All maps are axiomatically abstractions. Some of real places, others of imaginary ones... but they always have less information than a real place...
Let's suppose that this is true. Nevertheless, I don't think it bears directly on what I said:
Here's another mechanic that's not an abstraction: GM describes what the PCs can see by reference to where the PCs are in a place that the GM has a map of; the players say where there PCs move in that place; the GM tracks that movement on the map; the GM uses the map to tell the players more about what their PCs can see; etc. The map is a representation, used to help establish and coordinate fictional facts about where the PCs are and what they can see; it's not an abstraction.
The mechanic is not
a map. The mechanic is
using tokens on a map to track the movement, and location, of the PCs, and thereby to work out where they are and what they can see.
That mechanic is not an abstraction of anything. It's a representation - a pictorial representation.
Here's a contrasting approach to working out where the PCs are and what they can see, that also can incorporate a map: when I GM Prince Valiant, I work out where the PCs go by combining
statements by the players about where they, as their PCs, would like to go and what they would like to do and
my own ideas, normally informed by the scenario that I want to use, about what some place the PCs are in is like. Eg when the players had their PCs travelling to Constantinople from Sicily, I narrated a storm that required them having to land on the Dalmatian coast and then proceed overland, and then narrated (i) an encounter with Huns while they were trudging through Caparthian hills, and (ii) an encounter with undead warriors when they were making their way through Dacian/Transylvanian forest.
As the use of south-eastern European proper names indicates, these events were narrated by reference to actual European geography, and at some points I showed the players a map just to help them visualise where their PCs were, and how they were proceeding on their trek to Constantinople.
But the map was not itself a component of the resolution process. The resolution took place in the way I described, by the players saying where they wanted their PCs to travel to, and me saying a bit about what happened, and where, during that travel.
The contrast between these two different ways of resolving travel has nothing to do with whether or not maps are abstractions. It's to do with the contrast between (i) the use of a dynamic pictorial representation, and the movement of a token on the map as a core part of that representation, in order to work out where the PCs are and hence what they see, and (ii)
just talking about where the PCs go and what they can see.
Method (i) is of course associated with D&D (and wargames). Method (ii) is - more or less, and sometimes perhaps supplemented by a call for an appropriate test or check - the one I use not only in Prince Valiant, but also in Fantasy For Real, Marvel Heroic RP, Burning Wheel, and other RPGs that eschew map-and-key based resolution.
So more carefully, the question isn't "is this mechanic diegetic?" but "does this mechanic track something that is diegetic?" I.e. is it diegetically grounded? Dwarves in your campaign are diegetic; the Dwarf mechanics may or may not be diegetically grounded; if they are, presumably they're abstractions (but I guess not necessarily if the mechanics came first and the fictional Dwarf was designed around them); the Fighter class definitely is not diegetically grounded since there's no Fighter diegetic element at all.
If the fighter class is some combination of mechanical abilities, and those mechanical abilities all track elements of the fiction (and so are diegetically grounded), then I think that it follows tha the fighter is diegetically grounded, even though - in the fiction - there may be no such thing as
a fighter. This is because the set of fictional elements that (taken as a whole) is the ground of the fighter class abilities may not, in the fiction, be a group-defining or profession-defining bundle of elements.
This is one way of thinking about the original Ranger class: the ability to fight, the ability to track, the ability to use magic, the ability to use palantirs, etc - these are all mechanical elements of that class that (let's say, for the sake of the analysis) track stuff in the fiction. That is, in the fiction, this character who is, on the PC sheet, labelled a ranger, can fight, track, use magic, use palantirs, etc. So the class, being this bundle of abilities, is diegetically grounded - even though, let's say, in this setting there is no such thing as
a ranger in the way that Aragorn is a ranger of the north in JRRT's work.
The reason I think that (say) 4e classes are not diegetic is not because the abilities they grant aren't in some way grounded in the fiction (they may be, they may not be, but that's another story that isn't relevant to how I think about them in the context of this thread). The reason I think they're not diegetic is because a class establishes a bundle of abilities as markers of
being a particular sort of person/protagonist; but in the fiction, there may not be any such
type of person as a saliently existing thing.
As (I hope) you can see, I'm trying to hew closely to the cinematic usage of "diegetic": is a thing that the audience sense/interacts with in engaging with the fiction (in this context, this bundle of abilities being saliently bundled as a type of person) also a thing that exists in the fiction? And my answer, in the case of 4e D&D, is
no.
