Game dev also rewards people for applying the 80/29 rule effectively and you see less of that in commercial software.
In each game generation there’s a game that would be easy to write on the next or subsequent generation of hardware and is damned difficult to implement on the current one. Cleverness and outright cheating make it work, after all fashion.
80/29 rule is the paretypo principle?
Typo but yeah.
It reaches a dead end eventually. That's where we are, edge of speed, where the only mods left are aesthetics veering at photorealism.
The game simulation will get more detailed/granular as aesthetics dial down in perceived value. You can always go bigger/wider/more procedural/more multiplayer.
This is also why every hard problem eventually shows up — games are just simulation + interaction, and eventually everything that can be simulated will have some attempted implementation out there, struggling along. (For some reason, this does not appear to stop at “interesting” things to simulate — see all the literal simulators on steam)
The simulations have yet to release photo-realism in lieu of event-perception, where simulation parallels reality, but that's not really playable as a game, only as a view.