There's the issue of Fortune-in-the-Middle. Saving throws as described in 1e AD&D are explicitly FitM, and some approaches to 4e play are FitM too. As I understand it (and I might have this wrong as I'm not that up on narrativist play) the game mechanic that's in the middle of FitM can be used to support narrativist play, but it doesn't have to be. The game mechanic decides part of what happens in the game world. For many, probably most, players, what the mechanic describes - such as whether a saving throw succeeds or whether the Come And Get It maneuver works - is all that matters. My understanding is that for a narrativist, interpreting what that result means in the game world is an important part of play too - how does the hero escape, why do the antagonists rush forward to their doom?
The FitM approach doesn't necessarily give players metagame power, it depends on who has the power to interpret the mechanic. Maybe only the GM can interpret the results. I get the impression that Gary Gygax's concern is with justifying D&D's mechanics to critics, and giving D&D DMs some arguments to deploy against charges of lack of realism. FitM enables rpg participants to choose the most plausible interpretation of an abstract game mechanic, so it can be used for this purpose.
The main connection that Ron Edwards draws between FitM and narrativst play is that, in narrativist play, FitM preserves the player's conception of his/her PC - eg if you miss an attack, rather than narrating that as "I suck" - which you have to in RQ, say, because you can't narrate it as a parry by your enemy if the processing of the resolution never got to the parry stage - you can narrate it as "My powerful flurry of blows is parried by their equally awesome sword skills" - now instead of sucking my guy is so awesome that I'm in a duel with the best duelist in the country.
4e uses FitM to allow theme to emerge - eg when a PC goes down, we don't know yet what the "0 hp" means, because we don't know yet whether or not there will be a heroic recovery (from a warlord's inspiration, or a 20+ on the death save, or whatever). So the FitM allows the 0 hp to act as a prelude to/foreshadowing of either heroic recovery or tragic failure. I think that's another narrativst-ish deployment of FitM.
Ron Edwards identifies FitM as a common aspect of gamist play too - "the point is,
he says, "that Exploration as such can be deferred, rather than established at every point during play in a linear fashion." My feeling is that in a lot of Gygaxian play with pawn stance players, much of the exploration may never
be established ie we don't really care exactly what happened in that 1 minute combat round, just about the outcome (ie just as you, Doug, said in your post).
The dialogue between Sister Rebecca and morgan Ironwolf was used to illustrate the influence of alignment choice on player activity.
Sure, but that seems only to confirm my point. The influence of alignment choice on player activity is itself an expression of "actor stance" RPing. The fact that much play was pawn stance, and that some saw this as a problem that didn't fit well with the game's alignment system, is shown by the effort that writers of the time (eg Lewis Pulsipher in White Dwarf, Gygax in parts of the PHB and DMG) to insist that alignment matters in ways that go beyond clerical spell selection and interacting with intelligent swords.
Every time I see a poster in a paladin thread saying "Paladins have to be restricted by mechanical alignment or else paladin players will have their PCs kidnapping and torturing villagers at the drop of a hat" I see someone who has encountered pawn stance players, doesn't like them, but can't see any way of resolving the issue other than by wielding the stick of mechanical alignment against them.
Let's remember that there's a difference between narrative and story. Narrative is used to describe the scene (there are three orcs in the room) and story is what happens during (a battle ensues, dice are rolled, story emerges). Martial characters can interact with the story but they don't have the capacity to change or even shape the narrative (they cannot say that the orc are really chickens and the room is really a giant pit), only the story that emerges (they can choose what actions to take to fight the orcs, talk to them, etc). Casters have the ability to change and shape the narrative (those orcs really are chickens) as well as interact with the story (burn, lightening, ray beams).
I think this is the same distinction that I've drawn in these threads between "scene framing" and "action resolution".
"You see some orcs." "OK - I attack them." That's action resolution.
"You see some orcs." "OK - I polymorph them into chickens." That's scene-reframing.
The borderline can be murky, and depends heavily on tropes, mechanics and group expectations. For instance, in MHRP "You see some Skrulls." "OK, I teleport to another planet to get away from them." can be part of action resolution, because the game has mechanics to handle that sort of change of geographic location as part of the resolution of a single scene. On the other hand, in D&D "You see some demons." "OK, I Plane Shift to the Seven Heavens to get away from them." is scene re-framing rather than action resolution, both because the game lacks the mechanics to handle that sort of change of geographic location as part of the resolution of a single scene, and because the game lacks robust mechanics to regulate how much pressure the GM can place on a LG PC who Plane Shifts to the Seven Heavens, meaning that it can look like Viking Hat GMing to treat this as anything other than a "get out of jail free card", at least in the immediate term.
Another way to look at it is, narrative does not require dice rolls. It is a description
I think that's too strong a claim. A D&D wizard's teleport often requires a die roll. So does polymorphing the orcs into chickens (they get saves).
And not all action resolution requires dice rolls. Cast Cure Light Wounds in Basic D&D to unparalyse your ally who's been paralysed by a ghoul doesn't require a die roll; likewise a 4e paladin using Lay on Hands in the middle of combat. But both are action resolution rather than scene-reframing.
See, because I'm a fan of improvisational play, that kind of thing is toxic to me. One of the big rules of improv is "don't deny." You don't contradict or negate a previous person's contribution. You accept it, and move forward with it, and it becomes part of the scene. If it is established (by being said) that you enter a room full of orcs, then that's what happened, and your choices now are about what you do about that, not about the scenario itself.
But I don't think this means that only the wizard has the power of narrative change, because the wizard doesn't deny that the orcs were there, either. The wizard just changes them into chickens. Mechanically, this is roughly identical to the fighter killing them all (ie: it removes the entire threat from the scenario). I'm into letting fighters massacre more orcs. I'm not so into denial.
For at least some of those who are interested in "narrative otions" for non-spellcasters I think this somewhat misses the point. (I use the word "think" deliberately. It may be that I've not probably grasped your on-point point.)
I think everyone acknowledges that when a wizard uses polymorph to turn the orcs into chickens, that doesn't retcon the shared fiction (there's no denial, in your sense of that word). In the fiction, the orcs were orcs that got magically transformed into chickens. The point is that,
at the table, the player didn't actually engage with the situation involving the orcs. That situation was not explored. Rather, in practical terms it was rewritten. (In a movie, this would be a moment of light relief, as what looks like a threatening situation actually reveals itself to not be one at all.)
The analogue for a fighter would be a token, or a "kill em all dead kwik" skill roll, that allowed the player of a fighter to equally simply, in mechanical terms, declare "Nope, no orcs here. I've killed them all!"
There are RPG designs that exploit this contrast between engaging a scene and rewriting it. For instance, Burning Wheel and HeroWars/Quest (and probably other systems too) have simple resolution, in which a single roll from a player (whether of caster, or non-caster) resolves a scene and obliged the GM to frame a new one. The general advice in those systems is to use simple resolution to keep up your pacing and not get bogged down until something that really
matters comes along - then you switch to the complex resolution systems and we're not longer talking about rapid movement from scene to scene but the players actually engaging a scene that the GM has framed.
D&D seems unlikely to have this sort of simple resolution mechanic any time soon, given that 4e didn't fully embrace it despite coming the closest to these sorts of design sensibilities. The idea of a hit point threshold on spells like polymorph in early versions of D&Dnext seemed designed to make those spells lack their scene-reframing character - they become "closers" that you can only use after having engaged the scene for a bit and found out what it's about. But I think that's mostly gone now, hasn't it?