But how does bifurcation of consciousness happen and how does it diagonalize the state so that we never see a cat in a complex linear superposition of |alive> and |dead> at the same time? 1/sqrt(2) |alive> + 1/sqrt(2) |dead> should be an equally valid outcome of the experiment via purely unitary evolution.
Consciousness is not a thing that gets bifurcated. It is just a word for an information state that is aware. If you create two information states, you have two awareness states. That is not bifurcation any more than copying a movie and editing one pixel on one frame (or as a better analogy having an effectively infinite number of frames with every possible pixel combination some of which describe aware states).
I don't see how that would allow or require seeing a cat in a complex linear superposition. You just have one set of frames which describe a live cat and another set of frames that describe a dead cat.
I feel like this is a pretty reasonable question being met with non answers. And it's a shame bc I think MWI is the frontrunner as of 2022, but this is a pretty significant question with Many Worlds I had not considered before.
Well I stole it from Roger Penrose (same Penrose as in the title article) which I think he presented in either The Emperor's New Mind or Shadows of the Mind. I think its the latter book which has more concrete arguments like this diagonalization argument. Also, the idea that collapse of the wave function could be mediated by virtual gravitons or something like that once the superposition state got sufficiently "massive" (which is the genesis of the ideas in the title article here).
Both are worthwhile if you like thinking about these kinds of things.
I've generally found nearly zero people who have read those books and of them most people poo-poo it all, and don't really engage with some of the issues presented in them.
He may be wrong. Einstein was also wrong about the EPR paper. But it seems like he's trying to engage with the problems and do science rather than just provide philosophical justification.
The state and result of observation depends on the method of interaction, i.e. hamiltonian. In principle it's possible to observe the cat in a mixed state, we just don't use interaction that gives that result.
Maybe your copy in another branch even posted the same comment.