Simon: But you're still proposing a rule in which if you have a world which splits into world A and world B, they have probability 1/2 each, and then when world B splits into B1 and B2, it changes the probability of A to 1/3 - until an unobserved physical process turns the probability of A back to 1/2. Seems a little odd, no?

That's an excellently sharp compression of the problem with observer-moment anthropics the way I'm trying to do it - the conditional probabilities aren't invariant when you compose or decompose events. If you're anesthetized during the split between A and B, and wake up and observe yourself in B, before B splits, it seems the prior probability of B is 1/2. If you're anesthetized until after B splits into B1 and B2, it seems the prior probability of B is 2/3.

Huh. The probability theory I'm trying to use here really isn't giving sensible results at all.