pemerton said:But to have another go at it: if the rules are acknowledged as a metagame construction for resolving the game (ie the actions that players announce on behalf of their PCs) then there is not the least reason to suppose that the people of the gameworld could infer to those rules via experimentation and observation. That experiment and observation would tell them what the ingame causal processes are - and ex hypothesi the rules of the game are in no fashion a model or simulation of these.
Because the players WILL apply the rules as if they were the laws of physics, if they have any sense. If there's a cheap way to create magical long distance communication, the PCs will go into the telegram business, even if no one else in the world ever thought to do so before. If the players know minions die from one hit point of damage, they will rig traps that do minimal damage and set them off with themselves in range, knowing the minions will be fried and they'll be only slightly harmed -- and they'll ask why none of the locals ever thought of using very weak bombs to do the same thing. (City guards in 4e aren't minions, after all...a city guard with a 1d10 burst 2 explosive could clear out a dozen orc drudges and be barely scratched himself.)
In short, if there's an obvious disconnect between how the players can manipulate the world, using their knowledge of the rules, and the way the other people living in the world have manipulated it in the past, suspension of disbelief is ruined and it becomes much harder to care about the world or if you're saving it or not -- it's just too palpably unreal. (And the people in it are clearly idiots for not thinking of these things when they're obvious to the players after a short period of play. Who wants to risk their neck to save idiots?)
Asking players to ignore the implications of the rules in the name of obeying the spirit of the rules is asking them to behave like idiots and deliberately handicap themselves. Few players do this willingly.