All wanky metaphysical and non-metaphysical objections aside, I don't think we were really talking about the ~~~reality~~~ of triconsonantal roots. As far as I can tell, that was standing in for a discussion on the most elegant analysis of how these languages work.
Anyway, I liked Yng's comment, but my (possibly wrong) takeaway is not that Arabic roots don't exist, but that they are far messier (more irregular, more suffused with random semantic drift) than a simplistic conlanger would assume.
That is a totally fair takeaway. And the point of that post was specifically to give an insight into the morphology of Arabic and how it plays out so that people would get an idea of some of that complexity.
Basically, I just like to think of Arabic derivation as working just like any other language - on the word-to-word level. Perhaps this is all semantics (hohohohoho), but this way the meaning of a derived word X is constructed by analogy with the meaning of the original word Y from which X is derived. The root is then part of the morphological process of transforming a word into another word by derivational modification. It has no meaning in and of itself other than what is taken from the word it is derived from.
Understanding roots like this makes analysis of Arabic's morphology much more elegant, since it allows us to understand (for example)
tama7wara, which is derived from a noun which already has a derivational prefix applied and carries over that prefix, as the result of the normal extraction of a root from a base word. It also allows us to understand the incorporation of loanwords much more easily (since it doesn't require the creation of a root which then floats around in the intellectual ether in an undefined space, but allows us to see vocabulary as just that - vocabulary).
From a conlanging point of view, it also stops the weird cartesian ultra-regular approach where all words are straightforward combinations of a root and pattern and all roots have almost all patterns attested and vice versa.
Finally, from a pop-linguistic point of view, it runs counter to the prevailing weird romanticisation of Arabic which understands it as 'agglutinative' (not in the technical linguistic sense) and totally cartesian and logical in this way, when Arabic is really just like any other language.
That said, there is (as always) some messiness involved in that the
m- formant (for example) does not always carry over into new derivations -
m7ammad for example has a diminutive
7ammuud(e). Since this can't be straightforwardly explained by a process that, for example, deletes initial
m in derivations (or something), some phonological theories struggle with this - if the root doesn't exist in some sense, then how does the brain distinguish between this initial derivational prefix
m- and a 'root'
m- (as in
maaher, whose diminutive is
mahhuur?) But this is on the morphological level and not on the semantic level where I'd stand by my point.