I think superdeterminism doesn't need to conspire to affect your choices (other than asserting they are deterministic, but it doesn't need to guide away from an

I think superdeterminism doesn't need to conspire to affect your choices (other than asserting they are deterministic, but it doesn't need to guide away from any particular choices in a conspiracy). It just needs to simulate both sides together on both side to know whether the measurements should correlate or not (say both sides have hidden state that resimulates the entire universe to see which measurement the other side would choose, this gets around it for Kochen-Specker/Conway freewill theorem at least, I'm not sure about for Bell's). As long as everything was in causal contact at the start of the big bang, it can let every particle have machinery inside that resimulates everything else to decide whether a given measurements randomness is correlated with another's, without invoking non-locality. With non-locality allowed it can just do it superliminally with threads of entanglement. Both are equivalent unless something can leak out of the resimulation other than how correlated a measurement should be with another, xor if info can be sent through entanglement other than the degree of correlation, or if the numbers turn out to be truely random or at least not be algorithmically random, like maybe Chaitin's constant.