Pertinent sections of my previous post, where I feel I already addressed the questions you're now asking:
"the whole "success with a complication" as a central feature is one obvious thing to me. The mechanics are explicitly telling you, "the game needs to move forward and something interesting must happen at this point."
***
"At a more basic level, the philosophy that "nothing much happens; the status quo is maintained" is an unacceptable outcome -- instead, any interaction with the mechanics must happen in such a way that doing so moves the game forward somehow.
***
In a traditional game "the game mechanics themselves aren't intentionally being built to generate complications, challenges, dilemmas, demand action or what-have-you."
I've got a view on this: it's analytical and genealogical.
Classic D&D is, at its core, a game of puzzle-solving. At the start of the game, the GM has all the information (in the form of the map and the key), and the players have almost none (perhaps some rumours, not all of which they can rely on). Over the course of play, the players acquire more and more of that information - by moving through the dungeon and mapping it, by listening at doors, by using detection/scrying magic, by judiciously opening doors etc. They can then exploit this information to plan and undertake dungeon raids, in the way that Gygax describes in his PHB.
In classic D&D play, the most important categories of action are movement, listening/looking/detecting/scrying, fighting, and talking. Movement only requires mechanical resolution in special cases (eg climbing, perhaps some balancing, etc) and the resolution of that (i) follows common sense (eg unsuccessful climbing can lead to falling) and (ii) has as its more significant consequence that the movement doesn't occur, and hence the PC is not in the place that the player wanted them to be such that they could do whatever the thing is that the player wanted them to do (eg open a door).
Listening, look, detecting etc have a whole lot of baroque rules. Some permit retries (eg my recollection is that listening does, under a modest constraint, according to Gygax's DMG) and some of which are far more strict in this respect (eg looking for secret doors; or the flat inability of some detection magic to penetrate some materials). The significant consequence of failing these sorts of actions is that the information in question is not gained.
One important category of action that straddles both movement and looking is opening doors. Sometimes retries are allowed (eg opening an ordinarily stuck door) and sometimes not (eg opening a locked door). If a door isn't opened, then as per the general category of actions above, either the information is not gained or the movement is not accomplished. Thus the players either fail to progress in solving the puzzle, or fail to progress in their raiding of the dungeon.
That is the appropriate complication for this game.
Fighting and talking are a bit different. Talking can produce complications (vis the reaction table; and the loyalty subsystem in the DMG can be seen as an offshoot of this). And so can fighting: the players' position can be set back by the loss of hit points, the loss of their retainers (if morale checks fail and retainers flee), etc. These complications can affect the ability of the players to achieve their puzzle-solving goals (eg if they fail to acquire information by talking, or if they are driven off in a combat) and also can produce consequences from their raids.
So I don't really think that classic D&D eschews a mechanics-complication connection. It's just that the goals of play are very different from (eg) Apocalypse World or Burning Wheel, and so what counts as a complication, and the way the mechanics mediate the introduction of complications, is different.
What makes the overall situation more complicated, in my view, is the retention - over decades of D&D design, at least until 3E and arguably since then with the exception of 4e - of variants on these mechanics, although the goal of play is no longer the puzzle-solving and raiding that is at the heart of classic D&D. And many other RPGs - eg Rolmeaster, just to pick an example I know well - hew relatively close to this pattern. I think it is in this sort of play that the idea that mechanics don't mediate the introduction of complications gains more plausibility. That falls very much to the GM, who generally does it by drawing on and/or manipulating the backstory that they have prepped.
I don't know about
all modern RPGs, but there are certainly some (eg AW, BW) that very clearly and deliberately deviate from this pattern that has predominated in respect of D&D and similar RPGs.
This sort of play does not generally need mechanical resolution to drive it forward. It doesn't have a "forward", except in the sense of the players acquiring, and exploiting, more information. The important categories of action are