Models are not reality

It's The Data, Stupid Cont'd [Jonah Goldberg]
From a reader:

Your second point, "there's no lab that can recreate the world's climate." is so spot-on. I've been doing Modeling and Simulation for the DoD for about 15 years. In our case, we have empirical test data we can use validate the models — you know real flight-test data, etc. Whenever we get empirical data that breaks the model we first do a sanity check to verify that the data isn't some random out-lier, and if not then we determine the underlying cause and adjust the model. If I were caught bending the data to fit the model as the CRU was doing, I'm quite confident that I'd be unemployed and quite possibly facing time as a guest of the Feds.