"Your decision theory should (almost always) be the same, whether you suppose that there is a 90% probability of something happening, or if it will happen in 9 out of 10 worlds. "

I STRONGLY disagree here. If you suppose there is a 90% probability of something happening this usually means that you haven't updated your priors enough to recognize that it actually happens in approximately 100% of worlds, and less frequently (but sadly, probably not 9 times less frequently) that you haven't updated enough to recognize that it actually almost never or outright never happens.

"Average utilitarianism suddenly looks a lot more attractive - you don't need to worry about creating as many people as possible, because there are already plenty of people exploring person-space."

If many worlds meant infinitely many people this claim would be quite plausible to me, but why should aggregation stop mattering just because there are bignum people?