I can make the random setting choices for a Bell experiment by taking two photons from parts of the night sky that have not been in causal contact since very shortly after the big bang (assuming everything was in causal contact right at the big bang - which is reasonable, but I suspect not provable).

So fine - the superdeterminism conspiracy is local iff it can then be carried forwards billions of years in time, ensure that those specially correlated photons were emitted towards the right set of monkeys orbiring this insignificant star, and so on.

Alternatively you could also make the settings choices based on what the Dow Jones finishes on tomorrow and whether or not you personally went to the toilet before it closes. If those really are conspiratorially correlated in just such a way as to fool us into seeing "Bell violations", then why aren't you rich?

If thats the price of "saving locality" I'm not having anything to do with it, and most other physicists feel the same. The superdeterminists are vocal not persuasive.

Of course there are many other challenges (why does science even work if I can never "really" do independent replications of experiments and so on.) Scott Aaronson has a good rant about it somewhere...