This really doesn't make sense. He's treating all possible theories as if they're disjoint events in a single random variable, but most of them would be better

This really doesn't make sense. He's treating all possible theories as if they're disjoint events in a single random variable, but most of them would be better modeled jointly as separate variables, most pairs of which would be considered either independent or conditionally independent. It seems to me like this leads the math in a very different direction, one where even infinite theories don't drive any probabilities toward an asymptotic zero. So, what was the author's point? That Bayesian statistics are irrational? Or that they can be misinterpreted and misapplied in ridiculous ways?