For some time, when GG talks to folks about climate change, he avoids saying “models predict thus-and-so will happen,” instead focusing on some more basic information like the correlations of CO2 content and temperature in the geologic record. Part of the reason is that climate models have some big issues, and one of those is that they of necessity use empirical stopgaps to replace basic physics going on at scales below the resolution of the models. These stopgaps are at the most basic level fudge factors, and the deal is that they are chosen to match our instrumental record–and usually the instrumental record on temperature. This in a sense makes them backwards-looking; it is possible that the tuning values that make a model fit the last half of the twentieth century are wrong for slightly different conditions in the twenty-first century (and arguably that such fancy climate models are utterly out of touch with how the earth back in the Pleistocene or Eocene or Cretaceous really worked). It is also possible that an incorrect combination of fudge factors happens to work out in some situations but is highly misleading others.
Most of that tuning went on behind closed doors, making it hard to know what trade-offs different groups were accepting in making their models match some part of the climate record. Now, finally, Science reports that the lid on these internal deliberations is being taken off. Users will be able to see what model-creation groups have been assuming. It sounds as though part of the result of this new-found transparency is a recognition that you might want to tune these models differently if you are interested in different aspects of the climate system.
It is ironic that the presence of aggressive climate-change skepticism has prevented more public sharing of this information to this point because arguably some of this activity is what might most benefit from critical examination. One might hope that the exposition of these trade-offs, most of which are presumably well-known within the community, might focus research on the most critical of these tuned parameters, getting into the proper dependence of the desired factor on physical inputs.
Anyways, it is high time for this, and GG looks forward to seeing this improve the responsiveness of climate models to substantially different environmental conditions.