Computer models are fast becoming a permanent feature of policymaking in American government. But as the authors of this systematic, empirical study demonstrate, these models can be proven instruments of politics and power as well as aids to rational decision making.

Partisans use computer models to support their argument, the authors find, and gain a temporary advantage over the opponent in policy debates. But successful use of models by one side generates their subsequent use by the other. What results are "datawars" in which both sides use computer models to bolster their own arguments and shoot down the opposition's. Though datawars may lead to a better understanding of available policy options, political ideology and political power still determine which policy is chosen.

Based on the results of a cooperative research project on computerized planning models, Datawars studies the context of model use, focusing on the dynamics of the federal policymaking process. It uncovers how the "supply" forces of model characteristics and the "demand" forces of need and desire for models among policymakers interacted to spawn the truly impressive amount of modeling that now supports federal policymaking in the United States. The book's emphasis is on the determination of successful models; the data for this determination are case studies of federal agency use of two major modeling systems, one for macroeconomic modeling and the other for microsimulation modeling.

The authors show that while models may render a policy group temporarily victorious, the clear winners of datawars are the private firms and modeling professionals who supply the computer models and data used by both sides. And perhaps the book's most important finding is that the elegant features of models will not, by themselves, guarantee even a trial run in the halls of power.