> How do you look at that description, and not conclude that according to that argument, Bayesian reasoning is just strictly wrong?

I believe you're conflating epistemics with decision theory. Sure, the measure of all minds experiencing your current mind-state may be dominated by Boltzmann Brains, with observations that do not correspond to any local state of the world, and which will dissipate momentarily.

But, since your decisions as one of those BB's have no effect, you should make decisions based on the fraction of minds-like-you which are living in a persistent world where those decisions have effects which can, in principle, be predicted.