1 & 2 are probably a result of a number of different things, but the reason for doing 1 or 2 aren't why consistency matters, consistency matters because people doing 1&2 express concern when healing surges produce inconsistencies when pursuing 1 or 2. The problem healing surges create for players doing 1 & 2 is almost always an issue of consistency (which results in strained believability).
Yes, but the
reason they get this consistency problem is because, by doing 1 and 2, they are treating metagame rules as if they were process simulation.
This is why I think the simulation issue is more fundamental than the realism issue - not that I'm really trying to persuade you of this - it's obviously something on which reasonable minds might differ! - but to try and articulate a different perspective.
But, if I understand your use of process simulation (and I am not sure I do), we aren't talking about people treating the mechanics as such. We are talking about a basic desire for the game not to disrupt willing suspension of disbelief. This is very different from wanting detailed simulation of reality.
By "process simulation" I don't mean detailed simulation of reality. I mean the action resolution mechanics - the actual things the players do, like generating random numbers and applying modifiers from their character sheets - corresponding in a fairly tight way to events in the fictional gameworld.
In D&D, for example, the die roll to hit is traditionally seen as corresponding to something in the fiction - the PCs swinging of the sword or pulling of the bowstring, for example.
But not all action resolution mechanics are process simulation. In 1st ed AD&D, for example, Gygax states expressly in the DMG that a successful saving throw might correspond to finding a niche in an otherwise bare rock face - in this case rolling the successful saving throw doesn't really correspond to the PC doing anything in the fiction, but rather permits the player to narrate something that otherwise couldn't be narrated - namely, his/her PC's hiding behind a niched to avoid the dragon breath.
As CrazyJerome has explained here and elsewhere, hit points have been fuzzy between process simulation, and metagame. 4e pushes them strongly in the metagame direction. All the consistency problems with healing surges, though, come from treating them (and saving throws for dying) as process simulations.
If the above attempt to distinguish species of mechanics didn't make sense, here is a
passage from Ron Edwards that is what I personally have in mind:
In Simulationist play, cause is the key, the imagined cosmos in action...
Resolution mechanics, in Simulationist design, boil down to asking about the cause of what, which is to say, what performances are important during play. These vary widely, including internal states, interactions and expressions, physical motions (most games), and even decisions. Two games may be equally Simulationist even if one concerns coping with childhood trauma and the other concerns blasting villains with lightning bolts. What makes them Simulationist is the strict adherence to in-game (i.e. pre-established) cause for the outcomes that occur during play.
The mechanics - the things we do at the table to play the game - model ingame causal processes - hence (assuming I'm not misunderstanding CrazyJerome) "process simulation".