[lukifer]

One thing I've wondered about is the assumption that every possible universe exists independently, with an infinity of entire universes splitting at every fork (n states per picosecond per particle, compounding!); it seems to me that shared state(s) and a long-tail of micro-diffs (like git / Merkle trees / etc) could model the same phenomenon more efficiently/parsimoniously (disclaimer: not even slightly a theoretical physicist). This could potentially allow sufficiently overlapping states to merge, or exactly opposing states to cancel, yielding an equilibrium amongst the sea of micro-probabilities, converging either to a singular universe, or a much smaller patchwork of stable universes (either separate, or intertwined).

This parsimony question also overlaps the Simulation Hypothesis: the fact that most physical laws resolve at macro-scale, by aggregating probabilities, and only break down into odd behavior (where a particle seems to care whether or not you're looking), smells to this programmer like an optimization hack. :)

Just started reading Donald Hoffman's "The Case Against Reality", which posits some very bold answers to this question, theorizing that there is no reason to assume that base reality is anything like our perceptions (subjective or scientific), any more than there's a reason to assume the perception of a desktop icon has a direct relationship with its filesystem implementation. The implications of a "consciousness-first" rather than "matter-first" explanation of existence are staggering, and it isn't hard to set up a model in which that explanation is more parsimonious than Multiple Worlds.