Don't Guess. Test The Models!

The one thing that anyone can do to solve a dispute over things like math, science, or engineering is to test hypothesis and see what type of outcome occurs. If the data doesn't match the hypothesis, then the hypothesis is wrong and needs to be modified.

This same principle should be used to settle one of the biggest disputes taking place today, that being climate change, is to test the climate models present being used to show that we are all doomed by comparing them against actual data. Someone, specifically Larry Kummer, has suggest just such a thing.

Do you trust the predictions of climate models? That is, do they provide an adequate basis on which to make major public policy decisions about issues with massive social, economic and environmental effects? In response to my posts on several high-profile websites, I’ve had discussions about this with hundreds of people. Some say “yes”; some say “no”. The responses are alike in that both sides have sublime confidence in their answers; the discussions are alike in they quickly become a cacophony.


The policy debate turns on the reliability of the predictions of climate models. These can be tested to give “good enough” answers for policy decision-makers so that they can either proceed or require more research. I proposed one way to do this in Climate scientists can restart the climate change debate & win: Test the models! — with includes a long list of cites (with links) to the literature about this topic. This post shows that such a test is in accord with both the norms of science and the work of climate scientists.

Not just any data should be used. Use the unadulterated, unadjusted raw data, particularly that from what are referred to as “pristine” stations, meaning those sited where there is no influence by creeping development, parking lots, HVAC exhausts, or other man made sources of heat. There are plenty of them all over the world. Plug in the raw data to the models and lets see what they project. If they fail to project the actual results within some predetermined margin of error, then we know the models are junk and any predictions they make should be ignored. Some of this can be performed using hindcasting, starting the model projections at a date from 40, 50, 60 years ago or more and see if the projections even come close to predicting what actually happened.

My guess? They will fail miserably.