Simulation doesn’t have to obey the rules. For non-scientists it is enough to “tell” them how it works without actually rendering it, cause they’ll never check. For scientists, you have to make their brain believe that all is consistent with complex computations all the way down.

Imagine that you know physics and math very good. But then simulation suddenly fails due to segfault and you realize that all your knowledge was just a gibberish nonsense and your work sessions and discussions were dream-like experience. Otherwise it was just a pretty dumb 3d simulation slightly better than a modern AAA game.

Edit: I mean, a simulation argument opens a huge can of worms, if you consider the perspective of a lazy simulation developer them-self.

While a simulation doesn't have to obey the rules, any given simulation is likely to resemble a system that does adhere to rules, since a simulation is indeed simulating something else.

Even if some external observer can pause, rewind, intervene and violate rules, the reason a simulation exists is to model something else.

This "something else" will have some kind of rules, and the goal of the simulation will be similarity to the actual realm external to the simulation.

Meanwhile, a lazy replication of a model by an unmotivated author would likely impose finite quantities upon scales of interaction, since a simulation won't be able to recreate a real time version of something larger than the external reality itself.

We wouldn't notice a lagged simulation that tries to consume its footprint, but it's probable that an artificial creation would impose caps on aspects of a system to prevent runaway reactions that produce useless simulations. To us, those sorts of limitations would resemble extra physical laws, and while incontrovertible to the simulated entity, such limits might confront intuition in strange ways.