The point of absolutizing something is to incorporate contextual information into the thing itself.

No, the point of absolutizing something is to make it an absolute value, rather than being relative to something else. That's what the current spec text does. The fact that absolutizing such a selector produces something that can't be perfectly reproduced in a literal form is slightly inconvenient but irrelevant. (Had we decided the other way with :root to make it match anything without a parent, or had decided to accept :top-level, then it could be represented in a literal form. But the fact that it can't currently isn't relevant here.)

:not(*) is guaranteed to match nothing only if it's not in a negated context. If it's negated, then we match everything. So this is not a good way to represent a selector that matches nothing.

That's... fine? The point is that docFrag.query("+ div") can't match anything - nothing is the sibling of a document fragment. If this somehow gets negated (how?), then you're asking for "everything that's not an immediate sibling of the document fragment", and that is indeed every reachable element.

Currently foo as a relative selector with a non-virtual scoping root is absolutized to :scope foo. Since we don't consider it acceptable for absolutizing foo as foo in that case, then how is leaving foo as foo here considered acceptable? Either both cases need some kind of transformation, or neither of them does.

No, it's working exactly correctly here. Again, docFrag.query("div") returns exactly the same elements whether you interpret it as a relative or absolute selector; all the div elements inside the fragment are descendants of the fragment.

This is different from a non-virtual scoping root, which exists in a document with other elements outside of it, which could potentially be (incorrectly) matched by just treating the selector as absolute. Thus why you need to add :scope there, to correctly pipe in the knowledge that the matched elements must have the scoping element as their ancestor.

Give up on the concept of “absolutizing” a selector and define matching against relative selectors as its own special thing.

Possible but unnecessary - none of the "problems" you listed above are actually problems.

Define :scope as matching against virtual scoping roots [...]

We could do this. The concept of an invisible featureless parent element didn't exist when I first wrote this text; the Shadow DOM stuff came later. Going this way would let us entirely remove this section and just rely on the standard scoping-root stuff, which is probably a good idea.

We could do this. The concept of an invisible featureless parent element didn't exist when I first wrote this text; the Shadow DOM stuff came later. Going this way would let us entirely remove this section and just rely on the standard scoping-root stuff, which is probably a good idea.

I think it's not only a good idea but a better idea.