This critical answer on Academia.SE from Meta.mathoverflow deserves some attention, in my view:
The problem with academia.SE is that it is very different in style from MO and other SE sites. The paradigm of SE is that the questions are well-defined enough to have a correct answer, and the answers, well, attempt to provide it. Academia.SE is more of a discussion forum. Opinion-based questions, as well as non-questions flourish. Popular questions tend to have many answers repeating various talking points in different ways; it is telling that the answers almost never have any references/citations. The answers get upvoted not because they offer any useful insight, but because the upvoters agree with the opinions expressed there. Often, the highest upvoted answer is quite short and simply states an opinion.
My opinion is that this strikes an open wound and identifies an area where we need to improve.
I don't mean to strike down on subjective answers overall. Subjective answers can still have a positive role, when they identify good practices or suggestions, or are in areas that are not so well documented. And we are clearly not the only SE site with this concern; for instance, I would guess that most of Interpersonal.SE is subjective questions and answers where it is difficult or impossible to give a reference.
SE has some advice on good subjective and bad subjective questions, and suggests that these answers become a lot better if they share experiences over opinions, and are backed up with facts and references rather than just "because I'm an expert".
It seems to me that we often disregard these good practices on subjective topics, and do not worry about making our answers as data- and experience-backed as it would be possible, in many cases. Looking back at my post history I am myself guilty multiple times of this sin, so I do not claim to be better than the rest of the community. But I think that we should reflect on this feedback from an external user and try to improve in this area.
(Important: please do not go and downvote that Meta.MO answer --- it is difficult to get constructive criticism and suggestions for improvement if we mass-downvote those who provide them.)