Well, here's a counter-argument - not intended to be knock-down, given this stuff lies outside my main fields of expertise!Considering these two things... My approach to these sorts of ideas is to deny the entire Meinongian hypothesis at the start, and more relevantly that of @clearstream's proposal here as well. Meinong's Gold Mountain HAS NO HEIGHT, because 'to have a height' means it has a value, X, in meters, and lacking such an X, the property height cannot be said to be possessed by it. Since 'all mountains have a height', the Gold Mountain IS NOT A MOUNTAIN. Thus other attributes and entailments of 'mountainness' DO NOT APPLY TO IT.
The application to CS' argument is then straightforward, no imagined things have all the definite properties of real things, and thus cannot be classified as being members of the sets of those things.
It's a gold mountain. That is to say, a mountain constituted of gold. So it has a density of 19.3 times that of water. That's an entailment that does follow from it being gold.
So, similarly, it is a mountain. And all mountains have some height or other, and so the gold mountain has some height or other. But (analogously to the point about Holmes's handedness) there is no particular height that it can be said to have - although, being a mountain, perhaps we can confidently assert that it's more than 1 metre high (that would be a pile of gold or a lump of gold, not a gold mountain).
Actually, I think it does.While that won't particularly serve as a guide to which fiction you 'should' imagine, it does serve one important purpose, it tells us that the possible fictions are only bound by pure aesthetic criteria and nothing else, no logical or other sorts of constraints are binding.
<snip>
I won't say this all really effectively changes much
First, another philosophical example:
In mid-20th century philosophy of perception, there was discussion of the following issue: if a person is punched and "sees" stars, or if a person is drunk and "sees" pink elephants, is there some determinate number of stars, or of elephants, that that person sees?
The context for the discussion was sense-datum theories of perception, which seem like that might be committed to an affirmative answer, given that they think the sense data really exist, and hence must have the determinate properties of actually existing things. Unlike the Meinongians, they seem even more strongly committed: not just that there is some number or other of stars, or elephants, but that there is some particular number that is the number thereof.
Now, the relevance to RPGing:
Suppose that the GM tells the players "You see a mob/horde/gang of <whatevers>". In AD&D, even if the GM doesn't tell the players how many <whatevers> they see, and even perhaps hasn't worked it out yet, the rules of the game commit everyone at the table to there being some determinate number of them. Because otherwise the action resolution rules - in particular, combat - can't be applied.
Compare, say, 4e D&D. This requires that the mob occupy a certain space in combat, but not that there be some determinate number of mob-members. So my hobgoblin phalanxes and hordes of vrocks must have had some-or-other number of members, but nothing in the rules or the play of the game demands that any particular number be determined.
Generalising: a difference of boxes can demand a difference in the clouds. It might seem like a trivial point, but we've all suffered through years of tallying gold pieces, and tracking arrows and encumbrance, because its implications have not always been properly thought through!