In this issue of Climatic Change, Ron Brunner argues that the US Global Change Research Program (USGCRP) has to its peril overemphasized the potential policy significance of comprehensive, predictive models [1 ]. Brunner thinks that although comprehensive models are properly a central goal of global change science, they cannot produce effective global change policy. He argues for a 'modest alternative' to comprehensive modeling and policymaking. The alternative consists of geographically decentralized 'policy teams' which would generate and test a wide variety of limited policy formulations and action alternatives on a regional or even local scale over a short (2-3 year) cycle. The problem, in Brunner's view, is fourfold. First, predictive global models do not now have, and are unlikely soon to reach, very high levels of accuracy. While model quality is being debated and improved, public confidence in the USGCRP is eroding. Furthermore, authoritative validation of integrated assessment models (IAMs) is at least unlikely, if not altogether impossible, because they include socio-economic processes subject to large future modifications by intelligent social decision-making and by unpredictable human events. Second, predictions even excellent ones are not very important to the ways policies are actually established and evaluated in America (and perhaps, though he does not say so, in democratic societies in general). Thus even if highly accurate, thoroughly validated models existed, they would not matter much to the policy process. Third, if and when they materialize, actual global change policies are themselves unlikely to be comprehensive, simply because consensus on any comprehensive policy is very difficult to achieve. Finally, because it focuses primarily on national and international-level policy and because its rigorous demands limit the number of models it can produce, the comprehensive approach risks overlooking effective policy options at smaller scales. For these reasons, Brunner believes, policymakers would be better served by efforts to proliferate, experiment with, and evaluate a wide variety of limited, non-comprehensive policy alternatives. In this editorial, I reflect on the role of comprehensive models, such as IAMs and earth system models (ESMs), in politics and policymaking [2]. I distinguish between the latter terms because part of what I will address is the importance