Doing experiments in the social sciences isn't easy - you can hardly meddle with a society just to see what happens. But it's possible to create computer models of societies, to see which features lead to realistic societies. This is what Joshua M. Epstein and Robert Axtell have done with their Sugarscape model, and Growing Artificial Societies
is a report of some of the things they have observed. The model starts with a landscape in which sugar grows, with agents who roam about collecting sugar which they need to live. Even with such a simple model is it possible to see featues such as migration and uneven wealth distribution.
The model is then extended in various ways, for instance giving the agents various properties which can be passed on to offspring via sexual reproduction. Introduction of another commodoty, spice, allows the possibility of trade, an the authors point out the importance of local interactions, as distinct from believing that there must be a global equilibrium price. Introduction of disease to the model shows the advantages of such a cellular automaton based model, as distinct from models based on differential equations. In the last chapter the authors show what can be seen by putting eveything together, showing that such a model can be an invaluable tool in the social sciences, in particular in showing the links between different fields. There is a CD-ROM to go with the book, and it would be good to see some of the examples, but there is no problem with reading the book on its own. (Some parts of the Sugarscape model can also be found on the internet). I'd recommend this book to anyone wanting to see this new way of approaching the social sciences, as well as to those interested in the programming side of such models.