The answer to both of these questions ultimately depends on the prior. But your procedure doesn't care about the prior - it leaves the user to sneak in whatever prior is their favorite. Thus, different users will sneak in different priors and get different answers.

Yes, of course, but that's fine; I'm not claiming any particular prior. What I am saying is that the prior is over possible worlds not observer moments, just as it is not over planets. I refuse to assign probabilities between observer moments, and assert that it is entirely unnecessary. If you can show me how I'm nonetheless assigning probability between observer moments by some underhanded scheme, or even where it matters what probabilities I sneak in, go ahead, but I'm still not seeing it.