> there's no consensus on why this happens other than hand wavy evolutionary biology stuff

This reminds me of a comment about chess - there may be certain abstractions, lines/strategic patterns with meaningful relations "in the big", but in the small, like with endgame tablebases, small differences in state have an effect on the outcome of the system that is difficult to explain/predict because it seems to be random.

Like looking into an extremely compact fractal phase space instead of more predictable/seemingly geometrical ones.