Agent-based modeling is showing great promise in the social sciences. However, two misconceptions about the relation between social macroproperties and microproperties afflict agent-based models. These lead current models to systematically ignore factors relevant to the properties they intend to model, and to overlook a wide range of model designs. Correcting for these brings painful trade-offs, but has the potential to transform the utility of such models.
In Models, Simulations, and Representations, ed. by Paul Humphreys and Cyrille Imbert, pp. 115-144. New York: Routledge, 2011