"I'm not entirely positive that causality can't run backwards in time"

Part of me thinks maybe, part of me thinks the concept is incoherent, as there's no logical way to be able to recognize backward causality, almost as if causality can only be forward by definition, but I haven't been able to fully wrap my head around the logistics.

Like, it makes sense for the past to determine the future because the future isn't defined yet. But what does it mean for the future to determine the past? It can't *change*, because there's only one past, so the future event had no choice either because it had to be such that it caused that particular past state. So you can't verify that the past was determined with a counterfactual (like "if X hadn't happened, Y wouldn't have happened) which I think is the usual way of defining causation. So at best you measure some kind of correlation between past and future states. But correlation alone doesn't imply causation, and if you assume causation you could just as easily assume that the causation went from past to future.

One possible way it could work is if the past wasn't measured yet, and you assume it's undetermined until you measure it, and you find you can determine what's measured by present or future choices/randomized events. But you can't *really* measure the past, because it's gone, you can only measure the present (everything we observe is necessarily in the present) and deduce that the present was caused by a particular past state. (For example, you can watch a video that exists in the present and deduce that what it depicts happened in the past.)

But, if that past event was undetermined and unmeasured thus far, and you can only measure the present, how can you prove that what you're measuring really was the past and not just something in the present manifested on the spot?

Though that still leaves room for highly counterintuitive measurements, for example, if you can determine what happens in a video recording that you haven't watched yet through present or future choices/randomized events, that's really weird.

(I say choices/randomized events because if the present/future event isn't randomized then you could suppose a direct line of causation from the past to the present/future is what caused the correlation. Maybe depending on the situation. And there's a lot of randomness/unpredictability/indeterminism in choices too.)