i actually think the climate-change people wreck their position by the thoroughness of their defense. so desperate are they to convince you and make you do something about it that they seek to render their position impregnable, which only makes it pitiable. the position is this: (1) you must distinguish between weather and climate, so a cold winter doesn't show anything about climate change. (2) but a warmer than average winter is a clear harbinger of our hotter future. (3) but a series of colder-than-average winters is caused by global warming. etc. the evidence is unequivocal! because the hypothesis is consistent with any possible observation, or more strongly would actually be confirmed by the observation that p and by the observation that ~p.

here's the problem with modeling, i believe: if winter is cold over area x, we back-fill the model to produce the phenomenon by or compatibly with global warming. that's why modeling is not the same as experimentation: you are producing your own data, which amazingly enough confirms your expectations. see a modeled environment is compatible with contradictions in a certain way: so we start with a set of data showing colder-than-average temps in the northern hemisphere and then see if we can produce a model on which that occurs compatibly with global warming. yes! but then say you took a set of data that showed temps in the northern hemisphere getting warmer. could you produce a model that would lead to that result compatibly with global warming? of course. but then the fact that we produced the first model is adduced as scientific evidence that the earth is heating up. not what we'd call a good methodology; it's more a rhetoric of persuasion i suppose.

in a real world, global warming cannot have both the effect of heating up the northern hemisphere and cooling it down at the same time. but in a computer model wherein time is traversed again and again and the world consists of the data fed in by the staff, there's no reason why it can't. but you need a lot more than that to actually show anything.