Wheel of Fortune and Pictionary and Crossword Puzzles and Portal (inferring patterns/relationships + integrating with knowledge-base after examination and interaction and then solving the puzzle) absolutely require skill. There are clearly people who are better and worse at these types of games
<snip>
These games rely upon deeply structured, deeply coherent, deeply integrated puzzles where patterns/relationships can be inferred after both (a) GM foreshadowing and (b) interaction with/prompting the obstacle for more information.
Counter to what has been expressed, there are a LOT of obstacles/aspects of the imagined space that haven’t been pinned down/firmed up/granularity rendered even in very high resolution sandboxes. And characters are going to invariably bump up against them a fair bit (like the afformentioned guard).
let's look at how this works in play.
The player decided he wanted to bluff the guard and said X to the guard. The GM is deciding after knowing what the player is attempting what details he will add to the guard and which of those are ultimately going to determine whether the player succeeds or fails.
Since those details were determined after the player attempted his bluff, the player had no concrete information to base his action declaration on. All the player could do was to try to guess before the action declaration what details the GM will fill in on the fly and ultimately use to determine the guards reaction and then try to frame the action declaration accordingly. That's not the kind of skilled play that OSR is about as I understand it. OSR is about attempting to use the established fiction to have your characters act in ways to guarantee you accomplish your goal. Things like, there's a pit trap we know about on the map, let's get the orcs attention and lure them into it and then shoot our bows at any survivors till the group of orcs is dead. In the bluffing the guards case, there's no fictional details prior to the 'PC bluffing' to base any decision on. So even if we resolve the guard bluff with no skill checks, it still won't be OSR skilled play.
Skilled play in this sense often involves interacting with say the leader of one faction of orcs in the dungeon in order to pit them against another faction or another type of creature (so forming an alliance or finding some kind of shared goal). So this doesn't just have to be about bluffing a guard: this can be about involved negotiations with a group of monsters. And that does take skill, and it isn't something that usually gets left to die rolls in a typical OSR session (mechanics can certainly come up; for example the GM, if he or she is uncertain about how an orc chief might respond could call for a roll of some kind, but ideally the orc chief is responding to the specifics of what the PCs are saying because that is what makes the interaction skilled in this way of playing.
I've bundled these quotes together because I think they all bear upon what I believe is a fundamental limit in classic D&D play: that once the scope of the fiction moves beyond a rather narrow and somewhat stereotyped scope - the dungeon, its architecture and geography which itself follows some well-known conventions (levels and all those imply; doors and staircases of various sorts; etc) - the notion that the player is cleverly engaging the established fiction becomes increasingly strained.Skilled play is leveraging the system to achieve player goals within the scope of the game. Doing an improv scene with the GM is not leveraging the system -- it's sidestepping it. That you can do this with skill is well and good, and I'm not about to argue it cannot be done with skill as improv acting is very much a thing you can have lots of skill in, but it's not leveraging the system. This is detectable because this doesn't interact with the system at all, it's entirely outside the game system. You have inputs into the improv scene from the system (maybe) and the outputs possibly feed back into the system, but this function of acting out with the GM and having the GM then decide what their character does here is not within the system of the game
The classic version of negotiating with an Orc leader to pit factions against one another itself relies on some pretty radical simplifications: that the leader is motivated by treasure (always a staple in classic D&D), or has some other point of leverage that can be learned from prior engagement with the fiction (discovering a clue in the dungeon, or picking up a rumour at the inn, etc). It's because the leader lacks nuance as a character that skilled play is feasible.
If the situation becomes as @FrogReaver descirbes (which is what I take @Ovinomancer also to have in mind) then we no longer have the player ascertaining and then leveraging the fiction. We're much closer to something like a 4e skill challenge or Burning Wheel Duel of Wits, except that the GM has the liberty at each point to decide whether or not the player's efforts further the PC's goals or not. To me, that really doesn't look like skilled play in the Gygaxian sense.