Randall, that was an interesting read. My interpretation of this paper is that it picks up once fundamental models of climate have already been built and achieved ‘consensus’ adoption. If you start with a well-defined model, you can assign probabilities to the contributions of different potential causes. I particularly like the distinction between ‘necessary’ and ‘sufficient’ — global warming may be necessary for a particular heat wave or storm, but itself isn’t sufficient. It would be helpful if we all thought in those terms.
In writing this post I tried to go through some of the papers on counterfactual analysis, but I found them to be fairly dense, mathematical, and insufficiently broad to answer the kinds of questions I was asking. In most areas of science, we lack the maturity necessary to assign probabilities in a precise way. If a new study linking carbs to weight gain comes out, what am I supposed to think? Does this change the probabilities? What model do I use to update by probabilities? My intuition is this is outside a mathematical framework — we are not ready to assign an algorithm. There are just too many unknowns. Am I wrong in this assumption?