“In principle, agent-based modelling should make some of the grandest social and political questions of our time accessible to rational experiment, such as whether the globalization of the economy is likely to lead to greater cultural harmony or to cultural conflict. But some social scientists remain uneasy, suspecting that any particular agent-based model of a social phenomenon risks coming to conclusions that depend on the underlying assumptions of the model. How do we know whether any one set of rules or assumptions will lead to truly representative behaviour, and not to an excessively crude caricature of the real situation? In short, such models can hardly be expected to provide a sound basis for policy until we can distinguish what is contingent from what is robust: what a particular model will produce as opposed to what all good models will produce.”