Either they share narrative authority ("Here, create some fact that is true about this world that I can't overrule."),
Oh yeah, do that all the time. I run all
knowledge skills in a manner similar to how Burning Wheel's
Wises work, no matter the system. So most of the time when a player makes a knowledge skill check it's not a prompt for me as GM to provide more information, it can be, but usually isn't. In most cases it's to check the accuracy of a
fact the player has added to the shared fiction in a statement immediately prior to the roll. If the roll succeeds then the
fact the player has stated is true. If the roll fails then the
fact is mostly true (or sometimes false, but much less often as that is boring) with the details being slightly different. So, for example, a player might state that there is a Wizard's Tower in the nearby woods, then roll a Knowledge (Local Area) check. If the check succeeds, then there is in fact a Wizard's Tower in the nearby woods. If the check fails then something about the information the PC has is flawed. Perhaps it's not a Tower but a Hidden Grotto, that is still in the nearby woods, but must be searched for. Or perhaps there is indeed a Wizard's Tower nearby, but instead of being in the woods it's in the hills on the far side of the woods so a longer journey needs to be made to reach it.
Sometimes I literally just ask players to add things to the fiction, particularly if it is facts pertaining to aspects of the world directly tied to their PC.
or else they bow to established fiction ("This is already true so I can't overturn it."),
Well, once fiction is established it remains true. Once it's established that there is a Wizard's Tower located in the nearby woods, it will remain so unless circumstances within the ficition (the PCs burn it down; a demon teleports it into the Abyss) cause that fact to change.
or else they submit to the outcome of fortune ("The dice are in control.").
Sometimes the dice are in control. This is especially true in circumstances where conflict is happening. I as GM don't decide if an NPC successfully stabs a PC, the dice dictate the outcome of the NPC's attack.
As far as I can tell, you don't do the first two at all. Your theory is that nothing is true until you say it, which means you are never really beholden to established fiction.
I am beholden to the fiction, once it is established. That's were my whole "it doesn't matter if you prep it beforehand or make it up on the fly" comes from. You can have a thousand pages of prepped material, but, until that material actually becomes established as fact within the narrative, it's just as non-existent as something I haven't ad libbed into the narrative.
And in particular, you seem to make no differentiation between mechanics that are tied to the fiction or mechanics that have no relation to the fiction or which create the fiction. That leaves you only submitting to the dice, but only in the sense that in this exact moment the dice say the players fail or succeed, but you as a GM have full ability to interpret what that means ("A motorcycle comes into existence!"). This is from my perspective no real check on your ability to fully control the narrative at all. With no ways to share your narrative authority, you run a GM centric game in which you are in full control. The players give you idea prompts or they act as random number generators to prompt your story, but it is essentially your story. It could be a fun one, but agency as I see exists only as minor aesthetic of play.
Yeah, I think I failed to properly explain how I go about ad libbing narrative elements.
My definition of a railroad is, "Do the players have meaningful agency." Your definition of a railroad is "A preplanned series of encounters." I grant you that a preplanned series of encounters can be a railroad, but not that something isn't a railroad just because it isn't. I do think you probably are trying to give the players as much agency as they can have within your process of play, but from my perspective as a player that's not a lot. As I said, if I was inclined to railroad my players the easiest way to do that was prepare nothing and always just respond to what they do. Schrodinger's Dungeon is the most powerful railroading technique available.
I do agree that Schrodinger's Dungeon can be a powerful railroading tool. I actively strive to ensure that it does not become that.
Consider a case that we've already considered, picking from three chests to find the magic ring. This is a meaningful act only if the ring is already in one of the chests. If the ring doesn't come into existence until I open it, you are in the same position as a con artist who can put the shell into any cup he wants only after I choose. If I have no reason to believe the ring exists until I open the chest, it doesn't matter which chest I pick. I will always only be right because you decided at that moment if I was right. But if you at any time before I choose a chest write down where the ring is, then there is a sense in which my choice can be right or wrong. That is, there is now a sense in which my choice has meaning and your choice as the GM has been negated.
Okay, so, I would "write down" where the ring is by deciding where it is and then remember where I decided it was and not changing it's location on a whim. The only difference is that I wouldn't physically write it down on a piece of paper (I actually might as I take a buttload of notes during play). Also, if I were to ad lib the existence of the three chests I would also decide which chest the ring is at the same time.
I got so angry watching that game. I don't even like Wil Wheaton, but I felt so bad for him in that game. That constituted player abuse and was some of the worst GMing I've ever seen. I'm not sure what you picked up on because there were a lot of things wrong with that game, but one of the things I picked up on was just how random cloud cuckoo land the world they were playing in was. Any action could lead to any outcome whatsoever, at the GMs whim. I think the worst moment for me in the game was when they beat the bad guys trivially and then the GM improvised that the bad guys clothes animated and attacked them. This is straight up victory negation, where the GM feels his encounter turned out too easily and the PCs won to easily, and instead of going "good job" the GM fudges something to rob them of their victory ("More orcs arrive!", "The bad guy just got 30 more hit points!"). In effect, a roll that said "success" was turned into a roll that said "failure". That's one of the many problems with improvising things on the fly.
Yeah, I'm pretty sure that's the I watched too. I thought it was a great example of how NOT to GM a game! I felt sorry for Will, and there was another player too if I remember correctly. What a crappy experience that must have been, and a really bad way to promote TTRPGs IMHO.