I don't mean to discard the 'prescriptivist vs descriptivist thing' altogether, it's a great way of thinking about it; and it was new to me in this context.
In general my view of models/analogies is that they are as illustrative and useful in how they differ from what they aim to describe as how precisely they can match up.

but I do tend to be anal about those caveats, to the extent that I self-replied to my own arguments above. *cringe* lol