Spacetime as a Neural Network

Speed of light = speed of causality = computational constraint of the simulation

Wave function = preference

Point Particles = choice

Gravity = computational lag

During interactions when preference -> choice if one dimension of interaction takes longer to compute than another, instead of waiting the simulation favors the delayed dimension. (If this could be formalized it could integrate GR with quantum, but requires new math currently over my head)

Dark matter = hidden (emergent) complexity via long distance coupling, where force drops off from 1/r^2 to 1/r (fitting MOND).

This assumes that emergent complexity with higher coupled degrees of preferences is harder to compute than decoupled wave functions. It also assumes the universe is extremely computational efficient but has limits.

I've been trying to imagine an experiment that could falsify this. Lots of ideas but none are good enough to share yet.

I think the biggest issue with this theory is the current interpretation of the CMB, which I'm still trying to deeply understand.