Well, pretty broadly, game balance (a vast topic all to itself), simplicity, speed of use, practicality, and genre-/trope-support. None of these things actually intersects with "simulation" as I had understood the term. Hence why, even though 4e's rules are quite good at producing what one might call "high fantasy action movie"
tropes, people who like simulationism tended to be very sour on it because it was not much concerned about rigidly justifying the physical process of daily/encounter powers nor the internal logic of Second Winds or Action Points.
For example, 4e's "square fireballs" that people liked to poke fun of (and, amazingly, sometimes with affection rather than scorn). A grid is already somewhat un-simulationist (since we know space doesn't actually exist on a grid), but 4e goes a step further by using Chebyshev or "chessboard" geometry, where diagonal distances are
equal to rectilinear distances. This is often cited as outright anti-simulationist, at least by self-identified pro-sim folks I've known. It certainly doesn't follow your assertion of simulating a type of fiction, at least not as far as I'm aware.
Yep. This seems nonsensical if you've misconstrued the purpose of the simulation. We are not simulating a world with rational biology for academic study. We are at best simulating heroic action fiction for entertainment. There's a strong argument that we aren't even simulating a heroic action fiction world - just the fiction itself. Heroic action fiction has dramatic highs and lows - one way we simulate those is by driving the character down and up in hit points.
I had been given to understand that that was what "simulation" usually referred to: attempting to generate rationally-expected results, as though it were a real place with rules analogous to the rules that govern our world (just a different set thereof). Hence why pro-"simulationist" folks really dislike things like monsters using different rules than PCs, even though monsters rarely get the opportunity to actually use the vast majority of rules players care about. Consistency across manifestations, and the ability to rationally predict how things will work based purely on a
qualitative understanding of what they are (rather than a
quantitative understanding of the actual rules invoked). E.g., Wikipedia describes it as, "Simulationism maintains a self-contained universe operating independent of player will; events unfold according to internal rules."
That "self-contained universe operating independent[ly]" thing doesn't seem to jive with your description of "simulating heroic action fiction alone, no world involved--that instead sounds like
narrativism, where the goal is to produce certain kinds of narratives (in this case, heroic action narratives in a high fantasy milieu) even if that requires nonphysical or even non-independent operations within the world (e.g. "fate points," "bennies," etc.) Now, perhaps I've had a bad working understanding of "simulationism," but it's...more than a little surprising that it took over a decade to learn that I had been fundamentally misunderstanding the term this entire time.
Well, certainly not a realistic simulation, but it works pretty well for simulating action movie combat, for instance.
See above. This sounds to me like dissolving any difference that might be had between "narrativism" and "simulationism," turning it into one big mass of "
performing a thing" rather than "
playing a thing." A conflation of "performance" in the sense of
acting and "performance" in the sense of
system architecture, e.g. "that was a lovely performance of
Macbeth" and "if we can solve this problem, we can increase performance by 25%."
Don't get me started. 
