I'd be interested in seeing it.
Alrighty. Just remember, you asked for it!
The idea as I understand it with GNS "Narrative" play, is that the DM takes a back seat in terms of the narrative action, in the same way that in a gamist game, the DM takes a back seat in terms of the gameplay action. Hence it is considered a faux pas, at least among most D&D gamers I know, for the DM to unilaterally declare what the PCs choose to do, and unless the players are definitely okay with it, it is generally frowned upon to "force" things to end up where the DM wants them to be. (Even for folks who don't share my opposition to fudging and railroading, it is generally accepted that doing any of those things in outright defiance of what players are trying to do is...unwise, shall we say.)
In a GNS "Narrative" game, the players are the ones choosing the Value, the goal, the
purpose of play, and deciding how to deal with the Issues that result from pursuing that goal, with the GM acting only enough to frame the scenes sufficiently that the players actually do have Issues to deal with. In a GNS "Gamist" game, the players are evaluating their own performance by some metric of Score, e.g. surviving combats or collecting treasure etc., in order to obtain Achievement of some kind, e.g. defeating the opposition, saving the innocent dragon from the evil princess, what-have-you, and the GM acts only enough to ensure that the metric(s) of Score have actual meaning to the players and that the Achievements are motivating enough to pursue.
I break up GNS "Simulation" games into two categories, what are called "process" Sim and "genre" Sim by others. In GNS "Simulation" games of the "process" Sim style, the GM acts only so far as to establish sufficient Groundedness, rooting the imaginary sensory/perceptual experience of play as (metaphorically) tangible as possible so that it
feels like a living world that actually exists. The GM then turns the metaphorical crank of the world so that the Simulation of its future states proceeds (hence "process" Sim) logically and rationally from the initial data. From there, sufficiently comprehensive naturalistic reasoning on the part of the players should allow them to correctly predict and respond to the events in that world. In GNS "Simulation" games of the "genre" Sim style, the GM acts only so far as to establish a Conceit (or set thereof), the principal concept, theme, or tone that will color the experience of play, and to provide a milieu in which the players' Emulation of an appropriate genre, creator, style, etc. can occur.
I used the phrase "acts/acting only so far as to X" with a reason: if the DM/GM steps too far outside these limits, it tends to get a negative response from the players. A GM running a Groundedness-and-Simulation game that starts specifying too many rules outside of naturalistic reasoning will start to draw accusations of being "unrealistic," whether by injecting human-made
conventions like tropes (which are a Conceit-and-Emulation thing) or by being excessively
abstract in how the thing is handled (which is generally a Score-and-Achievement thing, as Scores are necessarily an abstraction.) A GM running a Score-and-Achievement game that expands the rules beyond what is needed, especially if so doing makes them particularly complicated and unbalanced, will draw criticism for weakening the
game just to make it (metaphorically) "look" nicer.
To be clear, here, I don't actually think that this means these game-purposes are necessarily mutually exclusive. I don't even claim that they're jointly exhaustive; there may be other game-purposes I have never considered. Instead, my point is there
is something of a "GM, stay in your lane!" element here, a "don't water down the X just to get some more (implicitly,
unnecessary) Y" attitude. And the type of "GM, stay in your lane" thinking that tends to show up in D&D discussion is of a GNS Gamist nature, or failing that, a GNS Simulationist nature (IME, almost always "process" Sim, much more rarely "genre" Sim.)
When these game-purposes are combined, I find that usually this is done in one of two ways: either the purposes involved are already reasonably amenable to blending (e.g. "genre" Sim is pretty friendly with everything except "process" Sim IME),
or the two purposes are engaged on different levels of play, and thus avoid getting in one another's way by being focused on different things (e.g.
combat being very Gamist, but
exploration being very "process" Simulationist--it's naturally difficult to do both things fully simultaneously, so they can coexist in their enclosed spaces.) If a game actually tries to combine two purposes in the very same activities, it does seem that this can lead to problems, as noted above, with players feeling dissatisfied by compromise rather than pleased by diversity.
Overall, I find "process" Sim/"Groundedness-and-Simulation" to be the most
picky of the four game-purposes I've articulated, because all three of the others generally agree that some kind of human-made convention is acceptable. Groundedness-and-Simulation wants to minimize all human-wrought contrivance, wants to match player reasoning to character reasoning as close as one can get to 1:1; there will always be
some contrivances, but they ought to be minimized as much as possible for this game-purpose. Again, this doesn't mean it is
incompatible with the others, but I find it is the one most likely to raise a stink about divergence.
Score-and-Achievement bends relatively easily because "succeed more" is an easy fallback Score and "do something worth remembering" is an easy fallback Achievement, and, well, there's a reason we call them roleplaying
games and not RP
puzzles or RP
performances--it's hard to totally divorce from having
some measure of evaluation and growth. Conceit-and-Emulation, my term for "genre" Sim, blends relatively well because genres can sometimes rest almost purely on aesthetics, and because "let's play a/an <X> game" where <X> is "supers" or "cyberpunk" or "wild west" etc. is just really easy for most people to conceptualize. Values-and-Issues, GNS "Narrativism," is a bit harder simply because it's relatively
new, but it can be done (with varying difficulty) if there's enough tools for the players to drive conflicts forward by themselves, rather than needing to rely on the GM to make it happen.