Iain below links to a very interesting post on the resistance on so much of the serious non-specialist scientific press (e.g., New Scientist, Scientific American, etc.) to attempts to conduct falsification tests of various claims of climate science.
Falsification testing is a crucial element of establishing scientific certainty. Any “science” that has a very limited willingness or ability to conduct such tests should be expected to have less certainty in its claims that what we typically think of as science.
I think that one of the underlying problems with the whole “science says” argument when it comes to climate predictions is that they resist falsification tests. It’s not only that journals resist reporting them, but they’re also hard to design and execute.
In classical physics you can normally isolate the phenomenon under consideration pretty well. For example, a scientists can drop a ball in a near vacuum to test F = MA. In various “systems science” fields, where the interaction of the parts is essential to the theory at hand, this isolation becomes more difficult, but doable if the system as whole is small enough. Think of needing to test a complete airplane wing in wind tunnel, or more realistically in a wind tunnel simulator that has been in turn subject to repeated falsification tests versus a real wind tunnel so that engineers can have confidence in its results. Now scale the system up from a wing to the whole Earth, so that it is almost impossible to really conduct a test. Now further exacerbate the problem by making predictions of long-term change that can’t be tested over short periods (climate vs. weather), so that even if you wanted to run a single falsification test it would require something like 30 – 40 years (the general estimate of climate modelers for how long it takes separate signal-from-noise in a climate model prediction).
At this point, you have to ask whether this is really the same things as, say, particle physics. It seems to me that there is a continuum of certainty that can be expected from various different means of theory building and testing that share the label “science”. Classical physics is at one end and climate science at almost the other. Unless you want to add political science, economics and so forth to this spectrum, in which case these would obviously provide even less certainty than climate science. In fact, I think that climate science shares many methodological constraints with economics (one world, systems perspective, etc.). The big advantage of climate science is that the atomic unit of analysis for climate science is the molecule, while for economics it is an even trickier thing called a human.
In effect, climate scientists are trying to borrow the historical record of amazingly accurate predictions made by physical scientists and apply this prestige to their pronouncements, when there is a lot of reason to think that “climate science” will have a hard time ever replicating the predictive accuracy of physics.