I was responding to someone who was asking me about it. That's why.
I guess I can see how you see that, but that's not what I pulled from
@pemerton's questions. C'est la vie.
I don't agree that just because the players rolled a good check, that the world suddenly changes to make the loyal Captain flip on his boss. Now, if the Captain is disloyal or I haven't determined his loyalty, that's one thing. But my game world doesn't necessarily change just because the players rolled well.
Well, no, if the Captain is established in the fiction (more on this soon), then this would violate narration of results must be grounded in the fiction (and genre appropriate, but betrayal is, so that's not the conflict). The issue I have here is how your characterize your game world changing -- and I think the conflict is what we consider to be established in the fiction.
It doesn't matter whether the players are aware of the information or not, if it is something I've established then it's unlikely to change just because they rolled well. If I haven't established it, then it's absolutely open to a good roll like you've described.
And, here's that conflict. You treat things that you've decided, but that players do not yet know, as established in the fiction. I don't. Unless it's in the world openly, it's up for grabs. Now, I may very well (and do) make all kinds of notes for myself, but these are aids to help me quickly make decisions in play -- kind of defaults, if you will. But, I unless those notes make it into play, they're not established. So, unless I've already presented that, say, the Captain is loyal to the Burgomaster, then that loyalty isn't set in stone. Only once it's in play does it become part of the fiction. Anything I have in my notes is more like the Pirate Code.
What does this buy me? Verisimilitude - an increased sense that the world exists outside of and isn't simply being generated for the PCs. Sometimes they try an approach that won't work and bounce hard off it. I don't generally call for rolls that are impossible, so the players are aware when they fail at something that couldn't be accomplished.
Unsurprisingly, I'm going to disagree with you. Not that you get what you want out of your method -- I believe you do and that's great. But that verisimilitude is capably of being defined as you have or that my method doesn't generate it in equal abundance.
First, about your definition. The world doesn't really exist without the PCs -- if there are no PCs, there's no game, and you've just been writing a story. So, if you have a world, it exists because of the PCs. Now, I get what you're driving at, and that's that there's fiction in the world that exists no matter what the PCs do, but, at that point, you're still writing fiction you're just telling it to your players and they have no opportunity to change it. If they do have an opportunity to change it in play, then we're back to it having been created as a challenge to the PCs, which would be because of the PCs. I don't think that you can have a coherent definition that is 'exists outside of the PCs.'
Semantics aside, though, I don't see how you writing down secret notes that you then tell the PC generates a feeling of realness or complexity that cannot be created in play by following PC actions. For instance, the example
@Manbearcat presents has the Captain telling the Burgomaster a hard truth. If this was written in the GM's notes beforehand, it would be indistinguishable to the players form a situation where the GM invented it on the spot. And it involves things that aren't the PCs. There's nothing special about notes that increases a feeling of realness or depth in a game.
All of that said, though, I do fully understand there's a different feel to these two methods, at least to a GM who sees behind the curtain. There is certainly a different GM feel to an adventure that has good notes and plays out well compared to a game more discovered in play and completely unscripted. These feel very different to GM, so I understand your point that the notes method feels better to you (arguably, given how most enter the hobby, it's more comfortable and familiar than better, but that's a different discussion). However, and this is my point, the fiction created is hard to impossible to distinguish from each other. Verisimilitude is equally obtainable in each.
I have not suggested that everything with all possible NPCs must be prepared. What I have said is that for prepared NPCs this is how I handle it. I don't change the fiction just because the players roll well. In the case where the NPC is improvised, I do handle it much as you suggest, where the uncertain fiction is open to being determined by a roll. Even then, certain thing may be impossible, like convincing the ancient red dragon to give you its hoard. It doesn't matter how well you roll, the fiction is not going to reconfigure itself to that extent.
I think perhaps notes was the wrong tack. You clearly think that NPCs should react according to the GM's ideas about that NPC rather than leaving things to the dice. Perhaps you don't have a strong feeling about a certain thing and so leave it to the dice, but that doesn't change that if you do have a strong feeling, your intent for the NPC dominates. Which was my intent when I said NPCs are scripted -- I think this was referred to upthread, I forget by who, as GM-simulation. The GM controls the simulation of the NPC at all times, even if they decide to occasionally cede control to the dice the authority for saying how an NPC reacts belongs solely to the GM. I prefer letting things be more open to the mechanics and then fitting the fiction to match.
And, no, there's no way that you could convince an ancient red dragon to give you it's hoard (absent extraordinary circumstance). That violates both being rooted in the fiction and genre expectations. You keep circling back to the argument that not doing GM-simulation means that anything is available to a roll, despite being told this is not the case. At some point, I hope you listen and stop making that argument.