Friday, July 04, 2008

Crooked Timberite and global warming zealot John Quiggin inadvertently asks a very important question: "I don’t see how AGW [anthropogenic global warming] differs [from] other examples like mainstream medicine v homeopathy and AIDS reappraisal, evolution vs creation." In other words, what is the difference between bucking the scientific consensus regarding global warming and, say, doubting that HIV causes AIDS?

The answer is quite simple, really, and has to do with the concept of parsimony in scientific theories. Pace Quiggin, what makes the HIV theory of AIDS compelling is not the overwhelming consensus of the scientific community--which is routinely wrong about all sorts of things, after all--but rather the difficulty of coming up with an alternative theory that explains the documented spectacular correlation between HIV infection and AIDS symptoms, as well as all the other evidence amassed in favor of the HIV theory. Similarly, the problem with homeopathy is not that mainstream medical opinion discounts it, but rather that accepting it requires reconciling its effectiveness with the entire body of physical, chemical and biological evidence which argues for its uselessness. (Creationism actually has the reverse problem: it's so parsimonious that it would easily "explain" anything--that is, absolutely any set of phenomena that might be observed--and therefore explains nothing.)

Now, I'm no expert on climatological research, but my impression is that the consensus predictions of global warming rely on elaborate computer models that incorporate and combine all sorts of influences on climate, whose likely effect and interaction are estimated using elaborate statistical methods applied to past data. Given the complexity of these models, I would be absolutely astonished if it weren't possible to introduce extra factors, or alter the behavior of existing ones, in a way that radically changes the predicted outcomes without creating any significant inconsistencies with the known inputs. In other words, the exceedingly complex theories embodied in the current models don't offer much parsimony, and hence one can likely embrace alternative models with significantly different outcomes that are very nearly as parsimonious as the widely accepted ones.

Now, perhaps I'm wrong, and the models are constructed so generally as to rule out any such alternatives. If so, though, then I would expect the claims for the models' predictions to be much stronger than I have seen--something along the lines of, "the predictions are scientifically irrefutable", or some such. More likely, they are nothing more than "best current guesses"--valuable, of course, but hardly conclusive. And skeptics of their predictions should certainly not be lumped together with doubters of various more established, more parsimony-generating scientific theories.