Well, I personally don't like "hidden backstory". Clearly plenty of D&D players and GMs do.
This thread wasn't started to serve as a warning or caution. The interest was in analysis of play techniques - there's a common technique of pre-authorship, what is it for?
What I feel created some struggles early in the thread was that many answers offered to that question were metaphorical (eg "The players explore the world") and it took quite a bit of time and effort to get them rendered in more literal terms (eg "The players make moves with their PCs to trigger the GM to narrate to them a certain bit of information which is recorded, either literally or notionally, in the GM's notes").
I don't necessarily think that the hobby is better served by having less hidden backstory in RPGing, but I do think it is better served by actually recognising, in literal rather than metaphorical terms, how various techniques work and what sort of play experience they might deliver. I think this helps break down ungrounded misconceptions: eg that the action declaration "I search the study for the map" is different from the action declaration "I attack the orc with my sword". Of course in real life, looking for something is a very different causal process from trying to decapitate someone. But in RPGing, both action declarations expressions of desire as to the future state of the shared fiction (I found a map or I killed an orc), they can both be declared from a purely first-person RP perspective ("actor stance", to use some jargon), and it's possible to establish rules for resolving either that don't rely on hidden backstory-based adjudication.
So if, in fact, we are going to resolve them differently - the map one by referring to the GM's notes; the orc one by rolling dice - well, it's worth asking why we do it that way.
Whether or not they are two different things depends heavily on (i) what is pre-authored, and (ii) what the goal of play is.
If the pre-authorship is a whole lot of chests with gp in them and monsters guarding them, and the goal of play is to get lots of gp out of chests so as to earn XP, then the pre-authorship doesn't establish the narrative. White Plume Mountain is a well-known module that illustrates this.
But (to recycle some examples from upthread) if the goal of play is to find the map, and the pre-authroship is that the map is in the kitchen and not the study, then the pre-authorship does significantly establish the narrative. Likewise if the goal of play is to avoid arrest and conviction, and the pre-authorship is of the dispositions of the police, the attitudes of the magistracy towards bribes, etc: the attempt by the players to bribe their way out of trouble will succeed or fail based on that pre-authored content.
I assume that you are referring to player agency rather than the agency of the PCs. The agency of the PCs is a purely imaginary property of purely imaginary people, and can be interesting as part of the shared fiction (one of the PCs in my BW game is dominated by a dark naga, and that's an important part of the current situation) but is completely orthogonal to player agency.
When you say "the world is just the backdrop" that is a metaphor, because RPGing does not take place on a stage with a literal backdrop. If the action of the game is constrained by the pre-established fiction of Candlekeep, so that player action declarations either (i) fail, because the GM reads the book and sees (eg) that the officials of Candlekeep cannot be bribed, and/or (ii) consist to a significant degree in trying to prompt the GM to tell them stuff about Candlekeep so they can establish the parameters for feasible successful action declarations. Neither (i) nor (ii) exhibits a great deal of player agency over the content of the shared fiction.
It's true that saying "yes" increases the odds of player success to 100%. Saying no, though, reduces them to zero - so I'm not sure how telling "yes" or "no" based on pre-authorship is meant to increase agency.
In "say 'yes' or roll the dice" games, there are a collection of techniques used to sustain player agency: framing that has regard to the signals (around theme, tropes, character motivation, etc) sent by the players in the build and play of their PCs; narration of failure consequences that likewise build on those signals to contribute to new framing that continues to "go where the action is"; and of course allowing the players to get what they were going for if their checks succeed.
The reason not to say "yes" all the time is to allow for a dramatic rhythm of success and failure; it's part of what distinguishes an RPG from straight-out collaborative storytelling. The reason for the framing and consequence-narration techniques is to allow player agency to feed into the game even when action declarations are not succeeding.
The idea that there is no difference between this sort of RPGing, and the GM declaring an action unsuccessful ("No, you don't find the map" "No, there are no bribeable officials") on the basis of his/her notes, is just silly. [MENTION=6696971]Manbearcat[/MENTION] has made the general point with the reference to Calvinball; but examples closer to home will do. No one thinks it makes no difference to combat resolution that the GM decides by fiat whether or not the attack hits, rather than the player rolling the dice. Very few D&D tables treat the declaration of an attack and the making of a to hit roll as simply having the status of a suggestion to the GM; it's a move in the game. Likewise for finding maps - the difference between being able to make a move, and being able to make a suggestion to the only player who actually enjoys the power to make moves, holds just the same.
That's not an argument that finding maps and attacking orcs should be resolved the same way; but it is an argument that resolving the differently has very obvious consequences for the degree of player agency in one or the other domains of action declaration.
As far as the PvP issue is concerned, there are any number of ways to resolve that: opposed checks are one common one. But two game participants competing with one another on a more-or-less level playing field is quite different from the GM - who at least notionally is not a competitor, and who enjoys vastly more authority to establish the fiction in any relatively mainstream RPG - using his/her authority to stipulate an action as unsuccessful.
Chess vs Checkers, again.
You've chosen to focus your analysis on the endpoint declaration "I search for the map" and, in doing so, miss the larger aspects of the play that lead to that declaration. In player-facing games, that declaration is, as you note, a big deal and will definitively resolve the action of the map or introduce new challenges. But this is only because the scene framing allows this declaration -- ie, you're already in the right spot for that declaration to be effective. Any previous scenes were either not appropriate for that declaration and need to be overcome prior to being properly positioned for that declaration. This is glossed over in the narrow focus on the declaration. However, the same issue, that in the DM-facing game the proper fictional positioning needs to be achieved for the action to be resolved successfully, is focused on because the knowing of that positioning isn't open. However, it is open if looked at more broadly, the DM-facing game is open that you must overcome challenges to find the correct fictional positioning to locate the map. That multiple such locations may exist is part of that presented challenge. As such, the declaration "I search for the map" is not the same scene staking resolution that it is in more player-facing games -- it occupies a different position. In player facing games, the play is focused on achieving the fictional positioning necessary to stake finding the map as the outcome of an action declaration.
The GM controls the scene framing so that achieving this fictional positioning is difficult. However, once achieved, the player can stake the map and determine if the map is found or if additional challenge is presented prior to getting another chance. Depending on the nature of the challenges, and the results, the map may become unable to be found. This is pretty much the same as in the DM-facing game -- the objective of play is the same, to achieve the fictional positioning necessary to find the map. The DM sets pacing by placing challenges before that positioning, and too many bad outcomes from those challenges may result in the map becoming unable to be found (if the party dies, for instance).
The net difference here is how those challenges are staged and addressed. In the player-facing game, the challenges are largely random, based on the results of checks on action declarations. Success moves you quickly to the necessary positioning, failure delays your ability to achieve that positioning. But, then, the objective of play (finding the map) isn't fixed, it's also essentially random (with maybe some ability to affect the relative odds), with failure moving the objective further away again. In the DM-facing game, the challenges are fixed, and so are determinable with smart play and effective mitigation of risk. The declaration of 'I search for the map' isn't the crux of the stakes here, it's just part of the larger play.
The agency assigned to that declaration doesn't mean that player-facing games have more agency, it means that the agency is more confined to that declaration -- ie, that the agency that players have is more tightly bound to such stake-risking declarations. In DM-facing games, that declaration doesn't contain the same amount of agency, because the agency the players have is more diffused through how they approach the challenge of finding the map. Player-facing games have big declarations that can dramatically affect the play and fewer to no moves that aren't impacting play in significant ways. So the agency assigned to those big moves is conversely much larger and apparent within those moves. So declarations like 'I search for the map' do have much more agency involved than a similar statement in DM-facing games. This is because DM-facing games spread the agency around through many more, less individually important moves. Which room you search, for instance, isn't a move in player-facing games, because the scene set of the GM determines the available moves -- it either allows for the finding of a map or must be navigated to get to the scene where finding the map is a valid move. If the scene allows it, it's always a valid move, but you don't ever have the move where you pick which room to search. So, in looking at relative agency, you need to sum the total of the decisions made in GM-facing games rather than compare the "search for the map" move as equivalent moves that should contain the same agency. They aren't, but that doesn't mean one play-style has more agency than the other. Chess and checkers.
Can DM-facing games be easily abused to further restrict player agency? Sure. Calvinball as a concept (@manbearcat) is useful to describe this, but it's an abuse, not a mode of play. (For those not following, Calvinball is a game where the rules are made up on the fly so as to confound the other players and advance yourself. Also a reference to Calvin and Hobbes.) Player-facing games can become degenerate, too, although the use of Calvinball tropes are not something that's possible without serious distortion of the play concepts. The degeneracy in player-facing games are weak DMs that don't push consequences or frame useful challenges, player vs player sniping (intentional play to disrupt other players' goals), and player lock-in, where a player can dominate player to determine of other players. These aren't unique to player-facing games, but I find it distasteful to harp on DM-facing games with degenerate play examples of the kind you can easily claim do not exist in player-facing games (like Calvinball). That's cherry-picking flaws in play so that one looks worse than the other.
As for finding a map and killing an orc being functionally similar moves, I pointed out earlier the flaws in this thinking - it's chess vs checkers again and assuming that since such things are similar in your chosen playstyle that this is a universal truth -- it's not. Knight takes pawn is functionally similar to Knight moves without taking in chess, but both are different from jumping a piece in checkers vs moving a piece in checkers. Please, don't return to the 'fiction doesn't exist' argument -- I left off of that because it passed by, but it's still an incoherent argument when you're using fiction to inform the act of authoring new fiction (like trope and positioning restrictions, both things that exist in the fiction and are used to constrain writing new fiction). If you want to get metaphysical about the existence of fiction, we can, but it will be boring, and my entry question is "How can Mickey Mouse be a pop-culture icon if he doesn't exist?" That should, at least, inform me as to which mode of thinking on the anti-realist side of the argument you follow and allow useful counterpoints.