Joshua Epstein: Generative Social Science

[Readings] (02.28.09, 8:10 pm)

This book is meant as a successor to Epstein and Axtell’s Growing Artificial Societies. I was highly critical of Growing Artificial Societies due to what I found to be a lack of necessary attention and inquiry into the models used for the social simulation. In the context of this book, I think those shortcomings are meant to be seen as not the central focus, rather that different models could be plugged-in to a simulation and the simulation would do something. This book takes teh previous as a call to arms, and actually fleshes out the theory of social simulation with a large variety of examples. These are diverse and do work to serve as analytical tests of the rules that govern the simulation. In particular, there is substantial discussion of a model of the Anasazi collapse, which is compared with actual archaeological findings. A number of other simulations discuss diverse topics such as retirement, class emergence, social norms, and civil violence. A few of these suffer from the problems found in the original book, that they are abstract to a degree that they are self enclosed and cannot relate to the real world, but the several chapters that discuss the Anasazi serve to redeem this somewhat.

The use of social simulation can be seen as a way to discuss and analyze a model of social behavior. The slogan from Growing Artificial Societies suggested that “If you didn’t grow it, you didn’t explain it.” This does not mean that if you did grow it then you did explain it, but rather that the growing is necessary for explanation to be possible. The shift in focus of the book is primarily about the nature of scientific explanation, that a theory that comes from observations should be simulated to verify that theory. Computationally, this makes sense. Simulation is widely used in natural science and engineering, but rarely in social science because of the enormous complexity of human culture. The use of the computer in the simulation is not the goal, but rather the explanation is the goal (nonetheless, computation generally makes simulation easier).

The goal of these simulations is thus to test and show the emergence of certain structures via demonstration. This is similar to my work in the simulation of literary worlds, but my goal is different. My goal could be made more similar if my focus were primarily in the construction of an accurate model that reflects the author’s narrative world. As it stands, this is really a secondary goal, where my primary one is in the experience of the player. It is interesting to compare the purely emergent and generative project with the structures of drama management, where the role of a drama manager is to take control and exert influence over a simulation, while the agent based approach is about providing simple rules and letting the larger phenomena attend to themselves.

Reading Info:
Author/EditorEpstein, Joshua
TitleGenerative Social Science: Studies in Agent-Based Computational Modelling
Tagsai, simulation, social simulation, specials
LookupGoogle Scholar, Google Books, Amazon

No Comments »

No comments yet.

RSS feed for comments on this post. TrackBack URI

Leave a comment

You must be logged in to post a comment.