Then why in hell would anyone ever GM such a game? Seriously.
@Ovinomancer has answered for his part; I can only answer for mine.
Finding out what happens in RPGing is fun for me. There is the action and colour; in more dramatic games, there is that too; sometimes there is comedy; there is mystery and revelation.
There is an obvious resemblance between these pleasurable parts of the activity, and watching a film or (less so, I think) reading a book. Compared to a film it is slower-paced and (related but not identical) less well edited. But compared to
watching a film it is creative. And compared to writing a story on one's own it has all the fun and surprise and sociality of doing something with one's friends.
The preceding few sentences aren't any sort of attempt to explain everything about what is fun for me in RPGing, but try to point to some of it.
(When I play rather than GM the creative aspect is a bit less evident, but the intimate inhabitation of my character, and the emotional experiences that flow from that, are present in the way that they are not when GMing.)
you can't present any sort of mystery for solving or later reveal, you can't present the players a living setting that has things happen - both now and in the past - independent of the PCs and-or their actions meaning said PCs and their players are largely operating in a vacuum beyond the here-and-now, and if by "naturalistic extrapolation" you mean "if the PCs do x, then y happens; if they do not, z happens" then their actions (or lack of) have no future consequences. What's the point?
Someone has to put the mystery or secret in place, be it the GM or a player, for there to be a mystery to solve; and whoever that person is must by default know the solution or what the secret is. Ditto puzzles, riddles, or anything else where there's a clear but not-immediately-obvious answer: someone's got to put it there, and that someone already knows the answer.
These claims are not true. What I mean by that is that they are claims about what is possible when adopting a certain approach to RPGing, and I know from my own experience of RPGing using that approach that the things you say can't be done, can be done; and that the things you say are necessary, are not.
Presenting mysteries, secrets, puzzles
The way that I do this, when I GM, is to introduce a situation - an event, a NPC, an object, etc - which does not yet have an explanation (in the fiction) known to anyone at the table. (This is just what
@Ovinomancer said upthread.)
For instance, in a reasonably recent session of my Classic Traveller game the PCs found an ancient alien pyramid complex with a pendulum apparatus in it.
What is the pendulum for?, they wondered. Good question!
In the Burning Wheel game I GM, the players learned that the sorcerer Jabal - the nemesis of at least one PC, the employer of two others - was going to marry the Gynarch of Hardby, heself a powerfl mage. What is the reason for this wedding? Good question!
Presenting a living setting that has things happen independent of the PCs or their actions
The wedding above would be an example of this. Another example would be our Classic Traveller game, where the PCs encounter various vessels that are travelling for reasons that are not caused by the PCs: eg an Imperial armada attacking the world the PCs were on.
It's true that
the GM is not doing much imagining of events in the fiction that are not narrated, in some fashion, at the table but that sort of imagining is not
presenting a living setting.
Presenting naturalistic consequences
In my Burning Wheel game, the players failed some sort of check as their PCs were fleeing a tower carrying the blood and head of a decapitated sorcerer. I narrated an encounter with some guards. The PCs failed to persuade the guards that they were just innocently out for a night-time stroll. The guards took them into custody.
That sequence of events is completely naturalistic. And is established via the narration of consequences of failed checks: had the players' checks succeeded, they would have naturalistically escaped observation (first check succeeds), or duped the guards (second check succeeds) and then events would have headed into a different direction instead of the imprisonment that the PCs suffered.
In my Classic Traveller game, the a PC used psionic power in front of a NPC who had the conventional Imperial hostility to psionics. The reaction roll indicated that she was hostile; then a player succeeded on a check made to see if his PC was able to calm her down for the moment, and so she was calmed. Then the psionic PC ended up being placed in charge of the overall situation (as Imperial Overseer), which the NPC reluctantly accepted (I can't remember if there was a check made for this or not). Then when that same PC ended up temporarily incapacitated due to an attack by an alien creature, the hostile NPC (being a noble of the same rank as the PC) asserted her authority over the situation (this was a GM decision to establish a complication that followed from the incapacitation).
That sequence of events is also completely naturalistic.
The GM approach of
first make a soft move that signals a risk, threat or danger and
then make a hard move that follows through with some irrevocable consequence is formalised in PbtA games but is easy to use even without that formalisation. The PCs are out at night doing unlawful things: on the first failure signal the risk (they meet guards) then on the second failure follow through (they are taken into custody). A PC uses psionics in circumstances where this might cause ire, and the reaction roll indicates as much: that's the threat. A series of events ameliorate the threat for the moment (the NPC is calmed down; the PC is put into a position of authority over her). But then when the PC (and thus his player) comes unstuck, I as GM follow through: the hostile NPC takes charge of the situation.
The upshot (as I've posted in another recent thread) was an attempt by some PCs to break into the NPC's base, which failed; and a consequent trial, which was brought to an end by the PC blowing everyone and everything up, and the PCs then abandoning their position.
The bottom line
Presenting mysteries, presenting a living world, and establishing consequences
do not depend upon treating things written in the GM's notes prior to play, and not yet revealed in play, as constraints on framing and on action resolution. I've just posted a few hundred words of examples that prove the point.