Complexity in World Politics: Concepts And Methods of a New Paradigm (Suny Series in Global Politics)

Complexity in World Politics Concepts and Methods of a New Paradigm

Edited by

Neil E. Harrison

COMPLEXITY IN WORLD POLITICS

SUNY series in Global Politics James N. Rosenau, editor

COMPLEXITY IN WORLD POLITICS Concepts and Methods of a New Paradigm

Edited by Neil E. Harrison

State University of New York Press

Published by State University of New York Press, 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. No part of this book may be stored in a retrieval system or transmitted in any form or by any means including electronic, electrostatic, magnetic tape, mechanical, photocopying, recording, or otherwise without the prior permission in writing of the publisher.

For information, address State University of New York Press, 194 Washington Avenue, Suite 305, Albany, 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. p. cm. — (SUNY series in global politics) Includes bibliographical references and index. ISBN 0-7914-6807-0 (hardcover : alk. paper) 1. International relations—Philosophy. 2. International relations—Methodology. 3. Complexity (Philosophy) I. Harrison, Neil E., 1949– II. Series. JZ1305.C657 2006 327.1'01—dc22 2005024118 ISBN-13: 978-0-7914-6807-4 (hardcover : alk. paper) 10 9 8 7 6 5 4 3 2 1

Contents

1

Thinking About the World We Make Neil E. Harrison

2

Complexity Is More Than Systems Theory Neil E. Harrison with J. David Singer

25

3

Complexity and Conflict Resolution Dennis J. D. Sandole

43

4

Understanding and Coping with Ethnic Conflict and Development Issues in Post-Soviet Eurasia Walter C. Clemens, Jr.

73

5

Beyond Regime Theory: Complex Adaptation and the Ozone Depletion Regime Matthew J. Hoffmann

95

6

Agent-Based Models in the Study of Ethnic Norms and Violence Ravi Bhavnani

121

7

Alternative Uses of Simulation Robert Axelrod

137

8

Signifying Nothing? What Complex Systems Theory Can and Cannot Tell Us about Global Politics David C. Earnest and James N. Rosenau

143

v

1

vi

CONTENTS

9

When Worlds Collide: Reflections on the Credible Uses of Agent-Based Models in International and Global Studies Desmond Saunders-Newton

165

10

Complex Systems and the Practice of World Politics Neil E. Harrison

183

Contributors

197

List of Titles, SUNY series in Global Politics

201

Index

205

CHAPTER 1

Thinking About the World We Make Neil E. Harrison

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

2

NEIL E. HARRISON

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

THINKING ABOUT THE WORLD WE MAKE

3

remedy them. They also illustrate problems and solutions, define the characteristics of each part, and the range of relations between them in exhaustive detail. As table 1.1 shows, a living system is complex in many ways that an automobile is not. The two primary differences between complex and simple systems are diversity and decentralization. In an automobile there are many diverse parts constructed for very specialized roles, but there is centralized coordination of their operations through mechanical or electronic management systems. In living systems, not only are the units diverse but each has a range of freedom of choice denied to parts in a mechanical system. Because units in a complex system have discretion in their choice of behavior, they are commonly called “agents.” Decentralized decision-making increases complexity. One measure of complexity is the length of the shortest possible message that fully describes the system (Gell-Mann 1994, 30–38). 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, system description is shorter; only one unit need be described in detail. Thus, heterogeneity among the units increases description length.2 But if the units also have behavioral discretion, system description requires description of the units (perhaps by class), of the range of their available choices, and of the rules of behavior that each will follow in making their individual choices. Centralization of decision-making simplifies complicated systems. Modern automobiles have sophisticated management systems that use miniature

TABLE 1.1. 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

4

NEIL E. HARRISON

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

THINKING ABOUT THE WORLD WE MAKE

5

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

6

NEIL E. HARRISON

The characteristics summarized in table 1.1 and described in this section are most commonly associated with each genus of system, simple or complex. No single descriptor defines either simple or complex systems. For example, simple systems may have many and diverse parts and complex systems (e.g., of bacteria) may have homogeneous units; and complex systems can be, at least temporarily, in equilibrium, while some simple systems appear dynamic. However, the more descriptors of one system genus that can be attributed to a specific system, the greater is the probability that that system is of that genus. Thus, complexity is an accumulation of the characteristics of complex systems. 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).

COMPLEX SYSTEMS IN WORLD POLITICS Intuitively, the social world seems complex in the sense described here, but current theories of world politics model it as a simple system. As Ruggie (1993) comments, world politics theories are “reposed in deep Newtonian slumber.” 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. These laws could be expressed exactly through mathematics (Capra 1982, 65–67; Ruggie 1993). For example, the properties of gases can be reduced to the mathematically describable motion of their atoms or molecules. Thus, the image is of a universe constructed like a perfect mechanical watch. Science, aided by mathematics, was the method for prizing open the watch case to see the workings inside (Hollis and Smith 1990, 47). 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, 69). A fixed human nature was presumed to determine human behavior, and “natural laws” governed spontaneous human society: “As the atoms in a gas would establish a balanced state, so human individuals would settle down in a society in a ‘state of nature.’” Natural laws included freedom, equality, and property rights (Locke 1980, 123–27; Kymlicka 1990, 95–159). The shadow of Newton’s universe continues to obfuscate knowledge in the social sciences. For example, while neoclassical economics remains the dominant explanation of economic phenomena, it is “an economic science after the model

THINKING ABOUT THE WORLD WE MAKE

7

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

8

NEIL E. HARRISON

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

THINKING ABOUT THE WORLD WE MAKE

9

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

10

NEIL E. HARRISON

Constructivism broadens “the array of ideational factors that affect international outcomes” and introduces “logically prior constitutive rules alongside regulative rules” (Ruggie 1998). The concept of internal models potentially extends the ideational content of world politics theories, while at the same time making analysis of agent motives more difficult. However, simulation of agent behavior now is possible. Internal models drive agent behavior, but those models may change when tested in a selective environment. Agents that consistently act in ways that are selected by their social environment as suboptimal face eradication. Because states are themselves systems, the process of matching internal model to external reality is one of trial and error. As Putnam (1988) has suggested, 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. If all states are adaptive complex systems, then the international system emerges from coevolution. International norms influence behavior through the internal process of internal model formation, one component of which is the desire to participate in a society of nation-states. In chapter 5, 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. Dynamic Systems Superficially, complexity appears to have some affinity with other world politics systems theories, like neorealism and world systems. As Waltz (1979, 91) describes the international system, it is “formed by the coaction of self-regarding units,” and its structure is “formed by the coaction of their units” and “emerge[s] from the coexistence of states. . . . International-political systems, like economic markets, are individualist in origin, spontaneously generated, and unintended.” However, the similarity is more perceived that real, as shown in more detail in chapter 2. The distinction is in the details: in conventional systems theories, structure is a fixed or only slowly changing determinant of agent behavior. In complex systems, structure is dynamic but “organization” is fixed. 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, a sunflower, a cat, or a human brain)” (Maturana and Varela 1980, 18). To describe the organization, it is only necessary to describe the relationships and not the components. For example, self-organization is “a general pattern of organization, common to all living systems, whichever the nature of their components.” The structure of a complex system is the actual relations among actual physical components: “[I]n

THINKING ABOUT THE WORLD WE MAKE

11

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

12

NEIL E. HARRISON

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-

THINKING ABOUT THE WORLD WE MAKE

13

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

14

NEIL E. HARRISON

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.

THINKING ABOUT THE WORLD WE MAKE

15

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.

16

NEIL E. HARRISON

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

THINKING ABOUT THE WORLD WE MAKE

17

phenomena in an interdisciplinary way. In addressing the epistemological issues surrounding computational methodologies, he argues that efficacy or usefulness is more important than the quality of model isomorphism and method. He then describes how several computational social science models (including ABMs), 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. The concluding chapter draws some general lessons from the four cases and shows how they illustrate important complexity concepts. It then assesses the validity of simulations and computational models and the epistemological questions they raise. It also shows that, from complexity concepts and ideas, complex systems theories for issue-areas can be specified and models for specific problems generated. Yet, because political systems are complex—and becoming more complex—a new epistemology and new methods are needed to understand them. Finally, it shows that recognition of complexity in politics suggests radically new policies for addressing international problems and pursuing national interests. NOTES 1. The terms “world politics” and “international relations” are used interchangeably throughout this book. I use “world politics” to better reflect the multilevel structure of the political world to which complex systems thinking is so well adapted. Patomäki and Wight (2000, 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. The terms “complex system” and “complex adaptive system” are often used interchangeably; the concepts described here principally derive from the latter. I use the term “paradigm” in the sense of a set of assumptions, concepts, values, and practices that comprise a view of reality, and in that sense it is quite comparable to “worldview” (Hughes 2000). 2. Common knowledge also shortens description. “Bicycle” conveys to most people a clear image of the system. 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. 3. Smith (2004) comments that world politics/IR theorists err in thinking of the state as solely political. 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. 4. In this book, the terms “internal model,” “mental model,” and “schema” are used interchangeably.

18

NEIL E. HARRISON

5. As a skeptic, Hume also argued that causation is a human construct. All we ever see is the conjunction of events, and we impute a causal relationship. But without an explanatory force linking cause and effect, causation cannot be “real” (Patomäki and Wight 2000).

REFERENCES Anderson, Philip W., Kenneth J. Arrow, and David Pines, eds. 1988. The Economy as an Evolving Complex System. Santa Fe Institute Studies in the Sciences of Complexity, 5. Redwood City, CA: Addison-Wesley. Arthur, W. Brian. 1989. “Competing Technologies, Increasing Returns, and Lock-in by Historical Events.” Economic Journal 99 (March 1): 116–31. ———. 1990. “Positive Feedbacks in the Economy.” Scientific American 262, no. 2 (February): 92–99. 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 M. 1997. The Complexity of Cooperation: Agent-Based Models of Competition and Cooperation. Princeton, NJ: Princeton University Press. Axtell, Robert L., Joshua M. Epstein, Jeffrey S. Dean, George J. Gumerman, Alan C. Swedlund, Jason Harburger, Shubha Chakravarty, Ross Hammond, Jon Parker, and Miles Parker. 2002. “Population Growth and Collapse in a Multiagent Model of the Kayenta Anasazi in Long House Valley.” Proceedings of the National Academy of Science of the United States of America, supp. 3, vol. 99, no. 10:7275–79. Barbier, Edward B., Joanne C. Burgess, and Carl Folke. 1994. Paradise Lost: The Ecological Economics of Biodiversity. London: Earthscan for the Beijer Institute. Buzan, Barry, and Richard Little. 2001. “Why International Relations Has Failed.” Millennium: Journal of International Studies 30, no. 1:19–39. Capra, Fritjof. 1982. The Turning Point. New York: Simon & Schuster. Cederman, Lars-Erik. 1997. Emergent Actors in World Politics: How States and Nations Develop and Dissolve. Princeton, NJ: Princeton University Press. Cederman, Lars-Erik, and Kristian Skrede Gleditsch. 2004. “Conquest and Regime Change: An Evolutionary Model of the Spread of Democracy and Peace.” International Studies Quarterly 48, no. 3 (September): 603–29. Cilliers, Paul. 1998. Complexity and Postmodernism: Understanding Complex Systems. New York: Routledge.

THINKING ABOUT THE WORLD WE MAKE

19

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

20

NEIL E. HARRISON

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

THINKING ABOUT THE WORLD WE MAKE

21

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

22

NEIL E. HARRISON

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

THINKING ABOUT THE WORLD WE MAKE

23

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

This page intentionally left blank.

CHAPTER 2

Complexity Is More Than Systems Theory Neil E. Harrison with J. David Singer

The previous chapter emphasized the differences between conventional theories of world politics and international relations and complexity. In 1971 Singer proposed a general systems taxonomy that potentially supported both systemic explanation and more limited theorizing. This chapter shows how the important concepts of general systems taxonomy compare with concepts in a complex systems taxonomy. In addition, 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. 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. CONCEPTS IN COMMON A taxonomy “has no descriptive, predictive or explanatory power, since it contains only definition propositions. But to serve as the basis for building models and theories, 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, 3). As a framework for theorizing about world politics and international relations, complexity modifies and expands on Singer’s general 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. Despite correcting

25

26

NEIL E. HARRISON WITH J. DAVID SINGER

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

COMPLEXITY IS MORE THAN SYSTEMS THEORY

27

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

28

NEIL E. HARRISON WITH J. 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. Yet, in his 1971 monograph Singer simply states that all social systems are open—that is, their boundaries are “permeable to information and energy from [their] environment”—and that no social system can realistically be treated as closed. There is, he writes, “clearly no such thing as a completely closed social system” (13).2 If so, assuming closure to simplify analysis and reduce empirical effort is not a reasonable parsimony but a gross distortion of reality. While every model is necessarily an “ersatz” reality, it must retain a recognizable link to the portion of reality it purports to represent or it will generate inaccurate “knowledge.” Complexity similarly treats all social systems as open, but, as Singer has commented, it is an empirical question as to how open each is (1971, 13). Some scholars approvingly describe rural communities’ openness to their “natural” environment, especially in poorer countries. However, such open communities often are both somewhat closed to their social environment and much simpler than more complex, modern societies that are more open to other social systems but less open to ecological systems (Harrison n.d.). Openness is not a measure of complexity, but complex systems usually are open. 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, 165). National elites use rhetoric for domestic political consumption that can incite potential enemies, the public and military desire the psychological comfort of discernible superiority, media amplify internation conflicts, 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. Selfrestraint within political elites and the media has diminished with the increase in the number of competing media companies, their geographical coverage, and the diminishing time lag between event and report, and corrective mechanisms within the international system have atrophied. Technologically induced immediacy reduces opportunities for editorial restraint. Similarly, technology has

COMPLEXITY IS MORE THAN SYSTEMS THEORY

29

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

30

NEIL E. HARRISON WITH J. DAVID SINGER

Austro-Hungarian and Ottoman Empires after World War I). Second, there is institutional memory: how decision-makers and the public perceive the benefits and costs of conflict. Success in conflict tends to bolster militarily adventurous groups, and recent failure may cause caution. Failure, however, may encourage a desire to regain position and respect. For example, defeat in World War I supported German aggression in the 1930s. 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. By collecting data for more than a century of interstate and civil wars, the project seeks patterns and commonalities among conflicts and avoids the historical fallacy that defines each conflict as a discrete and separable event. If the experience of conflict influences later conflict choices, feedback mechanisms within the state are the likely link. States have memories that influence future perceptions and choices. It is conventional wisdom, for example, that a “Vietnam syndrome” influenced decades of decision-makers, causing an apparent US aversion to armed conflict. Research has failed to find statistically significant linkages between war experience and later conflict choices (Singer and Cusack 1981). 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.” This does not mean that each conflict must be treated as a discrete event, but it does show that the feedback mechanisms within states are significantly more complex than is commonly believed. There is a randomness to the influences of memory and history that is not captured by simple theories. Through the concept of emergence (discussed further below), a complex systems theory of national security potentially allows for both path dependency through experience (state memory and capability) and randomness. Because state behavior emerges from domestic interactions, current conditions and institutions, and the variable distribution of power between politically influential groups, influence state internal models. But current social conditions and power relations are themselves historical artifacts. Thus, historical experience is perceived through the ever-changing lens of the present, which itself emerges from the past. Rationality Rationality assumptions are used as a convenient simplification in both orthodox theories and general systems approaches. But in the latter rationality is assumed only at the lowest level of aggregation: the individual human. At this level, rational-

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

32

NEIL E. HARRISON WITH J. DAVID SINGER

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

COMPLEXITY IS MORE THAN SYSTEMS THEORY

33

and if they are cultural they can be observed as individual psychological properties. 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, by observing the distribution and configuration of individual psychological properties” (Singer 1968, 144). 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. Singer also argued that “a system is nothing more than the sum of its parts and the relationships and interactions among them” (Singer 1971, 19) and that a system is “not composed of [external] systems, or of any other phenomena beyond its own component units and the relationships and interactions among them.” 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. In contrast to Singer (1968), the emergent effects of agent interactions are both dynamic and important and cannot be captured by observation of structure, behavior, or individual psychology. Although the emergent properties of a system cannot be captured by studying the system’s parts, emergence is real. It would be unscientific to reject a theoretically useful concept merely because accepted scientific methodologies cannot record the phenomenon. Fortunately, as elsewhere in the science, developments of new methods permit new thinking and empirical testing of novel hypotheses. With George Zweig, Gell-Mann had developed a theory of quarks. He was always skeptical that such partially charged subatomic particles would ever be found, and by the end of the 1960s it seemed that he was right; no evidence had been found for their existence (Riordan 2004). But by 1973, with new research techniques, “everything seemed to be coming up quarks.” It took most of the decade for the theory of quarks to become generally accepted, and it was later recognized with a Noble Prize. Similarly, one of the predictions of Einstein’s general theory of relativity was tested by astronomical observation for the first time in 2004, nearly ninety years after the theory was first proposed. 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. When Singer wrote his critique of emergence, it was not practical to model complex systems. 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). Not only are the predictions of that project now largely discredited, but also the design of the model, limited as it was by incomplete data sets and computing power, is considered crude.6 Social scientists who are not computer specialists can now program ABMs that model “mystical”

34

NEIL E. HARRISON WITH J. DAVID SINGER

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

COMPLEXITY IS MORE THAN SYSTEMS THEORY

35

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

36

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

COMPLEXITY IS MORE THAN SYSTEMS THEORY

37

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

38

NEIL E. HARRISON WITH J. DAVID SINGER

is to understand behavior, and the means are often hermeneutical, examining “human action from within, seeing it as intentional and meaningful behaviour” (S. Smith 1994, 400). The two approaches usually are assumed to be incompatible. Explanation from the outside, however scientific, is incomplete without consideration of the units: “The anarchical character of the international system . . . strongly suggests that the units affect the shape of relations, however firm the shove [from structures]” (Hollis and Smith 1990, 198). 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, unless they include a causal contribution from the units.” The inside approach is rejected by some as interpretive and, therefore, inherently unscientific. 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). Complex systems theories potentially integrate outside and inside orthodox views. The central concept of emergence marks complexity as favoring the inside view, yet its experimental methods are potentially as scientific as those of the revered “hard” sciences. 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, 197–98). If the system is real and must be analyzed as a whole, it must be shown that “wholes are more than their parts and that science is capable of establishing such a proposition” (198). 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. As discussed, ABMs also may be constructed to simulate historical reality. 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. The following chapters begin that task. NOTES This chapter was written by Neil Harrison based on David Singer’s selection from among his published works of those that anticipate complexity, and his comments and advice on earlier drafts. 1. 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.

COMPLEXITY IS MORE THAN SYSTEMS THEORY

39

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

40

NEIL E. HARRISON WITH J. DAVID SINGER

Kaplan, Morton A., 1957. System and Process in International Politics. New York: Wiley. Levin, Simon A. 1999. Fragile Dominion: Complexity and the Commons. Reading, MA: Perseus Books. Levins, Richard, and Richard Lewontin. 1985. The Dialectical Biologist. Cambridge, MA: Harvard University Press. Mäler, Karl-Göran. 2000. “Development, Ecological Resources and Their Management: A Study of Complex Dynamic Systems.” European Economic Review 44 (2000): 645–65. Mitrany, David. 1966. A Working Peace System. Chicago: Quadrangle. Palma, Gabriel. 1978. “Dependency: A Formal Theory of Underdevelopment for the Analysis of Concrete Situations of Underdevelopment.” World Development 6:881–924. Patomäki, Heikki, and Colin Wight. 2000. “After Postpositivism? The Promises of Critical Realism.” International Studies Quarterly 44, no. 2 (June): 213–37. Prigogine, Ilya. 1997. The End of Certainty: Time, Chaos, and the New Laws of Nature. In collaboration with Isabelle Stengers. New York: Free Press. Riordan, Michael. 2004. “Science Fashions and Science Facts.” Viewed at http://www.physicstoday.org/vol-56/iss-8/p50.html on March 22. Singer, J. David. 1961. “The Level of Analysis Problem in International Relations.” In The International System: Theoretical Essays, ed. K. Knorr and S. Verba, 77–92. Princeton, NJ: Princeton University Press. ———. 1968. “Man and World Politics: The Psycho-Cultural Interface.” Journal of Social Issues 24, no. 3:127–56. ———. 1970. “Escalation and Control in International Conflict: A Simple Feedback Model.” General Systems 15:163–73. ———. 1971. A General Systems Taxonomy for Political Science. New York: General Learning. ———. 1977. “The Historical Experiment as a Research Strategy in the Study of Politics.” Social Science History 2, no. 1 (Fall): 1–22. ———. 1990. “The Invisible Hand, Extra-Rational Considerations and Decisional Failure.” Paper presented at the Annual Convention of the American Political Science Association, San Francisco, September 2. Singer, J. David, and Thomas Cusack. 1981. “Periodicity, Inexorability, and Steermanship in International War.” In From National Development

COMPLEXITY IS MORE THAN SYSTEMS THEORY

41

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

This page intentionally left blank.

CHAPTER 3

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

43

44

DENNIS J. D. SANDOLE

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.

COMPLEXITY AND CONFLICT RESOLUTION

45

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)

46

DENNIS J. D. SANDOLE

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

COMPLEXITY AND CONFLICT RESOLUTION

47

We learn to be racist, therefore, we can learn not to be racist. Racism is not genetical. It has everything to do with power. (quoted in Coronel 1996, 2, emphasis added) In the language of complexity, Elliott has demonstrated with a simple experiment how mental models are socially constructed and can adapt to interactions with others, especially those with authority. Related to the Realpolitik-ethnocentrism nexus is the seductive totality and simplicity of the clash of civilizations idea of Samuel Huntington (1993, 1996), and earlier of Benjamin Barber’s (1992) jihad. Since September 11, 2001, 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. This idea may become, self-fulfillingly, more fact than fiction: a development that, especially if accompanied by nuclear weapons, would certainly not be in the interests of the United States or anyone else. Sources of Ethnocentrism Members of the conflict/conflict resolution community tend to be humanists, liberals associated with flexible, optimistic views of human nature. 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, which is associated with the Idealpolitik paradigm (see Sandole 1999a, 110–13), tends to ignore—on ideological, political, emotional and practical grounds—the role of biology in conflict, especially violent conflict behavior. Elsewhere (1990), I have argued that biological factors play a role in human behavior—as part of a complex constellation of social, political, economic, 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. I have also argued (1999a, 180–85) that it is not simply a question of “nature” or “nurture,” or indeed, in some simple additive sense, of “nature” and “nurture,” but of both interacting in complex ways, such that each may be affected by the other.2 Given the observations of, among others, Edward O. Wilson (1979), John Pfeiffer (1984), and Joseph Montville (1988) that our brains seem to be preprogrammed to bifurcate everything, including fellow human beings, into membership in in-groups and out-groups, it seems reasonable to conclude that “nature” has invested Homo sapiens with this

48

DENNIS J. D. SANDOLE

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

COMPLEXITY AND CONFLICT RESOLUTION

49

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

50

DENNIS J. D. SANDOLE

sniper incident (October 2–24, 2002) and the murder of three professors at the University of Arizona by a failing student (October 28, 2002); and as Stepp (2002, A3) points out, “The homicide rate for U.S. infants . . . [is] now virtually equal to the murder rate for teenagers, according to a new analysis of government data that revealed a surprising demographic milestone.” But what about the adults, 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, again, elevating the level of analysis to a more global version of us-them hostilities. President George W. Bush’s strident declaration that “you are either with us or the terrorists” has radicalized Muslims all over the world. It has also made many Americans feel a closer sense of community, but at the expense of the security of many American Arabs and Muslims, who feel threatened and victimized by governmental security services as well as by purveyors of hate crimes (see Pierre 2002a, 2002b). In other words, we are returning to the dangerous simplicity of a bipolar world, where, 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, and the worsening insurgency in Iraq, the Realpolitik-ethnocentrism nexus and the intragroup peace versus intergroup war dynamic are taking on a more global, civilizational, “jihadic” character. Hence, although Realpolitik has been conceived as a rational approach to the defense of individual and national interests, it has, in practice, 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; Sandole 1999a). Complex Systems Complexity offers insights in this regard. One of its major assumptions is that, among other things, everything is connected to everything else (see Waldrop 1992). Accordingly, any attempt at problem solving must be at least multi- if not interdisciplinary. But few of us have been educated that way: we receive our degrees usually in only one discipline, and therefore we as analysts may also be more a part of the problem than we are of the solution. Conflict Resolution As an interdisciplinary field, conflict resolution is intuitively similar to complexity and provides a conceptual basis for capturing the complexity of complex con-

COMPLEXITY AND CONFLICT RESOLUTION

51

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

52

DENNIS J. D. SANDOLE

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

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, provocable, forgiving, and clear. Its niceness means that it is never the first to defect, and this property prevents it from getting into unnecessary trouble. Its retaliation discourages the other side from persisting whenever defection is tried. Its forgiveness helps restore mutual cooperation. And its clarity makes its behavioral pattern easy to recognize; and once recognized, it is easy to perceive that the best way of dealing with TIT FOR TAT is to cooperate with it. (176, emphasis added, also see 54)

For TIT FOR TAT to work, however, “the future must have a sufficiently large shadow”; that is, 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). Extending Axelrod’s theory to the wars in former Yugoslavia during 1991–95 leads to the following scenario: 1.

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.

2.

Serbian military successes, plus the “nonprovocability” of the international community, stimulated the development and exacerbation of a violent, asymmetrical conflict-as-process, that is, “ethnic cleansing,” which was prosecuted by the Serbs against the major victims of the wars in former Yugoslavia, Bosnian Slavic Muslims.

3.

In the absence of the “provocability” of the Bosnian Slavic Muslims, the international community was effectively shamed into becoming “provocable” and retaliating against the Serbian “defection” from the previously stable TIT FOR TAT equilibrium, although in a very restrained way (as in the live-and-let-live system of trench warfare during World War I; see Axelrod 1984, chap. 4). Subsequently, the international community was “forgiving” toward the Serbs to avoid stimulating new or exacerbating ongoing violent conflict spirals.

54

DENNIS J. D. SANDOLE

4.

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, in the future, become appropriately “provocable.”

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

COMPLEXITY AND CONFLICT RESOLUTION

55

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

56

DENNIS J. D. SANDOLE

Apropos less lethal forms of conflict handling (e.g., mediation or arbitration), Coser tells us that such “Efforts . . . 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). 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,” especially in the relative absence of such on the part of the international community, thereby establishing a stable balance of power and ensuring that TIT FOR TAT succeeds in Bosnia without further international intervention. Indeed, according to James Pardew, the official originally in charge of the US program, the weapons “would be used for Bosnia’s defense and would contribute to stability in the region. The purpose of the trainand-equip program [therefore] is to prevent war by creating a military balance in Bosnia” (Pomfret 1996b, emphasis added; also see USIP 1997). There is, however, a problem with “balance of power,” as there is with Realpolitik in general. As Shaw and Wong (1989, 47) imply, “trials by ordeal” to determine “relative strength,” as manifested in former Yugoslavia, are associated with “groups as forces of selection [that] represent an emergent, proximate, environmental cause [of war]”: Since failure to maintain a balance of power could have resulted in extinction, groups and their expansion figure as forces of selection in our theory. Motivated by resource competition, conflict, and warfare, struggles to maintain balances of power [have given] rise to more complex societal units which [have] continued the legacy of intergroup warfare. . . . 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, to chiefdom, to nation-state. (45) Applying complexity concepts, therefore, involves more than stable balances associated with negative peace of an enforced temporary respite from violent conflict; it also involves building upon and transcending these and, in positive peace fashion, exploring the dynamic of deep-rooted processes and conditions that make, in the shorter run, the balances a “natural” consequence of social processes (see Galtung 1969, 1996). Seven years after the U.S. intervention, Bosnia was unable to function as a sovereign state. It depended on a North Atlantic Treaty Organization (NATO) force of twelve thousand foreign troops, and the functions of government relied on Western representatives with sweeping powers. And the same nationalist parties that incited the conflict had been

COMPLEXITY AND CONFLICT RESOLUTION

57

reelected (WP 2002). In December 2004, the NATO force was replaced by a British-led EU peace keeping force of some 7,000 personnel. 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), but also— with provocability still an issue—that the negative peace is not a stable one. TIT FOR TAT, therefore—and with it, complexity in general—still remain to be fully applied to Bosnia. THE “COMPLEXITY” OF COMPLEX SYSTEMS As implied thus far, Realpolitik philosophers, theorists, and practitioners tend to respond to the disorder, unpredictability, and insecurity inherent in “Hobbesian space” by advocating and/or pursuing the enhancement of predictability, regularity, and stability (the “PRS needs,” see Sandole 1984)—and, therefore, 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. Hence, Kenneth Waltz’s (1964) earlier defense of a “bipolar” international system as inherently more conducive to stability than a multipolar system; John Mearsheimer’s (1990a, 1990b) lamenting of the end of the Cold War and its simplicity; and following the terrorist attacks of September 11, 2001, President George W. 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. At the extreme “right-wing” end of the Idealpolitik-Realpolitik continuum, we find authoritarians who have a low threshold for uncertainty and insecurity and who, therefore, tend to find democracy too chaotic. But the irony here is that, as “extreme” Realpolitik practitioners implement more and more antidemocratic and threat- or force-based measures in pursuit of order and security, their efforts tend to become more and more counterproductive and self-defeating (see Burton 1972, chap. 6), generating “security dilemmas” (Herz 1950) and the eventual collapse of their own systems. 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, then we might have a situation as we did following the termination of World War II in Europe, when the erstwhile mortal enemies Germany and France established the basis for what has become the

58

DENNIS J. D. SANDOLE

European Union: one of the most illustrative, successful experiments in peacebuilding (still ongoing!) in pursuit of “positive peace.” But there is irony here as well: even Idealpolitik may contain the seeds of its own destruction, because “too much stability” may lead to boredom and atrophy. As Paul Sites (1973) and Kenneth Boulding (1962), among others, have argued, whatever else we as humans may “need” to develop effectively—physically, emotionally, psychologically, and socially—we also have a need for stimulation, for “drama.” 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. Quite a challenge, especially when people are stressed by threats at home and abroad. One of the major lessons of complexity, therefore, is to never take anything for granted for too long on either end of the Idealpolitik/voluntary order– Realpolitik/force-based order continuum. Hence, 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, 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). Frankel’s feat is an example of the potential for diversity in human mental models, Yugoslavia’s flip to violent conflict may reflect complex system sensitivity to initial conditions and small events. Given that complex systems can shift “catastrophically” from one end of the continuum to the other with apparently little effort, one implication here for the architects of globalization is the need to creatively influence the balance between order (“McWorld”) and disorder (“Jihad”), so that, for example, 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,” 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, motivated by fundamental ideology more than by a political agenda, are not deterred by traditional Realpolitik threats or the actual use of force. Consequently, their ultimate intention and effect are to generate maximum unpredictability, instability, and, therefore, disorder and insecurity in the West and those supported by the West. Once the “bite-and-counterbite” dynamic of terrorism versus counterterrorism reaches some critical threshold, another consequence is each side’s overperception of and overreaction to the actions of “the Other” (see Zinnes, North, and Koch 1961; Holsti, North, and Brody 1968). Hence, the self-fulfilling confirmation of Huntington’s (1993, 1996) otherwise contentious “clash of civilizations”

COMPLEXITY AND CONFLICT RESOLUTION

59

or of Barber’s (1992) “jihad,” with weapons of mass destruction in the bargain, all further enhanced by the U.S. rush to war against, and apparently long-term occupation of, another developing country in the Arab/Muslim world, Iraq. Three-Pillar Framework While complexity may have generated paralysis over Bosnia, 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. I have developed the “Three-Pillar” comprehensive mapping of conflict and conflict resolution (see Sandole 1998a; 1999a, chap. 6; 2003) as one such framework for identifying and integrating factors associated with traditionally competing frameworks. 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, chap. 7 and below). Basically, the three-pillar framework comprises pillar 1, conflict—latent conflict (pre-MCP); manifest conflict processes (MCPs); or aggressive manifest conflict processes (AMCPs)—while pillar 2 deals with conflict causes and conditions, and pillar 3, conflict (third-party) intervention (see table 3.1).

TABLE 3.1. 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

60

DENNIS J. D. SANDOLE

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

COMPLEXITY AND CONFLICT RESOLUTION

61

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

62

DENNIS J. D. SANDOLE

[2001]), conflicts developing at any local level could be responded to by a synergistic combination of track 1–9 resources at that level—plus, to the extent necessary and possible, societal, subregional, regional, and global levels as well. Should the vertical dimension fail to prevent “the house from catching on fire,” then there may be a need for the horizontal dimension to become operational. This would involve the judicious use of Realpolitik force, but basically within an Idealpolitik framework, 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. 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, inclusive of the Platform for Co-operative Security (see OSCE Istanbul 1999a, 1999b)—other developments are suggestive of the sole narrow use of Realpolitik force (e.g., the destruction of Grozny and killings of tens of thousands of Chechen civilians in the Russian Federation). 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. These and other developments—the relentless Israeli-Palestinian carnage; the “new” terrorism; the U.S.–Iraq war; the possibility of a nuclear war between India and Pakistan over Kashmir, with one or both sides trying to “preempt” the other; 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, via the law of unintended consequences, potentially very destabilizing. 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, for NATO to stop genocide. But this is not the same as declaring or insinuating, for example, that all Afghans, all Chechens, all Palestinians, all Saudis, all Wahhabis, all Arabs, or all Muslims are “terrorists,” 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. Paradoxically, this problem is being, in part, created self-fulfillingly by this perspective and corresponding behavior! Within the terms of the argument posed here, therefore, Realpolitik force must always take place within a bigger picture, a framework that also allows for and encourages conflict resolution (dealing with the underlying causes of the fire

COMPLEXITY AND CONFLICT RESOLUTION

63

at hand) and conflict transformation (dealing with the long-term relationships among the survivors of the fire), as well as (violent) conflict prevention (preventing the house from catching on fire in the first place), conflict management (if initial conflict prevention fails, preventing the spread of the fire), and conflict settlement (if management fails, forcefully putting out the fire). 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, 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, but to combine it synergistically with other, usually competing, ways of knowing and acting. CONCLUSION It would take extremely enlightened leadership—in the United States, Europe, Russia, China, and Japan, among others (e.g., Israel, Iraq, North Korea, Palestine, 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, but only in collaboration with others (see Sherif 1967). But for some inexplicable (perhaps, in part, “biological”) reasons, ecological degradation, exponential population growth, and a growing gap between haves and have-nots, among other compelling elements of the global problematique (e.g., AIDS), have failed to rise to the status of William James’s (1989) “moral equivalent of war.” Perhaps, “if we have time,” we can leave it to the children: the next generation of decision-makers. In the meantime, however, especially after the Bali bombings (October 12, 2002), Moscow Chechen hostage crisis (October 23–26, 2002), Madrid train bombings (March 11, 2004), and London transit bombings (July 7, 2005), global terrorism itself just might provide the motivation for the international community to come together, and not just to “root out” terrorism (Realpolitik), but to deal with its root causes as well (Idealpolitik). 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, 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.” Among the conceptual tools that could facilitate movement in this constructive (albeit ambitious) direction is the Three-Pillar Framework (3PF) or

64

DENNIS J. D. SANDOLE

3PF-generated new European peace and security system (NEPSS) discussed briefly in this chapter. Analysts working together with policy-makers could use either or both to capture the complexity of complex conflict situations. In this way, 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. To a very large extent, 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; furthermore, 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), Panel on “Global Complexity: Agent-Based Models in Global and International Studies,” Portland, Oregon, February 25–March 1, 2003. The author gratefully acknowledges comments and suggestions made by Neil Harrison and Patrick James. 1. Also see Rosenthal and Jacobson’s (1968) classic study “Pygmalion in the Classroom.” 2. For other discussions in this regard, see Cowley (2003) and Oldham (2003). 3. “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, 26–27, emphasis in the original). 4. This section reflects and builds upon parts of chapter 8 (especially pp. 193–201) of Sandole (1999a). 5. Sensitivity to initial conditions is a characteristic of complex systems that can also be found in some simple systems. In complex systems, however, small changes in initial conditions may lead to nonlinear system changes, a flip in the system to something quite different—as when a forest becomes a desert. But it does not follow that a system in which small changes in initial conditions “cause” large system changes is necessarily complex.

COMPLEXITY AND CONFLICT RESOLUTION

65

6. “There is no place for Turks [the derogatory term the Serbs use for Bosnian Muslims, whose ancestors adopted the Islamic faith of Turkish invaders] in Republika Srpska,” said General Milenko Zivanovic, the regional commander, who led the final assault on Srebrenica” (AP 1996, emphasis added). (Also see Honig and Both 1996 and Rohde 1997.) 7. Thus far, 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).

REFERENCES AP. 1996. “Serbs Mark a ‘Victory,’ Victims Are Exhumed.” Associated Press. International Herald Tribune, 12 July: 10. Axelrod, Robert. 1984. The Evolution of Cooperation. New York: Basic Books. ———. 1997. The Complexity of Cooperation: Agent-Based Models of Competition and Collaboration. Princeton, NJ: Princeton University Press. Bandura, Albert. 1973. Aggression: A Social Learning Analysis. Englewood Cliffs, NJ: Prentice-Hall. Barabasi, Albert-Laszlo, and Eric Bonabeau. 2003. “Scale-free Networks.” Scientific American, 288, no. 5 (May): 60. Barber, Benjamin R. 1992. “Jihad Vs. McWorld.” Atlantic Monthly 269, no. 3 (March): 53–65. Bateson, Gregory. 1973. Steps to An Ecology of Mind: Collected Essays in Anthropology, Psychiatry, Evolution, and Epistemology. Frogmore, England: Paladin. Bertalanffy, Ludwig von. 1973. General System Theory. Harmondsworth, England: Penguin. Boulding, Kenneth E. 1962. Conflict and Defense: A General Theory. New York: Harper and Row. Burton, John W. 1972. World Society. Cambridge: Cambridge University Press. ———. 1997. Violence Explained. Manchester, England: Manchester University Press. Coronel, Mercita. 1996. “Interview with Jane Elliott.” Contrast, 28 March, 1–2. http://www.magenta.nl/EyetoEye/contraste.html.

66

DENNIS J. D. SANDOLE

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

COMPLEXITY AND CONFLICT RESOLUTION

67

———. 1996b. “West’s About-Face on Karadzic: Familiar and Sad.” International Herald Tribune, 2 July, 1, 10. Herz, John H. 1950. “Idealist Internationalism and the Security Dilemma.” World Politics 5, no. 2 (January): 157–80. Hobbes, Thomas. 1950. Leviathan. New York: E. P. Dutton; London: J. M. Dent and Sons. (Orig. publ. in 1651.) Holsti, O. R., R. C. North, and R. A. Brody. 1968. “Perception and Action in the 1914 Crises.” In Quantitative International Politics: Insights and Evidence, ed. J. David Singer. New York: Free Press; London: Collier-Macmillan. Honig, Jan Willem, and Norbert Both. 1996. Srebrenica: Record of a War Crime. New York: Penguin Books. Huntington, Samuel P. 1993. “The Clash of Civilizations?” Foreign Affairs 72, no. 3 (Summer): 22–49. ———. 1996. The Clash of Civilizations and the Remaking of World Order. New York: Simon & Schuster. James, William. 1989. “The Moral Equivalent of War.” In International War: An Anthology and Study Guide, ed. Melvin Small and J. David Singer. 2nd ed. Homewood, IL: Dorsey Press. (Orig. publ. in McClure’s Magazine, August 1910, 463–68.) Jervis, Robert. 1997. System Effects:Complexity in Political and Social Life. Princeton, NJ: Princeton University Press. Kemp, Walter A., ed. 2001. Quiet Diplomacy in Action: The OSCE High Commissioner on National Minorities. The Hague: Kluwer Law International. Kral, Brigitta. 2000. “The Eyes of Jane Elliott—Blue Eyes Brown Eyes Exercise.” Horizon, 1–5. http://www.horizonmag.com/4/jane-elliott.asp. Kuhn, Thomas S. 1970. The Structure of Scientific Revolutions. 2nd ed. Chicago: University of Chicago Press. LeVine, Robert A., and Donald T. Campbell. 1972. Ethnocentrism: Theories of Conflict, Ethnic Attitudes and Group Behavior. New York: John Wiley. Lewis, Anthony. 1996. “Karadzic and Co. Have to Be Tried.” International Herald Tribune, 9 July, 11. Lund, Michael S. 1996. Preventing Violent Conflicts: A Strategy for Preventive Diplomacy. Washington, DC: U.S. Institute of Peace Press. Mearsheimer, John J. 1990a. “Back to the Future: Instability in Europe After the Cold War.” International Security 15, no. 1 (Summer): 5–56.

68

DENNIS J. D. SANDOLE

———. 1990b. “Why We Will Soon Miss the Cold War.” Atlantic Monthly, August, 35–50. Montville, Joseph V. 1988. Foreword to Vamik Volkan’s The Need to Have Enemies and Allies: From Clinical Practice to International Relationships, by Vamik Volkan. Northvale, NJ: Jason Aronson. Morgenthau, Hans J. 1973. Politics Among Nations: The Struggle for Power and Peace. 5th ed. New York: Alfred Knopf. Nagel, Ernest. 1961. The Structure of Science: Problems in the Logic of Scientific Explanation. London: Routledge and Kegan Paul. Nicholson, Michael. 1989. Formal Theories in International Relations. Cambridge Studies in International Relations, 3. Cambridge: Cambridge University Press. Oldham, Pamela. 2003. “A Flight of Mind: The Act of Dissociation Can Protect Children Emotionally from Trauma, but Repeated Use May Cause Lasting Harm.” Washington Post, 18 February, F1, F5. OSCE Istanbul. 1999a. OSCE Istanbul Summit Declaration. 18–19 November ’99. Istanbul: Organization for Security and Cooperation in Europe. ———. 1999b. OSCE Istanbul Summit. 18–19 November ’99: Charter for European Security. Istanbul: Organization for Security and Cooperation in Europe. Pagels, Heinz R. 1988. The Dreams of Reason: The Computer and the Rise of the Sciences and Complexity. New York: Simon & Schuster. PBS. 1985. “A Class Divided.” Frontline (PBS documentary produced by W. Peters). http://www.pbs.org/wgbh/pages/frontline/programs/categories/info/ 309.html. Pfeiffer, John. 1984. “Human Nature: The Universe Inside Your Skull.” Science Digest 92:92. Pierre, Robert E. 2002a. “Fear and Anxiety Permeate Arab Enclave Near Detroit: Muslim Americans Feel They Are Targets in War on Terror.” Washington Post, 4 August, A3. ———. 2002b. “Victims of Hate, Now Feeling Forgotten: Family of Man Killed After 9/11 Finds Little Charity, But Much Hardship.” Washington Post, 14 September, A1, A10. Pomfret, John. 1996a. “Croats and Muslims Edge Toward a Joint Army.” International Herald Tribune, 8 July, 5. ———. 1996b. “U.S. Starts Delivery of Heavy Weapons to Bosnia’s Muslim-Croat Forces.” Washington Post, 22 November, A44.

COMPLEXITY AND CONFLICT RESOLUTION

69

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

70

DENNIS J. D. SANDOLE

———. 1998b. “A Peace and Security System for Post-Cold War Europe: Preventing Future Yugoslavias.” In Encyclopedia of the European Union, ed. Desmond Dinan. Boulder, CO: Lynne Rienner. ———. 1999a. Capturing the Complexity of Conflict: Dealing with Violent Ethnic Conflicts of the Post-Cold War Era. London: Pinter/Cassell (Continuum International). ———. 1999b. “A Design for Peace and Security in Post-Cold War Europe.” Ethnic Conflict Research Digest (University of Ulster) 2, no. 1 (February): 41–42. ———. 2002. “The Causes of Terrorism.” In Terrorism: Concepts, Causes, and Conflict Resolution, ed. R. Scott Moore. Fort Belvoir, VA: U.S. Defense Threat Reduction Agency. ———. 2003. “A Typology.” In Conflict: From Analysis to Action, ed. Sandra Cheldelin, Daniel Druckman, and Larissa Fast. London: Continuum International. ———. 2004. “Building Peace in Post-NATO Bosnia: A Recommended Action Plan.” In From Peace Making to Self Sustaining Peace—International Presence in South East Europe, ed. Predrag Jurekovic, Frederic Labarre, and Ernst M. Felberbauer. Vienna: National Defence Academy. www.bmlv.gv.at/wissenforschung/publikationen/publikation.php?id=160. Saperstein, Alvin M. 1995. “War and Chaos.” American Scientist 83, no. 6 (November–December): 548–57. Shaw, R. Paul, and Yuwa Wong. 1989. Genetic Seeds of Warfare: Evolution, Nationalism, and Patriotism. Boston: Unwin Hyman. Sherif, Muzafer. 1967. Group Conflict and Cooperation: Their Social Psychology. London: Routledge and Kegan Paul. Sherif, Muzafer, and Carolyn W. Sherif. 1953. Groups in Harmony and Tension. New York: Harper and Row. Simmel, Georg. 1955. Conflict and the Web of Group-Affiliations. Trans. Kurt H. Wolff and Reinhard Bendix. New York: Free Press; London: CollierMacmillan. Sites, Paul. 1973. Control: The Basis of Social Order. New York: Dunellen. Stepp, Laura Sessions. 2002. “Infants Now Murdered As Often As Teens: Actual Rate May Be Higher, Experts Say.” Washington Post, 10 December, A3, A17. Stern, Paul C., and Daniel Druckman. 1994/1995. “Has the ‘Earthquake’ of 1989 Toppled International Relations Theory?” Peace Psychology Review 1, no. 2 (Autumn/Winter): 109–22.

COMPLEXITY AND CONFLICT RESOLUTION

71

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

This page intentionally left blank.

CHAPTER 4

Understanding and Coping with Ethnic Conflict and Development Issues in Post-Soviet Eurasia Walter C. Clemens, Jr.

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

73

74

WALTER C. CLEMENS, JR.

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

ETHNIC CONFLICT AND DEVELOPMENT

75

The behavior of social systems emerges from the interactions of their members. This means that their fitness is a function of the interaction of individuals within the social and political parameters of the society. Because the fitness of countries is an emergent property, it is not possible to predict with precision how countries will react to changes in the environment. Up to a point, countries that are more decentralized are expected to be more adaptable and, therefore, fitter. The fitness of the United States hovers close to the edge of chaos, while that of Singapore teeters on the brink of rigidity. Fitness Landscapes. 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. As in an arms race, the peaks of a predator and its prey may gain or decline according to changes in their offensive and defensive capabilities. If attackers acquire more lethal weapons, the fitness peak of the prey will drop. If individuals among the prey population acquire characteristics that reduce their vulnerability, their peaks will rise. Self-organized Criticality. Balanced between order and chaos, a fit being is like a sandpile that, if one more grain of sand is added, may collapse in an avalanche. This fragile equilibrium is called self-organized criticality. The sandpile metaphor, however, is not universally accepted and is not essential to complexity theory. Punctuated Equilibrium. The concept of punctuated equilibrium underscores that evolution is often marked by surges of speciation and avalanches of extinction (Gould 2002).4 Species often develop quickly, endure with little change for a long time, and then die out suddenly—not gradually. Thanks to mutation and self-organization, members of the species find their niche and hang on to it. When their environment changes, they must adapt or disappear. How long the system is stable and endures is difficult to predict—especially in politics. 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. 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. Adapting Snyder’s (2000) analysis, 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. In zone A was a

76

WALTER C. CLEMENS, JR.

set of countries that benefited from ethnic calm and enjoyed gradual economic and political development; in zone B, a shatterbelt of ethnic conflict and material regress; in zone C, 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). Finally, we may distinguish a hybrid zone D where major countries—Russia and Ukraine—shared some but not all characteristics of the other regions. 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. From the former Yugoslavia, the exemplar is Slovenia. From erstwhile Soviet allies in Eastern Europe, the leaders are the Czech Republic, Poland, Hungary, and a late rising star, Slovakia.5 Of former Soviet republics, only Estonia, Latvia, and Lithuania belong in zone A (Clemens 2001). Zone B comprises societies that became embroiled in severe ethnic fighting in the 1990s—Chechnya, most of former Yugoslavia, and the erstwhile Soviet republics of Armenia, Azerbaijan, Georgia, and Moldova. Each showed a very low capacity for coping with ethnic differences and the problems of establishing a viable economy and a stable democracy. In each case, as Snyder says, partial democratization probably aggravated ethnic tensions. Thus, “democracy” made it harder for Armenia’s leaders to negotiate any kind of compromise with Azerbaijan over Nagorno-Karabakh, because nationalist firebrands could mobilize votes against them.6 Zone C refers to Central Asia and Belarus, where dictators suppressed ethnic or other challenges to their rule. In the 1990s Tajikistan experienced much fighting between political rivals, but ethnic differences were not at issue. In the former Soviet republics of Central Asia, erstwhile Communist leaders became dictators claiming to be both nationalist and democratic. Kyrgystan had a free press for a time, but this ingredient of a true democracy disappeared in the mid-1990s.7 President Aleksandr Lukashenko tried to russify Belarus and negotiate its union with the Russian Federation. His opponents sought to establish and maintain a clear Belarusian identity, but Lukashenko repressed them with little overt violence. 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. There were signs that Bulgaria, Romania, Croatia, Montenegro, and perhaps even Serbia might follow suit. But the scales teetered. Each of these countries could readily drop into zone B or C. Thus, Serbia made major strides toward real democracy and peace with Montenegro in 2001–2, but could still become embroiled in

ETHNIC CONFLICT AND DEVELOPMENT

77

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

APPLYING COMPLEX ADAPTIVE SYSTEMS THEORY TO EXPLAIN PAST AND PRESENT FITNESS Adopting the language of CAS, this chapter argues that countries such as Slovenia, the Czech Republic, and Estonia in the 1990s demonstrated a high level of

Norway Belgium United States Canada Japan Israel Greece Cyprus Singapore Slovenia Korea, Republic of Brunei Darussalam Argentina Estonia Cuba Belarus Malaysia

Country 1 6 7 8 9 22 24 25 28 29 30 31 34 41 52 53 58

HDI Rank 2003, n⫽175 3 2 6 5 11 22 25 25 28 29 30 31 34 38 n.a. 48 53

GDI Rank 2003, n⫽175 Free Free Free Free Free Free Free Free Partially Free Free Free Not free Partly free Free Not free Not free Partly free

Freedom Index 2003 28 (MF) 19 (MF) 6 (F) 18 (MF) 35 (MF) 33 (MF) 56 (MF) 22(F) 2(F) 62 (MF) 52 (MF) n.a. 68 (MF) 6 (F) 155 (RE) 151 (RE) 72 (MU)

Economic Freedom Rank 2004, n⫽153

TABLE 4.1. THE HIGHEST-RANKING COUNTRIES IN “HUMAN DEVELOPMENT” AND OTHER VALUES FROM VARIOUS CIVILIZATIONS

8 17 18 11 21 21 50 27 5 29 50 n.a. 92 33 43 53 37

Honesty Rank 2003, n⫽133

Protestant Catholic Protestant Catholic Japanese Israeli Orthodox Orthodox Mixed Catholic Asian Muslim Catholic Protestant Catholic Orthodox Muslim (continued)

Cultural Tradition

74 103 104 112 127 133

61 82 83 91 103 107

GDI Rank 2003, n⫽175 Free Free Not free Partly free Free Not free

Freedom Index 2003

70 n.a. 43 n.a. 83 n.a.

Honesty Rank 2003, n⫽133

Buddhist African Chinese Muslim-Hindu Hindu-Muslim African

Cultural Tradition

Sources: “HDI” and “GDI” are from U.N. Development Programme, Human Development Report 2003 (New York: Oxford University Press, 2003), tables 1 and 22. “Freedom Index” is from Freedom House at www.freedomhouse.org. “Economic Freedom” is from Heritage Foundation at www.heritage.org. “Honesty Rank” is from Transparency International at www.transparency.org.

40 (MF) 89 (MU) 127 (MU) 99 (MU) 119 (MU) 72 (MU)

Economic Freedom Rank 2004, n⫽153

Code: For economic freedom, F ⫽ free; MF ⫽ mostly free; MU ⫽ mostly unfree; RE ⫽ repressed.

Thailand Cape Verde China Indonesia India Swaziland

Country

HDI Rank 2003, n⫽175

TABLE 4.1. (continued)

80

WALTER C. CLEMENS, JR.

fitness. As we see in table 4.1, 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, Romania, and Belarus. Each country in zone A joined both NATO and the European Union. In zones C and D, by contrast, few countries showed much interest in or had much prospect of joining NATO or the EU in the foreseeable future. Societies in zone A achieved high levels of fitness on many fronts after the demise of the Soviet empire. Success in one domain helped them cope with problems in others. Ethnic peace made it easier to raise living standards, consolidate democracy, and nourish creativity. Economic advances in Estonia, for example, make it easier for Tallinn to provide welfare benefits for Russian-speakers residing in Estonia but who were not citizens. Countries in zones B, C, and D displayed low levels of overall fitness even though many possessed assets lacking in zone A. Thus, Azerbaijan, Kazakstan, and Russia possess energy resources far superior to those in any lands in zone A. Parts of Ukraine and Russia have better soil as well as much richer mineral deposits than any country in zone A. Georgia, Armenia, Ukraine, and Russia have evolved from states and cultures dating back more than a thousand years. Slovenia, by contrast, was never an independent state before 1992. Estonia and Latvia had only two decades of independence between the two world wars. Most countries in zones B, C, and D faced simpler ethnic challenges than in many zone-A countries, because they were more homogeneous. Ethnic minorities were very small in Belarus, Moldova, the South Caucasus, and in most of Central Asia (except for Kazakstan). About four-fifths of the Russian Federation’s population was Russian but most other groups in the federation spoke Russian. A million or so Chechens occupied only a dot on the federation’s periphery. Still, the governments in zones B, C, and D experienced great difficulty in dealing with ethnic minorities. By contrast, Estonia and Latvia in the 1990s faced minorities of Slavic speakers that made up more than one-third of the resident population. Estonian and Latvian leaders espoused a kind of ethnic nationalism tempered by civic moderation. They instituted a naturalization process that required aspiring citizens to pass residency, language, and civic tests. 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. Children and young adults, of course, learned Estonian or Latvian more readily than most of their elders. Still, ethnic tensions produced no deaths in the Baltic. Estonia even permitted noncitizens to vote in local elections. Indeed, 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.

ETHNIC CONFLICT AND DEVELOPMENT

81

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

82

WALTER C. CLEMENS, JR.

Whatever the shortfalls of the European Union, it is a triumph of cooperation compared to the beggar-thy-neighbor behaviors of ex-Communist societies. Indeed, 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. 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. This is not quite “order for free,” which Kauffman’s version of CAS attributes to established ecosystems (such as coral reefs). Still, it resembles the positive results that Adam Smith expected if individuals were allowed to do what they do best, as if guided by an “invisible hand.” Self-Organized Criticality CAS warns that societies may be less fit than they appear. Fitness depends on the harmony of many factors. Just as an extra grain of sand may cause a sandpile to collapse, a new or heavier burden could seriously weaken an apparently fit society. 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, 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. Punctuated Equilibrium The concept of punctuated equilibrium warns us not to expect steady progress in fitness. West European unification did not emerge gradually but in sharp jumps and with some steps backward. Meaningful social change often requires a period of preparation. New generations can be educated. In Estonia and Latvia many young persons who speak Russian at home are learning the official state language. Accumulating experiences may tip even middle-aged Russian-speakers toward integration with native Balts. Long plateaus without improvement may drive some people to depart or to take drastic steps to effect change. But regress

ETHNIC CONFLICT AND DEVELOPMENT

83

is also possible. 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. Can we graph changing patterns of fitness among the societies of post-Soviet Eurasia? This is not a simple task, if only because fitness among humans is multidimensional. The UN Human Development Index provides a solid starting point to measure public health, education, and material living standards.8 If we focus on ethnic problems, we would also study measures of ethnic harmony and its opposite—injury, dislocations, and deaths caused by ethnic unrest. We expect that low fitness in this domain will tend to correlate with low scores in overall human development, lack of political and civil liberties, low technological achievement, and corruption. Though it is difficult to show all these variables in a single peak, a cobweb graphic could illustrate the correlations suggested here.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. But they offer only general principles for anticipating future outcomes or prescribing constructive policies. In this regard, however, it does no worse than most competing theories—few of which provide useful handles for predicting or shaping the future. Indeed, if CAS is correct about the role of self-organization in fitness, social Darwinists and ultrarealists are wrong: success in politics does not derive from raw power plus cunning. 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 high creativity is most frequently found close to the edge of chaos. This insight helps explain why Central Asia is frozen in time, why the Caucasus explodes, and why Russia resorts to an iron fist to overcome chaos, and why Slovenia and Estonia adapt well to their new freedoms. This insight has clear policy implications: avoid the extremes of dictatorship and anarchy. To generate a healthy and innovative society, cultivate selforganization—not a system steered from on high. Western policymakers and investors should not count on authoritarian regimes in Kazakstan or Azerbaijan to

84

WALTER C. CLEMENS, JR.

maintain order forever. They should not prop up local dynasties in the hope of securing privileged access to oil and gas. Outsiders cannot compel internal reforms but should do what they can to nudge these societies toward greater selforganization. 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. 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. 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. The stronger and more diverse the independent agents shaping the formerly Communist societies, the healthier and fitter they will be. Constructive policies will cultivate creative individuals, businesses, and NGOs that enlarge public goods and are not dominated by government. These independent agents face a difficult struggle against the moral legacies of Communism—corruption, groupthink, and a welfare mentality that discourage initiatives from the bottom up. Even if the goal of self-organization seems clear, questions arise about the road to this goal. What if democracy terminates democracy—as happened in Germany in the 1930s? Is self-organization desirable if the majority votes against the minority, 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, as in parts of Nigeria? HOW TO ACQUIRE AND NURTURE FITNESS Culture matters.10 All the societies in zone A became oriented toward universal literacy, free thought, and open debate (relative to most other societies) long ago. The societies in zones B, C, and D moved toward universal literacy only in the past 100 years.11 Many regimes in these zones still discourage or try to prevent open debate on policy and other important issues. Following the leads of John Wycliffe, Jan Hus, Martin Luther, and other reformers, each society in zone A acquired its sacred religious texts in the vernacular between the fifteenth and seventeenth centuries. For the first time in history, some princes and religious leaders also urged individuals—female as well as male—to read and interpret sacred texts on their own. This twin revolution helped to liberate all who experienced it (Clemens 2005). After the Peasants’ Revolt, however, Luther feared that he was provoking chaos. He then wrote his Short Catechism instructing people what to believe. But Luther could not stop the transformation he had

ETHNIC CONFLICT AND DEVELOPMENT

85

Figure 4.1. Date the Bible Published in Vernacular Correlated with HDI Rank

unleashed. The synergies of literacy and individualist thinking were empowered by the printing press, the Renaissance, the discovery of the New World, and growing refinement of scientific methods. Catholic France and Italy had Bibles in the vernacular even before Luther’s challenge to Rome. In the seventeenth century Sweden’s monarchy and state church wanted their subjects—even servant girls—to read and discuss the Bible. Bibles in the vernacular also helped cultivate a sense of national identity (Hastings 1997; Lepore 2002). Certainly many factors shape human development, but figure 4.1 shows a strong correlation between high HDI scores and early publication of Bibles in the vernacular. Where Orthodox Christianity prevailed, Bibles in the vernacular were not widely published until the late nineteenth century or the twentieth century. (The sole exception was Romania, which published both the New and Old Testaments in the seventeenth century.)12 Wide-scale literacy came to the Orthodox countries much later than in Protestant and Catholic countries or in Jewish communities.

86

WALTER C. CLEMENS, JR.

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

ETHNIC CONFLICT AND DEVELOPMENT

87

promised to replace corruption and chaos with a rule of law. By 2004, however, he was promoting his own cult of personality. Saakashvili had learned the rhetoric of democracy, but—in a society that wants a strong, charismatic leader— he gravitated toward the national norm. Would closer ties with America improve fitness in former Soviet republics? For countries such as Georgia and Uzbekistan, closer ties with the U.S. hyperpower might bring material gains but could also weaken self-organized fitness. Lacking internal strength, 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, 57). 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, metaphors, models. Thus, a major insight of CAS is the concept of societal fitness. Unlike neorealists who believe that relative material power—missiles and GDP—is the best guide to world politics, CAS suggests each actor’s most basic need is a capacity to cope with challenges at home and abroad, including ethnic diversity. How could we operationalize these concepts? Let us assume that HDI rank is an approximate indication of societal fitness, and that societal fitness in a large, modern society depends on universal literacy and free expression. 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. 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. To really understand these relationships, however, each variable would have to be studied in depth. 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. 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?

2.

Trace the evolution of literacy in these countries. Develop a common standard for measuring literacy. 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?

88

WALTER C. CLEMENS, JR.

3.

Trace the growth of independent thinking. What indicators— in science, the arts, 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.

4.

Adapt the approach used to study traditionally Christian countries to those in the Muslim and other religious traditions.

5.

Determine a way to prove causation rather than mere correlation. How should we weigh the contribution of one factor, such as literacy, against others, such as growing wealth?

6.

Analyze the chicken-and-egg. Ask which came first: individual freedom or the twin revolution? Long before Luther, conditions favoring individual freedom were stronger in some regions (such as Bohemia) than in Byzantium or Russia.

7.

Distinguish the kinds of ethnic/national consciousness that existed in previous centuries (for example, among Bohemia’s Hussites) and that of the last century or two.

8.

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?

9.

Learn from outliers: Romania (Orthodox but closer to Rome than to Russia) had the Bible relatively early but nonetheless has a low HDI ranking. Slovaks got the Bible relatively late but achieved a fairly high HDI score in the late 1990s. 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; they are accessible online from the U.N. Development Programme.

10.

Consider the shortcomings of the twin revolution. If literacy and independent thinking conduced to human development, why has the West shown so much intolerance, violent nationalism, and war? Perhaps the twin revolutions were necessary but not sufficient for overall fitness.

11.

Consider the policy implications: If high levels of human development may be traced to a twin revolution that began five hundred years ago, how can they be fostered in societies that have experienced one or both revolutions only in recent decades?

12.

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?

ETHNIC CONFLICT AND DEVELOPMENT

89

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

90

WALTER C. CLEMENS, JR.

5. See the studies done for Freedom House by Adrian Karatnycky, Alexander Motyl, and Aili Piano (2001) and the country reports published regularly by Human Rights Watch, Amnesty International, Transparency International, the United Nations Development Programme, and the U.S. Department of State—all accessible on the Internet. 6. Citing several studies, Snyder concludes that the ethnic content of the Moldovan conflict was ambiguous. The Moldovan government in the early 1990s was nationalistic, but Russian-speakers in the breakaway “Transdniestr Republic” were driven more by nostalgia for the Soviet empire than by nationalism (Snyder 2000, 250–51). 7. Abutting the former USSR, Moscow’s one-time client state Mongolia is a special case. In the 1990s Mongolia moved quickly toward democracy, even though it was poorer than most parts of the USSR and had a weak infrastructure for education and communication. The country had few internal ethnic problems (90 percent of the population is Mongolian; 4 percent Kazak; 2 percent Russian; 2 percent Chinese; 2 percent other) and did not clash with China despite the potential for expansionist claims by each side. 8. 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 capitals, see Clemens 2001, 110–11. 9. For a model, see Maruca 2000, 24. 10. For a range of viewpoints, see the essays in Lawrence E. Harrison and Samuel P. Huntington (2000); Mariano Grondona (1996); and Dominique Jacquin-Berdal et al. (1998). 11. Literacy rates are difficult to track and measure, but estimates for many formerly Communist countries are at Snyder 2000, 200–202. 12. Dates of Bible publication in many languages, including those native peoples of Siberia and North America, are given in www.world scriptures.org, which also reproduces the opening lines of St. John’s Gospel in each language. 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, each spoken now by only a hundred or so persons. 13. An Azeri in Moscow showed me his family Qur’an written in Arabic, for him a completely unknown tongue. 14. Some translations into Malay have been so literal that they were not intelligible without prior knowledge of Arabic. But Bosnians could have read a translation into Serbo-Croat in 1875 (Swartz 2004).

ETHNIC CONFLICT AND DEVELOPMENT

91

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

92

WALTER C. CLEMENS, JR.

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

ETHNIC CONFLICT AND DEVELOPMENT

93

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

This page intentionally left blank.

CHAPTER 5

Beyond Regime Theory Complex Adaptation and the Ozone Depletion Regime Matthew J. 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. This is not a wholesale indictment of regime theory. As there is no single regime theory, it is not possible to indict the whole enterprise (Hasenclever, Meyer, and Rittberger, 1997). Rather, I focus on one aspect of a complexity approach that is missed by traditional (especially neoliberal) regime theory—namely, 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. I illustrate how complexity concepts can help us to move beyond regime theory with a brief examination of the ozone depletion regime. In its traditional, neoliberal instantiation, regime theory is mostly concerned with bargaining. 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, 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; Krasner 1983; Hasenclever, Meyer, and Rittberger, 1997; Koremenos, Lipson, and Snidal 2001). The goal of regime theory is simple: to explain when rules, procedures, principles and norms (Krasner 1983) are likely to occur in a particular issue-area.

95

96

MATTHEW J. HOFFMANN

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

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. A complexity approach provides insight into how this coevolution takes place, thus providing insight into the formation of regimes and their evolution through time. Through complexity, 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. 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). To demonstrate such an approach, I turn to the ozone depletion regime. This oft-studied regime (see, e.g., Litfin 1994; Haas 1992; Benedick 1991; Tolba 1998; Parson 1993; Rowlands 1995) has been hailed as a singular success in the realm of environmental politics. 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. Specifically, I explore two aspects of this complicated regime. First, 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, and what rules would be encompassed in the regime itself. It is impossible to understand the ozone depletion regime without understanding the underlying requirement for universal participation that developed. Second, why did the United States, which could have ignored this change, adapt its understanding of the system rules to fit the change?. 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. 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. First, I discuss the aspects of complexity used to analyze the ozone depletion regime. I then give a brief overview of the ozone depletion regime for background purposes. The discussion of the two main puzzles is next, and I conclude with some thoughts about further empirical testing of the hypotheses presented.

98

MATTHEW J. HOFFMANN

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

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. Very simply, the logic of complexity entails the coevolution of actor understandings and the political context. States are embedded in a system with rules, and each state has an internal understanding of the system rules. At any time states “know” what the global response to environmental problems like ozone depletion should be, though not all states will share the same understanding. States act on the basis of these internal rule models. Their actions (bargaining, for instance) shape the system rules, reifying or changing the rules that dominate in an issue-area. There is, thus, 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. Change is driven by interpretation of scientific information, domestic politics, and social interaction with other states—the evaluation of actions and the altering of rule models.5 The ozone depletion regime provides fertile ground for a demonstration of the empirical utility of a complexity perspective. After a brief background description of the ozone depletion regime, I trace the process of complex adaptation and discuss how universal participation emerged, what impact it had on the ozone depletion regime, and why the United States adapted to this emergent participation rule.

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, chap. 2).6 Politically, this “global” problem entailed varied responses. Some states, notably the United States and the Nordic states, took unilateral domestic action and began to regulate CFCs. International activity began in the late 1970s, and negotiations resulted in a succession of agreements (Vienna Convention, 1985; Montreal Protocol, 1987; London Amendments, 1990; Copenhagen Amendments, 1992) that moved the state of the ozone depletion regime from a call for research through a complete phaseout of CFCs. 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. At first the EU was opposed to reductions in ozone-depleting chemicals, because collectively the European states had come to dominate the

100

MATTHEW J. HOFFMANN

worldwide CFC market when US domestic actions served to decrease its output. With the EU and US at loggerheads, the Vienna Convention accomplished little, except for an important call for further negotiations and a statement of guiding principles. The discovery of the ozone “hole” in 1986 dramatically heightened the sense of urgency surrounding the ozone depletion problem, and though the international community lacked scientific proof of the connection between CFCs and ozone depletion, they forged ahead with negotiations to reduce CFC emissions. These negotiations culminated in the Montreal Protocol and were again mainly comprised of US/EU bargaining. 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. The Montreal Protocol itself laid out a compromise on binding reductions that would see CFC emissions decrease by 50 percent by the year 2000. 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 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. Regime dynamics did not end with the Montreal Protocol, however. After the protocol was signed, new scientific findings solidified the proof of the CFCozone depletion connection, and the urgency surrounding the issue ramped up once again. The move to amend the Montreal Protocol and strengthen the ozone depletion regime was complicated by the now universal attendance at the ozone negotiations. Post-Montreal negotiating sessions routinely included upwards of one hundred states, and the North-South dimension of these negotiations altered the regime dynamics significantly. At the London meeting of the Parties to the Montreal Protocol bargaining was much more North-South than it was United States–European Union. 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. Thus, in five short years, 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. Two parts of this regime formation story demand our attention. First, in 1987 we see a breakpoint change in how the international community perceived the proper response to ozone depletion. Before this time, the proper response consisted for the most part of Northern states alone. The system rule defined

BEYOND REGIME THEORY

101

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

102

MATTHEW J. HOFFMANN

Southern states, now operating with similar rule models again, instantiate a newly stable system rule for universal participation. This is quintessential complex adaptation. Agents are driven by rule models and their actions (re-)create the system in which they are embedded. Innovation in a subset of the population of agents disrupts the system rules and leads to adaptation across the whole system. The microprocess of agent-adaptation and internal rule model change is linked to the macroprocess of evolution in system rules. The story is profound because of the implications of the transitions. The system rules can be considered the political context (or “episteme”—see Johnston 2005) within which regimes are formed; they are the rules of the game and structure the process of regime formation. Understanding the system rules is a key aspect of analyzing regimes, because they determine what gets bargained over and how such bargaining takes place. In traditional regime theory, the system rules are assumed to be static and unproblematic. This is not always a poor set of assumptions, but it reifies a misleading understanding of politics as relatively stable. 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. With a complexity perspective we can account for change in system rules and more fully explain the process of regime formation. In the following discussion, 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.9 Stage 1: North-Only Participation In 1985 there was a stable system rule calling for North-only participation. Both Northern and Southern states had internal rule models that defined the ozone depletion problem as one requiring a Northern negotiated regime, and the actions of both Northern and Southern states reified the North-only system rule. It was obvious to both Northern and Southern states that ozone depletion was a Northern problem (Sims 1996, 201–14). Only four of the original twentyone signers of the Vienna Convention were Southern states, and by 1987, only two developing nations had ratified the convention (Benedick 1991, 265–69). 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. [The] group thought they owned the issue.” 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-

BEYOND REGIME THEORY

103

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

104

MATTHEW J. HOFFMANN

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

BEYOND REGIME THEORY

105

In participating, Southern states altered the stability of the system rule calling for North-only participation. With Northern states still following the Northonly rule and Southern states following a new internal model that called for their own participation, 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. The flux had little impact on the Montreal negotiations themselves. There was a great deal of inertia behind the proceedings to this point, and Southern participation at Montreal did not greatly alter the work that had been accomplished over the previous two years. In addition, as the Northern states were still operating on a North-only participation rule, the substance of the discussions at Montreal remained focused for the most part on Northern issues of CFC reductions and substitutes. While Southern issues were raised, the South started ‘playing’ too late to be a major factor at Montreal, as most of the provisions of the MP had already been worked out (Miller 1995, 78–79). 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. 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. As Northern states would soon discover, 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. 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. Stage 3: US Adaptation and the Emergence of Universal Participation Southern states were not satisfied with a protocol that was negotiated essentially without them, and they signaled this dissatisfaction by declining to sign the Montreal Protocol. This action, while further eroding the North-only system rule, blindsided the United States and other Northern states. 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, 169–70). 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. Yet Northern adaptation to the flux in the participation system rule was not inevitable. If Northern states deemed the Montreal Protocol sufficient to

106

MATTHEW J. HOFFMANN

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

BEYOND REGIME THEORY

107

is virtually no chance that the current protocol will be sufficient to solve the problem” (International Environment Reporter 1988a, 111), and the NRDC prepared to force the EPA to take action through the courts once again (Global Climate Change Digest 1988). Even industry joined the calls for further international action, hoping to delay or defeat unilateral measures (Carnevale 1988). It is hard to overestimate the importance of industry’s role in pushing universal participation in ozone depletion. Though industry did not create this vision of a global response, once it was in place industry played a crucial role in fostering the evolution of US rule models toward universal participation.15 These evaluations of the Montreal Protocol significantly weakened the US internal rule model calling for a North-only response to ozone depletion. The North-only rule had driven US behavior in the negotiations of a landmark agreement, but the Montreal Protocol was a landmark only as a bare beginning in the fight against ozone depletion. With the participation rule weakened by negative evaluation, the United States adopted a universal participation rule. The United States adapted to the new system rule and began advocating universal participation. The new consensus on the system rule calling for universal participation made a significant difference in how the bargaining proceeded. Participating in earnest in the negotiating process for the first time in large numbers, the South began pushing for developmental assistance at a 1989 London conference on the ozone depletion problem attended by 123 states (UNEP 1989, paragraph 11). As Litfin observed, “[T]he treatment of developing countries, which hitherto had been considered a minor issue, became a central concern” (Litfin 1994, 129). It was at this conference that the South became vocal. Daniel Arap Moi of Kenya voiced the generic Southern concern when he stated, “Some nations will not find it easy to forego the use of CFCs in their quest for industrialization” (International Environment Reporter 1989a, 106). Representatives from India and China were more direct. 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, 169). A Chinese official protested that Southern nations “resented the rich ‘telling them what to do and not to do’” (International Environment Reporter 1989b, 169). The work ahead was clear-cut. As Richard Smith of the US State Department reported: “At Helsinki [the first meeting of the parties to the Montreal Protocol], 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, 77). It was clear that if the Northern states were to achieve Southern cooperation, they would have to accede to an international

108

MATTHEW J. HOFFMANN

fund and technology transfer mechanisms—they would have to make side payments to the South. This was the bargain that the South, now fully engaged in the negotiating process, was driving. The United States was loath to submit to Southern demands, and the impasse threatened to scuttle the ozone negotiations. The South made it clear that they would not participate without a fund, and the United States, in particular, held on to its opposition to a new fund, advocating instead the use of the World Bank and other existing institutions (Weisskopf 1990a, A21). Just prior to the London meeting of the parties to the Montreal Protocol, 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, A1). At London, compromise ruled the day, and by June 29, 1990, a deal had been struck (UNEP 1990). The London Amendment to the Montreal Protocol contained the ultimate phaseout by 2000 desired by all parties (and most nonparties), but the bigger accomplishment of the London negotiations was the establishment of the Multilateral Fund and technology transfer mechanisms. The United States and other Northern states agreed to pay the full incremental costs of the transition away from CFCs incurred by the South, and more importantly, they agreed to do this with new and additional funds administered by a new institution. This was a huge victory for the South, because the new institution was to be jointly controlled, rather than solely administered by donor states. With the funding provision in place, the South (crucially India and China) agreed to the Montreal Protocol, and the fight against ozone depletion became a truly global affair. Transition through Complex Adaptation Universal participation was not a natural or inevitable rule for the ozone depletion regime. Nothing about ozone depletion or the interests of the major actors radically changed between 1986 and 1988. What did change were the internal rule models of the Southern states—how they perceived and acted toward the ozone depletion problem. This change set off a coevolutionary transition in both the internal rule models of the Northern states and the system rule for participation. Regime theory misses these dynamics and is thus hampered from understanding the bargaining that takes place in regime formation and beyond. BEYOND REGIME THEORY? Three questions remain, however. First, why should we care about the emergence of universal participation and the US adaptation to this new rule? The second is

BEYOND REGIME THEORY

109

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

110

MATTHEW J. HOFFMANN

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

BEYOND REGIME THEORY

111

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

112

MATTHEW J. HOFFMANN

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

BEYOND REGIME THEORY

113

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

114

MATTHEW J. HOFFMANN

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

BEYOND REGIME THEORY

115

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.

116

MATTHEW J. HOFFMANN

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), www.globalchange.org/gccd/gcc-digest/d88sep3.htm. 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.

BEYOND REGIME THEORY

117

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

118

MATTHEW J. HOFFMANN

Miller, Marian A. L. 1995. The Third World in Global Environmental Politics. Boulder, CO: Lynne Reiner. Mills, Mike. 1988. “But Some Say ‘Too Little Too Late’: Ratification of Ozone Pact Recommended.” Congressional Quarterly Weekly Report, February 20, 370. Mitchell, Ronald, and Patricia Keilbach. 2001. “Situation Structure and Institutional Design: Reciprocity, Coercion, and Exchange.” International Organization 55, no. 4: 891–917. Morrisette, Peter M., Joel Darmstadter, Andrew Patinga, and Michael Toman. 1991. “Prospects for a Global Greenhouse Gas Accord: Lessons from Other Agreements.” Global Environmental Change 1, no. 3 (June): 209–23. Parson, Edward. 1993. “Protecting the Ozone Layer.” In Institutions for the Earth: Sources of Effective International Environmental Protection, ed. Peter Haas, Robert Keohane, and Marc Levy, 27–74. Cambridge, MA: MIT Press. Pierson, Paul. 2000. “Increasing Returns, Path Dependence, and the Study of Politics.” American Political Science Review 94, no. 2: 251–68. Rajan, Mukund Govind. 1997. Global Environmental Politics: India and the NorthSouth Politics of Global Environmental Issues. Delhi: Oxford University Press. Risse, Thomas, Stephen Ropp, and Kathryn Sikkink, eds. 1999. The Power of Human Rights: International Norms and Domestic Change. Cambridge: Cambridge University Press. Rowlands, Ian. 1995. The Politics of Global Atmospheric Change. New York: Manchester University Press. Sims, Holly. 1996. “The Unsheltering Sky: China, India and the Montreal Protocol.” Policy Studies Journal 24, no. 2:201–14. Tolba, Mostafa. 1998. Global Environmental Diplomacy: Negotiating Environmental Agreements for the World, 1973–1992. Cambridge, MA: MIT Press. UNEP. 1986. Draft Report of the Ad Hoc Working Group on the Work of its First Session, Geneva, 1–5 December 1986. UNEP/WG.151/1.4/draft_rep_of_the_ first_sess. ———. 1987a. Report of the Ad Hoc Working Group on the Work of its Second Session, Vienna, 23–27 February 1987. UNEP/WG.167/2/Report_of_wg_on_its_ 2nd_session.

BEYOND REGIME THEORY

119

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

This page intentionally left blank.

CHAPTER 6

Agent-Based Models in the Study of Ethnic Norms and Violence Ravi Bhavnani

During the nightmarish April, May, and June of 1994 following the assassination of President Juvénal Habyarimana, some five hundred thousand to eight hundred thousand Tutsi were killed by civilian militias and the Hutu-dominated army in a matter of weeks. Moderate Hutu who failed to participate in the violence were, time and time again, instructed to kill their Tutsi neighbors or face death at the hands of Hutu militias. When confronted with the choice of killing ethnic others or being killed by members of their own ethnic group, individuals with no prior disposition to engage in ethnic violence were turned into efficient killing machines. Few refused to participate in the killing; the result that “neighbors hacked neighbors to death in their homes, and colleagues hacked colleagues to death in their workplaces . . . doctors killed patients, and school teachers killed their pupils” (Gourevitch 1998, 115).1 The end result was that among the Hutu, killing Tutsi became the norm, and similar behavioral norms have motivated mass participation in, or complicity with, group violence in settings as diverse as Cambodia, Guatemala, Northern Ireland, and the former Yugoslavia. While Rwanda’s culture has been described as one of fear and conformity, 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; Mamdani 2001)—or the vehemence with which Tutsi were massacred, leading events in Rwanda to be described as the “fastest killing spree of the 20th Century” (Power 2001, 84). As Gourevitch (1998, 96) notes, “Every Rwandan I spoke with seemed to have a favorite, unanswerable question. For Nkongoli, it was how so many Tutsi had allowed themselves to be

121

122

RAVI BHAVNANI

killed. For Francois Xavier Nkurunziza, a Kigali lawyer, whose father was Hutu and whose mother and wife were Tutsi, the question was how so many Hutu has allowed themselves to kill.” The scale of violence was simply unprecedented, and stands in marked contrast to violence in 1963 in which roughly thirteen thousand lives were lost. Thus, “To believe that ordinary Rwandans killed, in their hundreds and thousands, and perhaps more, because of a congenital transhistorical condition—‘a culture of fear’ or of ‘deep conformity’—would require stretching one’s sense of credibility” (Mamdani 2001, 200). In Rwanda, Hutu resistance to the killing was evident at both the individual and community level (Des Forges 1999). Rather than being driven by fear of Tutsi, it was fear of fellow Hutu that drove the reluctant to participate in the genocide. Likewise, there is little doubt that structural factors that pertain to the economy, state capacity or penetration, or international aid flows—to name but a few— have important implications for the nature and onset of violence. 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. For one thing, prior levels of violence are inadequate predictors of future levels of violence. Cities, regions, states, and countries are not inherently peaceful or prone to interethnic violence. Rather, 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 1963.2 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, but need to clarify why violence can erupt in the absence of such a catalyst, or why similar catalysts can lead to different outcomes. 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, how they were persuaded to participate, and the effect of widespread participation on the scale and duration of violence. In contrast to explanations that point to a culture of conformity or highlight the importance of structural factors, 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. 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, prevail in some ethnic groups but not in others, and why these norms can either promote interethnic violence or cooperation.

AGENT-BASED MODELS IN THE STUDY OF ETHNIC NORMS

123

In the next sections of this chapter, I briefly describe the heterogeneity of Hutu beliefs and behavioral dispositions, the manner in which intragroup interaction increased behavioral conformity, and the subsequent emergence of a violence-promoting norm during the genocide in 1994. I then provide a brief introduction to ABMs, discuss my preference for an exploratory as opposed to a consolidative modeling approach (Bankes 1994; Casti 1997), and describe how ABMs may be used to study processes of norm formation and change within ethnic groups. I conclude the chapter with a comparison of agent-based, game-theoretic, and equation-based approaches.

RWANDA 1994: HETEROGENEITY, INTERACTION, AND ADAPTATION Rwanda is a densely populated, heterogeneous society, characterized by relatively frequent interactions (and intermarriage) between members of the two major ethnic groups. As a result, 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.3 Given the variation in levels of Hutu extremism, complicity with the state’s genocidal agenda was initially low (Des Forges 1999). 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. The end result was that large numbers of Hutu killed Tutsi who were acquaintances, colleagues, or, in extreme cases, relatives although individuals adapted to behavioral expectations in different ways (Des Forges 1999). For instance, 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, only to go and join the killing in another. . . . Could they have killed under duress— knowing that if they refused or even appeared reluctant, they would surely be killed—and saved a life when the opportunity presented itself ? (Mamdani 2001, 221) In other instances, behavioral adaptation was more complete: Everyone was called to hunt the enemy. . . . But let’s say someone is reluctant. Say that guy comes with a stick. They tell him, “No, get a masu.” So, OK, he does, and he runs along with the rest, but he doesn’t kill. They say, “Hey, he might denounce us later. He must kill. Everyone must help to kill

124

RAVI BHAVNANI

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

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, to be capable of explaining phenomena at the aggregate or group level, and to incorporate feedback mechanisms from the aggregate back to the individual level. ETHNIC NORMS As baffling as the scale of violence in Rwanda is the number of those purported to have participated in the killing. How, then, 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). In the context of interethnic rivalry, a norm thus defined either compels or dissuades members of a group from engaging in violence against ethnic rivals. By implication, the strength of a norm dictates how typical this behavior is of the group as a whole. Individual members of an ethnic group are motivated to participate in or oppose violence against nominal ethnic rivals for a variety of reasons. Individuals may follow their own convictions or beliefs, seek revenge for prior acts of violence, follow or go with the crowd, or simply derive entertainment value from participation. Where individual motives are absent or insignificant, 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.4 These schemas may emphasize obedience (an inherent rationale that orders are to be followed, with an allowance, albeit small, for disobeying orders that are contrary to any logic); detachment (cutting off one’s feelings for an “enemy”); paranoia (fear of being killed by an enemy if the enemy is not killed first); revenge (“tit-for-tat” behavior); patronage or work (a duty with a quid pro quo of material rewards); honor (saving face); or religious duty (such as a jihad or the building of a Hindu Rashtra). Each of these schemas may transform or undermine traditional constraints on violence or (in the case of cooperation) introduce constraints where none previously existed.5 Yet while group consciousness is necessary to ensure collective action, it often proves insufficient when the individual costs of participation are high— a point that is underscored by Mamdani, who notes that “fear could silence opposition, but it could not generate enthusiasm (for killing) . . . as they grew in scope, the massacres targeted anyone, peasant or professional, who refused to join in the

126

RAVI BHAVNANI

melee” (2001, 219). Norms, on the other hand, clearly prescribe appropriate or expected behavior, and leave less room for individual choice. Given that the existence of an ethnic norm is a matter of degree—norms may be strong, weak, or simply absent—the severity of punishment directly affects the strength of the norm. In addition to punishments, 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, interact with, and enforce each other’s behavior). It follows that while punishment, group composition, and social structure all influence norm formation and change, it is difficult to attribute the emergence of a norm to any specific factor, since norms are an emergent property of social systems and their existence depends upon complex patterns of interaction, influence, and internalization among individuals.6 To capture the emergent properties of norm creation and change, I turn to complexity theory and its simulation by ABM. AGENT-BASED MODELS Generally, ABM are comprised of one or more types of agents, as well as a nonagent environment in which the agents are embedded. The profile or “state” of an agent can include various characteristics and preferences, as well as particular social connections (i.e., identities, memberships, networks) and a memory of recent interactions and events.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. Agents may also possess adaptive mechanisms (learning or evolutionary) that lead them to change their heuristics based on their own experience. Each agent’s behavior affects other agents as well as the nonagent environment, resulting in behavioral change at the group or system level. The nonagent environment can encompass any variables external to the agents that are relevant to behavior, ranging from physical features such as geography or topography to things comprising states of the world like political, economic and social conditions. An environment, therefore, is specified in terms of various entities or dimensions, each with an associated “state.” The environmental entities in a model usually have their own dynamics, describing how they change over time independent of agent behavior. These variations can reflect natural progressions (or regressions) according to logical rules and also involve uncertainty or noise. In addition, they could represent the effects of shocks or “triggers” such as sudden economic collapse, the mobilization of ethnic rivals, or

AGENT-BASED MODELS IN THE STUDY OF ETHNIC NORMS

127

a military invasion. Besides following its own dynamic rules, the environment may also adapt in response to agent behavior. The existence of interlaced feedback relationships—agent-agent and agent-environment interaction—leads to the nonlinear, 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. One then runs the program to simulate the behavior of the agents and the dynamics of the environment. When an ABM is simulated on a computer, agent behavior is generated as agents determine which other agents to interact with, 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, as well as the emergent macrolevel structures, relationships, and dynamics that result from the aggregation of this microlevel activity. In principle, 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. 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. For instance, an exploratory model can serve as an experimental device to examine how members of an ethnic group might behave under a variety of assumptions, while stopping short of offering precise and detailed forecasts of how they will act given a particular set of circumstances.8 Consolidative modeling, on the other hand, usually involves the development of “model” systems that represent “real-world” systems with easily measurable physical characteristics and components. These models often require exhaustive inputs, which are then processed with computer programs that can run to millions of lines of code. Ideally, this large amount of data can be transformed into a useful or manageable form, but often the outputs are quite detailed as well. Although most often applied in settings where each component has clear physical properties, examples of a consolidative approach include complex multiagent models that use agent architectures based on “naturalistic decision-making.” The consolidative approach, however, is largely inappropriate for studying violent conflict, which is characterized by significant information uncertainties and practical barriers to experimental validation. 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. The specification of agent attributes, decision rules, adaptive mechanisms, and the agent-interaction topology all require decisions on the

128

RAVI BHAVNANI

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

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

130

RAVI BHAVNANI

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

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. This flexibility is perhaps the fundamental reason why ABMs are capable of replicating phenomena commonly exhibited by complex social systems. ABM vs. Game-theoretic Models Likewise, 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). 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, yielding certain nonlinear results. Given the nature of the social problem, however, ABMs offer some distinct advantages. First, GTMs typically assume that systems go to equilibria as limiting states, and they do not focus on processes that unfold over time. 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. Scholars have attempted to capture the richness of interaction of real-world problems by increasing the number of players, permitting nonsimultaneous play, introducing the option to “exit” and adding noise, altering the payoff matrix, and working on finite repetition (Axelrod and Dion 1988). While these refinements have led to a number of interesting insights, they still remain equilibrium-centered. Moreover, the solutions generally fall within the class of structure-induced equilibria, 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. Second, GTMs do not lend themselves well to studying agent and environmental heterogeneity. It is feasible to specify larger numbers of agents with varying characteristics, but only with multiplayer games that quickly become intractable. In fact, most applications of GTMs employ two or, at the most, three types of agents, partly out of the overriding concern for being able to derive equilibrium solutions. Meanwhile, environmental differences can only be obtained by specifying many separate models, each with their own particular structure. ABM, by contrast, can capture both forms of heterogeneity within the same model, enable greater scope for variation, and implement dynamic changes—so that these model components are also heterogeneous over time. Third, in GTMs the characteristics of players are typically determined exogenously, and player “types” typically tend to be fixed—for example, “weak” or “strong” states in deterrence games. It follows that it is difficult for the players in GTMs to change their defining characteristics over time, unlike the agents in

132

RAVI BHAVNANI

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.

AGENT-BASED MODELS IN THE STUDY OF ETHNIC NORMS

133

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

134

RAVI BHAVNANI

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.

AGENT-BASED MODELS IN THE STUDY OF ETHNIC NORMS

135

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.

136

RAVI BHAVNANI

Opp, K. 1982. “The Evolutionary Emergence of Norms.” British Journal of Social Psychology 21 (2): 139–49. Opp, K., and C. Gern. 1993. “Dissident Groups, Personal Networks, and Spontaneous Cooperation: The East German Revolution of 1989.” American Sociological Review 58, no. 5:659–80. Parunak, V., R. Savit, and R. Riolo. 1998. “Agent-Based Modeling vs. EquationBased Modeling: A Case Study and Users Guide.” Proceedings of Workshop on Multi-Agent systems and Agent-Based Simulation. MABS’98. Posner, R., and E. Rasmusen. 1999. “Creating and Enforcing Norms, with Special Reference to Sanctions.” International Review of Law and Economics 19, no. 3:369–82. Power, S. 2001. “Bystanders to Genocide: Why the United States Let the Rwandan Tragedy Happen.” Atlantic Monthly September:84–108. Prunier, G. 1995. The Rwandan Crisis: History of a Genocide. New York: Columbia University Press. Sherif, M. 1936. The Psychology of Social Norms. New York: Harper. Strauss, S. 2004. “The Order of Genocide: Race, Power, and War in Rwanda.” Doctoral diss., Department of Political Science, University of California, Berkeley. Tilly, C. 1978. From Mobilization to Revolution. Reading, MA: Addison-Wesley. Ullmann-Margalit, E. 1977. The Emergence of Norms. Oxford: Oxford University Press. Varshney, A. 2002. Ethnic Conflict and Civic Life: Hindus and Muslims in India. New Haven: Yale University Press. Veerhagen, H. 2001. “Simulation of the Learning of Norms.” Social Science Computer Review 19, no. 3:296–306. Watts, D., and S. Strogatz. 1998. “Collective Dynamics of Small World Networks.” Nature 393:440–42.

CHAPTER 7

Alternative Uses of Simulation Robert Axelrod

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

137

138

ROBERT AXELROD

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

ALTERNATIVE USES OF SIMULATION

139

those assumptions. The discovery of equilibrium results in game theory using rational choice axioms is a good example of deduction. Simulation is a third way of doing science. Like deduction, it starts with a set of explicit assumptions. But unlike deduction, it does not prove theorems. Instead, a simulation generates data that can be analyzed inductively. Unlike typical induction, however, 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, and deduction can be used to find consequences of assumptions, simulation modeling can be used as an aid to intuition. Simulation is a way of doing thought experiments. While the assumptions may be simple, the consequences may not be at all obvious. The large-scale effects of locally interacting agents are called “emergent properties” of the system. Emergent properties are often surprising, because it can be hard to anticipate the full consequences of even simple forms of interaction.2 There are some models, however, in which emergent properties can be formally deduced. 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. But when the agents use adaptive rather than optimizing strategies, deducing the consequences is often impossible; simulation becomes necessary. Throughout the social sciences today, the dominant form of modeling is based upon the rational choice paradigm. Game theory, in particular, is typically based upon the assumption of rational choice. In my view, 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, since its unrealistic assumptions undermine much of its value as a basis for advice. The real advantage of the rational choice assumption is that it often allows deduction. The main alternative to the assumption of rational choice is some form of adaptive behavior. The adaptation may be at the individual level through learning, or it may be at the population level through differential survival and reproduction of the more successful individuals. Either way, the consequences of adaptive processes are often very hard to deduce when there are many interacting agents following rules that have nonlinear effects. Thus, simulation is often the only viable way to study populations of agents who are adaptive rather than fully rational. While people may try to be rational, they can rarely meet the requirement of information or foresight that rational models impose (Simon 1955; March 1978). One of the main advantages of simulation is that it allows the analysis of adaptive as well as rational agents.

140

ROBERT AXELROD

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

ALTERNATIVE USES OF SIMULATION

141

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

This page intentionally left blank.

CHAPTER 8

Signifying Nothing? What Complex Systems Theory Can and Cannot Tell Us about Global Politics David C. Earnest James N. Rosenau Life’s but a walking shadow, a poor player That struts and frets his hour on the stage And then is heard no more. It is a tale Told by an idiot, full of sound and fury, Signifying nothing. —The Tragedy of Macbeth, act 5, 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, “life” for most of us connotes roseate meanings: dynamism, growth, learning, evolution, and adaptation. So perhaps it is no surprise that the complexity sciences—explicitly concerned with these properties of a variety of systems, from physical to social—not only invoke the metaphor of life but also have postulated the idea of “artificial” life (Langton et al. 1991; Langton 1994, 1995; Kauffman 1995; Langton and Shimohara 1997). It is unremarkable, furthermore, that social scientists observe in human societies the dynamism, adaptability, and unpredictability of organisms and ecological systems. Corning (2002), for one, argues that human societies are “superorganisms” and global politics are becoming a “super-superorganism.” 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.” Smith and Stevens (1997) reduce social organization to the attachment behavior regulated by the brain’s cyclic production of neurotransmitters known as “opioids.” More provocatively, Corning notes that the biologist Edward O. Wilson argues “the humanities and social sciences shrink to specialized branches of biology” (Wilson 1975, 547, quoted in 143

144

DAVID C. EARNEST AND JAMES N. ROSENAU

Corning 2000, 103). Indeed, there is a growing field of “biopolitics” (see Somit and Peterson 1997). Rather than a walking shadow in our analyses, then, life and its connotations of dynamism are central to an important line of contemporary thought about social systems. This includes the many IR theorists today who accept complexity and nonlinearity as a metaphor for the inordinate intricacy of global and international politics. The proliferation and influence of supra- and subnational actors, surprising cascading events like the Eastern European revolutions of 1989 or the current crisis of multilateralism, the transformative effects of global information technologies; 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, unpredictability, and change in the international system. Complexity appears at first glance to be precisely the paradigm we need to understand global politics today. Furthermore, 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. Yet, by and large, international relations scholars who use complex systems theories—not to mention complex systems theorists who study international politics—are few and far between. Clearly, 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. Why? Macbeth might claim complex systems theories are tales told by an idiot, though we are more optimistic about complexity’s prospects. 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, while complexity is a meaningful metaphor, complex adaptive systems—at least as conventionally formulated by theorists like Holland (1992, 1995, 1998) and, in political science, Jervis (1997), Axelrod (1997), and Axelrod and Cohen (1999)—differ in important ways from social and political systems. Although they may behave in complicated and confusing ways, social systems have structures of authority that may be inconsistent with the definition of complex adaptive systems. These differences are more than mere definitional or typological differences; we argue that in social systems, authority serves to minimize complexity. 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. We argue, furthermore, 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

SIGNIFYING NOTHING?

145

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

146

DAVID C. EARNEST AND JAMES N. ROSENAU

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

SIGNIFYING NOTHING?

147

mont 1994). Clearly, students of IR and global politics have embraced the paradigm of complexity. Yet the paradigm of complexity holds greater sway than the theory does. There are both methodological and, more importantly, epistemological reasons for this. To understand why, we first discuss how complexity theorists use computer models, with their attendant limitations, to model complexity in global politics. We then explore the deeper ontological assumptions complexity theorists must make about political actors and their motivations. The Third Way The overall behaviors of a social system, its dynamics and temporal trajectories— what complexity researchers call its “emergent” properties—generally are what most interest complexity theorists in the social sciences. Usually, social researchers are most interested in emergent properties such as cooperation, trade, warfare, disease transmission, and other social phenomena. By the definition of a complex adaptive system, these emergent properties result from the local interactions of numerous autonomous, independent agents pursuing local decision rules. This massively parallel structure of the complex adaptive model usually begs for the use of computing technologies that allow either parallel processing (that is, 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).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, dimes, and a ruled piece of paper to study residential segregation—the volume of calculations the researcher needs to undertake may be prohibitive. Complexity theorists thus often rely on computers to conduct their social simulations. Such computer-based models face a number of barriers to their acceptance among IR theorists. Though our discipline’s cultural aversion to new methods may be one of them, we choose to focus instead on the broader question of how a researcher may use these computer simulation methods within a broader research program. Two problems are immediately apparent. Computer simulations of complex adaptive systems are, first of all, neither a deductive nor an inductive method. 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. Although complexity simulations seek in principle to discover patterns, they do so through one of two different ways. First, researchers may observe an empirical phenomenon that they hypothesize is the emergent property

148

DAVID C. EARNEST AND JAMES N. ROSENAU

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

SIGNIFYING NOTHING?

149

test hypotheses about global politics. Complex systems theory, therefore, lacks both the empirical appeal of induction and the disconfirmative value of deduction. These observations about the complexity sciences are not novel, of course. Practitioners of complex systems theory have defended their work against these charges. Axelrod (1997) acknowledges that the simulation of complex adaptive systems is neither inductive nor deductive. Rather, he calls this method a “third way” of doing science, in which “simulated data . . . can be analyzed inductively” (Axelrod 1997, 4). 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, emphases in original). In this respect, both Axelrod and Epstein and Axtell argue complexity simulations combine elements of inductive and deductive methods as an aid to the researcher’s intuition. Despite these advantages, the methods of complex systems theory alone cannot prove or disprove hypotheses about global politics—although it is a theory of process, it cannot be a theory of politics. To make sense of the intricacies of contemporary global politics, then, researchers who use simulations of complex adaptive systems must supplement these efforts with empirical investigations. Of course, 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. Yet until a computer simulation can disprove a hypothesis, complex adaptive systems are little more than thought experiments on a computer—much ado about nothing.3 It is unclear, furthermore, that empirical tests of computer-simulated processes can in fact test our hypotheses about actual dynamic systems. Although Elliott and Kiel (1997) advocate such a complementary approach, 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. For one, while humans may follow simple decision rules, they also may not. Humans are both adaptive and habitual, capable of both learning and misapprehension, and paradoxically irrational yet calculating in their interactions with other human beings. We recognize that complex systems theorists argue they can simulate learning and strategic behavior in their computer-based worlds; for reasons we enumerate below, however, we remain yet to be convinced. A second, and perhaps more important, feature of societies is the role of authority. Unlike in physical complex adaptive systems, 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. A social system may have authority present and be complex; it may be complex and adaptive; but it cannot have both

150

DAVID C. EARNEST AND JAMES N. ROSENAU

authority and adaptation through complex interactions of autonomous agents, since centralized authority and individual autonomy, by definition, are mutually exclusive. As we argue below, the conceptual conundrum of authority in complex systems presents a fundamental challenge to the application of complex systems theory to questions of world politics. The Role of Expectations The three witches prophesied to Macbeth his murderous actions. Human agents typically rely, by contrast, on their own, less Delphic knowledge to inform their behavior. A substantial literature suggests that human beings are not perfectly rational; we rely upon heuristics and other simplifications in our everyday decisionmaking. 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). In short, our decisions depend, at least in part, on who we are. At first glance, 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. We suspect, however, that this simplicity itself may gloss over some important characteristics of expectations and interests in human actors. The appeal of the rationality assumption is its simplicity, 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. While complexity theorists can reasonably assume that biological or physical agents pursue these simple decision rules, it is unclear that this assumption holds for human agents, because humans’ expectations are contingent in part on their identities. If so, then a simulated complex adaptive social system will make assumptions about expectations and interests that are unlikely to capture this dependence. Again, it is useful to understand the methods complexity researchers use to simulate the adaptive and learning behavior of agents. Typically, complexity researchers capture the adaptive behavior of agents through genetic algorithms (see Holland 1995, 1998). These are software routines that each agent in the simulation follows to learn, evolve, or adapt to his or her environment. Agents themselves may adapt by changing their attributes or passing advantageous characteristics to future generations, the agents’ decision rules may adapt through learning, or both. Since students of global politics typically are interested in actors’ expectations and interests rather than in their biological characteristics, 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. Through iterative feedback from the simulated social environment, the agents adjust their decision rules through mutation, through

SIGNIFYING NOTHING?

151

selection, or through reproduction. Fitter agents come to dominate the system, 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. Through these processes, agents typically narrow the population of decision rules; from a randomly generated set of rules, for example, the researcher may find that the genetic algorithm produces a small population of “fit” strategies.4 For example, Axelrod (1997, chap. 1) has shown how tit-for-tat emerges as an optimal strategy in a complexity simulation of an iterated multiplayer prisoner’s dilemma. An important question is, then, whether or not genetic algorithms can emulate the processes of how human beings make decisions. There are two reasons to suspect that they cannot, both of which arise from how a political actor’s identity can influence his or her expectations. First, the literature on interest groups tells us the spatial organization of interests matter: a political actor’s identity as a passive or active participant, for example, depends in part on how the interests of others are spread throughout a system. Building on Olson’s seminal work (1965) on collective action problems, James Q. 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. When both costs and benefits are concentrated, interest group politics will result; when both are diffuse, majoritarian politics will result; diffuse costs and concentrated benefits yield client politics; and concentrated costs but diffuse benefits create entrepreneurial politics. In this respect, an agent’s interests and, in turn, expectations will be shaped by the spatial organization of interests in the system as a whole. While complex systems theory in principle embraces precisely this type of contingency in decision rules, the problem arises with the a priori assumption of the spatial organization of costs and benefits. A randomly generated population of interests, expectations, or decision rules merely assumes that interests are diffuse and that politics, therefore, are either majoritarian or entrepreneurial. Likewise, the modeler’s choice of fitness criterion makes an assumption about the concentration or diffusion of costs and benefits from the environment. If the modeler assigns specific interests or decision rules, however, he or she places the model in either the world of client politics or interest group politics. 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. The inferential problem is, therefore, determining to what degree the emergent behavior of the social system results from the dynamics of the system, and to what degree from the modeler’s assumptions. If multiple and divergent outcomes may result from the same set of assumptions, as complex system theory posits, it becomes exceedingly difficult to disentangle those results which emerge endogenously from those that are true by construction. The modeler’s

152

DAVID C. EARNEST AND JAMES N. ROSENAU

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

SIGNIFYING NOTHING?

153

to local governments, civil society, terrorist cells, and others. Authority therefore is problematic; most of us no longer assume the primacy of the nation-state. This feature of global politics poses a double challenge to the application of complex systems theory to international politics. First, we need a complex adaptive system that shows how authority shifts from one authoritative actor to another and how “layers” (or “spheres,” “nodes,” or “attractors”) of authority may result. Second, and formidably more important, we need a complex adaptive system that shows how authority emerges in the first place from the interactions among autonomous agents. As we have already noted, the pattern of authority in a complex adaptive system is one of its distinctive features: it has none. Authority is perfectly decentralized; each agent decides and acts on the basis of internal rules that evolve in response to environmental feedback. This is the logical antithesis of social authority, in which a privileged agent makes allocative decisions for a group of other actors. Unlike in a complex adaptive system, political authority often compels individuals to act contrary to their internal rules: the beggar, no matter how needy, will go to jail for stealing a loaf of bread. 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, classical thinkers clearly thought so. A complex adaptive social system, one that derives its dynamism and adaptability from its precarious balance on the edge of chaos, is nasty, brutish, and short according to Hobbes. The “natural” response to such systems, according to classical thinkers and organizational theorists alike, is centralized authority: the state, the firm, the hegemon, or the “leviathan.” Of course, the role of authority in a complex system may be one of degree: some complex systems are characterized by little if any authority, 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. But if we take the definition of a complex system at its word, in which decision-making is perfectly diffuse (or to extend our example, a market in which agents can break contracts or steal from others), how can a social system with its attendant structures of authority be a complex system? And if a complex system can have “some” hierarchy, or alternatively some balance of centralized and decentralized decision-making, 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, 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

154

DAVID C. EARNEST AND JAMES N. ROSENAU

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

SIGNIFYING NOTHING?

155

tion of how authority cascades through political systems; how it shifts from one authority “attractor” to another; and if, when, and where it may achieve a degree of dynamic equipoise. But these processes and questions require us to make some initial assumptions about authority in a complex adaptive system. For some questions of global politics—such as the study of market transactions, information dynamics, and voting behavior—researchers can make some reasonable assumptions. But for other questions these assumptions risk making any findings about authority highly problematic. Before we can model shifts in authority, a complex adaptive social system must have political authority in the first place. One possible way to do this is to endogenize authority at the genesis of a complex adaptive system. 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. Setting aside the question of whether or not this is truly a complex adaptive social system, this approach cannot tell us about the sources of authority. If the emergence of authority is, furthermore, path dependent (and there are good reasons to suspect it is), this exogenous approach may assume away important evolutionary dynamics. The alternative method—and in our minds more challenging but theoretically more fruitful—is to grow authority from the bottom up, as the emergent property of a complex adaptive system. A number of complexity theorists already have tackled this challenge. Axelrod’s (1997) tribute model shows that aggregate collectivities may emerge from the behavior of decentralized agents in a complex adaptive system. Likewise, Cederman (1997) seeks to endogenize the processes of the constitution of states in the international system. Epstein and Axtell (1996) show how markets may emerge and how actors will assume specialized roles as creditors or debtors. Kollman, Miller, and Page (1997) simulate “instability” or variations in the effectiveness of political institutions. While these approaches are an important first step, they may not truly simulate the emergence of the authoritative allocation of values for a population of agents. Axelrod’s tribute model purports to demonstrate how collective action arises through coercion, how power creates its own authority. Yet, the tribute model produces only quantitatively different actors; some states in the model develop more power than others. But the emergent actors are not qualitatively different: they do not make decisions for other agents in the system. Epstein and Axtell’s sugarscape model is similarly devoid of agents that are qualitatively different. Though the economy of the sugarscape creates debtors and creditors, both debtors and creditors are functionally identical. Each follows local decision rules. One might argue that creditors on the sugarscape, or strong states in the tribute model, in fact are qualitatively different actors or authorities, since their power or wealth deprives weaker agents of viable choices. Though these weaker

156

DAVID C. EARNEST AND JAMES N. ROSENAU

agents technically are not compelled in their enactment of their individual decision rules, the argument might go, the power of stronger actors presents them with Hobson’s choices, and hence no choice at all. After all, 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. Yet, this analogy shows how inappropriately the tribute model or the sugarscape model treat the concept of an authoritative actor. Authority is not merely the force to make allocative decisions for a collectivity. It is also the force to define and shape the collectivity as well. After all, if we do not pay our taxes, we go to jail. Social authority therefore includes the power to remove any given actor from society itself; it therefore not only allocates values but affects the aggregation of interests. This is where complex adaptive approaches to authority and the emergence of new political actors have fallen short. Because decision-making is by definition perfectly diffuse in a complex adaptive system, a complex adaptive system can neither simulate social authority nor describe how social authority constitutes the very actors that are its subjects. This argument anticipates, furthermore, how complex systems theory’s approaches to authority fail to engage the constructivist critique of structuralism in international relations theory. As Wendt (1987, 1994) and others have argued, it is theoretically groundless for scholars to assume that any agents are ontologically primitive in global politics. Spruyt (1994) shows that even the nation-state is a contingent social construction. Ruggie (1986, 1993) similarly argues that differentiation among nation-states is a historically path-dependent constitutive process. Whether they be states, voters, or organizations, political agents themselves therefore are contingent and indeterminate. In politics, whether domestic, international, or global, the rules of identity also may be issuespecific—the political issue may determine who the “actors” are. In this respect, 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 we noted above, social authority itself can constitute political actors by adding them to society (through enfranchisement, liberation, or a host of other processes) or removing them (conquest, imprisonment, and so on). Societies are not merely open systems, with political agents entering and exiting. They are, rather, “protosystems” that include an infinite number of latent actors and dormant systems. These actors and systems emerge, adapt, erode, shift, and dissolve with extraordinary speed. In politics, therefore, there are no “agents” per se; rather, there are latent identities, attributes, or values that are context and spatiotemporally dependent, which other actors may invoke to mobilize or remove political actors. Lustick (2000) and Cederman (1997, particularly chap. 8) each have attempted to endogenize this type of latency in political agents. Both approach the

SIGNIFYING NOTHING?

157

question of agency by endowing the agents in their complex adaptive social systems with latent identity attributes. As their models run, contexts may emerge in which the actors’ latent identities are activated. These are promising approaches to the issue of the contingency of agency, but they are less developed on the question of the contingency of authority. It remains to be seen whether complexity theorists can endogenize fully these constitutive processes. In contrast to Lustick and Cederman, most complexity simulations of politics treat these questions axiomatically. By assigning rules of identity exogenously—that is, 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. Complexity theorists make not just procedural assumptions, but also an important explanatory assumption: the very existence of political actors.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, dynamic, nonlinear interactions of the agents, but from the researcher’s own assumptions about authority and agency at the model’s genesis. The simulative methods of complexity thus risk obfuscating important assumptions made by complexity researchers about actors and authority in global politics today. It is little wonder, therefore, that students of global politics remain skeptical about complex systems theory’s methods. THE SOUND AND THE FURY, SIGNIFYING NOTHING? “Can machines think?” the mathematician Alan Turing (1950) asked more than a half century ago, 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. One way to respond to our criticisms of complex systems theory is, we believe, to ask a similar question about “artificial authority”: can a machine command compliance? Toward this end, the works of Axelrod (1997), Cederman (1997), Lustick (2000), and others hold some promise in “growing” authority as the emergent property of a complex adaptive system. Soon, 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 a few guidelines may help develop a research program in artificial authority. Following the example of the Turing test, we propose an experimental design that incorporates human subjects into the virtual world of the simulated social system. If and

158

DAVID C. EARNEST AND JAMES N. ROSENAU

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

SIGNIFYING NOTHING?

159

tems theory can create artificial authority from the ground up, therefore, we fear its hour on our conceptual stage is drawing to a close.

NOTES 1. The best-known simulation software is, of course, the Santa Fe Institute’s Swarm software, available for a number of different operating systems (http://www. swarm.org). Other software packages have lower learning thresholds, particularly NetLogo (developed at Northwestern University, http://ccl.northwestern.edu/net logo/) and StarLogo (developed at MIT, http://education.mit.edu/starlogo/). RePast (University of Chicago, http://repast.sourceforge.net/) and Ascape (the Brookings Institution, http://www.brook.edu/dybdocroot/es/dynamics/ models/ascape/ReadMe.html) are two other simulation software packages. All are freely available with numerous sample simulations. 2. For example, Taber (2001) recalls a conference at which two modelers of fractals confessed they had “no earthly idea” how their model worked. See Taber 2001, 24. 3. According to methodological falsificationism, the utility of a method lies not only in its ability to disconfirm a hypothesis, but also in its failure to disconfirm hypotheses once it has demonstrated the ability to do so. This is the essence of logical positivism: while we cannot prove hypotheses, we can derive some confidence in our knowledge when we fail to disprove them. In this sense, complex systems theory faces two challenges. First, it must establish its ability to test and disconfirm hypotheses. For example, to establish its bona fides we may use it to disconfirm hypotheses that we have already falsified with other methods. Second, only after it has established its ability to disconfirm can complexity theory be used to probe and test new hypotheses. This second challenge is arguably more difficult to achieve. Otherwise, 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. 4. 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. 5. See Hoffman and Johnson 1997. Indeed, to many constructivists the assumption that a political system exists is not procedural, the assumption that actors have interests is not procedural, and the assumption that these interests are related to—or unrelated to—the actors’ position in space and time is not procedural. 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.

160

DAVID C. EARNEST AND JAMES N. ROSENAU

6. 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. Since complex systems theory posits that agents’ behaviors are correlated with each other, we cannot assume the independence of the behavior of silicon agents when human agents participate in the system. In other words, we cannot assume that silicon agents would necessarily have complied with authority in the absence of human agents, or alternatively that they would not have.

REFFERENCES Alberts, David S., and Thomas J. Czerwinski, eds. 1997. Complexity, Global Politics, and National Security. Washington, DC: National Defense University. Anderson, Peter J. 1996. Global Politics of Power, Justice and Death: An Introduction to International Relations. London: Routledge. Axelrod, Robert. 1997. The Complexity of Cooperation: Agent-Based Models of Competition and Collaboration. Princeton, NJ: Princeton University Press. Axelrod, Robert, and Michael D. Cohen. 1999. Harnessing Complexity: Organizational Implications of a Scientific Frontier. New York: Free Press. Beaumont, Roger. 1994. War, Chaos, and History. Westport, CT: Praeger. Cederman, Lars-Erik. 1997. Emergent Actors in World Politics: How States and Nations Develop and Dissolve. Princeton, NJ: Princeton University Press. Corning, Peter A. 2000. “The Sociobiology of Democracy: Is Authoritarianism in Our Genes?” Politics and Life Sciences 19 (March): 103–8. ———. 2002. “Synergy and the Evolution of ‘Superorganism,’ Past, Present, and Future.” Paper presented at the annual meeting of the Association for Politics and the Life Sciences. Montreal, August 11–14. Earnest, David C. 2001a. “The Complexity of Globalization: Meeting the Challenges of Structuration Theory.” Paper presented at the Forty-second Annual Convention of the International Studies Association. Chicago, February 20–24. ———. 2001b. “Will No One Rid Me of This Meddlesome State? Social Inequality and the New Social Contract.” Paper presented at the Forty-second Annual Convention of the International Studies Association. Chicago, February 20–24.

SIGNIFYING NOTHING?

161

Elliott, Euel, and L. Douglas Kiel. 1997. “Nonlinear Dynamics, Complexity, and Public Policy.” In Chaos, Complexity and Sociology: Myths, Models, and Theories, ed. Raymond A. Eve, Sara Horsfall, and Mary E. Lee, 64–78. Thousand Oaks, CA: Sage. Epstein, Joshua M., and Robert Axtell. 1996. Growing Artificial Societies: Social Science from the Ground Up. Washington, DC: Brookings Institution. Harrison, Neil E. 2001. “The Value of a Complexity Metaphor for International Political Economy,” Prepared for the 42nd Annual Convention of the International Studies Association in Chicago, February 20–24. Hoffman, Matthew J., and David Johnson. 1997. “Change and Process in a Complex World: Using Complexity Theory to Understand World Politics.” Paper presented at the International Studies AssociationMidwest meeting. Cleveland, October 3–5. Holland, John H. 1992. “Complex Adaptive Systems.” Daedalus 121, no. 1 (Winter): 17–30. ———. 1995. Hidden Order: How Adaptation Builds Complexity. Reading, MA: Addison-Wesley. ———. 1998. Emergence: From Chaos to Order. Cambridge, MA: Perseus. Hughes, Barry. 1997. Continuity and Change in World Politics: Competing Perspectives. Upper Saddle River, NJ: Prentice-Hall. ———. 1999. International Futures: Choices in the Face of Uncertainty. Boulder, CO: Westview. Jervis, Robert. 1997. System Effects: Complexity in Political and Social Life. Princeton, NJ: Princeton University Press. Johnson, Steven. 2001. Emergence: The Connected Lives of Ants, Brains, Cities and Software. New York: Scribner. Kauffman, Stuart. 1995. At Home in the Universe: The Search for the Laws of SelfOrganization and Complexity. New York: Oxford University Press. Kollman, Ken, John H. Miller, and Scott E. Page. 1997. “Political Institutions and Sorting in a Tiebout Model.” American Economic Review 87, no. 5 (December): 977–92. Kuran, Timur. 1991. “Now Out of Never: The Element of Surprise in the East European Revolutions of 1989.” World Politics 44 (October): 7–48. Langton, Christopher G., ed. 1994. Artificial Life III: Proceedings of the Workshop on Artificial Life, held June 1992 in Santa Fe, New Mexico. Reading, MA: Addison-Wesley.

162

DAVID C. EARNEST AND JAMES N. ROSENAU

———. 1995. Artificial Life: An Overview. Cambridge, MA: MIT Press. Langton, Christopher G., and Katsunori Shimohara, eds. 1997. Artificial Life V: Proceedings of the Fifth International Workshop on the Synthesis and Simulation of Living Systems. Cambridge, MA: MIT Press. Langton, Christopher G., et al., eds. 1991. Artificial Life II: Proceedings of the Workshop on Artificial Life Held February, 1990 in Santa Fe, New Mexico. Redwood City, CA: Addison-Wesley. Levy, Jack S. 1997. “Prospect Theory, Rational Choice, and International Relations.” International Studies Quarterly 41, no. 1 (March): 87–112. Lodge, Milton, Charles Taber, and Aron Chase Galonsky. 1999. “The Political Consequences of Motivated Reasoning: Partisan Bias in Information Processing.” Paper presented at the Annual Meeting of the American Political Science Association. Atlanta, September 2–5. Lohmann, Susanne. 1994. “The Dynamics of Information Cascades: The Monday Demonstrations in Leipzig, East Germany, 1989–1991.” World Politics 47, no. 1 (October): 42–101. Lustick, Ian. 2000. “Agent-Based Modeling of Collective Identity: Testing Constructivist Theory.” Journal of Artificial Societies and Social Simulation 3, no. 1 (January). Available at http://www.soc.surrey.ac.uk/JASSS/3/1/ 1.html. Marion, Russ. 1999. The Edge of Organization: Chaos and Complexity Theories of Formal Social Systems. Thousand Oaks, CA: Sage. Meadows, Dennis L., et al. 1974. Dynamics of Growth in a Finite World. Cambridge, MA: Wright-Allen Press. Miller, John H. 1998. “Active Nonlinear Tests (ANTs) of Complex Simulation Models.” Management Science 44, no. 6 (June): 820–30. Olson, Mancur. 1965. The Logic of Collective Action: Public Goods and the Theories of Groups. Cambridge, MA: Harvard University Press. Opp, Karl-Dieter, and Christiane Gern. 1993. “Dissident Groups, Personal Networks and Spontaneous Cooperation: The East German Revolution of 1989.” American Sociological Review 58, no. 5 (October): 659–80. Rosenau, James N. 1990. Turbulence in World Politics: A Theory of Change and Continuity. Princeton, NJ: Princeton University Press. ———. 1997. Along the Domestic-Foreign Frontier: Exploring Governance in a Turbulent World. Cambridge: Cambridge University Press. ———. 2003. Distant Proximities. Princeton, NJ: Princeton University Press.

SIGNIFYING NOTHING?

163

Ruggie, John Gerard. 1986. “Continuity and Transformation in the World Polity: Toward a Neorealist Synthesis.” In Neorealism and Its Critics, ed. Robert O. Keohane, 131–57. New York: Columbia University Press. ———. 1993. “Territoriality and Beyond: Problematizing Modernity in International Relations.” International Organization 47, no. 1 (Winter): 139–74. Saperstein, Alvin M. 1996. “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.” In Chaos Theory in the Social Sciences: Foundations and Applications, ed. L. Douglas Kiel and Euel Elliott, 139–63. Ann Arbor: University of Michigan Press. Schelling, Thomas C. 1978. Micromotives and Macrobehavior. New York: W. W. Norton. Smith, Thomas S. 1997. “Nonlinear Dynamics and the Micro-Macro Bridge.” In Chaos, Complexity and Sociology: Myths, Models, and Theories, ed. Raymond A. Eve, Sara Horsfall, and Mary E. Lee, 52–63. Thousand Oaks, CA: Sage. Smith, Thomas S., and Gregory T. Stevens. 1997. “Biological Foundations of Social Interaction: Computational Explorations of Nonlinear Dynamics in Arousal-Modulation.” In Chaos, Complexity and Sociology: Myths, Models, and Theories, ed. Raymond A. Eve, Sara Horsfall, and Mary E. Lee, 197–214. Thousand Oaks, CA: Sage. Somit, Albert, and Steven A. Peterson. 1997. Darwinism, Dominance, and Democracy: The Biological Bases of Authoritarianism. Westport, CT: Praeger. Spruyt, Hendrik. 1994. The Sovereign State and Its Competitors: An Analysis of System Change. Princeton, NJ: Princeton University Press. Taber, Charles S. 2001. “Of Spells, Potions, and Computational Social Science.” Political Methodologist 10, no. 1 (Fall): 23–26. Turing, Alan M. 1950. “Computing Machinery and Intelligence.” Mind 59, no. 236 (October): 433–60. Urry, John. 2003. Global Complexity. Cambridge: Polity. Wendt, Alexander E. 1987. “The Agent-Structure Problem in International Relations Theory.” International Organization 41, no. 3 (Summer): 335–70. ———. 1994. “Collective Identity Formation and the International State.” American Political Science Review 88, no. 2 (June): 384–96. Wilson, Edward O. 1975. Sociobiology: The New Synthesis. Cambridge, MA: Harvard University Press. Wilson, James Q. 1980. The Politics of Regulation. New York: Basic Books.

This page intentionally left blank.

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, more precisely stated, agency-level computational models—will reshape the practice and use of social science/inquiry. When contrasted with methodological approaches that solely focus on the aggregate levels of analysis, 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. Moreover, as alluded to in the title, ABMs allow for a metaphorical collision of worlds, or rather an ability to combine, compare, and decompose theoretical views of the world not easily accomplished in the past. But before exploring these inquiry possibilities, it is prudent to bound these possibilities by considering the theoretical challenge to the credible use of computational models of agency. The preceding chapter by Earnest and Rosenau questions whether there can be a complex systems theory of political systems. They raise two objections. The first is that complex systems theory “lacks both the empirical appeal of induction and the disconfirmative value of deduction.” Therefore, they argue, it is not really a theory. Their second objection is that complex systems theory cannot capture and model the essence of political systems: the central role and importance of authority. For issues in which authority is less pervasive, like those in some of the earlier chapters in this volume, the complexity paradigm offers new thinking about old problems and suggests new hypotheses and explanatory

165

166

DESMOND SAUNDERS-NEWTON

methods. But Earnest and Rosenau argue that in those issue-areas in which authority is more influential, complex systems theory cannot describe the importance of authority and ABMs cannot simulate its effects. In this chapter, I do not directly challenge either of these criticisms of the usefulness of complex systems theory in explaining or predicting political systems. Earnest and Rosenau’s objections have merit in that they raise concerns about our ability to rigorously create knowledge about our world, as well as the appropriate substantive focus of the disciplines that constitute the field of study known as international relations or world politics. On the other hand, their concerns about our ability to reason from models, either via deduction, retroduction,1 abduction,2 or induction, can also be raised when we consider the “status quo” methodological approaches used by scholars in this field. In my opinion, exchanges related to methodological adequacy will likely fall into the category of reasonable scholars may differ. Instead, I argue that the epistemological problems associated with ABMs are worthy of note but that they are overstated; I show that ABMs can improve policymaking even if both of these criticisms are true. Several US government agencies, inclusive of the Defense Advanced Research Projects Agency (DARPA) and the Center for Technology and National Security Policy, 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. Even before theorists have worked out the ontological and epistemological problems associated with using complexity to explain political systems, 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. Such emergent applications not only reflect a desire to bring social science knowledge to bear on “problems of the day,” but also are consistent with the innovation generated by the tools of this intellectual approach. As noted by Frederick Turner, the new science (chaos and complexity) has “placed within our grasp a set of very powerful tools—concepts to think with. We can use them well or badly, but they are free of many of the limitations of our traditional [methodological] armory” (Turner 1997, xii.) With this in mind, this chapter explores a scheme for improving our ability to use these new perspectives and methods. 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.

EFFECTIVE COMPUTATIONAL INQUIRY The introductory chapter defined agent-based modeling as the simulation of complexity in the social sciences. Such a methodology offers the prospect

WHEN WORLDS COLLIDE

167

of representing the interactions of agents comprising a heterogeneous system of autonomous actors. This capability affords an opportunity to explore theories of complex international and global social phenomenon. I would like to deepen this definition by asserting that agent-based modeling—or, more broadly considered, 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, but also to pursue such inquiry in a more interdisciplinary fashion. Agency-based modeling reflects a methodological approach that allows for considering phenomena at levels of aggregation, or resolution, more granular than that of the Westphalian nation-state. Several methodologies afford the analyst a means of studying the forms and dynamics of a social system, and they vary by distinctive methodological styles and model ontologies (disciplines). Regardless of pedigree, what these models have in common are an ability to explicitly consider, and relate, agents, interactions, and environment. At least three distinct approaches potentially explain ecologies of agents and their interactions: multiagent simulation, computational social network analysis, and sociophysics. A potential typography for envisioning agency-based methodology is shown in figure 9.1.

Figure 9.1. A Typology of Agency-Based Methods

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. As a result, computational ontologies, or rather the “what is” that is instantiated as algorithm, will vary across methodological approaches, and this gives rise to epistemological implications: the ability to “know” will vary. Even though “what you can know” may well differ between agency-based methodological approaches, the ability to consider intuitively resonant levels of resolution is a true benefit of the approaches. Put simply, this modeling approach makes sense to important sets of constituents beyond the analytic community—for example, modeling neophytes with decision-making authority and analysts trained in traditional modeling traditions and shaped by hard-earned expertise. Of course, no methodological approach comes cheap or free of costs. All tools used by individuals involved in inquiry come with constraints such as theoretical assumptions or computational artifacts or a lack of ontological isomorphism.3 Thus, a prime concern as we consider the use of such a methodology is its credibility (Dewar et al. 1996). 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. Put another way, does the model result in a convincing or resonant narrative of analysis or inquiry? To that end, I address the general issue of computational epistemology and methodology as it relates to the use of computational models of agency. Further, I consider how such concerns are more specifically related to the examination of complex international and global social phenomenon. COMPUTATIONAL EPISTEMOLOGY AND METHODOLOGY As a modeling formalism, 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 Bankes, Lempert, and Popper suggest, 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). The following paraphrases suggest how Bankes, Lempert, and Popper (2002, 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, individuals, groups, or institutions.

WHEN WORLDS COLLIDE

•

The decision algorithms that these agents use can be based on knowledge and data available regarding the decision behavior, perceived or actual, of the associated individuals, groups, or institutions.

•

What we know about the relationships between such agents can easily fit into the agent-based mechanism.

•

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.

169

In summary, and in comparison with competing modeling frameworks, such as numerical models based on systems of differential equations or symbolic logic or linear models of behavior, the agent-based approach provides much greater facility for capturing the information that is available. Moreover, 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, Lempert, and Popper 2002, 377). While this modeling approach to representing human complex systems can be very effective, representation and verisimilitude are only a part of our epistemological concerns. More precisely, 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; 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. Whatever the shortcomings of non-ABM methods, it is still necessary to consider the primary principle of epistemology: how do we credibly learn “things” about the world by using ABMs? Furthermore, given the nature of ABM as a computational method, how do we deepen our knowledge by performing computations? While philosophical in their nature, these questions point to the necessity of identifying and developing criteria for performing research based on computational modeling. To that end, 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, we are drawn to consider the nature or use of models of social phenomena. I suggest that there are at least two important uses of models: prediction and exploration. With respect to predictive models, we immediately are faced with challenges that have harried users of analytic methods for studying social processes

170

DESMOND SAUNDERS-NEWTON

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

WHEN WORLDS COLLIDE

171

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

172

DESMOND SAUNDERS-NEWTON

Based upon Ed Waltz, Knowledge Management inthe Intelligence Enterprise (Boston Artech house, 2003), p. 177, as modified by Desmond Saunders–Newton.

Figure 9.2. 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), but each provides useful insights into understanding social behaviors in isolation, collaboratively and integratively (Saunders-Newton and Graddy 2001). With respect to specification, it is easy to envision how changes in model specification, which can be viewed as a separate model, can give rise to very different results. As for scenario differentiation, this speaks to coupling a given model with certain expectations about the world in which it operates. For the policy community, the scenario describes how a represented system may behave as a result of changes in policy or the social environment. Regardless of the basis upon which these alternative models are generated, the ensemble of models will likely contain more information than one single model; by conducting large numbers of modeling experiments, it is possible to derive insights through the exploration of the properties of an ensemble of

WHEN WORLDS COLLIDE

173

alternative models that no single model computer model could reveal. This approach lies at the heart of the exploratory modeling approach (Bankes 1993, 1996; Dewar et al. 1996; Saunders-Newton and Scott 2001). Further, by inferring across this ensemble as opposed to seeking insights from one model, we can induce invariant properties and abduct robust responses to these model outcomes. It is worth further emphasizing that these benefits are made possible by the availability and use of computational experiments. Moreover, the use of induction and abduction in a collaborative fashion reflects an intriguing interaction between human and machine “intelligences.” As illustrated by figure 9.2, the reasoning process reflects an interesting variation on the efforts of Charles Peirce’s consideration of reasoning.In this instance, the graphic illustrates the various interrelations between reasoning approaches. The integration of these approaches is very dependent on the ability to exploit computational approaches—for example, experimentation to expansively consider new hypotheses, models and responses. Further, one can use figure 9.2 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, abduction, and induction. As well as this ability to more effectively use alternative reasoning or inferencing approaches, the relationship between human “wetware” and algorithmic “software” can be structured to leverage human ability to adeptly make inferences from complex patterns. Thus, we move further along in the process of making use of algorithmic approaches to aid in decisionmaking and inquiry. 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. In anticipation of an algorithmic social science with ABM as a prominent methodological approach, a number of technologist, scientists, 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, and how this approach is being integrated into the policy community. 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. With respect to motivation, it is not difficult to assert that the issues of interest in the realm of international security can be categorized as not only complicated, but also often complex. While for many these terms are often synonymous,

174

DESMOND SAUNDERS-NEWTON

Mark Lazaroff (BAE Systems Advanced Information Technologies, Intelligence Innovation Division) and David Snowden (IBM).

Figure 9.3. Complicated and Complex Problems

they have very distinct meanings for research methodologists. As illustrated in figure 9.3, a complicated problem can often be viewed as one with identifiable casual relationships and known uncertainties. However, a complex problem is one much more consistent with emergent patterns. It is not difficult to further assert that the current international regime is easy to represent as complex, and often viewed by practitioners and laypersons as near-chaotic. Unfortunately, if the problem is defined as chaotic, or near-chaotic, its supposed incomprehensible nature will often result in the unwillingness to use rigorous, structured approaches to consider the problem. If such approaches are used to consider a “wicked” problem, the necessary simplifications often make the model and its associated results of questionable use to decision-makers. The objective then is to frame near-chaotic problems so that they are classified as complex, and to use a methodology that is appropriate to the problem domain. Complexity science moves the problem from the chaotic to the complex domain, while ABMs provide a means to assess phenomena in such regimes.4

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, 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. Bridging these two “factions” is important as a means of exploiting the insights derived about individuals and their artifacts, such as institutions and technologies, in order to deal with ever-present challenges such as violent conflict and complex humanitarian disasters. Computational models, as defined by Taber and Timbone (1996, 3), 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. More to the point, models serve as a language for expressing theory. Thus, a computational (algorithmic) social science is the algorithmic instantiation of social science theory. As shown in figure 9.4, the language of algorithmic social science (computational symbolic processing) falls between natural language and mathematical formalism. As a general statement, it is able to provide insights into the deep narrative and rich detail associated with natural language, and maintain the rigor associated with mathematical dialects. Another way of envisioning such a language is as analytic narrative. 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. While the efforts are fairly numerous, a few

Figure 9.4. The Language of Algorithmic Social Science

176

DESMOND SAUNDERS-NEWTON

Figure 9.5. System-Level Architecture of the Pre-Conflict Management Tools Program

directly resonate with a substantial amount of the literature in international and global studies. These include efforts in the realm of preconflict and postconflict operations—that is, actions necessary to avoid violent conflict or more effectively deal with the consequences of using military force. 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. 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, and identifying actions that could preclude or reduce the likelihood of such an event. 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, and also enhance the likelihood of postconflict success. The general conceptual format for this effort is shown in figure 9.5. 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. While the database—which is concerned with the near-

WHEN WORLDS COLLIDE

177

real-time collection, categorization, and visualization of local- and regional-level data, as well as traditional nation-state-level data—is of interest, 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. As originally envisioned, 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, ABMs of local-, regional-, and national-level phenomena and system-level models. In light of budgetary constraints, 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. To that end, the initial PCMT effort made use of three models of social vulnerability and authoritative network dynamics. In terms of modeling classes, the models include an econometric model of development and civil war, a qualitative computational model of internal conflict and state failure, and a model of authoritative network dynamics. The econometric model is based on the work of the World Bank’s Paul Collier, who focused on the relationship between civil war and development policy (Collier et al. 2003), and the qualitative model is a computational instantiation of the Fund for Peace’s Conflict Assessment System Tool (CAST). The network dynamics models of authority were developed by the Institute for Physical Sciences. These models comprise a collection of disparate modeling approaches, including a single ABM as of the summer of 2005, and are integrated by the use of computer-assisted reasoning methods.6 It is important to note that ABMs are of importance in thinking about the issues of societal formation and fragility, as well as the networks of authority capable of redressing or addressing conditions that give rise to social fragility. In addition, ABMs have proven quite helpful in the collaboration effort, 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, such as coalition partners, nongovernmental organizations, and multinational corporations. To craft information and knowledge so to increase its use in these decision-making environments, it is necessary to consider an appropriate means of translating model output into decision-enabling information. The PCMT researchers employ computer-assisted reasoning methods in conjunction with simulation models such as ABMs to create large ensembles of plausible future scenarios. This particular stratagem supports a robust adaptive planning (RAP) approach to reasoning under conditions of complexity and “deep uncertainty”

178

DESMOND SAUNDERS-NEWTON

that typically defeat analytic approaches (Lempert, Popper, and Bankes 2002).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. ABMs are an important component of this effort, particularly for the aspect of the PCMT architecture that is concerned with identifying “proponents and opponents to peace,” and are viewed as credible ways of “knowing” in their own right. It is fully expected that once the PCMT program moves from the technology demonstration stage into operational use, a majority of all future models used in the social-vulnerability aspect of the program will be ABMs.8

CLOSING THOUGHTS In closing, 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. 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 their related models, because they cannot have the predictive accuracy of models of fluid dynamics or classical dynamics. However, it is evident that proposed solutions generated by these highly accurate models are not particularly effective in addressing the problems of greatest concern. For example, more accurate weapons do not change the conditions that give rise to terrorist or transnational criminal networks. Interestingly enough, 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. The greater challenge may be less about ABMs than about their use. 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. However, an effort such as the PCMT program suggests that such a shift may be less than a generation away. In fact, a number of efforts are currently exploring the use of ABMs to better understand how actions propagate forward into time as consequences and externalities. Such a capability underpins operational concepts such as “effects-based operations.” 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, and Bankes 2003).

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.” Implicit in this assertion is our general inability to make generalizable statements based on a sample size of one. Conversely, if we had a larger sample of planets to observe, we could likely infer generalizable insights that would aid us in thinking about actions on our own world. 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. However, societies or cultures created in-silica are now becoming increasingly possible. 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, or rather the possible consequences of actions as they propagate into time. 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. By instantiating, growing, and maintaining large ensembles of these societies and cultures, it becomes possible to assess the impact of strategies, policies, and actions across extremely rich/deep computational models of different groups. Such a technical approach would be quite consistent with the earlier discussion of experimental design and computational epistemology, and would likely be impossible to realize without the use of agency-based models of culture. The challenges for developing and using such modeling artifacts are daunting, because they require thinking deeply about mappings between real-world and virtual-world ontologies, as well as means of considering the epistemological rigor and methodological sufficiency. These challenges, however, 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 considerations are the price we must pay in order to understand the answers we receive to the questions we ask. NOTES 1. While the concept of retroduction as originally defined by Charles Sanders Peirce has multiple—seemingly contradictory—meanings, 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. It should be noted that some suggest that Peirce’s ultimate

180

DESMOND SAUNDERS-NEWTON

intent was to relate deduction, abduction, and induction under the general conceptual frame of retroduction (Chiasson 2001). 2. Abduction, or inference to the best explanation, 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. 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). 3. 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. 4. In actuality, methods beyond ABMs will allow complexity perspectives to address problems in the chaotic domain—for instance, nonlinear dynamics. 5. “Analytic narrative” is a phrase meant to suggest a rigorous, yet compelling, means of sharing complex information or knowledge supportive of the decision-making process. 6. Future variants of the PCMT effort will move toward including additional ABMs into the modeling suite. 7. As defined by Bayesian decision theorists, deep uncertainty is the condition where the decision-maker does not know, or multiple decision-makers cannot agree on, the system model, the prior probabilities for the uncertain parameters of the system model, and/or the value function. 8. 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. 9. In July 2003, DMSO managed a workshop exploring the ability to model and simulate personality and culture. 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).

REFERENCES Bankes, S. 1993. “Exploratory Modeling for Policy Analysis.” Operations Research 41, no. 3 (May–June): 435–49.

WHEN WORLDS COLLIDE

181

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

182

DESMOND SAUNDERS-NEWTON

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

C H A P T E R 10

Complex Systems and the Practice of World Politics Neil E. Harrison

The study and practice of world politics has for too long been distorted by rational choice. This conveniently simple model has misled generations of scholars and policy-makers (Smith 2004). 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. . . . Rational choice preserves the status quo. . . . Thus, public policy as is becomes the public policy interest as it ought to be” (Petracca 1991, emphasis in the original). 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, thereby, further elaborated the complexity concepts and ideas that may be used to construct complex systems theories of issue-areas in world politics. Chapters 3 through 6 illustrated complexity and its benefits by applying complexity concepts and sketching complex systems theories for specific issue-areas. Chapters 7, 8, and 9 debated the epistemology and methods of complex systems. In this chapter, I first show how complex systems concepts can improve how we think about and understand world politics. In the next section, I consider Earnest and Rosenau’s epistemological critique of complex systems theory in chapter 8. In the third section, I show how complexity could reform policy in world politics. In a short coda, I summarize the many benefits of the complexity 183

184

NEIL E. HARRISON

paradigm and the theories it can spawn. Throughout this chapter I indicate several paths for further development and application of the complexity paradigm.

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. Complex systems concepts can improve current theory, generate novel insights, and allow exploration of new possibilities. Improving Current Theory Complex systems concepts can extend or elaborate current knowledge, a critical measure of a new taxonomy (see chapter 2). For example, Hoffmann shows how a complex systems theory can provide the microfoundations to international negotiations and agreements. Constructivism argues that interests and identity are constructed through interaction between states. Structure does not determine agent choices, because “agents and structures are produced or reproduced by what actors do” (Wendt 1994, 390). Thus, agents have a degree of freedom that introduces potential for dynamic system change. But as states are treated as units (Wendt 1994, 385), constructivism cannot explain how states exercise this freedom of choice. Thus, there is no explanation of the microfoundation of macroprocesses (and so of the sources of change in the international system). Conceptualizing the state as a complex system that is an agent in the international system (a meta-agent), Hoffmann fills this gap in constructivist theory. He treats international negotiations as the coevolution of adaptive states, 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. 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, 397). This language is close to complexity concepts of internal models. Constructivism sees identity as formed and changed through socialization in the international system, but complexity expands behavior to include both the emergent domestic processes and international coevolution, much as suggested, though more statically, in Putnam’s two-level games (1988). Bhavnani’s analysis of the Rwanda genocide also uses complexity concepts as an adjunct to conventional theories. He does not deny the usual explanations of the causes of the violence. The ethnic hatreds, government propaganda, and

COMPLEX SYSTEMS AND THE PRACTICE OF WORLD POLITICS

185

the death of the Hutu president are well recognized. He uses complexity concepts to answer a puzzle that conventional theories cannot touch: the initial conditions of the Rwanda conflict, however reasonable, cannot explain the rapidity and magnitude of the killing. However, the dynamic evolution of the killing can be captured by complex systems concepts that fill out the narrative of simpler models. As most conventional theory is at the level of the international system, there are many opportunities to add microfoundations with complexity concepts (which do not preclude greater simplifications at higher levels of analysis). Issuearea theories also can be adapted. For example, complex systems could explain how epistemic communities actually influence states’ policies in international environmental issues.1 Generating Novel Insights In a dynamic world, a constant stream of new ideas and hypotheses is essential to understanding. Paradigms change because explanatory failures of the old paradigm have accumulated, a new paradigm is available that explains more than the old, and the majority of scholars in the field recognize these two conditions (Kuhn 1970). 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). Complex systems concepts can generate radically novel hypotheses. 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. Reminiscent of Weber’s (1958) explanation of capitalism, this is an unexpected but very plausible hypothesis generated using complexity concepts. 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. Informal institutions developed from religious conviction generate politically effective behaviors that manage ethnic differences and support economic development. A complex systems theory of development led Clemens to ask different questions and search for data in new places. While correlation is not causation, Clemens’s argument is novel and certainly plausible, and deserves further investigation. Because it shows similarities with some theories of economic development (e.g., De Soto 2000), it may have application elsewhere. Clemens offers some empirical evidence in support of his hypothesis and outlines how to more fully test it. Whether or not the hypothesis is disproved, it illustrates complexity’s possibilities for innovative thinking about old and new problems.

186

NEIL E. HARRISON

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

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. Thus, a statement is meaningful only if it can be observed or can, in theory, be observed (Ayer 1936).4 The best empirical verification of a proposition is correct prediction of observable events. Positivism is problematic in the social sciences because of problems of observation and of correspondence between abstract concepts and real objects. The problem of identifying valid empirical indicators for theoretical concepts is broadly recognized but overlooked in deference to the simplicity of positivism. Arthur (1994) similarly rejects positivist/rationalist theory in economics as a beautiful model that bears little relationship to the reality of human behavior. Humans are less deductive and more inductive than positivism posits. Evolutionary epistemology is one explanation of knowledge accumulation that should be considered for complex systems in social science. Blind variations (almost guesses) in knowledge are selected through biological, psychological, and social processes and retained according to their contribution to individual or group survivability (Campbell 1960).5 In this view hypothesis generating simulations and exploratory modeling are as legitimate sources of knowledge variations as any other. Once selected, knowledge variations become part of the selection mechanism for further variations, in effect mimicking selection by reality. Organized in a nested hierarchy of selectors, the knowledge system becomes ever more intelligent and adaptive. 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. The whole is partly constrained by the behavior of the parts (emergence), but the parts are partially constrained by the whole (downward causation). Finally, an epistemology can be designed to specifically accommodate simulations generating “justified true belief.” For example, 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. The purpose of science, they argue, is to map theoretical constructs onto observable reality (much like positivism). But mapping in the social sciences is shaky at best, because (unlike the physical sciences) social science reality does not fit into binary logic categories. Physical objects are either members or not members of a category. An electron is nothing but an electron; it is never a molecule. 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. 5.16). Thus, 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).

188

NEIL E. HARRISON

Authority in Complex Systems Taking authority as the critical variable in world politics, as do Earnest and Rosenau, is a value-laden move. 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. Politics has also been defined as the formation and rivalry of groups (Schmitt 1976) and, more generally, as the generation of the structures and norms that govern human collectivities (for example, Arendt 1958). Following this latter view, politics is the process by which the institutions governing collective life are organized. Understanding politics this way, it is eminently reasonable to model it as a complex system. Some agents in the system always have more influence over the form and function of institutions than other agents do. But authority is not the whole measure of politics and, except for subscribers to the realist view, not the sole object of research in world politics. For three reasons, the existence of authority is not fatal to complex systems theories of world politics. First, authority operates through formal and informal institutions. Informal institutions, like cultural practices, are shared meanings and emerge from agent interactions mediated through prior states of such institutions. Second, social systems are not binary—for example, either authoritarian or not. It is always a matter of degree. I suggest that all societies are complex and can be modeled with complexity concepts, but some have more central control, and thus less complexity, than others. Third, for authority to be fatal to complexity it must be centralized. As Earnest and Rosenau acknowledge, globalization is diffusing authority from the state to other organizations (see Strange 1996). While individual agent decisionmaking may be as or more limited than before, it is because of less centralized authority. Freedom House reports that in 1900 no state was an electoral democracy with universal suffrage. By 2000, 120 of 192 countries were rated as electoral democracies. Coupled with economic liberalization, proliferating democracy diminishes centralized authority and increases self-organization and complexity. The decentralization of authority from the state to a large number of diverse private and public organizations competing economically and politically itself creates complexity, 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.6 Even within highly centralized authority systems, agents always have choices. They can choose to follow orders or they can refuse, accepting punishment. Often in social situations there is a third way: not to follow or refuse

COMPLEX SYSTEMS AND THE PRACTICE OF WORLD POLITICS

189

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

190

NEIL E. HARRISON

Every paradigm comes with its own language; for some scholars the linguistic gymnastics of postmodernism is as challenging as formal theory and mathematical notation are for others. The modeling challenge of complexity really is no different from the difficulties of researching across language and culture barriers, a skill quite common in comparative politics and IR. As with any paradigm, the choice of method depends on the research puzzle being investigated. 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, Swarm, etc.). Still others may wish only to specify units and interaction rules and “outsource” model construction to computer programmers, applied mathematicians, or statisticians.

REFORMING POLICY The most dangerous aspect of positivist, 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. The most important service of a complexity paradigm might be to free policy-makers from this overweening hubris and inculcate a sense of uncertainty. 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; Allison 1971). 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. Thus, success in the War on Terror demands accepting a more inclusive and nuanced interpretation of events than offered by simple Realpolitik. To intervene effectively in complex systems requires, first, that policymakers recognize the inherent uncertainties in their understanding of both the system and the effects of our interventions therein. Second, policy must seek out points of leverage that may be well hidden. Complex systems in world politics demand policy caution. Brian Arthur suggests that when intervening in complex systems “you want to keep as many options open as possible. You go for viability, something that’s workable, rather than what’s “optimal”. . . because optimization isn’t well defined anymore. What you’re trying to do is maximize robustness, or survivability, in the face of an illdefined future. And that, in turn, puts a premium on becoming aware of nonlinear relationships and causal pathways as best we can. You observe the world very, very carefully, and you don’t expect circumstances to last” (quoted in Waldrop

COMPLEX SYSTEMS AND THE PRACTICE OF WORLD POLITICS

191

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

192

NEIL E. HARRISON

model and through informal institutions select behaviors that are socially appropriate. Industrial societies will continue to blithely consume nonreplaceable resources as long as their paradigm is anthropocentric mastery of “nature” (Harrison 2000). In world politics, 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). In Gramscian language, a social paradigm or “mind-set” is supported by a hegemony of economic and political goals and intellectual and moral discourse (Hoffman 1984). Redirecting political trajectories requires change in both structures and ideas; in the language of complexity, it means reinforcing changes in both institutions and internal models. Gramsci’s approach was incremental, but ideational change may be better accepted when it is so sudden and substantial as to be an epiphany. In complex systems, problems are unclear, solutions have uncertain effects, and points of leverage are never simple. But complex systems can be pushed and prodded, and changed; yet, caution is required and instruments are imprecise. Because institutions and social systems are influenced by human perceptions of the world and how it works, dethroning the rational choice paradigm is the best way for scholars to positively influence world politics. But policy under complexity opens many other avenues of research, and the benefits are likely to be great.

TOWARD THEORY Earnest and Rosenau argue that there cannot be a general “complex systems theory” in world politics. There are no acceptable standards by which we can know knowledge of complex political systems, and ABM simulations, the methodological basis for a complex systems theory, exogenize authority, the essence of political systems. In this chapter, 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. But I do accept their general position, though for a different reason. 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. 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. Yet, complex system theories of individual issue-areas in world politics are possible. Differences between issue-areas in agents’ values and beliefs and in informal institutions (norms) and formal ones (organizations, etc.) may prevent an allencompassing complex systems theory. But theories—sets of premises, assump-

COMPLEX SYSTEMS AND THE PRACTICE OF WORLD POLITICS

193

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

194

NEIL E. HARRISON

drive up to ten miles an hour over the speed limit (seemingly accepted as a “safe” margin by drivers and police), and perhaps 10 percent of GDP is unmeasured and belongs to the “black economy.” 8. This is “satisficing” (Simon 1955). One mechanism for finding an excellent (but not the very best) solution is “simulated annealing.” See Kauffman 1995, 248–52.

REFERENCES Allison, Graham T. 1971. The Essence of Decision. Boston: Little, Brown. Arendt, Hannah. 1958. The Human Condition. Chicago: University of Chicago Press. Arthur, W. Brian. 1994. “Complexity in Economic Theory.” American Economic Review 84, no. 2:406–11. Ayer, A. J. 1936. Language, Truth, and Logic. New York: Oxford University Press. Campbell, Donald T. 1960. “Blind Variation and Selective Retention in Creative Thought as in Other Knowledge Processes.” Psychological Review 67, no. 6:380–400. ———. 1974. “Evolutionary Epistemology.” In The Philosophy of Karl Popper, ed. Paul Arthur Schilpp, 413–643. The Library of Living Philosophers, 14. LaSalle, IL: Open Court. Cioffi-Revilla, Claudio. 1998. Politics and Uncertainty: Theory, Models and Applications. Cambridge: Cambridge University Press. De Soto, Hernando. 2000. The Mystery of Capital: Why Capitalism Triumphs in the West and Fails Everywhere Else. New York: Basic Books. Easton, David. 1981. The Political System: An Inquiry Into the State of Political Science. Chicago: University of Chicago Press. Gettier, Edmund L. 1963. “Is Justified True Belief Knowledge?” Analysis 23:121–23. Harrison, Neil E. 2000. Constructing Sustainable Development. Albany: State University of New York Press. Henrickson, Leslie, and Bill McKelvey. 2002. “Foundations of ‘New’ Social Science: Institutional Legitimacy from Philosophy, Complexity Science, Postmodernism, and Agent-Based Models.” Proceedings of the National Academy of Sciences of the United States of America 99, Suppl. no. 3 (May 14): 7288–95 (accessed at www.pnas.org on May 15, 2005).

COMPLEX SYSTEMS AND THE PRACTICE OF WORLD POLITICS

195

Hoffman, John. 1984. The Gramscian Challenge: Coercion and Consent in Marxist Political Theory. New York: Basil Blackwell. Janis, Irving. 1982. Groupthink. Boston: Houghton Mifflin. Kauffman, Stuart. 1995. At Home in the Universe: The Search for the Laws of SelfOrganization and Complexity. New York: Oxford University Press. Kuhn, Thomas. 1970. The Structure of Scientific Revolutions. 2nd ed. Chicago: University of Chicago Press. Marney, J. F., and F. E. Tarbert. 2000. “Why do Simulation? Towards a Working Epistemology for Practitioners of the Dark Arts.” Journal of Artificial Societies and Social Simulation 3, no. 4 (accessed at http://www.soc. surrey.ac.uk/JASSS/3/4/4.html). Meadows, Donella H. 1997. “Places to Intervene in a System.” Whole Earth, Winter (accessed at http://www.wholearthmag.com/ArticleBin/109.html on February 7, 2003). Nelson, Richard R., and Sidney G. Winter. 1982. An Evolutionary Theory of Economic Change. Cambridge, MA: Harvard University Press. Petracca, Mark P. 1991. “The Rational Choice Approach to Politics: A Challenge to Democratic Theory.” Review of Politics 53 (Spring): 289–319. Putnam, Robert D. 1988. “Diplomacy and Domestic Politics: The Logic of TwoLevel Games.” International Organization 42, no. 3 (Summer): 427–60. Richards, Diana. 2000. Political Complexity: Nonlinear Models of Complexity. Ann Arbor: University of Michigan Press. Risse-Kappen, Thomas. 1994. “Ideas Do Not Float Freely: Transnational Coalitions, Domestic Structures and the End of the Cold War.” International Organization 48, no. 2 (Spring): 185–214. Rosenau, James N., and Mary Durfee. 2000. Thinking Theory Thoroughly: Coherent Approaches to an Incoherent World. Boulder, CO: Westview Press. Schmitt, Carl. 1976. The Concept of the Political. New Brunswick, NJ: Rutgers University Press. Simon, Herbert A. 1955. “A Behavioral Model of Rational Choice.” Quarterly Journal of Economics 69 (February): 99–118. Smith, Steve. 2004. “Singing our World into Existence: International Relations Theory and September 11.” International Studies Quarterly 48, no. 3 (September): 499–515. Strange, Susan. 1996. The Retreat of the State: The Diffusion of Power in the World Economy. Cambridge: Cambridge University Press.

196

NEIL E. HARRISON

Waldrop, M. Mitchell. 1992. Complexity: The Emerging Science at the Edge of Order and Chaos. New York: Simon & Schuster. Weber, Max. 1958. The Protestant Ethic and the Spirit of Capitalism. Trans. Talcott Parsons. New York: Charles Scribner’s Sons. Wendt, Alexander. 1992. “Anarchy Is What States Make of It: The Social Construction of Power Politics.” International Organization 6, no. 2 (Spring): 391–425. ———. 1994. “Collective Identity Formation and the International State.” American Political Science Review 88:384–96.

Contributors

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

198

CONTRIBUTORS

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

CONTRIBUTORS

199

Balkans, Middle East, Caucasus, Central Asia and Southeast Asia; e.g., 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. His most recent book is Capturing the Complexity of Conflict: Dealing with Violent Ethnic Conflicts of the Post-Cold War Era (Thomson Learning 1999). 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. He also is an Adjunct Associate Professor in the University of Southern California’s School of Policy, Planning & Development and a member of the External Advisory Board of George Mason University’s Center for Social Complexity. He has held appointments as a consulting scientific advisor at DARPA, a senior scientist and program director at the National Defense University’s Center for Technology & National Security Policy, 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. In addition to his work in the areas of international security and science/technology policy, Dr. Saunders-Newton has over a decade of experience in research methodology and social science research (domestic and international) in a variety of research agencies. He holds degrees from the Pardee RAND Graduate School (Ph.D., M.Phil), University of Michigan (MPP), and Lawrence University (B.A.), and has pursued post-doctoral studies at RAND and the Santa Fe Institute. J. DAVID SINGER earned his Ph.D. in 1956 from New York University and is Professor Emeritus at the University of Michigan. He has taught at Vassar and in Oslo, Geneva, Groningen, Taipei, and Mannheim. He founded the Correlates of War Project and was its Director from l963 to 2003. 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, 1816–1980 (Sage 1982, with Melvin Small), and over one hundred articles including “The Level-of-Analysis Problem in International Relations” published in World Politics in 1961.

This page intentionally left blank.

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

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. Persaud Global Limits: Immanuel Kant, International Relations, and Critique of World Politics— Mark F. N. 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. S. Arms Control Policy—Steve Breyman Agency and Ethics: The Politics of Military Intervention—Anthony F. Lang, Jr. 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. Rosenau and J. P. Singh (eds.) Theories of International Cooperation and the Primacy of Anarchy: Explaining U. S. International Monetary Policy-Making After Bretton Woods—Jennifer Sterling-Folker Technology, Democracy, and Development: International Conflict and Cooperation in the Information Age—Juliann Emmons Allison (ed.) 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. J. Barry Jones (eds.) Crisis Theory and World Order: Heideggerian Reflections—Norman K. 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. 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 Regimes, and Political Community—Rodger A. Payne and Nayef H. Samhat

SUNY SERIES IN GLOBAL POLITICS

203

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

This page intentionally left blank.

Index

3PF. See Three Pillar Framework 4⫹2 framework, 55 abduction, 166, 171, 173, 180 ABM programs in DOD, 175, 176, 180 ABMs. See Agent-based models actor sensitivity, 51 adaptive behavior compared to rational choice, 139 adaptive planning, robust, 177, 178 agency, 27 and independent thinking, 88 and structure, 7 contingency, 157 agency-level computational models. See agent-based models agent-based modeling as transdiscipline, 167 agent-based models, 5, 27, 33, 38, 113–14, 122–23, 126–27, 140, 167, 173–74 advantages of, 177 and aggregate studies of war, 134 and decision rules, 150 and dynamic processes, 127 and emergence, 132 and equation-based models, 130–31 and game-theoretic models, 131–32 and understanding processes, 140 appropriate ontology, 168 as aid to policy-making, 166–68 as experiments, 171

assumptions, 168 contingent identities in, 152 defined, 126, 127 DOD research of, 175, 176, 180 endogenizing authority in, 155 epistemological concerns, 171 epistemological implications of, 168 epistemology of, 169 exploration by, 171 identity in, 150 incomplete, 158 implementation, 129 information capture by, 169 inquiry possibilities of, 165 method problems, 145 methods of, 151, 157 modeler assumptions, 144 modeler expectations, effect of, 150–52 problem of simple behavior rules, 152 typology, 167 users of, 168 uses of, 169 agent-based systems, 82 agents, 3, 114, 158, 168 adaptation by, 98 adaptive, 99 adaptive mechanisms, 126 and social system, 98 basic needs of, 87 behavior of, from internal model, 9–10, 102 defined, 3, 9, 98

205

206 agents (continued) diversity of, 132 ecologies of, 167 in social relations, 128 modeling adaptive behavior of, 129 ahistorical theory, 192 algorithmic social science, 173–78 language of, 175 analysts need for retraining, 178 analytic narrative, 180 ant queen, myth of, 143 Antarctic ozone “hole,” 100 arms races, 51 Arthur, W. Brian, 13, 98, 190 assumptions by modeler, effect of, 148 autarky, 39 authority and coercion, 154 and complexity, 144, 146, 149 breakdown of, and CAS, 154 central in politics, 152 diffused by globalization, 188 emergence of, 152–57 in complex systems, 188–89 problem of endogenizing, 155 problematic, 189 “Turing test” for, 157 authority cascades, 154 Axelrod, Robert M., 13, 52, 53, 54, 96, 144, 149, 154 balance of power, 11, 56 Baltic States, 80 behavioral adaptation, 123 Bible date in vernacular, 85–86 publication dates, 90 biology and prejudice, 48 effect on group identity, 48–49 biopolitics, 144 Bin Laden, Osama, 12 blue eyes/brown eyes test, 46 Bosnia, 52–57, 65 bottom-up modeling, 140

INDEX Bush, George W., 31, 50, 57 butterfly effect, 74 causation, 11–13, 18, 27, 34–36, 64 in open systems, 8 in world politics, 13 cause and effect. See causation CAS. See complex adaptive systems Cederman, Lars-Erik, 13, 157 CFCs. See chlorofluorocarbons chaos, 166 “edge of,” 58, 63, 74, 89, 153 chaotic problems, 174 Chechnya, 86 chlorofluorocarbons, 99–101 industry, 107 link to ozone depletion, 103 technology, 106 clash of civilizations, 12, 47, 58 Club of Rome. See “Limits to Growth” coethnic punishment, 133 coevolution, 74, 81, 184 in North and South, 108 of internal models, 113 coevolutionary processes, 98, 110 Cold War, 1, 29, 43 Commonwealth of Independent States, 81 complex adaptation, 96 complex adaptive systems (CAS), 73–90 (Also see complex systems) and constructivism, 156 and politics, 144 as thought experiments, 149 breakdown of authority, 154 concepts of, 74–75 conceptual tools of, 89 contingency in, 148 fundamental insight of, 83–84 no authority, 153 world politics as, 146 complex systems (Also see complex adaptive systems) and conflict studies, 51–52 and rules, 5–6 characteristics of, 3 compared to simple systems, 2–5 different from general systems, 32–36

INDEX leverage points in, 190–92 nonlinearity in, 5 predictability of, 4 complex systems theories, 165 and induction, deduction, 149 challenge of, for IR/world politics, 145 defined, 2 from simple rules, 5 limits to methods of, 149 not deductive, problems from, 148 potentially integrative, 38 complexity, 28 aggregate studies of war, compared, 134 and authority, 144, 146, 149, 188–89 and change, 58 and conflict resolution, 52–57 and constructivism, 7, 9, 10 and decision-making, 3–4 and equilibrium, 4–5, 52 and exploration, 186 and living systems, 3 as generative science, 149 as metaphor, 144 causation in, 11–12, 34–36 compared to orthodox IR/world politics theories, 6–13 compared to complicated, 174 conceptual tools of, 87, 89 conditions, small changes in initial, 52 defined, 2–6 epistemology of, 186–87 epistemology of, non-positivist, 158 from simple rules, 5 generates novel insights, 185 improving orthodox theory, 184–85 in social sciences, 143 in world politics, 6–13 measures of, 3–4 new thinking from, 33, 166, 185 paradigm or theory, 147 parsimony in, 37 patterns in, 147 policies proposed by, 84 policy implications of, 190–92 problems of metaphor in politics, 153 proposed empirical testing, 87–89, 111–14, 127–30

207

superior to orthodoxy, 112 complexity methods, 189–90 complexity science, 11, 73, 174 defined, 2 computational epistemology, 179 computational models, 133, 175 as methodological pillar, 174 computational social network analysis, 167 conflict and initial conditions, 63 as balancing mechanism, 55 as process, 52, 53, 63 conflict resolution, 50–51 and complexity, 52–57 conflict transformation, 63 consolidative modeling, 127 constructivism, 7, 9, 10, 29, 184 assumptions of, 158 contingency in complex adaptive systems, 148 complex adaptation, 97 Correlates of War, 30, 34 COW. See Correlates of War credibility of methods, 168 culture, importance of, 84–86 Dayton Peace Agreement, 52, 54, 55, 57 decentralization of authority, 188 decision algorithms, 169 decision rules, contingency in, 151 decision-making and complexity, 3–4 and simplification, 174 deduction, 148, 166, 180 deep narrative, 175 deep uncertainty, 180 Defense Advanced Research Projects Agency, 178 Defense Modeling and Simulation Organization, 178 democracy, 76 and self-organization, 81 in form only, 86 Department of Defense research in ABMs, 175, 176, 180 dissipative structures, 4 dissipative systems. See dissipative structures

208

INDEX

DOD. See Department of Defense domestic politics, 7, 9, 28 dynamic systems, 10–11 dynamics at microlevel, 122 early warning system, 61 “edge of chaos,” 58, 63, 74, 89, 153 Einstein, Albert, 193 elites, political, 28 Elliott, Jane, 46–47 emergence, 7, 32–34, 81–82 and ABMs, 132 of authority, 152–57 of universal participation rule, 109 of ethnic norms in ABMs, 128 emergent properties, 147 of state, society, 74 emergent phenomena, 74 empirical testing, of complexity, proposed, 87–89, 111–14, 127–30 EPA. See US Environmental Protection Agency epistemic communities, 185, 193 epistemology, 171, 179 evolutionary, 187 for simulations, 187 of complexity, 186 non-positivist, 158, 186–87 positivist, 186 equilibrium, unstable, 4–5, 52 ethnic consciousness, 88 ethnic entrepreneurs, 125, 129 ethnic groups, 132 ethnic norms, 45–51, 125–26 defined, 125 emergence of, 126 ethnic violence, 121–26 and bottom-up processes, 124 ethnocentrism, 43, 45–51 defined, 45 in-group, 45 sociobiology of, 48 sources of, 47–49 EU. See European Union European Union, 61, 80, 82, 99, 103 evolutionary epistemology, 187 experimentation, 34

exploratory modeling, 127, 173 external, internal view, 38 extra-rational, 31 feedback, 28–29, 30–31, 98 negative, positive, 4 positive, 29 feedback, iterative, 150 fitness, 11, 74–75, 80, 82 acquiring fitness, 84–86 inclusive, 48, 64 of countries, 75, 84–86 of society, 73, 89 fitness criterion modeler’s choice of, 151 fitness landscape, 75, 83, 86 flip. See Systemic change formal institutions, 188 formal modeling, 113 Freedom House, 80, 188 Gell-Mann, Murray, 3 general systems, theories, 27 compared to complex 32–36 genetic algorithms, 150, 151, 158 globalization, 188 Gramsci, Antonio, 192 HDI. See UN Human Development Index Hobbes, Thomas, 153 Holland, John H., 5, 27, 98, 144 homeostasis, 5, 11 human behavior, 149 and biology, 47–49 Human Development Index. See UN Human Development Index human nature, 6, 44, 47–48, 49 human social networks, 74 Idealpolitik, 47, 55, 57 identity, 43, 158 issue specific, 156 induction, 140, 148, 166, 171, 173 inference-making, improving, 173 informal institutions, 188 in-group, 45, 128 initial conditions, small changes in, 52

INDEX inquiry efficacy and quality, 170 institutions, 11, 27, 35, 84, 188, 193 interest groups, 151 interests, 9, 44, 158 accidental, path dependent, 152 interhamwe, 123 internal models, 9–10, 35, 47, 49, 98, 132, 191 adaptation of, 108 and agent behavior, 102 change in South, on ozone depletion, 104 diversity in, 125 operationalizing, 114 problem of simple behavior rules, 152 US change in, on ozone depletion, 106, 110 internal rule models. See internal models international organizations, 36 international relations. Also see world politics international relations theories, orthodox, 7, 12, 27, 32–36 failures of, 1–2 compared to complexity, 6–13 international system, 57 international theorists. See international relations theories Islamic societies, and literacy, 86 Jervis, Robert, 13 jihad, 133 Jevons, W. Stanley, 7 Kaplan, Morton, 27 Kauffman, Stuart A., 89, 143 Koran. See Qur’an Kuhn, Thomas S., 57 learning, by agents, 9, 35 levels of analysis, 26, 27, 35, 165 leverage points in complex system, 190–92 “Limits to Growth,” 33 literacy evolution of, 87 importance of, 84–86 without debate, 86

209

Locke, John, 6 logical positivism, 158 London Amendment, 108 Luther, Martin, 84–86, 88 macrobehavior and microevents, 132, 189 mapping, 187 Markov chains, 12 and Monte Carlo methods, 189 mastery of nature, 192 mathematical models, of complexity, 189 Maturana, Humberto, 10 measures of theory quality, 171 media, 28 mental model. See internal model meta-agents, 8–9, 27, 34–35, 114, 184 method, for research puzzle, 190 methodology adequacy of, 166 falsification as, 158 simulative, 157 styles of, 167 methods of complexity, 33, 185, 189–90 microevents and macrobehavior, 189 micro-macro linkages, 146 models. Also see Agent-based models defined, 170 differentiation between, 171 ensemble of, 171 in social science, predictive accuracy of, 178 isomorphism of, 170 mathematical, computational, 189 nature of, 28 predictive, defined, 169–70 reasoning from, 166 specification of, 172 use by analysts, 178 modeler assumptions, not theoretically neutral, 151 modeling exploratory, 173 rules or attributes, 150 Montreal Protocol, 29, 100, 101, 105, 110 and US internal model, 106–8 multi-agent simulation, 167

210

INDEX

national consciousness, 88 National Defense University, 176 national interests, 32 nation-state. See state North Atlantic Treaty Organization, 54, 56, 61, 80 Natural Resources Defense Council, 106 nature and nurture, 49 negative peace, 56 neoclassical economics, 6 neorealism, 10 New European Peace and Security Systeem (NEPSS), 60–63, 64, 65 Newton’s universe, 6 nonlinear rules of behavior, 130 nonlinear systems, 189 nonlinearity, 12 as metaphor, 144 norm of universal participation, 107 norms, 30, 99 emergence of, 126 ethnic, 125–26 of behavior, 121 North Atlantic Treaty Organization. See NATO NRDC. See Natural Resources Defense Council obedience, 133 ontology, 26 of agent-based models, 168 open systems, 4, 8, 28 order for free, 82 organization, compared to structure, 10 Organization for Security and Cooperation in Europe, 54, 61, 62 orthodox theories, improving, 37–38, 108–11, 184–85 Osama Bin Laden, 12 OSCE. See Organization for Security and Cooperation in Europe out-group, 45, 128 ozone depletion regime, 95, 99–108, 110 paradigm shift, 57 parsimony, 36, 170 and complexity, 37

parsimony and complexity, 113 path dependence, 12, 29–30, 148, 152, 170 PCMT. See Pre-Conflict Management Tools Program peace negative, 56, 63 positive, 63 peace-keeping, 57 Peirce, Charles Sanders, 173, 179 perception, of decision-makers, 35 points of leverage, in complex systems, 190 policy effects, uncertainty of, 192 politics, and complex adaptive systems, 144 Popper, Karl, 193 positive sum common security, 61 positivism, 1, 158, 186–87 dangers of, 190 in social sciences, 187 positivist methods. See positivism postmodernism, 190 post-Soviet Eurasia, 73, 75–77 and complexity, 81–83 democracy, market capitalism in, 77 ethnic issues in, 80 zones of, 76 power, 44, 56 Pre-Conflict Management Tools Program, 176–78, 180 prejudice, and biology, 48 preventive diplomacy, 61 Prigogine, Ilya, 4, 29 Prisoner’s Dilemma, 53 problem solving, global, 63 problem, moving from chaotic to complex, 174 problems, unclear, 192 process tracing, 114 processes, 13 constitutive, 157 of decision-making, 31 simulation, insight into, 138 understanding of, with ABMs, 140 provocability, 53, 54, 56 punctuated equilibrium, 75, 82–83 Putnam, Robert D., 10

INDEX Qur’an, 86 racial prejudice, learned, 46 rational choice, 7, 9, 96, 109, 115, 139, 183, 185 compared to adaptive behavior, 139 rationality, 30–32 objective, 9, 32 of social aggregations, 31 subjective, 9–10 rationality assumption, appeal of, 150 realism, 11, 32 complexity compares, 2 hegemonic theory, 54 Realpolitik, 43–47, 55, 57 and War on Terror, 50 reasoning computer-assisted methods, 177 non-Peircian process, 172 recursive interaction, 38 refutation, as selection mechanism, 193 regime emergence, 96 formation, for ozone depletion, 99–108 ozone, 110 participation, 97 regime theory, 95, 102, 109 and complexity, 108–11 challenges to, 96 changes from complexity, 108 system rules exogenous, 112 Republic of Srpska, 54 research puzzle and choice of method, 190 retroduction, 166, 179 Robbers Cave experiment, 46 robust adaptive planning, 177, 178 rule model. See internal model rule of participation, 103 emergence of, 109 rules, 5 contingency in decision, 151 of behavior, nonlinear, 130 rules, system, 97, 111, 191 dynamic compared to static, 102 evolution of, 101 for participation, 109

211

North only, 106 Northern problem, 101 of ozone depletion, 101 operationalizing, 114 South demands change, 107–8 universal participation, 101 Russian Federation, 80 Rwanda, 121–25 characteristics, 123 satisficing, 150, 194 scenario differentiation, 172 Schelling, Thomas, 147 schema. See internal models security, positive sum common, 61 selection, 56 selection mechanism, 193 self-organization, 2–6, 83 and democracy, 81 and diversity, 191 and free thought, debate, 86 and hierarchy, 84 in world politics, 6–13 self-organized criticality, 75, 82 sensitivity of actors, 51 to initial conditions, 64 September 11, 2001, 1, 12, 43, 49, 50, 51, 64 Serbia, 52–57 Serbs, 65 “shadow of the future,” 53, 55 Simon, Herbert, 9, 179 simple systems, 2–5 and complex compared, 2–5 automobile as, 2 characteristics of, 3 defined, 2 predictability of, 4 static, 4 simplification and rigorous problem analysis, 174 by common knowledge, 17 simulated annealing, 194 simulation, 113 as experiment, 34 as research methodology, 138, 139

212

INDEX

simulation (continued) definition, 137 epistemology of, 187 incomplete, 158 multi-agent, 167 of CAS, testing theorems, 148 of complex systems, examples, 5 principal value of, 138 uses of, 137, 138 with computers, 147 simulation software, 158 simulative methods, 157 simulative-empirical research design, 149 Singer, J. David, 25 social capital, 134 social networks, 134 social organization, as attachment behavior, 143 social science algorithmic, 173–78 and positivist epistemology, 187 generalizable statements in, 179 problems of, 179 social science models, predictive accuracy of, 178 societal fitness, 73, 83, 89 sociophysics, 167 solutions, robust compared to optimal, 170 state, 17, 30–31, 35 as complex adaptive actors, 97 as emergent system, 7 as meta-agents, 8–9 feedbacks within, 30–31 in orthodox theories, 9 state of nature, 6 stratospheric ozone depletion. See ozone depletion structural factors, 122 structure, 11 and agency, 7 compared to organization, 10 surprise, in world politics, 1 systems, and psychology, 33 defined, 2, 33 emergent properties in, 33

general, 32 general and complex compared, 32–36 goals of, 191 nested, 26–27 paradigm of, 191 social, continuously evolve, 29 system rules, 111, 191 operationalizing, 114 systemic change, 29 in Yugoslavia, 58 systemic trajectory, 170 taxonomy, 27, 32 criteria of good, 36–37 defined, 25 taxonomy of complexity, 6 technology, effect on social science, 34 terrorism, 58, 62, 64 theories, general systems, 27, 32–36 theories, international relations. See theories, world politics theories, world politics, 10, 12 compared to complexity, 6–13 failures of, 1–2 orthodox, 7, 12, 27, 32–36 theory quality, measuring, 171 theory, ahistorical, 7, 192 Three-Pillar Framework, 59–60 defined, 60 time, arrow of, 4 Tit-for-Tat, 53, 55, 125, 191 top-down modeling, 140 Transparency International, 77 Ukraine, 80 umuganda, 124 UN Development Programme, 88 UN Human Development Index, 78–80, 83, 85–86, 87, 88, 90 countries ranked by 79–80 uncertainties, inherent, 190 uncertainty, 51 United Nations Environment Programme, 104, 193 United States, 99, 103 internal model on ozone depletion, 102, 106, 110

INDEX unstable equilibrium, 4–5, 52 U.S. See United States U.S. Environmental Protection Agency, 106, 110 Varela, Francisco, 10 veridicality, 170 Vienna Convention, 99, 100, 101, 103, 115 Waldrop, M. Mitchell, 5 Waltz, Kenneth, 10 War on Terror, 77, 190 and Realpolitik methods, 50 wicked problems, 174

213

World Bank, 177 world politics as complex adaptive system, 146 as processual flow, 13 as self-organizing complex system, 2, 6–13 orthodox and complex compared, 6–13 orthodox theories of, 7, 12, 27, 32–36 world systems theory, 10 World Trade Organization, 193 worldview, 17 Yugoslavia, 60, 76 internal conflict, 52–57 Yugoslavian war, 53

This page intentionally left blank.

POLITICAL SCIENCE

Complexity in World Politics Concepts and Methods of a New Paradigm Neil E. Harrison, editor Despite one hundred years of theorizing, scholars and practitioners alike are constantly surprised by international and global political events. The collapse of communism in Europe, the 1997 Asian financial crisis, and 9/11 have demonstrated the inadequacy of current models that depict world politics as a simple, mechanical system. 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. In language free of jargon, the book’s distinguished contributors explain and illustrate a complexity paradigm of world politics and define its central concepts. They show how these concepts can improve conventional models as well as generate new ideas, hypotheses, and empirical approaches, 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. “This book is well written and easily accessible, with essays by some of the major thinkers in the field of complexity science. It makes a number of intellectual contributions and helps fill a gap in the existing literature.” — Scott E. Page, coeditor of Computational Models in Political Economy Neil E. Harrison is Founder and Executive Director of the Sustainable Development Institute and the author of Constructing Sustainable Development, also published by SUNY Press. A volume in the SUNY series in Global Politics James N. Rosenau, editor

State University of New York Press www.sunypress.edu