Wednesday, January 14, 2009

Agent-based models (new book by Nigel Gilbert)

Nigel Gilbert contributed recently to the "Quantitative Applications in the Social Science" series by writting about 'agent-based models'. The series editor (Tim Liao) writes:" there are two general approaches to the study of social behavior. collect observational, survey, or other forms of data and analyze them, possibly by estimating a model; or begin from a theoretical understanding of certain social behavior, build a model of it and then simulate its dynamics to gain a better understanding of the complexity of a seemingly simple social system". Gilbert's recent work belongs in the second tradition which is usually termed Generative Social Science after Epstein and Axtell (1996)

I found chapter 3 (using agent-based models in social science) really useful as it gives practical steps in developing an agent-based model. Chapter 2 is also important as it discusses explicitly the concepts of time (as well as the concepts of agents and environment). Although he discusses time briefly, his treatment is important because it is usually underestimated. Within the GIScience literature, Dona Peuquet' s book (Representation of Space and Time) provides a comprehensive discussion that may benefit the development of agent-based models.

My personal view is that time has been trivialized in most of the agent-based models that I know. Time should be added to the list of significant research areas (e.g.,space/GIS, Learning and Simulation of language - chapter 5) that agent-based modellers need to address.