I find this account of an additional standard that applies in D&D not the most plausible account! The closest I can get to finding it plausible is to rewrite it as "the thing that the GM thinks is most plausible", or even better "the thing the GM thinks is the best fit with what they are imagining in the situation".
This I think is true.
My view is that this is not about standards but about techniques. The techniques that are typically used in D&D - reading consequences more-or-less directly off the notes (say "Save vs Petrification or take 2d6 damage from the pendulum blade trap), or the GM extrapolating more-or-less immediate causal consequences from their notes and their conception of the fiction - tend to produce fewer twists and turns.
For instance, consider the Dreadnought hoax that was mentioned upthread. In my experience, the typical GM of a D&D game is not going to regard that as the best fit with their conception of the fiction, and so it won't happen. Or for a fantasy version, Bilbo's hijinks with the spiders in Mirkwood.
If we go beyond hoaxes and trickery, let's think of chance meetings: Merry and Pippin meet Treebeard; Aragorn, Gimli and Legolas meet Eomer; Conan meets Pelias the wizard; the whole drama of Turin Turambar is driven by chance meetings.
I think the encounter and reaction rules in classic D&D are intended, among other purposes, to help produce these sorts of events as components of the game - and not unlike the player's use of their Noble background, these often require the GM to invent some account, in the fiction, of why things happen as they do (eg the roll of 12 tells us that the Ent enthusiastically befriends the Hobbits - what explains this in the fiction?). The difference between my preferred approach and the classic approach is that the random elements are more tightly anchored to resolution of player-declared actions, and the elements of the fiction are more tightly related to priorities the players have established for their PCs.
