But whether or not something is important for informing a decision has nothing to do with whether it is part of the world or not. It has everything to do with tailoring to the gameplay of the player. Something which has been consistently seen as unacceptably "gamist" and fundamentally incompatible--indeed, actively antagonistic--to "sim".
Regardless though, this blows a MASSIVE hole in the alleged bright-line, hard-binary distinction claimed by a large number of people in this thread, AIUI including yourself. That is, this exception now allows for crafting an enormous amount of the world both (a) only after rolling, and (b) specifically because of rolling, and in particular, things that explain why failure occurred, but which are prior to the action and indeed part of the world itself, not part of the action.
Your action cannot create crumbly handholds. That kind of separation has been used, repeatedly, by numerous posters in this thread, as the key thing which differentiates "traditional GM" sim-based play from other things: actions not only do not, but cannot result in something in the world being revealed to be true when it was unspecified before. And yet only now, long after this alleged bright-line distinction, this extremely hard binary separation between "X may happen" and "X may not happen", we get exactly the opposite. Now, a player's failed roll can in fact be the thing which causes the GM to narrate that there were crumbly handholds in the otherwise solid wall. It's not the world causing this. It's not conclusions drawn from or extrapolated from the world's established contents. It's specifically a rule, a mechanic, an abstraction telling us that failure has occurred, and then the GM developing some new fact about the world which is the thing that caused the failure to occur.
A fact about the world is only established after the player rolled a failure. Had the player not done so, no such fact would be established.
I was told, many, many, many, MANY times, that such a thing is utterly verboten. Now you are telling me it's not only okay, it's commonplace and most "traditional GM"-preferring, sim-focused players will in fact go with that no problem.
This calls into question the entire alleged distinction between this approach to play and the things to which it was (allegedly) contrasted, like PbtA. Because now facts CAN be established after the fact, in response to a rule-adjudication, which invert the causative order (action occurs -> rule is applied -> failure results -> the in-world cause of the failure is declared, NOT action occurs -> the in-world cause of failure is determined -> rule is applied -> failure results), as was explicitly required by multiple posters, IIRC including yourself.
In the example given, two contrasting things were described. One was claimed to be of a PbtA-like origin (despite coming from someone who doesn't play such games, doesn't like such games, and admittedly knows little about them beyond this), the other was claimed to be the "traditional GM" sim-focused type result.
The first, allegedly PbtA-like result was that the failure reveals that the whole wall was always crumbly to begin with, even though previous fiction had established that it was sturdy. This specific overt unrealism, where a past fact is simply outright negated by a new fact established because of a roll, was claimed to be utterly unacceptable to the "traditional GM"-preferring, sim-focused player. (I will address why this example was deeply flawed in a moment; for now, simply know that it's a jaundiced and mostly false characterization.)
The second, "traditional GM"-style sim-focused result was that the failure reveals that the specific handholds the player used were more crumbly than the rest of the wall, which is otherwise sturdy and climbable, the PC just picked handholds unwisely. Despite this also being a point of unrealism--only establishing the causative agent of failure after the action has completed, thus inverting the causal chain--this has been defined as not merely acceptable, but maybe even desirable.
In other words, the only difference between the two I can see--because both still involve the (previously-claimed-to-be utterly unacceptable) inversion of the causal chain and establishing failure-causing determinations after the action has been completed--is that the first establishes what one might call a "big" fact, while the second establishes what one might call a "small" fact; the first has a fact which contradicts prior established fiction in an extensive way, while the second contradicts prior established fiction only in a narrow way. Hence, it is not contradicting prior established fiction that is the problem; it is not the inversion of the causal direction that is the problem; it is not rules inducing changes to the world that is the problem; because both paths do that. It is only and simply that certain degrees of fact-establishment are "too much", all of which comes down to the arbitrary (and I do mean arbitrary, as in capricious) feelings of the people at the table. The exact same determination, in the same campaign, with the same GM and players, could be a "big" fact in one context and a "small" fact in another simply based on the emotional state of the player, the degree to which they have chosen to inform themselves (e.g. by investigating further, by doing in-world research, by talking to experts or witnesses, etc., etc.), the scene in which the fact is revealed, and the degree to which the players' attention is drawn to the fact vs drawn away from it.
I believe a lot of what you have read has been based on unstated assumptions we haven't shared, and that is hard to express.
