Which means the abstraction can, and should, change if doing so gets closer to the actual state of affairs.
Well, we can get into the weeds about how to define what the state of affairs in fact is, but that'll likely get us nowhere.
Not at all. It follows that the rules should prioritize efficacy and functionality. "A foolish consistency is the hobgoblin of little minds, adored by little statesmen and philosophers and divines." Consistency is not an unalloyed good. It has many benefits, and thus we should not dismiss a wise consistency. I myself am an advocate for a wide variety of particular kinds of consistency, e.g. unified resolution mechanics, not on the brute fact that they are consistent, but rather because their consistency leads to something else that is valuable in itself, such as making it easier, simpler, and more natural to adjudicate, or to speed up the learning process (which is, was, and always will be the single greatest hurdle to getting people into the hobby.) Requiring that the rules rigidly produce only and exactly one description of something, when that thing's value is necessarily relative to the context in which it appears, is a foolish consistency, pulling us away from efficacy and functionality.
Disagree in both directions.
Consistency in resolution mechanics IMO often tends to blur the abstraction rather than sharpen it, as that which is abstracted then has to be shoehorned into fitting with the unified mechanic. Discrete subsystems for the win, here.
As for consistency of description: absent illusion magic (which I'll henceforth ignore for these purposes), any given thing is what it is. Thus, the description of that thing shouldn't change. The 8'-high by 3'-wide stone door in front of you with the leering gargoyle face carved on it is what it is whether you're viewing it in bright light, dim red torchlight, or through a telescope from 300 yards away. Even if you can't see it at all, it's still there and its properties haven't changed.
The same is true of a living creature. Take our old friend the ogre: he has a set of intrinsic properties, which we describe as best we can by using numbers: 88 hit points, strength 21, 9'3" tall, AC 12 unless wearing armour, 640 lbs weight, etc. etc.; and those things don't change simply based on who/whatevr he happens to be interacting with at the time.
Same is true of your PC. You've got a whole bunch of numbers on your character sheet describing her in some detail; and those numbers are locked in no matter what she's doing in the fiction or who she's interacting with at the moment. Her maximum hit points don't suddenly drop to 1 when she meets Orcus at 6th level; she still has all 48 of 'em, and even though Orcus might be capable of hammering her for 60 points a round if he wants to, she might get lucky and survive a few rounds if he rolls crap for damage.
The other IMO ludicrous possibility is that a single creature might present different mechanics to two or more other creatures interacting with it
at the same time! Let's take our trusty ogre again, and put him up against Xena - a 17th-level killing machine - and Gabrielle, her 6th-level protege*. In your model, to Xena our ogre is a 1-point wonder (even though Xena at her best can only hope to give out 60 points damage a round, barring criticals), while Gabrielle - standing right next to her - has to chop through all 88 of its hit points no matter what.
And if another ogre comes up during all this and punches our poor original ogre in the nose for being such a lousy watchman, what mechanics does the new ogre get to deal with while violently interacting with the first one?
Bleah, says I! Far simpler to just say the ogre has the 88 hit points it has and let its foes - no matter who they are - chop through them as they may. And if that means it on average takes high-level PCs a bit longer to mow down low-level opponents (thus giving those low-level opponents a bit longer to pose a threat), I count that as a benefit rather than a drawback: the power curve has been flattened.
* - yeah, yeah, 4e doesn't like mixed-level parties; but master and sidekick working as a team is a trope as old as the hills, and if a supposedly big-tent game can't handle such things I'd call that a rather epic design fail.
But that was exactly my point...?
??? I thought you were suggesting that 54% was a very poor intend-to-adopt rate. If I misread that, sorry.
The map is not the territory--and different maps actually do have different information on them. The territory always remains whatever it is (I assume you grant that we're looking at the territory only in one particular moment.) But which parts of the territory are in fact on the map, and which ones are intentionally left off the map, varies by context. It is precisely the same with the level 2 solo ogre and the level 14 (or whatever) minion ogre. Different data has been represented in the abstraction, because we have proverbially "zoomed out."
So not only do the PCs mechanically grow as they level up, the monsters mechanically "shrink" as well. The latter part of that would seem to be a large part of why 4e's power curve is so (IMO unnecessarily) steep.
Except that there is a reason. The 88-HP ogre cannot produce the kind of experience you intend with this. It just can't. That's the whole point. You are hoping and praying that coincidence will fall in your favor. We can do better; we can design better.
What's the "intended experience", though; and why are we designing toward an intended experience rather than designing agnostically and letting things happen as they may?
