Challenging local realism with human choices

Maybe it's worth noting one other proposal here. The paper by Leifer and Pusey was published in Proceedings of the Royal Society A nearly a year ago, but for some reason an article about it just appeared and was picked up by Daily Grail. There's a clearer discussion at phys.org here:
https://phys.org/news/2017-07-physicists...uture.html
And an earlier draft of the paper is here:
https://arxiv.org/pdf/1607.07871.pdf

This provides an alternative way around Bell's Theorem, which would allow both locality and realism, namely for information to pass between the entangled particles by retrocausality. The authors take an earlier result showing that under a realist assumption, time-symmetry implies retrocausality, and generalise it by replacing the realist assumption with a different one which they call lambda-mediation. If I understand correctly, the idea is that there would be nothing in the laws of physics to forbid signalling from the future to the past, but only something in the state of the universe - analogous to the view that the thermodynamic "arrow of time" is a result of boundary conditions rather than time-asymmetric physical laws.

Of course, there's a strong underlying assumption here that signalling from the future to the past must be prevented somehow, because it is never observed. In other words, the experimental evidence for precognition is assumed to be bogus.