Are all the worlds in the wavefunction from the beginning of time or do they somehow spring out from one world? This is called overlap vs non-overlap (first discussed by David Lewis).

So, by "world" in this post I'll mean "basis sate for the universe." The basis is arbitrary, so what "world" means will still depend on how I'm choosing what "worlds" are - there's the energy basis, for instance, where nothing ever changes if you look at just one of those "worlds." But you can have animals or computers in your basis states if you want - they aren't energy eigenstates, so they change with time.

Anyhow, currently the universe is spread out over a very wide variety of energy eigenstates, which is a fancy way of saying that lots of stuff changes. If we only allow quantum mechanics (that is, strictly follow MWI), this spread over "energy-worlds" is how the universe has been since the beginning of time. But if we look at the exact same state a different way, you could just call the initial state of the universe a basis state, and then, lo and behold, the universe would have sprung from one world, and the distribution of worlds then changed over time. This way of looking at things is probably pretty useful for cosmology. Or you could use worlds that change over time but don't include the original state of the universe, giving you overlap again. This is what we do unintentionally when we choose worlds that have humans in them, which is also pretty useful :)

For overlap vs. non-overlap to get more complicated than "both are valid pictures," you'd need some model where there weren't any static worlds to talk about - this would be a change to QM though. Also, this does raise the interesting question of how complicated that initial world (if we look at it that way) was. It doesn't have to be too complicated before we see interesting stuff.

Anyhow, it's pretty likely I was too hasty in my mistake-detection. But meh, I rarely regret putting off reading things. And I only occasionally regret putting my foot in my mouth :)