Optimizing Models of Social Systems

The utility of optimizing models of social systems is examined, using Forrester's world model as an example. One ``optimal'' policy set was tested for its sensitivity to uncertainties in the formulation of both the model assumptions and an assumed long-term objective function. It is shown that the normal uncertainties associated with social system modeling make the search for an optimal policy unwarranted. The inaccuracies inherent in the assumptions of long-term social system models allow the modeler to explore with confidence only the general dynamic characteristics of the modeled system.