Copied from a chat where I tried to explain interpretations 3 and 4 a bit more:
I’m not sure what it means for a world to be more real either, but to the extent the idea makes sense in the many worlds interpretation of quantum mechanics (where some Everett branches are somehow “more real” or “exist more”) it seems reasonable to extend that to other mathematical structures. One intuition pump is to imagine that the multiverse literally consists of an infinite collection of universal Turing machines, each initialized with a random input tape. So that’s #3 in my post. I see #4 as sort of a fallback position, where if there is no fact of the matter about which worlds have more “measure” or “reality fluid” that can be discovered through philosophical reasoning or other kinds of investigation, you can still have UDT “add up to normality” (e.g., make a UDT-based AI behave in a way that a typical human would find reasonable) by hard coding a “prior” into its utility function, which can be interpreted as it simply caring more about some worlds more than others.
Copied from a chat where I tried to explain interpretations 3 and 4 a bit more: