The thing that makes superdeterminism hard to believe is the same thing makes time travel plots where destiny is fixed hard to believe. The level of coincidence and orchestration required to make things play out in just the right way, despite determined humans trying to make them come out another way, becomes ridiculous.
The angle settings in a Bell test can come from basically anywhere. For example, you can point telescopes at opposite sides of the universe, feed the frames into a cryptographic pseudo random number generator, and use the output to decide the angles. For superdeterminism to be true, that would mean the initial state of the universe had to be such that millions of years later billions of stars on opposite sides of the universe would be in just the right configuration to make the SHA1 hash of a picture of them come out in a way that correlated in a simple way with the polarization angle of the photons you were generating locally. And then they'd also correlate for the next one. And the next one. And the next trillion after that. And also it has to be just right for all the other ways you can set up the angle selection.
I won't say it's literally impossible to satisfy that many constraints. There's a lot of degrees of freedom to work with, and only finitely many constraints before heat death. But it seems really really contrived to me. Like explaining why someone has rolled five hundred natural 20s with "Oh that's just how things were always going to play out" instead of "that seems suspicious. Let me see that die."
The angle settings in a Bell test can come from basically anywhere. For example, you can point telescopes at opposite sides of the universe, feed the frames into a cryptographic pseudo random number generator, and use the output to decide the angles. For superdeterminism to be true, that would mean the initial state of the universe had to be such that millions of years later billions of stars on opposite sides of the universe would be in just the right configuration to make the SHA1 hash of a picture of them come out in a way that correlated in a simple way with the polarization angle of the photons you were generating locally. And then they'd also correlate for the next one. And the next one. And the next trillion after that. And also it has to be just right for all the other ways you can set up the angle selection.
I won't say it's literally impossible to satisfy that many constraints. There's a lot of degrees of freedom to work with, and only finitely many constraints before heat death. But it seems really really contrived to me. Like explaining why someone has rolled five hundred natural 20s with "Oh that's just how things were always going to play out" instead of "that seems suspicious. Let me see that die."