[jfengel]

I don't think the problem with superdeterminism is the lack of free will, but with the way it doesn't really give you anything mentally to work with. It posits some early state from which everything could be deterministically extrapolated... except that state is both very complicated and completely hidden. It takes all of the probabilities and shoves them in a black box and says, "The answers exist, and they're in there. But you can't actually look in the box for the answers. You have to go do the experiment and wait for the speed of light to propagate the answer to you."

Like all interpretations, it's mathematically equivalent to any other. It's just a question of what helps you think about the problem, and I don't think many people find it very edifying. You can replace the box with a random number generator, which is at least small enough to fit in your pocket. The superdeterminism box appears to have been crammed full of untold centillions of answers... none of which are accessible beforehand.

If there were reason to think that the superdeterminism box were somehow smaller -- if it all really came down to just one random bit, say, that had been magnified by chaotic interactions to appear like more -- that would attract some attention. And I suppose it would be conceptually testable, by running Laplace's demon in reverse, except that that's not possible either from inside the universe.

So it doesn't really come as a surprise that superdeterminism falls behind MWI or Copenhagen or even pilot wave, because each of those hands you something that you can use to mentally organize the world. Superdeterminism just seems to hand you a catchprase, "As it was foretold in the Long Ago -- but which I just found out about".

[zazaraka]

Whats wrong with a pseudo-random number generator? You start the universe with 1 million random bits and then just iterate your function on them. How would we detect repetition at the 2^1 mil level? Maybe the universe would repeat itself after a while, but how would we know?

Superdeterminism also plays nicely with the simulation hypothesis. You seed the virtual machine with some randomness and the physical laws and then you run the simulation.

[jfengel]

I don't believe you'd need even a million random bits. It's conceivable that only a few random bits are actually required, and then let iteration take care of the rest.

There's nothing wrong with that. I just don't think people find it very useful as an organizing principle, so it doesn't attract a lot of attention.