In Newcomb's problem an agent picks either one-box or two-box and finds that no matter which option they picked, a predictor predicted them in advance. I've gone to a lot of effort to explain how this can be without requiring backwards causation (The Prediction Problem, Deconfusing Logical Counterfactuals), yet now I find myself wondering if backwards causation is such a bad explanation after all.
Unfortunately I'm not a physicist, so take what I say with a grain of salt, but there seems to be a reasonable argument that either time or its direction is an illusion. One prominent theory of time is Eternalism in which there is no objective flow of time and terms like "past", "present" and "future" can only be used in a relative sense. An argument in favour of this is that it is often very convenient in physics to model space-time as a 4-dimensional space. If time is just another dimension, why should the future be treated differently than the past? Nothing in this model differentiates the two. If we have two blocks X and Y next to each other, we can view either X as the left one or Y as the left one depending on the direction we look at it from. Similarly, if A causes B in the traditional forwards sense, why can't we symmetrically view B as backwards causing A, where again if we viewed it another way A to B would be backwards causation and B to A would be forwards causation.
Another relativistic argument against time flowing is that simultaneity is only defined relative to a reference frame. Therefore, there is no unified present which is supposed to be what is flowing.
Thirdly, entropy has often been the arrow of time with other physical laws claimed to be reversible. We are in a low-entropy world so entropy increases. However, if we were in a high-entropy world, it would decrease, so time and causation would seem to be going backwards (from our perspective). This would seem to suggest that backwards causation is just as valid a phenomenon as backward causation.
I want to remind readers again that I am not a physicist. This post is more intended to spark discussion that anything else.
(Another possibility I haven't discussed is that causation might be in the map rather than the territory)