Domino Valdano

And speaking of holes, in my explanation of how I view it, I think there may be a big hole in that I said we can know that being a Bayesian leads to more accurate predictions because you can run a Bayesian side by side a non-Bayesian in a controlled experiment and see that the Bayesian does better. But it seems like there is some circular reasoning going on there, since in order to conclude anything from that, you'd have to be a Bayesian in the first place. I think this may be the "problem of induction" rearing its ugly head. Although I'm not sure that my way of looking at norms is any more or less susceptible to that. In some ways, I think I'm a bit of a coherentist in that I know my brain works a certain way, and I have a coherent description of how it works, and everything I experience and think seems to hang together in a certain way... but whenever I try to explain what that way is, I have to start somewhere.

One of the problems I see with your description of belief, is that if the very idea of belief does really entail being a Bayesian, then I'm not sure how you can consider that a norm that is good or bad... how can linguistic tautologies be good or bad?