Unless otherwise stated, a question and answer pair is presumed to be ontological. It is "about" phenomena; it is not about the opinion regarding a phenomena independent of the arguable truth or falsity of this opinion but depending solely on the popularity of the opinion. (n)
(n) is how I understand the approach to answering questions on this site. Scientific questions are purely about phenomena: ontological. But apparently not everyone agrees?
To put it concretely: if on, say, mathoverflow, somebody asks whether a theorem $T$ is true or false, the assertion that "many people pretend it's true, it's easier to do so" (u) is NOT an answer. That treats the question as not "about" the nature, not about truth or falsity, but rather about preference of observers of nature ... which is off topic unless specifically requested.
Consider, again, what would happen, if on tex.se or stackoverflow, somebody ask how to write a code that gives a certain output, and somebody else answers "here is a definition that is most common, although it doesn't actually give anything like the required output". That wouldn't work, would it?
I suggest that a Q and A site functions best if it follows approach (n) instead of (u).
Now, does anybody really consider approach (u) to be preferable to approach (n) when answering some questions, especially there where a definite ontological answer exists in the literature?
Consider the example: a question asks about a theorem, and this theorem is actually known to be false in the technical literature, and references exist. (I) They are given as an answer. To this a reply is given: (II) a plenty of other literature conventionally assumes for convenience, when teaching, that it is true, and suggests treating it as true. Approach (n) states that (I) is the answer preferred. Approach (u) states that (II) is the answer preferred.
If in this case anyone DOES prefers approach (u) over (n) when answering questions: why?