A major rift in views is that I am pedantic on what outcomes a system formally prescribes (I rolled a damage dice, I got 5, I decrement your HP by 5) and what a DM decides (the troll attacks Jacky instead of Paula). I don't differentiate between big acts of deciding and small acts of deciding: it's all deciding. I don't care if the DM has a story in mind, or decides on the spot, it is still authoring or narrating.
Therefore were I to reject the possibility of skill on the basis of such deciding, there would be little of skill left in RPG. But I think (and observe!) that we can play RPG skillfully. I thus must either abandon the above position on deciding, or say instead that deciding can be done skillfully, can evoke skill, can elevate skill. Deciding can be skillful and can create opportunities for skill. Seeing as I believe all those things, for me my view is consistent.
I'm not sure it's helpful to frame a discussion of
skilled play in terms of
artful GMing, at least until a bit more has been said about the asymmetry of participant roles in a RPG.
Here's one way into that, presented under the heading "Doing Away with the GM":
You need to have a system by which scenes start and stop. The rawest solution is to do it by group consensus: anybody moved to can suggest a scene or suggest that a scene be over, and it's up to the group to act on the suggestion or not. You don't need a final authority beyond the players' collective will.
You need to have a system whereby narration becomes in-game truth. That is, when somebody suggests something to happen or something to be so, does it or doesn't it? Is it or isn't it? Again the rawest solution is group consensus, with suggestions made by whoever's moved and then taken up or let fall according to the group's interest.
You need to have orchestrated conflict, and there's the tricky bit. GMs are very good at orchestrating conflict, and it's hard to see a rawer solution. . . . In our co-GMed Ars Magica game, each of us is responsible for orchestrating conflict for the others, which works but isn't radical wrt GM doage-away-with. It amounts to when Emily's character's conflicts climax explosively and set off Meg's character's conflicts, which also climax explosively, in a great kickin' season finale last autumn, I'm the GM. GM-swapping, in other words, isn't the same as GM-sharing.
A GM may artfully frame scenes and manage the pacing within and between them. The GM may artfully manage the process whereby
suggested fiction becomes
established fiction - a lot of this is about action resolution, but not all of it: sometimes its just adding colour to a scene (eg the GM frames a situation, resolution is moving along, and then the player asks of a NPC "Is she tall or short?" and the GM stipulates an answer to the question). The GM may artfully orchestrate conflict (as in your example of deciding which player's PC comes under pressure in a fight scene).
But noticing this doesn't take us any closer to analysing
skilled play, I don't think. That is about
how players constrain the shaping and pacing of scenes (see eg
@Manbearcat's thread about skilled play earning a long rest); how players
oblige the GM to make
X rather than
Y part of the fiction (eg in my Green Knight game, the PCs were able to shed Dishonour points at the end of Encounters by having made choices that established this rather than this other outcome of the situation); and often the previous two things will be fallout of
how the players respond to conflict. (Not always, I think - a lot of dungeon crawling might rely on relatively low-conflict scenes, like the gelatinous-cube-in-a-pit trap discussed in the recent "fair trap" thread - but often.)
Certain approaches to how GM's frame scenes, establish fiction and orchestrate conflict are not really compatible with skil;ed play, because they make it hard for players to constrain and oblige, and they tend to make the fall out from conflict independent of the choices the players make when responding to it. Roughly speaking, these are the approaches that
@Ovinomancer has called "Force" not too far upthread; and these are the approaches that
@Manbearcat has called "participationism" or (perhaps a bit less neutrally) "rudderless system setting tourism".
So I don't think there is any rift. I think that Ovinomancer and Manbearcat (and most other posters in this thread) are well aware that RPGing requires people to make suggestions about the fiction, and requires a process to make these "true" (ie part of the shared fiction) - I don't think anyone dissented from Vincent Baker's remarks to this effect upthread. The point they are making is that not all GM-side processes of this sort are consistent with the exercise of skill on the player side.