity are constructed through interaction between states. HARRISON paradigm and the theories it can spawn. a critical measure of a new taxonomy (see chapter 2). there is no explanation of the microfoundation of macroprocesses (and so of the sources of change in the international system). 397). Hoffmann fills this gap in constructivist theory. constructivism cannot explain how states exercise this freedom of choice. agents have a degree of freedom that introduces potential for dynamic system change. in Putnam’s two-level games (1988). Complex systems concepts can improve current theory. generate novel insights. government propaganda. Structure does not determine agent choices. but complexity expands behavior to include both the emergent domestic processes and international coevolution. For example. and . Thus. But as states are treated as units (Wendt 1994. Thus.

it illustrates complexity’s possibilities for innovative thinking about old and new problems. As most conventional theory is at the level of the international system. A bifurcation in the cultural paths of some peoples seems to have made them more resilient under empire and more successful at exploiting the opportunities presented by its demise. and the majority of scholars in the field recognize these two conditions (Kuhn 1970). Because it shows similarities with some theories of economic development (e. Not only have rational choice paradigms failed to explain much of the institutionalism of the modern international system but they also have misled policy-makers (Smith 2004). the dynamic evolution of the killing can be captured by complex systems concepts that fill out the narrative of simpler models. Reminiscent of Weber’s (1958) explanation of capitalism..g. a new paradigm is available that explains more than the old. it may have application elsewhere. a constant stream of new ideas and hypotheses is essential to understanding. Paradigms change because explanatory failures of the old paradigm have accumulated. there are many opportunities to add microfoundations with complexity concepts (which do not preclude greater simplifications at higher levels of analysis). A complex systems theory of development led Clemens to ask different questions and search for data in new places. Complex systems concepts can generate radically novel hypotheses. however reasonable. Clemens offers some empirical evidence in support of his hypothesis and outlines how to more fully test it. . However. Whether or not the hypothesis is disproved. For example. Issuearea theories also can be adapted. Walt Clemens hypothesizes that the Protestant practice of debating holy texts institutionalized activities essential to effective democracy and open markets: literacy and open and civilized debate on policy matters.COMPLEX SYSTEMS AND THE PRACTICE OF WORLD POLITICS 185 the death of the Hutu president are well recognized. cannot explain the rapidity and magnitude of the killing. He uses complexity concepts to answer a puzzle that conventional theories cannot touch: the initial conditions of the Rwanda conflict. and deserves further investigation. this is an unexpected but very plausible hypothesis generated using complexity concepts.1 Generating Novel Insights In a dynamic world. complex systems could explain how epistemic communities actually influence states’ policies in international environmental issues. Informal institutions developed from religious conviction generate politically effective behaviors that manage ethnic differences and support economic development. While correlation is not causation. De Soto 2000). Clemens’s argument is novel and certainly plausible.

”3 Positivism asserts that there are two sources of knowledge: deductive logic and empirical evidence. The meaning . Using metaphor and argument rather than models. There is agreement on what epistemology is but not on how to pursue it. and Einstein’s thought experiments helped to develop relativity theory. THE NATURE OF POLITICS AND EPISTEMOLOGY AND METHODS Earnest and Rosenau’s critique of complex systems theory is rich and detailed. Saunders-Newton argues that exploratory simulations through an ensemble of models can generate more information. systems. Epistemology Earnest and Rosenau’s critique is based in a positivist epistemology. it is problematic in social science and rejects simulation as neither logical nor empirical and. the latter must be verifiable by experience.186 NEIL E. than a single model can. by definition. there can be no theories of complex systems. thus. Abstract mathematical models are used to work through the logic of relationships among physical variables. as without meaning. they question whether world politics in reality is a complex system as commonly understood. HARRISON Exploration In addition to consolidative or predictive modeling that represents reality with extensive data sets. they note that there is no epistemology of complex systems in world politics. they argue. First. Second. limits (constrains) self-organization. Sandole shows that contending views can be reconciled and policies crafted to prevent ethnic conflict or mitigate its effects. So. Without one. Knowledge is conventionally defined as something like “justified true belief. Because politics is about authority that.2 complexity can generate exploratory models to test ideas about the relations of agents. more flexibly. There is not space here to respond fully to every matter they discuss. By reducing uncertainty and generating novel insights into social behavior and multiple action options. and environment. The former is a priori true. Epistemology is the effort to distinguish true knowledge from false and helps to determine if one theory is better than alternate theories. exploratory modeling can improve policy. they criticize the potential for isomorphism between model and reality in world politics. in this section I assess the reasonableness of the two premises on which they found their arguments. While positivism is the dominant world politics methodology.

Humans are less deductive and more inductive than positivism posits. the knowledge system becomes ever more intelligent and adaptive. and social processes and retained according to their contribution to individual or group survivability (Campbell 1960). . Once selected. is to map theoretical constructs onto observable reality (much like positivism). Finally. Blind variations (almost guesses) in knowledge are selected through biological. Organized in a nested hierarchy of selectors. because (unlike the physical sciences) social science reality does not fit into binary logic categories. in the social sciences the “cross-mapping and redundancy” that simulation can provide are an especially valuable contribution to knowledge accumulation (a point Saunders-Newton also makes). But human subjects can “belong to one or more of a number of referent groups or be in any one or more of a number of psychological states” (para. knowledge variations become part of the selection mechanism for further variations.16). Positivism is problematic in the social sciences because of problems of observation and of correspondence between abstract concepts and real objects. The principle of downward causation—“all processes at the lower level of a hierarchy are restrained by and act in conformity to the laws of the higher level” (Campbell 1974)—completes a complexity model of knowledge formation. 5. The problem of identifying valid empirical indicators for theoretical concepts is broadly recognized but overlooked in deference to the simplicity of positivism. it is never a molecule. in theory.” For example.4 The best empirical verification of a proposition is correct prediction of observable events. but the parts are partially constrained by the whole (downward causation). But mapping in the social sciences is shaky at best. The purpose of science. The whole is partly constrained by the behavior of the parts (emergence).5 In this view hypothesis generating simulations and exploratory modeling are as legitimate sources of knowledge variations as any other. Evolutionary epistemology is one explanation of knowledge accumulation that should be considered for complex systems in social science. Marney and Tarbert (2000) discuss an epistemology in which simulation in social science is a valuable “third leg” supplementing and complementary to theorizing and empirical testing. Arthur (1994) similarly rejects positivist/rationalist theory in economics as a beautiful model that bears little relationship to the reality of human behavior.COMPLEX SYSTEMS AND THE PRACTICE OF WORLD POLITICS 187 of a statement is equivalent to a definition of the empirical conditions under which it is true or false. psychological. Thus. in effect mimicking selection by reality. an epistemology can be designed to specifically accommodate simulations generating “justified true belief. An electron is nothing but an electron. they argue. Thus. a statement is meaningful only if it can be observed or can. be observed (Ayer 1936). Physical objects are either members or not members of a category.

and the influence of authority in the modern world political system can be better captured through complexity concepts than through a simple model more relevant in a past era of state dominance. but some have more central control. By 2000. Second. it is because of less centralized authority. and thus less complexity. proliferating democracy diminishes centralized authority and increases self-organization and complexity. Informal institutions. Arendt 1958). social systems are not binary—for example. politics is the process by which the institutions governing collective life are organized. Often in social situations there is a third way: not to follow or refuse .188 NEIL E. Following this latter view. Freedom House reports that in 1900 no state was an electoral democracy with universal suffrage. than others. HARRISON Authority in Complex Systems Taking authority as the critical variable in world politics. Some agents in the system always have more influence over the form and function of institutions than other agents do. They can choose to follow orders or they can refuse. authority operates through formal and informal institutions. Third. Coupled with economic liberalization. I suggest that all societies are complex and can be modeled with complexity concepts. Politics has also been defined as the formation and rivalry of groups (Schmitt 1976) and. is a value-laden move. either authoritarian or not. While individual agent decisionmaking may be as or more limited than before. the existence of authority is not fatal to complex systems theories of world politics. Understanding politics this way. Easton’s (1981) definition of politics as the authoritative allocation of valued things is not the sole conventional definition of politics nor necessarily the most appropriate for our purpose. are shared meanings and emerge from agent interactions mediated through prior states of such institutions. accepting punishment. agents always have choices. The decentralization of authority from the state to a large number of diverse private and public organizations competing economically and politically itself creates complexity. as the generation of the structures and norms that govern human collectivities (for example. As Earnest and Rosenau acknowledge.6 Even within highly centralized authority systems. it is eminently reasonable to model it as a complex system. globalization is diffusing authority from the state to other organizations (see Strange 1996). more generally. First. for authority to be fatal to complexity it must be centralized. It is always a matter of degree. as do Earnest and Rosenau. For three reasons. except for subscribers to the realist view. 120 of 192 countries were rated as electoral democracies. like cultural practices. But authority is not the whole measure of politics and. not the sole object of research in world politics.

working around the rules to personal advantage. some complex systems can be described by Markov chains in which the distribution of future events is independent of the history of the system. However authority is defined and handled within complex systems theories. and phenomena are “independent entities. the Soviet state did not absolutely control the behavior of its people. This permits investigation of how authority emerges. it is an advance in world politics that the effect of authority is even considered problematic. the future state of the system is a probabilistic function of the present. and popular resistance became informally institutionalized. and survival and event history analysis. Common mathematical techniques for analyzing stochastic dynamic systems in several disciplines (including economics) can be adapted for complex systems. For example. Cioffi-Revilla (1998) uses several mathematical methods to develop a formal theory of politics and uncertainty. the relevant statistical techniques are generally referred to as “Markov-chain Monte Carlo. At any time. Under these conditions. This human ability to adapt to environmental conditions reduces the importance of formal authority and makes system response to authority less predictable.7 For all its control and punishment. nonlinear and maximum likelihood estimation. modified Monte Carlo methods for estimating sample distributions can reduce computational complexity. model. . and dissolves and how it influences formal and informal institutions. how macrobehavior comes from microevents. Earnest and Rosenau rightly argue that authority should be treated endogenously as a property of the system. and the influence of macrocontext on microlevel phenomena.COMPLEX SYSTEMS AND THE PRACTICE OF WORLD POLITICS 189 but to “game” the system. In rational choice theories. His methods go beyond ordinary least-squares regression analysis to include Boolean logic. In this and related epistemologies. evolves.” Richards (2000) reviews mathematical methods for modeling nonlinear political systems. Models of complex systems may be mathematical or computational. So. without regard to the past. theory.” and science comprises analytical and ontological activities that relate theory to phenomena with models—and theories become sets of models (Henrickson and McKelvey 2002). he discusses nonlinear systems. Although he does not specifically refer to complexity or complex systems and he does not theorize from individual agents. Complexity Methods Evolutionary epistemology rejects positivism’s deductive creation of a theoretical world from self-evident truths (axioms) in favor of models that represent reality. the role of authority is clear but clearly wrong.

rational choice theories is that they foster the belief that the causes of problems and the consequences of alternate responses can be known with a high probability. You observe the world very. Still others may wish only to specify units and interaction rules and “outsource” model construction to computer programmers. applied mathematicians. Thus. puts a premium on becoming aware of nonlinear relationships and causal pathways as best we can. rather than what’s “optimal”. HARRISON Every paradigm comes with its own language. a skill quite common in comparative politics and IR. that policymakers recognize the inherent uncertainties in their understanding of both the system and the effects of our interventions therein. Some students of world politics will build custom programs or learn advanced mathematics but others may use several proprietary or open source software simulation packages (Agent Sheets. or survivability.). Sandole argues that Realpolitik policy can never end ethnic conflict. because it feeds off the biological belonging that bifurcates the social world into “us” and “them” and also drives ethnocentrism and ethnic conflict. As with any paradigm. the choice of method depends on the research puzzle being investigated. To intervene effectively in complex systems requires. in the face of an illdefined future. REFORMING POLICY The most dangerous aspect of positivist. or statisticians. for some scholars the linguistic gymnastics of postmodernism is as challenging as formal theory and mathematical notation are for others. Second. Allison 1971). something that’s workable. Swarm. etc. and you don’t expect circumstances to last” (quoted in Waldrop . success in the War on Terror demands accepting a more inclusive and nuanced interpretation of events than offered by simple Realpolitik. .190 NEIL E. Complex systems in world politics demand policy caution. very carefully. What you’re trying to do is maximize robustness. The most important service of a complexity paradigm might be to free policy-makers from this overweening hubris and inculcate a sense of uncertainty. first. because optimization isn’t well defined anymore. The greatest contributor to the success of the Kennedy administration’s nuanced response to the Cuban Missile Crisis was the failure of the overly simplistic decision-making for the Bay of Pigs (Janis 1982. The modeling challenge of complexity really is no different from the difficulties of researching across language and culture barriers. . in turn. policy must seek out points of leverage that may be well hidden. You go for viability. And that. Brian Arthur suggests that when intervening in complex systems “you want to keep as many options open as possible.

Positive and negative incentives work. If it’s wildly variable. Because complex systems are counterintuitive. good policy requires thinking broadly about problems and finding leverage points for intervention. A TIT FOR TAT strategy reflects complexity by combining the need for order and security with “niceness” and “foregiveness. neither Realpolitik nor Idealpolitik can stand alone. changing parameters in the system—is “diddling with details. malfunction of systems can often be traced to the rules and “who has power over them. As mutations drive evolution. the most important leverage is the “paradigm or mindset out of which the system arises. a social group’s behavior comes from the interaction of its internal model with external reality. This is a powerful and dangerous point of intervention that governments concerned with control and predictability rarely consider.” This is simple policy. technological innovation (Nelson and Winter 1982). The best that policy-makers can hope for are policies that are excellent rather than optimal.” Second. As scholars of world politics have long realized.” But parameters are where we put “probably ninety-five percent of our attention. If the system is chronically stagnant.”) and Reagan’s call to get government off the backs of the people. self-organization drives economic processes.8 While consistent pursuit of ideological purity may enhance reelection. . . 331–34).” Cultural norms work on each individual’s internal . First. Realpolitik advocates regulating the conflicting groups’ environment to increase selection pressure for appropriate changes in their internal models. pragmatic policy is more realistic. Changing the goals of the system works through the internal models of system agents. But internal models can also be influenced by education. as Idealpolitik advocates. In complexity terms. this is Kennedy’s inaugural address (“ask not what government can do . . arranging the deck chairs on the Titanic. In Sandole’s analysis of ethnic conflict. A third and bigger point of leverage is the goals in the system. .COMPLEX SYSTEMS AND THE PRACTICE OF WORLD POLITICS 191 1992. Combining education and selection increases the rapidity of learning and norm change. Cautious intervention demands a broad conceptualization of possibilities. But building Idealpolitik’s positive peace usually requires Realpolitik’s negative peace of militarily imposed stability. and they need to be prepared to change. In political terms. Meadows’s four most effective intervention points are the most relevant to the present discussion. changing the rules of the system changes behavior. and other social changes.” Order and security alone cannot allay underlying tensions and change hearts and minds. increasing social diversity increases self-organization and emergence: “Let a thousand flowers bloom and ANYTHING could happen” (Meadows 1997). . Meadows (1997) suggests nine possible leverage points. parameter changes rarely kick-start it. Finally. they usually don’t stabilize it. The least useful—namely.

it means reinforcing changes in both institutions and internal models. HARRISON model and through informal institutions select behaviors that are socially appropriate. In Gramscian language. Gramsci’s approach was incremental. and points of leverage are never simple. In this chapter. solutions have uncertain effects. and ABM simulations. But I do accept their general position. Redirecting political trajectories requires change in both structures and ideas. Yet. But complex systems can be pushed and prodded. in the language of complexity. etc. exogenize authority. and changed. but ideational change may be better accepted when it is so sudden and substantial as to be an epiphany. dethroning the rational choice paradigm is the best way for scholars to positively influence world politics. Industrial societies will continue to blithely consume nonreplaceable resources as long as their paradigm is anthropocentric mastery of “nature” (Harrison 2000). Differences between issue-areas in agents’ values and beliefs and in informal institutions (norms) and formal ones (organizations. Because institutions and social systems are influenced by human perceptions of the world and how it works. caution is required and instruments are imprecise. But policy under complexity opens many other avenues of research. the essence of political systems. But theories—sets of premises. I have argued that epistemologies for complex systems theories can be found and that authority in modern industrial democracies does not defeat complexity and its simulation. the increasing acceptance of the rational choice paradigm in policy circles makes scholars complicit in the policies that form the reality of world politics (Smith 2004). There are no acceptable standards by which we can know knowledge of complex political systems. complex system theories of individual issue-areas in world politics are possible. TOWARD THEORY Earnest and Rosenau argue that there cannot be a general “complex systems theory” in world politics.192 NEIL E.) may prevent an allencompassing complex systems theory. yet. and the benefits are likely to be great. Complexity is more than a metaphor—the taxonomy outlined in chapters 1 and 2 forms a basis for theory-building—but less than an encompassing theory. a social paradigm or “mind-set” is supported by a hegemony of economic and political goals and intellectual and moral discourse (Hoffman 1984). problems are unclear. Some of the central tenets of complexity—emergence and the importance of context and initial conditions—do make a generally applicable and ahistorical complex systems theory of world politics logically impossible. In complex systems. In world politics. though for a different reason. assump- . the methodological basis for a complex systems theory.

Strange 1996). For example. 3. This book has defined complexity concepts and ideas and introduced many potential theoretical challenges. most cars on the interstates . the institutions that mediate agents’ interactions are unique to each issue-area. 7. At the international level. The general population breaks official rules regularly and with relative impunity. Humans have different expectations in regard to environmental issues—more ethical and inclusive ones (Harrison 2000)—than in regard to economic issues. Popper’s view of science as accumulating knowledge by the selection mechanism of refutation operating on randomly evolving conjectures. where they are more individualist. Einstein’s conclusion that gravity bends light was only a proposition. but the conditions under which it could be verifiable were clear. Risse-Kappen (1994. until the proof by Eddington in 1919. the conventional or short-form definition of knowledge is adequate. the World Trade Organization and the United Nations Environmental Programme are structured and operate very differently. and statements of their relations from which testable predictions can be made—that endogenize these issue-area characteristics are not only possible but preferable.” 2. The complexity paradigm offers a novel perspective on world politics at all levels that will generate new theories and models of issue-areas. and they pursue their goals differently in each issue-area.g. concepts. It was conventional to define knowledge so until Gettier (1963) showed that this is an incomplete definition without some additional conditions. 5. 6. For example. Similarly. The IRS audits about 1 percent of all returns. and throughout this chapter I have noted many directions for theoretical development of theories of complex systems and their empirical testing. The extent to which states have lost authority or other organs have gained authority is much debated (e. 4. It also encourages innovative methods for understanding political reality and advising policymakers. 187) argues that the theory of epistemic communities has failed to specify “the conditions under which specific ideas are selected and influence policies while others fall by the wayside. Historical experimentation discussed in chapter 2 would be of this type if the right data could be found.. NOTES 1. For our present purposes.COMPLEX SYSTEMS AND THE PRACTICE OF WORLD POLITICS 193 tions. This paradigm can increase our understanding of the complexity of world politics and reduce the probability of surprising events.

“Evolutionary Epistemology. ———. no. Easton. David. Leslie. 2:406–11. 1994. Ayer. The Mystery of Capital: Why Capitalism Triumphs in the West and Fails Everywhere Else. Cioffi-Revilla. 413–643. Arendt. 1963. Hernando.” See Kauffman 1995. Boston: Little. 1958.” 8. 3 (May 14): 7288–95 (accessed at www. W. Hannah.” Psychological Review 67. 2000. no. Claudio. and Logic. Gettier. Campbell.pnas. IL: Open Court. Henrickson. Postmodernism.” American Economic Review 84. . HARRISON drive up to ten miles an hour over the speed limit (seemingly accepted as a “safe” margin by drivers and police). 1960. Language. A. “Blind Variation and Selective Retention in Creative Thought as in Other Knowledge Processes. 2000. 2005). New York: Basic Books. Graham T. Brown. 14. no. Cambridge: Cambridge University Press. 1936. The Library of Living Philosophers. “Foundations of ‘New’ Social Science: Institutional Legitimacy from Philosophy. Paul Arthur Schilpp.194 NEIL E. De Soto. Politics and Uncertainty: Theory. LaSalle. 2002.” In The Philosophy of Karl Popper. Chicago: University of Chicago Press. REFERENCES Allison. “Is Justified True Belief Knowledge?” Analysis 23:121–23. J. Models and Applications. “Complexity in Economic Theory. Complexity Science. Edmund L. and Agent-Based Models. 1971. Neil E. 1974. The Essence of Decision. Arthur. and Bill McKelvey. Brian. Truth. This is “satisficing” (Simon 1955). and perhaps 10 percent of GDP is unmeasured and belongs to the “black economy. Albany: State University of New York Press. New York: Oxford University Press. The Political System: An Inquiry Into the State of Political Science. 1981.” Proceedings of the National Academy of Sciences of the United States of America 99. One mechanism for finding an excellent (but not the very best) solution is “simulated annealing. 6:380–400. ed. Harrison. The Human Condition. 248– on May 15. Donald T. Suppl. Constructing Sustainable Development. Chicago: University of Chicago Press. 1998.

Risse-Kappen. 2003).. Stuart. 1982.” Journal of Artificial Societies and Social Simulation 3.soc. E. “Singing our World into Existence: International Relations Theory and September 11. Schmitt. 1995. Donella H. 1991. NJ: Rutgers University Press. Richard R. At Home in the Universe: The Search for the Laws of SelfOrganization and Complexity. Kuhn. 1996.html). New York: Oxford University Press.. “Why do Simulation? Towards a Working Epistemology for Practitioners of the Dark Arts. Cambridge: Cambridge University Press. John. Nelson. 3 (Summer): 427–60. “The Rational Choice Approach to Politics: A Challenge to Democratic Theory. Winter. Steve. 2004. Thomas. . Herbert A.” Whole Earth.. Marney. Thinking Theory Thoroughly: Coherent Approaches to an Incoherent World. Thomas. “A Behavioral Model of Rational Choice. Chicago: University of Chicago Press. Boulder. New York: Basil Blackwell. no. New Brunswick. Strange. CO: Westview Press. and Mary Durfee. Kauffman. J. 1970. An Evolutionary Theory of Economic Change. 2000. 1982. 1997. Richards.” International Organization 48.COMPLEX SYSTEMS AND THE PRACTICE OF WORLD POLITICS 195 Hoffman. no. “Diplomacy and Domestic Politics: The Logic of TwoLevel Games.wholearthmag.html on February 7. no. Winter (accessed at http://www. Smith. Janis. F. 2 (Spring): 185–214. 1976.” International Organization 42.” International Studies Quarterly 48. and F. Carl. Robert D. Putnam. 1994.” Review of Politics 53 (Spring): 289–319.” Quarterly Journal of Economics 69 (February): 99–118. MA: Harvard University Press. The Gramscian Challenge: Coercion and Consent in Marxist Political Theory. Ann Arbor: University of Michigan Press. 1988. Groupthink. 3 (September): 499–515. and Sidney G. Political Complexity: Nonlinear Models of Complexity. Meadows. “Ideas Do Not Float Freely: Transnational Coalitions. James N. Susan. 1955. Simon. surrey. Domestic Structures and the End of the Cold War. Boston: Houghton Mifflin. 4 (accessed at http://www. Rosenau. Diana. Cambridge. The Concept of the “Places to Intervene in a System. Tarbert. 2000. 1984. Mark P. 2000. The Retreat of the State: The Diffusion of Power in the World Economy. The Structure of Scientific Revolutions. 2nd ed. no.

” American Political Science Review 88:384–96. Wendt. Trans. “Anarchy Is What States Make of It: The Social Construction of Power Politics. 1992. no. . Max. HARRISON Waldrop. Alexander. M. Talcott Parsons.” International Organization 6.196 NEIL E. “Collective Identity Formation and the International State. 1994. 1992. 2 (Spring): 391–425. New York: Charles Scribner’s Sons. Weber. New York: Simon & Schuster. Mitchell. Complexity: The Emerging Science at the Edge of Order and Chaos. 1958. ———. The Protestant Ethic and the Spirit of Capitalism.

and popular rebellion in Sub-Saharan Africa and South Asia. The Complexity of Cooperation: Agent-Based Models of Competition and Collaboration and (with Michael D. initiated in Europe circa 1500. 2004). 2003). His current research interests include complexity theory (especially agent-based modeling). He is best known for his interdisciplinary work on the evolution of cooperation which has been cited in more than five hundred books and four thousand articles. and Associate. is Professor of Political Science. JR. He has taught at the University of California at Santa Barbara and at M. and along the Pacific Rim.T. and lectured widely in Eurasia. Boston University..B. RAVI BHAVNANI. Ph. Magna Cum Laude.Contributors ROBERT AXELROD is Arthur W. Bromage Distinguished University Professor of Political Science and Public Policy at the University of Michigan. in political science (University of Michigan-Ann Arbor.D. explore the application of complexity to understanding politics and organizations. ed. His most recent books. with an emphasis on agent-based modeling. from Notre Dame University. His research focuses on the micro-foundations of mass participation in ethnic violence.I. 197 . and Ph. from Columbia University. He has been elected president of the American Political Science Association for 2006-07. His current research focuses on the ways that revolutions in literacy and free thought. He has written or edited fifteen books. Harvard University Davis Center for Russian and Eurasian Studies.D. CLEMENS. and international security. including The Baltic Transformed: Complexity Theory and European Security (2001) and Dynamics of International Relations: Conflict and Mutual Gain in an Era of Global Interdependence (2d. is an Assistant Professor at Michigan State University. WALTER C. the Americas. He received his doctoral degree in comparative politics and methodology. have contributed to societal fitness as understood by complex systems theory. Cohen) Harnessing Complexity: Organizational Implications of a Scientific Frontier. Clemens studied in Vienna and Moscow but received an A. civil war.

international relations theory and political methodology. and social constructivism. D. His recent publications include the book Ozone Depletion and Climate Change: Constructing a Global Response from SUNY Press (2005) and a coedited volume Contending Perspectives on Global Governance with Alice Ba (Routledge Press 2005). His book Constructing Sustainable Development (SUNY 2000) linked both research interests. DC. DENNIS J. He also has co-edited Science and Politics in the International Environment (Rowman and Littlefield 2004) with Gary Bryner and has published other articles and chapters on international environmental politics. with James N. Norfolk. NEIL E.198 CONTRIBUTORS DAVID C. HARRISON’s research interests include complex systems and international environmental politics. 2005. EARNEST is an Assistant Professor of Political Science and International Studies at Old Dominion University. Ole R. Virginia. he has authored a number of books and articles. and Taiwan. A founder-member of ICAR. Along the Domestic-Foreign Frontier: Exploring Governance in a Turbulent World (Cambridge 1997). where he teaches international political economy. while his methodology interests are in the application of agent-based models to problems of international politics. His research focuses on global environmental governance. He received his doctorate from the Graduate School of International Studies at the University of Denver in 1994 and has taught in Colorado. Among his recent books three stand out retrospectively as a trilogy: Turbulence in World Politics: A Theory of Change and Continuity (Princeton 1990). He is professor of conflict resolution and international relations at the Institute for Conflict Analysis and Resolution (ICAR) at George Mason University. He is co-author of On the Cutting Edge of Globalization: An Inquiry into American Elites (Rowman and Littlefield. JAMES N. His main areas of interest include “identity” conflict/conflict resolution in the . SANDOLE received his Ph. Scotland in 1979. Holsti and Yale H.D. Rosenau. Ferguson). in Politics from the University of Strathcyde in Glasgow. Previously he held an appointment as a Fellow in Political-Military Studies at the Center for Strategic and International Studies in Washington. and Distant Proximities: Dynamics Beyond Globalization (Princeton 2003). MATTHEW J. he worked closely with conflict resolution pioneer John Burton in Britain and the US. ROSENAU is University Professor of International Affairs at The George Washington University. His substantive research focuses on the political incorporation of migrants in democratic societies. Wyoming. HOFFMANN is an Assistant Professor in the department of Political Science and International Relations at the University of Delaware. complexity theory. A former president of the International Studies Association.

understanding and dealing with the complex etiology of what has come to be called in the Western media “suicide bombing” and other acts of terrorism. He founded the Correlates of War Project and was its Director from l963 to 2003. Caucasus. DAVID SINGER earned his Ph. and Lawrence University (B. In addition to his work in the areas of international security and science/technology policy.. DESMOND SAUNDERS-NEWTON is a member of the senior management staff of BAE SYSTEMS Intelligence Innovation Division where he leads the Social Computation and Complexity Directorate.). in 1956 from New York University and is Professor Emeritus at the University of Michigan. a senior scientist and program director at the National Defense University’s Center for Technology & National Security Policy.A. M. J.Phil). He also is an Adjunct Associate Professor in the University of Southern California’s School of Policy.D. Central Asia and Southeast Asia. and Mannheim. He has held appointments as a consulting scientific advisor at DARPA. and over one hundred articles including “The Level-of-Analysis Problem in International Relations” published in World Politics in 1961.CONTRIBUTORS 199 Balkans. Taipei. . with Melvin Small). His most recent book is Capturing the Complexity of Conflict: Dealing with Violent Ethnic Conflicts of the Post-Cold War Era (Thomson Learning 1999). Groningen. University of Michigan (MPP). Saunders-Newton has over a decade of experience in research methodology and social science research (domestic and international) in a variety of research agencies. e.D. He holds degrees from the Pardee RAND Graduate School (Ph. He has published more than twenty books including Nations at War: A Scientific Study of International Conflict (Cambridge University Press 1998 with Daniel Geller) and Resort to Arms: International and Civil War. and has pursued post-doctoral studies at RAND and the Santa Fe Institute. and served as the American Association for the Advancement of Science S&T Advisor to the Deputy Under Secretary of Defense for Advanced Systems and Concepts. He has taught at Vassar and in Oslo. Planning & Development and a member of the External Advisory Board of George Mason University’s Center for Social Complexity. 1816–1980 (Sage 1982. Middle East. Dr..g. Geneva.

This page intentionally left blank. .

and Power: Successful Sanctions in United States Foreign Policy—George E. and Tony Porter (eds. Duina Economic Interdependence in Ukrainian-Russian Relations—Paul J.) Which Lessons Matter? American Foreign Policy Decision Making in the Middle East. 1979–1987—Christopher Hemmer (ed. Peace. Singh States. Crawford and Darryl S. Hobbs (ed. Weiss (eds. Editor American Patriotism in a Global Society—Betty Jean Craige The Political Discourse of Anarchy: A Disciplinary History of International Relations—Brian C.) 201 .) Harmonizing Europe: Nation-States within the Common Market—Francesco G. Desch. and Global Politics in the Twenty-First Century—Ronnie D. Sinclair (eds. and Theoretical Integration in International Studies—Rudra Sil and Eileen M. Dominguez. D’Anieri Leapfrogging Development? The Political Economy of Telecommunications Restructuring— J. Jorge I. Kacowicz Private Authority and International Affairs—A. A.) Beyond Boundaries? Disciplines. Schmidt Power and Ideas: North-South Politics of Intellectual Property and Antitrust—Susan K. L. Jarvis (eds. Firms.SUNY series in Global Politics James N. Rosenau. Sell From Pirates to Drug Lords: The Post–Cold War Caribbean Security Environment— Michael C.) Collective Conflict Management and Changing World Politics—Joseph Lepgold and Thomas G. P. Lipschutz Pondering Postinternationalism: A Paradigm for the Twenty-First Century?—Heidi H. Doherty (eds.) After Authority: War.) International Relations—Still an American Social Science? Toward Diversity in International Thought—Robert M. Virginia Haufler.) Zones of Peace in the Third World: South America and West Africa in Comparative Perspective—Arie M. and Andres Serbin (eds. Claire Cutler. Paradigms. Shambaugh Approaches to Global Governance Theory—Martin Hewson and Timothy J.

Franke Money and Power in Europe: The Political Economy of European Monetary Cooperation— Matthias Kaelberer Why Movements Matter: The West German Peace Movement and U. Singh (eds. Payne and Nayef H. Persaud Global Limits: Immanuel Kant.202 SUNY SERIES IN GLOBAL POLITICS Hierarchy Amidst Anarchy: Transaction Costs and Institutional Choice—Katja Weber Counter-Hegemony and Foreign Policy: The Dialectics of Marginalized and Global Forces in Jamaica—Randolph B. and Political Community—Rodger A. Lang. Jr. Samuel Barkin What Moves Man: The Realist Theory of International Relations and Its Judgment of Human Nature—Annette Freyberg-Inan Democratizing Global Politics: Discourse Norms. International Relations. Swazo Political Identity and Social Change: The Remaking of the South African Social Order— Jamie Frueh Social Construction and the Logic of Money: Financial Predominance and International Economic Leadership—J. Democracy. S. Barry Jones (eds. and Development: International Conflict and Cooperation in the Information Age—Juliann Emmons Allison (ed. Samhat .) Systems of Violence: The Political Economy of War and Peace in Colombia—Nazih Richani The Arab-Israeli Conflict Transformed: Fifty Years of Interstate and Ethnic Crises— Hemda Ben-Yehuda and Shmuel Sandler Debating the Global Financial Architecture—Leslie Elliot Armijo Political Space: Frontiers of Change and Governance in a Globalizing World—Yale Ferguson and R. Arms Control Policy—Steve Breyman Agency and Ethics: The Politics of Military Intervention—Anthony F. Life After the Soviet Union: The Newly Independent Republics of the Transcaucasus and Central Asia—Nozar Alaolmolki Information Technologies and Global Politics: The Changing Scope of Power and Governance—James N. P. and Critique of World Politics— Mark F. J.) Theories of International Cooperation and the Primacy of Anarchy: Explaining U. International Regimes. International Monetary Policy-Making After Bretton Woods—Jennifer Sterling-Folker Technology. S.) Crisis Theory and World Order: Heideggerian Reflections—Norman K. N. Rosenau and J.

Democracy. and Global Civil Society—Elisabeth Jay Friedman.) Globalization. Berejikian Globalization and the Environment: Greening Global Political Economy—Gabriela Kütting Sovereignty. Rosenau (eds. Security. with assistance from Gregor Walter) International Regimes for the Final Frontier—M. Matthew. and Kenneth R. Bryan McDonald. Democracy. Rutherford (eds. and the Nation State: Paradigms in Transition—Ersel Aydinli and James N. J. Cortell The Multi-Governance of Water: Four Case Studies—Matthias Finger.SUNY SERIES IN GLOBAL POLITICS 203 Landmines and Human Security: International Politics and War’s Hidden Legacy— Richard A.. Hoffman Global Capitalism. Ludivine Tamiotti.) Identity and Institutions: Conflict Reduction in Divided Societies—Neal G.) Collective Preventative Diplomacy: A Study of International Management—Barry H. eds Building Trust: Overcoming Suspicion in International Conflict—Aaron M. Peterson Ozone Depletion and Climate Change: Constructing A Global Response—Matthew J. and Jeremy Allouche. Williams Globalizing Interests: Pressure Groups and Denationalization—Michael Zürn (ed. Kathryn Hochstetler. Schmidt (eds. and Civil-Military Relations in Colombia—Williams Avilés . Hoffmann States of Liberalization: Redefining the Public Sector in Integrated Europe—Mitchell P. Jesse and Kristen P. and Ann Marie Clark United We Stand? Divide and Conquer Politics and the Logic of International Hostility— Aaron Belkin Imperialism and Nationalism in the Discipline of International Relations—David Long and Brian C. Steiner International Relations Under Risk: Framing State Choice—Jeffrey D. Smith Mediating Globalization: Domestic Institutions and Industrial Policies in the United States and Britain—Andrew P.

.This page intentionally left blank.

150 and dynamic processes. 178 agency. 51 adaptive behavior compared to rational choice. 98 basic needs of. 122–23. 144 modeler expectations. 169 inquiry possibilities of. 176. 180 ABMs. 3. 150 incomplete. 33. 38. 157 modeler assumptions. 127 and emergence.Index 3PF. 175. 127 DOD research of. 168 epistemology of. See Agent-based models actor sensitivity. 167. from internal model. 126–27. 169 exploration by. 167 users of. 98 adaptive. 155 epistemological concerns. 134 and decision rules. 168 adaptation by. 165 method problems. 177. 126. 150–52 problem of simple behavior rules. 88 and structure. 129 information capture by. 171 epistemological implications of. 180 ABM programs in DOD. 168 uses of. 139 adaptive planning. 7 contingency. 177 and aggregate studies of war. 158. 9. 55 abduction. 173–74 advantages of. effect of. 171 identity in. 169 agent-based systems. 166–68 as experiments. 5. 171 assumptions. 113–14. 27. 152 defined. 102 defined. 173. 171. 176. 130–31 and game-theoretic models. 99 adaptive mechanisms. 166. 158 implementation. 157 agency-level computational models. 168 as aid to policy-making. 131–32 and understanding processes. 168 contingent identities in. See Three Pillar Framework 4 2 framework. 98 205 . robust. 114. 87 behavior of. 152 typology. See agent-based models agent-based modeling as transdiscipline. 180 endogenizing authority in. 82 agents. 140 appropriate ontology. 151. 140. 126 and social system. 145 methods of. 175. 9–10. 27 and independent thinking. 3. 167 agent-based models. 132 and equation-based models.

156 and politics. 153 chaotic problems. George W. 108 of internal models. 103 technology. 96. 143 Antarctic ozone “hole. 11–13. 154 balance of power. 173–78 language of. Robert M. 144. 144. 48 effect on group identity. 64 in open systems. 51–52 and rules. 154 and complexity. 56 Baltic States. 34–36. 152 diffused by globalization. 157 CFCs. 129 ahistorical theory. 8 in world politics. 74. 152–57 in complex systems. 83–84 no authority. 192 algorithmic social science. 74 causation. 29. 184 in North and South. 12. 13. 58 Club of Rome. 31. 113 coevolutionary processes. 12 blue eyes/brown eyes test. 146 complex systems (Also see complex adaptive systems) and conflict studies. W. 140 INDEX Bush. 47. See chlorofluorocarbons chaos.” 58.. 144 Bin Laden. 57 butterfly effect. 1. 90 biology and prejudice. 98. 149 breakdown of authority. 3 compared to simple systems. 144 as thought experiments. 106 clash of civilizations. 86 chlorofluorocarbons. 154 Axelrod. 89 contingency in. 43 Commonwealth of Independent States. effect of. 52. 50. 133 coevolution. 13. 46 Bosnia. 107 link to ozone depletion. 148 fundamental insight of. 175 analysts need for retraining. 157 authority cascades. 65 bottom-up modeling. 80 behavioral adaptation. myth of. 13 cause and effect. 190 assumptions by modeler. 110 Cold War. 51 Arthur. 52–57. 27. 99–101 industry. 155 problematic. 74–75 conceptual tools of. 32–36 . 154 concepts of. See causation CAS. 123 Bible date in vernacular. 81. 132 ecologies of. 39 authority and coercion. 54. 89. 154 central in politics. 189 “Turing test” for. Brian.” 100 arms races. 188–89 problem of endogenizing. 178 analytic narrative. 166 “edge of. 73–90 (Also see complex systems) and constructivism. 63. 188 emergence of. 13. 2–5 different from general systems. 174 Chechnya. 148 autarky. 5–6 characteristics of. Lars-Erik. 128 modeling adaptive behavior of. 153 world politics as. 167 in social relations.. 149. See complex adaptive systems Cederman. See “Limits to Growth” coethnic punishment. and CAS. Osama. 18. 146. 48–49 biopolitics. 96 complex adaptive systems (CAS).206 agents (continued) diversity of. 180 ant queen. 98. 81 complex adaptation. 74. 149 breakdown of. 11. 53. 85–86 publication dates.

127–30 207 superior to orthodoxy. 55 as process. 9. 185 improving orthodox theory. 6–13 compared to complicated. 57 decentralization of authority. 147 parsimony in. 34–36 compared to orthodox IR/world politics theories. 188 decision algorithms. 6–13 measures of. 180 Defense Advanced Research Projects Agency. 52. 134 and authority. 2 from simple rules. 167 conflict and initial conditions. 112 complexity methods. 52 defined. 84–86 Dayton Peace Agreement. 2–6 epistemology of. 146. 175 as methodological pillar. 58 and conflict resolution. compared. 158 from simple rules. 52–57 and constructivism. small changes in initial. 174 computational social network analysis. 5 limits to methods of. 50–51 and complexity. 73. problems from. 174 deduction. 168 culture. 52–57 conflict transformation. 10. 149 challenge of. 3–4 and equilibrium. 186 and living systems. 148. 144 causation in. 144. 84 policy implications of. 52. 3–4 and simplification. 169 decision rules. 158 contingency in complex adaptive systems. See dissipative structures . 29. 9. contingency in. 178 Defense Modeling and Simulation Organization. 97 Correlates of War. 37 patterns in. 3–4 new thinking from. 7. 176. 180 dissipative structures. for IR/world politics. 149. 63 as balancing mechanism. 55. 7. 33. 4 complex systems theories. 143 in world politics. 38 complexity. 151 decision-making and complexity. 189–90 complexity science. 190–92 problems of metaphor in politics. 148 potentially integrative. 52 and exploration. non-positivist. 153 proposed empirical testing. 188–89 and change. 166. 63 consolidative modeling. 148 complex adaptation. 190–92 nonlinearity in. 165 and induction. 133. 11. 76 and self-organization. 175. 145 defined. 11–12. 127 constructivism. 28 aggregate studies of war. 4–5. 184 assumptions of. 179 computational models. 34 COW. 175 deep uncertainty. 111–14. 147 policies proposed by. 10 and decision-making. deduction. 185 paradigm or theory. 166. 54. 86 Department of Defense research in ABMs. 89 conditions. 81 in form only. 63 conflict resolution. 87–89. 3 as generative science. 4 dissipative systems. 53. 5 generates novel insights. 186–87 epistemology of. 174 defined. 87. 174 conceptual tools of. 149 as metaphor. importance of. 2 computational epistemology. 30. See Correlates of War credibility of methods. 5 predictability of. 180 deep narrative.INDEX leverage points in. 149 not deductive. 184–85 in social sciences. 178 democracy.

86 flip. 149 and biology. 43. 11. 156 induction.” 58. 127. society. 125. 187 for simulations. 99. 193 epistemology. See Department of Defense domestic politics. See European Union European Union. 187 of complexity. 128 emergent properties. 88 ethnic entrepreneurs. 186 non-positivist. 84–86 inclusive. 80. 7. 38 extra-rational. internal view. 28 Elliott. 45–51 defined. 49 human social networks. 55. 144 homeostasis. 89 fitness criterion modeler’s choice of. 45 in-group. 147 of state. 109 of ethnic norms in ABMs. 48 sources of. 57 identity. 188 in-group. 4–5. 129 ethnic groups. 47–49 Human Development Index. 27. 48. 27 compared to complex 32–36 genetic algorithms. 10–11 dynamics at microlevel. unstable. 103 evolutionary epistemology. See US Environmental Protection Agency epistemic communities. 64 of countries. 124 ethnocentrism. 132 ethnic norms. 171. 186–87 positivist. 61. 87–89. of complexity. 173 informal institutions. 47. improving. 166. 28 dynamic systems. 80. 126 ethnic violence. 74 Idealpolitik. 11 human behavior. 186 equilibrium. 73. small changes in. theories. 153 Holland. 45–51. 173 inference-making. 173 external. 47–49 EU. 188 formal modeling. 158. 152–57 of universal participation rule. 43. 75. 45 sociobiology of. Thomas. 5. 158 issue specific. 82 acquiring fitness. 46–47 emergence. 151 fitness landscape. 188 Gramsci. political. John H. 83. 7. 80. 3 general systems. 188 Gell-Mann. 148. 45. 185. 31 feedback. Albert. 153 Einstein. 84–86 of society. See Systemic change formal institutions. 193 elites. 6. 151. Murray. 179 evolutionary. 74–75. 52 ethnic consciousness. 47–48.. 128 initial conditions. 98 negative. 140. 74 empirical testing. 28–29. iterative. 132 of authority. 61 “edge of chaos. 127–30 EPA. 44. See UN Human Development Index Hobbes. 158 globalization. 63. 89. 82. 34 . 192 HDI. 122 INDEX exploratory modeling. Jane. 75. 29 feedback. 125–26 defined. 74.208 DOD. Antonio. 5. 81–82 and ABMs. 187 experimentation. 125 emergence of. 111–14. 171. 52 early warning system. 121–26 and bottom-up processes. proposed. positive. 74 emergent phenomena. 98. 9. See UN Human Development Index human nature. 150 fitness. 150. 32–34. 4 positive. 113 Freedom House. 30–31.

165 leverage points in complex system. 132. 7 Kaplan. 170 mathematical. 184 method. See internal models international organizations. 28 mental model. 114 problem of simple behavior rules. 12 and Monte Carlo methods.” 33 literacy evolution of. 27. 133 Jevons. orthodox. 110 and US internal model. 158 accidental. 98. 187 Markov chains. 101. Humberto. by agents. 158 London Amendment. 150 Montreal Protocol. 49. 36 international relations. 84–86 without debate. 27. 27 Kauffman. 166 specification of. on ozone depletion. computational. 100. on ozone depletion. 28 predictive. 13 jihad. 151 modeling exploratory. 189–90 microevents and macrobehavior. 47. 171 ensemble of. Also see world politics international relations theories. for research puzzle. 9. Thomas S.INDEX inquiry efficacy and quality. 192 mathematical models. 104 diversity in. not theoretically neutral. 8–9. defined. 44. path dependent. and literacy. 123 internal models. 152 US change in. 105. 9–10. 27. 32–36 failures of. 171 media. John. 87 importance of. 27. 1–2 compared to complexity. See international relations theories Islamic societies. 172 use by analysts. 170 institutions. 35 levels of analysis.. 110 internal rule models. 12. W. 114. 189 nature of. 189 micro-macro linkages. 158 simulative. 89. Robert. Morton. 178 modeler assumptions. 9. 167 methods of complexity. 10 measures of theory quality. 189 mastery of nature. 35. 86 Jervis. of complexity. 108 and agent behavior. 173 rules or attributes. 166 falsification as.. 189 mapping. 106–8 multi-agent simulation. predictive accuracy of. 108 Luther. 84. 132. 191 adaptation of. 169–70 reasoning from. 189 Maturana. 86 Locke. 171 in social science. 152 interhamwe. 29. 88 209 macrobehavior and microevents. 146 models. 157 styles of. 57 learning. 125 operationalizing. 178 isomorphism of. 185. Also see Agent-based models defined. 170 differentiation between. 57 international theorists. 102 change in South. 6 logical positivism. 84–86. 106. 167 . 151 interests. 143 Koran. 34–35. See internal model meta-agents. Stuart A. 33. 193 interest groups. 11. 190–92 “Limits to Growth. 190 methodology adequacy of. 6–13 international system. 26. 7. See Qur’an Kuhn. 35. Stanley. 35. Martin. 188.

56 neoclassical economics. 95. 192 politics. 53. 114 processes. 61 positivism. 44. 76 power. uncertainty of. 99–108. 82–83 Putnam. See NATO NRDC. 144 norm of universal participation. 8. 30. 10 Organization for Security and Cooperation in Europe. market capitalism in. of decision-makers. 6 neorealism. compared to structure. Robert D. 190 in social sciences. 73. 152. moving from chaotic to complex. See positivism postmodernism. 189 nonlinearity. 190 policy effects. 53 problem solving.210 INDEX parsimony and complexity. 48 preventive diplomacy. improving. 176–78. 110 paradigm shift. 12 OSCE. 184–85 Osama Bin Laden. 144 Popper. 126 ethnic. 63 problem. 88 National Defense University. 37–38. 107 norms. 1. 63 positive. 60–63. and biology. global. 12 as metaphor. Charles Sanders. 65 Newton’s universe. 140 provocability. 148. 133 ontology. 158. 57 Peirce. 106 nature and nurture. 99 emergence of. 4. and complex adaptive systems. 75. 61. 170 and complexity. 49 negative peace. 82 organization. 28 order for free. Ilya. 168 open systems. 81–83 democracy. unclear. 128 ozone depletion regime. 12. 13 constitutive. 57 parsimony. 29 Prisoner’s Dilemma. 4. 173. 61. 62 orthodox theories. 170 PCMT. 187 positivist methods. 36. 10 national consciousness. 26 of agent-based models. 10 New European Peace and Security Systeem (NEPSS). 56. 157 of decision-making. 54. See Organization for Security and Cooperation in Europe out-group. 64. 113 path dependence.. 56 punctuated equilibrium. See Pre-Conflict Management Tools Program peace negative. See Natural Resources Defense Council obedience. 186–87 dangers of. 77 ethnic issues in. 193 positive sum common security. 174 problems. 29–30. 63 peace-keeping. 54. 56 Pre-Conflict Management Tools Program. 192 process tracing. 130 nonlinear systems. in complex systems. 138 understanding of. 56. 32 nation-state. 121 North Atlantic Treaty Organization. 37 . 179 perception. 190 post-Soviet Eurasia. 180 prejudice. 35 points of leverage. 6 nonlinear rules of behavior. 45. Karl. 75–77 and complexity. 61 Prigogine. 80 Natural Resources Defense Council. 125–26 of behavior. See state North Atlantic Treaty Organization. insight into. 31 simulation. 108–11. with ABMs. 80 zones of. 54. 176 national interests.

82 sensitivity of actors. 115. system. Herbert. 34 as research methodology. 101 for participation. 95. 55. 43. 108–11 challenges to. 151 of behavior. 96 changes from complexity. 32 of social aggregations. 190 retroduction. 2–5 automobile as. 75. 138. 7. 166. 64 Serbia. 50 reasoning computer-assisted methods. 43–47. 2 predictability of. 139. 11. 30–32 objective. 9–10 rationality assumption. 113 as experiment. 54 research puzzle and choice of method. 102. 2–5 and complex compared. 130 rules. 5 contingency in decision. 172 recursive interaction. 17 simulated annealing. 109. 38 refutation. learned. 177 non-Peircian process. 107–8 universal participation. 54 Realpolitik. 183. 96. 9. 6–13 self-organized criticality. 2–6. 108 system rules exogenous. See internal models security. 4 simplification and rigorous problem analysis. 193 regime emergence. 84 in world politics. 179 Robbers Cave experiment. 1. 49. 9. 52–57 Serbs. 150 realism. debate. 97. 194 scenario differentiation. 112 Republic of Srpska. 57 and War on Terror. 185 compared to adaptive behavior. 109 and complexity. 46 robust adaptive planning. See internal model rule of participation. 12. 2 characteristics of. 4 static. 109 North only. 56 selection mechanism. 191 and free thought. 97 regime theory. appeal of. 179 simple systems. nonlinear. 51. 83 and democracy. 109 rules. 61 selection. Thomas. 177. 103 emergence of. 139 . 32 complexity compares. 101 operationalizing. 121–25 characteristics. 81 and diversity.” 53. 178 rule model. 50.INDEX Qur’an. 174 by common knowledge. 123 211 satisficing. for ozone depletion. 31 subjective. as selection mechanism. 3 defined. 86 and hierarchy. 101 of ozone depletion. 193 self-organization. 150. 65 “shadow of the future. 2001. positive sum common. 46 rational choice. 194 simulation. 110 participation. 111. 99–108 ozone. 80 Rwanda. 86 racial prejudice. 55 Simon. 139 rationality. 102 evolution of. 147 schema. 114 South demands change. 64 September 11. 172 Schelling. 2 hegemonic theory. 96 formation. 106 Northern problem. 9. 51 to initial conditions. 191 dynamic compared to static. 101 Russian Federation.

inherent. 171 theory. 104. 34 terrorism. 30–31. and psychology. 73. 187 incomplete. 35 as complex adaptive actors. 158 multi-agent. measuring. 25 taxonomy of complexity. 58 systemic trajectory. 4 Tit-for-Tat. 55. 36–37 defined. 111. 27. 64 theories. 87. 140 Transparency International. 32–36 goals of. 88 UN Human Development Index. 179 social science models. robust compared to optimal. See ozone depletion structural factors. 7. international relations. 157 simulative-empirical research design. 90 countries ranked by 79–80 uncertainties. 10. 170 taxonomy. 11 and agency. 191 social. 89 sociophysics. 167 of CAS. 27. 6 technology. 191 nested. 137 epistemology of. 88. 80 umuganda. 32 general and complex compared. 77 Ukraine. 137. 167 solutions. 27. 143 social science algorithmic. 33 emergent properties in. 62. 8–9 feedbacks within. 58. 25 social capital. See theories. David. 106. 1–2 orthodox. 59–60 defined. 32–36 theory quality. 32 criteria of good. 134 social networks. 134 social organization. 10 surprise. 29 in Yugoslavia. 32–36 theories. 17. 179 problems of. 102. 33 defined. as attachment behavior. 173–78 and positivist epistemology. 1 systems. predictive accuracy of. world politics theories. 103 internal model on ozone depletion. 7 compared to organization. 26–27 paradigm of. 148 of complex systems. 78–80. 5 principal value of. 9 state of nature. 149 Singer. 99. 29 system rules. general systems. 187 generalizable statements in. 12 compared to complexity.212 INDEX general. ahistorical. testing theorems. world politics. 192 Three-Pillar Framework. 85–86. 53. 138 with computers. in world politics. 7 as meta-agents. 83. 83. 170 state. 125. 6–13 failures of. 97 as emergent system. 110 simulation (continued) definition. 124 UN Development Programme. J. 191 operationalizing. 2. 138 uses of. 33 . 190 uncertainty. 158 simulative methods. 60 time. arrow of. continuously evolve. 51 United Nations Environment Programme. effect on social science. 193 United States. examples. 122 structure. 147 simulation software. 12. 191 top-down modeling. 6 stratospheric ozone depletion. 7. 178 societal fitness. 114 systemic change. 30–31 in orthodox theories.

77. 60. 10 War on Terror. 106. 110 Varela. 146 as processual flow.INDEX unstable equilibrium. 10 veridicality. 5 Waltz. 174 213 World Bank. 115 Waldrop. Mitchell. 12. 13 as self-organizing complex system. 6–13 orthodox theories of. 190 and Realpolitik methods. Francisco. 76 internal conflict. See United States U. M. 10 World Trade Organization. 53 . 27. 170 Vienna Convention. 32–36 world systems theory. 103. Kenneth. Environmental Protection Agency. 99. 6–13 orthodox and complex compared. 177 world politics as complex adaptive system.S. 7. 2. 100. 52 U. 101. 193 worldview. 52–57 Yugoslavian war. 17 Yugoslavia. 4–5.S. 50 wicked problems.

This page intentionally left blank. .

“This book is well written and easily accessible. scholars and practitioners alike are constantly surprised by international and global political events. Complexity in World Politics shows how conventional theories oversimplify reality and illustrates how concepts drawn from complexity science can be adapted to increase our understanding of world politics and improve policy. and empirical approaches. In language free of jargon. also published by SUNY Press. Page.sunypress. Rosenau. Harrison. They show how these concepts can improve conventional models as well as generate new ideas. The collapse of communism in Europe. It makes a number of intellectual contributions and helps fill a gap in the existing literature. editor Despite one hundred years of theorizing. editor State University of New York Press www. the book’s distinguished contributors explain and illustrate a complexity paradigm of world politics and define its central concepts. A volume in the SUNY series in Global Politics James N. mechanical system. hypotheses. and conclude by outlining an agenda of theoretical development and empirical research to create and test complex systems theories of issue-areas of world politics. with essays by some of the major thinkers in the field of complexity science. coeditor of Computational Models in Political Economy Neil E.POLITICAL SCIENCE Complexity in World Politics Concepts and Methods of a New Paradigm Neil E. and 9/11 have demonstrated the inadequacy of current models that depict world politics as a simple. the 1997 Asian financial crisis.” — Scott E. Harrison is Founder and Executive Director of the Sustainable Development Institute and the author of Constructing Sustainable .