OK, but if you own the machine, you can just pick the outcome you want, or draw it from the distributions at random. We (observers inside the machine) cannot know the future of course.

Yes. I referred to the randomness that would prevent this, "once that is solved".

Guess I'm in the camp that eventually we'll find some model or discover something new, to discover what is behind the randomness, so it is no longer just random. But, yes, that is big IF.

Until then, with current theories, we couldn't do these calculations. They'd just be approximations accounting for some randomness.

Many-Worlds doesn’t contain or require any randomness.

I guess for whatever reason you don’t consider that to be the correct discovery?

Because it doesn't remove the randomness from our universe. It punts it to other universes. That is great, but doesn't allow us to predict things in our individual universe.

Or another way of saying it. We have 2 answers, they are determined. That is great, we know the 2 answers, one in each universe. Now the problem is we don't know what universe we are in. Now which universe we are in is random.

We didn't move the ball towards doing something useful in our own universe.

The 'universes' are loose abstractions, not a defined part of the theory; there's no actual hard distinction between timelines, in much the same way as coastlines don't have a defined length. They all blend into each other if you look closely enough.

That said, isn't the obvious answer 'all of them'?