Chapter 1 of this research-oriented book is a detailed introduction, which points out that it is difficult to use the behavior of individuals to predict the behavior of a society. Indeed, David McClelland’s early contention that the “need to achieve” in a number of countries could serve as a prognosticator of world position, or even dominance, in relation to other countries was grandiose in its theoretical scope. I do not know whether this hypothesis is reasonable today. Even if an enhanced replication of McClelland’s work were attempted, it would have to be considered simplistic as compared with the aggregate multidisciplinary models with which Epstein and Axtell are concerned.

After an excellent overview of previous artificial society models and methods, the authors state the theme of their research. They will use the computer to “grow” social structures and collective behaviors, by programming an artificial society consisting of agents, an environment, and rules. Some of an agent’s characteristics (such as sex and metabolic rate) will be invariant, while others (including wealth and health) can change, depending on the programmed rules of behavior for the agents and the environment. The authors suggest that object-oriented programming languages are particularly apropos for agent-based modeling. Appendix A gives a nice synopsis of the object-oriented design features that they used. The introduction ends with a brief summary of each chapter, which nicely details the development and outcomes of their “sugarscape model.”

Chapter 2 discusses “Life and Death on the Sugarscape.” The sugarscape is a landscape of resources that agents like to eat. Agents seek sugar to eat, but may die if they burn up all of their intake. When seasons are introduced, migration occurs. This results in a highly skewed distribution of wealth, which mimics human societies. Behavioral rules may then be altered to look for effects on the distribution of wealth over time. In addition, the authors create agent social networks and trace their evolution through time and space. The software allows the specification of either von Neumann or Moore neighborhood rules. These differences can be viewed as animations by using a CD-ROM that is said to be available to accompany the book. At the end of the chapter, Epstein and Axtell state that the sugarscape model can function as a laboratory in which fundamental social structures can be grown “in silico.”

The aim of chapter 3, “Sex, Culture, and Conflict: The Emergence of History,” is to grow a history of an artificial civilization. Initially, fertility rules are established. With sexual reproduction, combat, cultural exchange, and disease transmission, evolutionary processes that alter the genetic composition of the artificial society can be observed. In this artificial society, the interplay between biology and economics may be studied. Rules may be altered to view the effects of such manipulations on the society.

In chapter 4, “Sugar and Spice: Trade Comes to the Sugarscape,” spice is added as a second commodity agents will seek. Agents can then trade commodities for mutual benefit. It becomes reasonable to evaluate a central theoretical question of economic behavior, based on the running of the artificial societal model.

In chapter 5, “Disease Agents,” the agent-based society is used to study epidemiology and immunology. Initially, this seemed reminiscent of Conway’s computer game of Life, but it turned out to be much more sophisticated.

Chapter 6 presents “Conclusions.” In this chapter, all dimensions are turned on to explore a complex, multidimensional artificial society. Dimensions include movement, resource gathering, sexual reproduction, combat, cultural transmission, trade, inheritance, credit, pollution, immune learning, and disease propagation. Other interesting collective structures emerge in the artificial society, such as tribes of agents, stationary wealth distributions, and collective patterns of movement.

Epstein and Axtell have opened the door to global simulation and prediction. Students or researchers interested in this topic would be well served by this concise treatment. This book has potential as a text for a graduate course. It is well written and presented, and contains an excellent index, list of references, appendices, and numerous figures and animation stills. Overall, this book is extremely well conceived and executed. It would be interesting to factor in possibility distribution scores for each of the initial variables in order to attempt some real-world predictions. In their conclusions, however, the authors say they are working on a modest archeological reconstruction model, which they feel offers the greatest utility from a policy standpoint. Also, they appropriately note the influence of computational limits on artificial societies.