But if the world already is as you describe here, then the plot is already controlling the pace at a level that matches the 15MAD "problem", so 15MAD becomes moot on that.
And if the world is already as you describe here, then the key encounters should anticipate a full strength party. So again the issue of 15MAD is moot.
Not entirely moot, because there may be issues of PC-balance which presuppose a non-15 MAD.
The solution in my first Rolemaster campaign was that, above 10th level, everyone played a spellcaster of some sort. The solution in my second Rolemaster campaign was that we changed the suite of options and interpretations in play (Rolemaster is very much a toolkit style of game) in order to strengthen warriors relative to spellusers.
I'm sure other solutions are possible, too, that don't rely particularly on timeline scenarios.
You're right. It is a playstyle thing. If your style of play is to have the NPCs always and consistently do absolutely nothing if the PCs attack them and then withdraw, then you're going to have 15MAD.
Who said anything about "attacing and then withdrawing"? You may be projecting your own conceptions of scenario design here.
Suppose that the PCs are trying to rescue some prisoners. And they know that the NPCs are going, in due course (ie sooner than weeks, not necessarily tomorrow) going to sacrifice those NPCs in a demonic ritual.
And suppose that, en route to rescuing the prisoners, the PCs find a cat stuck in a tree. Do the players have their PCs stop to rescue the cat, or not? And if they do, and if it takes longer than they thought, and uses up more of their resources - so they have to rest a bit, or go back to town to reequip, or whatever. Ought the GM to kill the prisoner's offstage, or not? My strong preference is not to - because it is anticlimactic, and subordinates thematic concerns to operational concerns.
Other play groups (and probably Lewis Pulsipher) might think the GM ought to kill the prisones if that's what the timeline says, because doing so emphasises the importance of operational concerns. This is a question of playstyle preference.
And this is all orthogonal to a reactive world. You can play a game with highly reactive NPCs, in which failure will only happen onscreen. The simplest example of this would be the contrast between the NPCs killing the hostages offscreen, and the NPCs sending assassins to kill the PCs who are pursuing them. Both are forms of NPC reactions. Either might be suitable to reinforce the importance of operational play, in an operationally-oriented game. The second, but not the first, would be a suitable sort of NPC reaction in a "no failure offscreen" game.
If your preferred style of play is a sequence of static, non-responsive encounters with no meaningful impact on the game world or its inhabitants... then, yeah, the 15MAD is a problem for you and you shouldn't play a game that has any sort of strategic play to it.
There's no need to be insulting. It's a fairly narrow approach to play to assume that "meaningful impact" is confined to "events happening in a timeline predetermined by the GM", or that unless failure can't happen offscreen, there can be no meaningful impact. That is to narrow "meaningfulness" to operational or strategic matters. For some RPGers (perhaps many?) these aren't the most interesting dimensions of possible significance. The question of whether the PCs live up to their ideals, for example, may be more significant, and that is often better determined by making sure that climaxes occur onscreen.
One thing I've noticed is the time it takes to resolve combats in 3e/Pathfinder/4e offers a powerful disincentive to using random encounters. They simply took too long. The resource attrition wasn't worth the session play time-attrition.
Agreed. For me this is not a problem, because I don't particularly care for random encounters, or "filler"/attrition encounters, in my game. But it does mean that 4e (I won't comment on 3E/PF) is not well-suited to a certain sort of D&D game.