Harnessing Complexityby Robert Axelrod, Michael D. Cohen
Harnessing Complexity will be indispensable to anyone who wants to better comprehend how people and organizations can adapt effectively in the information age. This book is a step-by-step guide to understanding the processes of variation, interaction, and selection that are at work in all organizations. The authors show how to use their own paradigm of "bottom up"… See more details below
Harnessing Complexity will be indispensable to anyone who wants to better comprehend how people and organizations can adapt effectively in the information age. This book is a step-by-step guide to understanding the processes of variation, interaction, and selection that are at work in all organizations. The authors show how to use their own paradigm of "bottom up" management, the Complex Adaptive System-whether in science, public policy, or private commerce. This simple model of how people work together will change forever how we think about getting things done in a group."Harnessing Complexity distills the managerial essence of current research on complexity. A very valuable contribution to the emerging theory of competition and competitive advantage."-C.K. Prahalad, University of Michigan, coauthor of Competing for the Future"A brilliant exposition that demystifies both the theory and use of Complex Adaptive Systems."-John Seely Brown, Xerox Corporation and Palo Alto Research Center
- Basic Books
- Publication date:
- Sold by:
- Hachette Digital, Inc.
- NOOK Book
- File size:
- 541 KB
Read an Excerpt
Preface This is a small book about a large question: In a world where many players are all adapting to each other and where the emerging future is extremely hard to predict, what actions should you take?
We call such worlds Complex Adaptive Systems. In Complex Adaptive Systems there are often many participants, perhaps even many kinds of participants. They interact in intricate ways that continually reshape their collective future. New ways of doing things -- even new kinds of participants -- may arise, and old ways -- or old participants -- may vanish. Such systems challenge understanding as well as prediction. These difficulties are familiar to anyone who has seen small changes unleash major consequences. Conversely, they are familiar to anyone who has been surprised when large changes in policies or tools produce no long-run change in people's behavior.
When managers and policy makers hear about complexity research, they often ask, "How can I control complexity?" What they usually mean is, "How can I eliminate it?" But complexity, as we shall see, stems from fundamental causes that cannot always be eliminated. Although complexity is often perceived as a liability, it can actually be an asset. The thesis of this book is that complexity can be harnessed. So, rather than seeking to eliminate complexity, we explore how the dynamism of a Complex Adaptive System can be used for productive ends. Therefore, we ask how organizations and strategies can be designed to take advantage of the opportunities provided by complexity.
In a world of mutually adaptive players, even though prediction may be difficult, there is quite a bit that you can do.Complexity itself allows for techniques that promote effective adaptation. When there are many participants, numerous interactions, much trial-and-error learning, and abundant attempts to imitate each other's successes, there will also be rich opportunities to harness the resulting complexity. And there will be things to avoid. To take a simple example: Even though one action seems best, it usually pays to maintain variety among the actions you take so that you can continue to learn and adapt. The purpose of this book is to help managers and policy makers harness complexity.
We address a variety of readers. Some of our readers may simply be interested in learning more about the exciting research frontier called complexity. For these readers we provide a nontechnical introduction to the field. Other readers may have more specific interests in how to design social systems, or in how to make better policy for existing social systems. For these readers we draw on a wide range of applications from business, political, and cultural settings. We make no assumptions about the backgrounds of our readers, except that they are curious about how social systems work and how they can be made to work better.
As we were developing the organizational implications of complexity, we saw the need to bring order to the vast range of research in the field. To do this, we constructed a framework that provides a systematic way to analyze a particular setting and thereby suggest useful questions and promising possibilities for action. We found that our framework helped to clarify some deep relationships among many hitherto separate lines of complexity research. Moreover, the framework uncovered some important gaps. To fill them, we made a number of specific contributions to complexity research. These include the critical role of nonrandom interactions in adaptation, the contrast of biological with informational copying, and the relationships between credit allocation and measures of performance.
The foundations of this book lie in three distinct fields: evolutionary biology, computer science, and social design.
From evolutionary biology come the insights of Darwinian evolution, particularly that extraordinary adaptations can come about through the selection and reproduction of successful individuals in populations. Even though moths in England could not understand or predict that the Industrial Revolution would turn white-barked trees into soot-covered trees, it did not take very long for selection by predatory birds to transform the population of moths near a factory from white to black.
From computer science come insights about how systems with many artificial agents can be designed to work together and even adapt over time to each other and to their ever-changing environment. Two areas of computer science have been especially important to us. First, there is the field of evolutionary computation, which has fostered an engineering approach to adaptation. With an engineering approach, one asks how systems can be designed to become more effective over time. By making evolution and adaptation an engineering problem, evolutionary computation has shed light on how complex systems can be adaptive. Second, there is the rapid growth of distributed and network-mediated computing (including the Internet), which has led computer science into deeper analyses of just what it takes to make systems of many agents work together and grow.
From social design come insights into people and their activities in political, economic, and social systems. Entire disciplines -- such as political science, economics, sociology, psychology, and history -- have been devoted to understanding human beings and the settings they build and live in. Among the approaches that have concentrated on social design are organization theory and game theory. Organization theory provides insights into how institutional structure matters. Game theory provides insights into how people can choose strategies to maximize their payoffs in the presence of other people who are doing the same.
While the foundations of this work come from evolutionary biology, computer science, and social design, our analysis differs from all three of these in important ways.
Unlike evolutionary biology, we are primarily interested
- in the shaping of evolutionary processes rather than just observation and explanation,
- in intelligent individuals with language and culture, rather than plants and animals that rely primarily on their genetic heritage, and
- in different measures of success rather than taking the ability to have offspring as the sole measure of success.
Unlike computer science, we are primarily interested
- in systems composed of people or organizations rather than pieces of software,
- in systems with long and rich histories rather than systems that have little or no history, and
- in systems in which the costs of trials needed for adaptation are measured in terms of efforts and even lives of people rather than in cycles of computer time.
Unlike some approaches to social design, we are primarily interested
- in problems in which the preferences and even the identities of the participants can evolve over time, rather than situations in which the players and their preferences are fixed, as they are in game theory, and
- in problems in which decentralization is both promising and problematic, rather than situations in which decentralization is seen as practically a panacea, as in some forms of neoclassical economics.
In our analysis there are three key processes in a Complex Adaptive System. These key processes provide the basis of our three central chapters: Variation, Interaction, and Selection. We see variation, interaction, and selection as interlocking sets of concepts that can generate productive actions in a world that cannot be fully understood. We show how the very complexity that makes the world hard to understand provides opportunities and resources for improvement over time.
We are often asked how "complexity" differs from "chaos." The simple answer is that chaos deals with situations such as turbulence (Gleick, 1987) that rapidly become highly disordered and unmanageable. On the other hand, complexity deals with systems composed of many interacting agents. While complex systems may be hard to predict, they may also have a good deal of structure and permit improvement by thoughtful intervention.
Our approach is not just an extended "evolutionary metaphor," nor is it part of Social Darwinism (Hofstadter, 1955) or sociobiology (Wilson, 1975). We view processes of biological change as wonderful examples in the larger set of Complex Adaptive Systems. However, they have special kinds of agents, particular sorts of strategies, distinctive patterns of interaction, and their own special processes of selection. The patterns one sees in biology are not always found in other Complex Adaptive Systems. Copying a strategy for stock trading (such as a computer algorithm) involves only digital information and so is nearly costless compared with producing a new organism that contains a copied gene. Evaluating a business strategy (say, the introduction of a new product) can be enormously expensive compared with making a random variation of a fruit fly. Variation, interaction, and selection are at work in a population of business strategies, but the detailed mechanisms are often distinctly unbiological. To harness complexity effectively, many kinds of Complex Adaptive Systems must be considered.
We have paid special attention to the role of information in Complex Adaptive Systems. The continuing fall in the costs of copying and recombining information often results in the very rapid spread of strategies. An increasing penetration of information technology into social processes will therefore change those processes fundamentally.
We have emphasized the contextual forces determining interaction patterns. This too takes our work away from the traditional approaches in economics as well as biology, where there is often scant attention to the important consequences of patterns of interaction.
These additional aspects of our framework make it richer, able to incorporate realistic aspects of the world, though it still leaves out many factors, as good frameworks must do. The richer framework is not able to make detailed predictions, of course. There are too many things that might happen. We are willing to bear the lack of detailed prediction because we are interested in situations in which accurate prediction has always been difficult. In return for accepting complexity, we have a more systematic approach to harnessing it. This intellectual tactic reminds us of a guiding principle of the martial art of judo: "Throw with your opponent's own weight."
Our emphasis on harnessing complexity will, we hope, prove to be an important contribution. We chose "harness" for our title to convey a perspective that is not explanatory but active -- seeking to improve but without being able fully to control. This orientation determined many features of the book but most basically the focus on aspects of Complex Adaptive Systems that suggest guiding rules of thumb and leverage points of intervention.
The Complex Adaptive System approach is a way of looking at the world. It provides a set of concepts, a set of questions, and a set of design issues. By itself, it is not a falsifiable theory. Such a theory would have to specify the operational meaning of the key concepts and mechanisms in a particular domain. For example, to apply the Complex Adaptive Systems approach to economic markets, one would have to specify who the economic actors are, what they can see and do, how they generate variety in their behavior, how they interact with each other, and how the actors and their strategies are selected for retention, amplification, or extinction.
This book provides our personal view of how complexity research can be made relevant to problems of social design. We have benefited greatly from the many researchers working on complexity. Our task here is not to provide a textbook that surveys this dynamic field. Instead, we present our view of what complexity research offers to those who want to improve the world as well as marvel at it.
We are indebted to James March and John Holland for laying the foundations on which this book builds, and for providing personal as well as intellectual leadership. In the 1970s, James March began to write articles about a provocative topic: "the technology of foolishness" (March, 1976). He forced into the open an issue that remains hidden in a more conventional view of choice or decision making in social systems: the hard reality that the world in which we must act is often beyond our understanding. He began to draw out the implications of this fact when others were mostly in denial. It implies that each action we take is partly an instrumental step and partly a learning experience.
From John Holland we learned how adaptation can be regarded as an engineering problem. His inventions, starting with the genetic algorithm (Holland, 1975), provided a systematic way to design and study Complex Adaptive Systems with computer simulations.
For providing a wonderful network of connections to fellow workers in complexity theory, we are indebted to the Santa Fe Institute and the University of Michigan's Center for the Study of Complex Systems.
The immediate impetus to write this book was provided by a report commissioned from us on national information policy by the Highlands Forum. The Highlands Forum is a group of people from industry, government, and academia that deals with information issues and is sponsored by the Department of Defense. We decided to write our report from the point of view of what the Complex Adaptive Systems approach can say about information policy. In writing that report, we developed our own vision for how complexity can be harnessed in information policy. We then saw that this vision could be applied to many different areas of social design.
An indirect source of inspiration for this book was provided by a project sponsored by the Intel Corporation through a grant to the two of us and Rick Riolo. This project uses large-scale computer simulation experiments to study self-organizing social structures. The work we did on this project, along with the modeling work that the two of us have been doing for several decades, helped us better understand how complex systems can be built and analyzed.
For valuable advice (not always followed) we would like to thank David Axelrod, Arthur Burks, Corinne Cohen, Rachel Cohen, Paula Duffy, Edmund Durfee, George Furnas, John Holland, Christopher Lee, Ann Lin, James March, Jeffrey Mackie-Mason, Melanie Mitchell, Stephen Morrow, Scott Page, Paul Resnick, Rick Riolo, Douglas Ross, Raphael Sagalyn, Amy Saldinger, Carl Simon, Robert Wallace, Michael Wellman, and Marina von Neumann Whitman. For editorial help we thank Maria Bonn, Loretta Denner, and Donna George. For financial assistance, Robert Axelrod thanks the University of Michigan LS&A Faculty Enrichment Fund. Michael Cohen thanks the Ameritech Foundation.
For continual inspiration over more than fifteen years, we are indebted to the BACH research group at the University of Michigan. The BACH group is named after its original members: Arthur Burks, Robert Axelrod, Michael Cohen, and John Holland. The group also came to include long-term members William Hamilton, Rick Riolo, and Carl Simon and shorter-term members Stephanie Forrest, Douglas Hofstadter, Melanie Mitchell, Michael Savageau, and Reiko Tanese. Like any good complex system, it has become more than the sum of its parts. We are delighted to dedicate this book to the BACH group.
Copyright © 1999 by Robert Axelrod and Michael D. Cohen
Meet the Author
Robert Axelrod is Professor of Political Science and Public Policy at the University of Michigan. A MacArthur Prize Fellow, he is a leading expert on game theory, artificial intelligence, evolutionary biology, mathematical modeling, and complexity theory. He lives in Ann Arbor, Michigan. Michael D. Cohen, Professor of Information and Public Policy at the University of Michigan, has served as an External Faculty Member of the Santa Fe Institute and as a long-term consultant at the Xerox Palo Alto Research Center.
and post it to your social network
Most Helpful Customer Reviews
See all customer reviews >