The problem of evidence is only one of the problems: there is also the problem of hypotheses.
The fact that Bayes can be shown to work in toy problems involving dice and coins, but that doesn’t mean that it can work in cases where the hypothesis space isn’t finite or known—and it is neither in science. Realistic Bayesians dealing withr p realistic problems wil not be able to .think ofof every hypothesis. Newton didnt think of General Relatively It is no good having the right rule to update a hypothesis on evidence, if you don’t have the hypothesis. Realistic Bayesians need to exchange information about hypotheses in order to and up for their shortcomings.
The problem of evidence is only one of the problems: there is also the problem of hypotheses.
The fact that Bayes can be shown to work in toy problems involving dice and coins, but that doesn’t mean that it can work in cases where the hypothesis space isn’t finite or known—and it is neither in science. Realistic Bayesians dealing withr p realistic problems wil not be able to .think ofof every hypothesis. Newton didnt think of General Relatively It is no good having the right rule to update a hypothesis on evidence, if you don’t have the hypothesis. Realistic Bayesians need to exchange information about hypotheses in order to and up for their shortcomings.