I'm not sure refuting is the correct term. I think the most fundamental problem for climate science is that everything is based on models, and so nothing is fal

I'm not sure refuting is the correct term. I think the most fundamental problem for climate science is that everything is based on models, and so nothing is falsifiable. This means that nothing is refutable. Everything is based on models which are retrofitted the past data. This means they, generally, do an effective job of mapping to old data (as that is what they're created on), and then extrapolate from those old data into the future. But the big problem is that those future results are generally plagued by very poor rates of predictability. But since there's no falsifiability, it doesn't matter. There are major examples in both directions here. In the overstating direction, between 2001 and 2010 the warming slowed to 0.11C per decade. [1] For the 17 years before then was 0.17C. Because the models just map past data to future, nobody predicted this. There is of course a matter of long term vs short term. But at the same time, I think it's important to ensure we don't use this as never-fail defensive tool. Everybody missing a decadal long period of dramatic change is something that should not be ignored and I think we've been disconcertingly quick to marginalize. In the other direction many climate models underestimated the rate of antarctic sea ice melting. This is, paradoxically, then used as a defense of the models. 'See? We were just being conservative. The sea levels are rising even faster than we said!' But this is again a very disconcerting form of handwaving, that also seems to indicate severe bias. Something happening faster than a model predicts is no better than it happening slower than the model predicts. What it means is the same in both cases - the model is broken. [1] - https://www.scientificamerican.com/article/did-global-warming-slow-down-in-the-2000s-or-not/ https://www.scientificamerican.com/article/did-global-warmin...