upvote
>The MWI has no explanation for how this other infinity of possible states disappears into just two states. You still wind up with the collapse of the wave function--involving a lot of hand-waving or appealing to some nebulous property of consciousness.

The real issue here is the preferred basis problem, not an infinity of worlds: a single state vector can be decomposed many ways, and the bare math doesn't say which decomposition counts as "the branches." Decoherence picks it out: once the cat is entangled with its environment, the alive/dead decomposition becomes effectively unique and stable, through ordinary unitary dynamics. No collapse, no consciousness (that's von Neumann–Wigner, not Everett). MWI does have an open problem, but it's a different one: the Born rule, i.e. why |c|^2 should count as probability when every branch is real. Also, the normalization is |c1|^2 + |c2|^2 = 1.

reply
Sure, the MWI still has an open problem which is why it looks like there is a collapse of the wave function.
reply
No, it explains that well enough, it just struggles to explain why you would expect to see the collapse with the probabilities that you do (though, to be fair, most other interpretations don't presume to explain this at all, it's just a rule. MWI just reframes it in a way that makes it odd that there is a rule).
reply
AI comment
reply