Thanks for the links, I really respect engagement with argument. I see a lot of evidence that the planet has been/is warming, but still see lots of issues with

Thanks for the links, I really respect engagement with argument. I see a lot of evidence that the planet has been/is warming, but still see lots of issues with the attribution / prediction / catastrophe papers. Take your KPMG report: it has lots of predictions of catastrophic trends, but cannot be used to make an actual prediction since "he LtG scenarios were thus not meant to produce point predictions, but rather to help us understand the behavior of systems in the world over time." "Although the steepness of a scenario’s decline cannot be used for predictive purposes (Meadows et al., 2004), it can be said that BAU2 shows a clear collapse pattern" I believe the Lakatosian solution to the demarcation problem is best: judge a scientific programme by its progressiveness over time. Evidence of a progressive programme comes about through repeated confirmation of theory against evidence without adding arbitrary auxilliary hypothesis. I worry that some corners of climate change science are degenerate: when a model doesn't match predictions there is always an explanation using an auxillary hypothesis ("there are less fires than we predicted, but we now use a different metric. And our new metric is getting worse!!", "the record temps didn't appear in our model because we didn't include X, but will come eventually", "the coral didn't die because Y, but will die eventually", "the ice caps haven't melted yet because Z, but will eventually"). I'm not saying these auxillary hypothesis are necessarily wrong, just that they are used as excuses for wrong predictions. This years record heat was an opportunity for climate scientists to say: "see! We predicted this in 99% of our models, now you should trust us", instead we got "our models missed this, but because we didn't include X,Y,Z; so you should still trust the models." Even worse X,Y,Z have been known about for years so either the modeller thought they weren't important and was wrong, or thought they were important but didn't include the. Both scenarios worry me and feed my skepticism.