Back in 2003 I wrote:

Those global warming models remind me of the hyped up expert systems of the 80s. These were supposed to accumulate and organize information and make inferences from that information. The favorite application was medical diagnosis. Each time the system made a mistake, the developer would say, “I just need to tweak it one more time”. [These systems are still around and are being used in specific application areas. But the dream of the perfect diagnostic tool that could replace an experienced physician has not come to pass.]

That “one more tweak” syndrome hints at the major problem of these models. They are inductive. By modifying the assumptions and parameters you change the results. Eventually, you can fit it to the data points you have observed. However, there is no guarantee that the model’s predictions will hold in the future; each wrong result will require further tweaking of the assumptions and parameters. The question will always be – is the latest tweak based on hard science and real evidence, or did it simply produce the answer you needed?

A deductive computer model, on the other hand, closely replicates the subject being modeled according to the laws of physics. You can’t change the way the model operates by changing assumptions because you would be changing the laws of physics. Such models are used in automotive design, for example.

Theoretically, one could build a deductive model of the earth’s climate, but that would require modeling trillions of objects in faster than real-time. That isn’t going to happen. In the meantime, we should remain very skeptical of predictions and prescriptions based on computer models of the climate.

The blogosphere, led by Instapundit, has been very interested in a TCS interview with Freeman Dyson, a professor emeritus of physics at the Institute for Advanced Study, in Princeton. This passage reminded me of my earlier post on climate computer models:

Benny Peiser: In a Winter Commencement Address at the University of Michigan two years ago you called yourself a heretic on global warming, the most notorious dogma of modern science. You have described global warming anxiety as grossly exaggerated and have openly voiced your doubts about the reliability of climate models. These models, you argue, “do a very poor job of describing the clouds, the dust, the chemistry and the biology of fields, farms and forests. They do not begin to describe the real world that we live in.” There seems to be an almost complete endorsement of the world’s scientific organisations and elites of these models together with claims that they reliably epitomize reality and can consistently predict future climate change. How do you feel belonging to a tiny minority of scientists who dare to voice their doubts openly?

Freeman Dyson: I am always happy to be in the minority. Concerning the climate models, I know enough of the details to be sure that they are unreliable. They are full of fudge factors that are fitted to the existing climate, so the models more or less agree with the observed data. But there is no reason to believe that the same fudge factors would give the right behavior in a world with different chemistry, for example in a world with increased CO2 in the atmosphere.

Take it from me and Professor Dyson: Computer Models of the climate are dangerous when people believe the model rather than the unpredictable reality. Climate is always changing and there isn’t much we can do about it, except move.