Pull the ol' entity system and only simulate things "close by" (or raise the granularity of simulation for things further away). What do I care about a metane m

Pull the ol' entity system and only simulate things "close by" (or raise the granularity of simulation for things further away). What do I care about a metane molecule in Alpha Centauri but its contribution to a bigger, planet-spanning or galaxy-spanning effect? Same idea. Edit: this has an obvious issue: if (when?) humanity (or rather the most ubiquitous "observer" species, whatever it is) "saturates" the cosmos, these optimization opportunities will get rarer and rarer. If it leads to localized "slowdowns" of physical phenomena then we'll have positive evidence of the simulation hypothesis.