I like the top answer to the question linked to (in the comments) on chemistry a lot. I also go to the meta post How do I properly use the “Not an Answer” flag?
The main thing to take from the above meta post regarding this particular question is:
When should I not use this flag?
Do not use this flag when:
- The user posts a partial answer
- The answer makes an attempt to answer the question, even if it is wrong or inaccurate or you disagree with it
- The problem with the answer is subtle and would benefit from additional explanation [...]
That is, just because an answer is wrong doesn't necessarily mean it should be deleted.
My personal reasoning for this is simply that showing someone what's wrong/what not to do is often as important as the right way to do something. We signal that an answer is wrong by downvoting and comments can be used to clarify what exactly is wrong about the answer so that (if the answerer doesn't improve their wrong answer, which is the ideal case), this can be used to tell the OP (and future readers) what not to do if something similar comes up again, as well as why not to do so.
The above is all well and good but we also have other policies, in particular, the one equivalent to the one linked to on chemistry: non-mainstream questions are not allowed.
My view (and indeed, the principle I've been trying to work on until now) is in how the above are combined: if an answer is just wrong (i.e. 'blatantly incorrect' or 'misleading'), it shouldn't be deleted purely for being wrong, although should be downvoted. If however, an answerer starts pushing a non-mainstream theory in an answer, it should be deleted.
The difference between 'wrong' and 'non-mainstream'
I'm not aware of any hard-and-fast rules that help determine this and to me at least, this is a judgement call - if something's wrong, it shouldn't be published in a mainstream journal, so you could argue is non-mainstream. However, this is what I use as a basis for defining the difference/making the appropriate action:
If an answer contains a relatively small conceptual/mathematical error, typo or similar but overall is right, then it's 'not actually wrong' and the appropriate course of action is to make/suggest an edit and/or leave a comment explaining the problem (I don't even downvote these).
If an answer contains a fundamental problem that can be fixed with editing (i.e. that makes the overall answer wrong), then suggest/make an edit or leave a comment. If the answerer refuses to accept the edit/comment, then downvote as the answer is wrong, or flag for mod attention if it's causing an 'edit war'.
If an answer contains a fundamental problem that can't be fixed with editing (or would require far too much editing to fix), then it's simply wrong and should be downvoted. The answerer may or may not wish to either try and salvage their answer, or just delete it entirely.
However, if an answer links to a non-mainstream theory in any way, it should be deleted. If the non-mainstream theory is that of the answerer and the answerer doesn't disclose that it's their theory, it's actually spam.
If an answer contains content that's part of a non-mainstream theory that the answerer has published (e.g. online, in a blog, on vixra, in a non-reputable journal etc.) but doesn't actually link to the non-mainstream theory, this is still based on a non-mainstream theory (as opposed to the answerer merely making a mistake or getting confused or mislead), so will similarly be deleted.
Essentially, my view is that, if it's clear that an answer is based on or otherwise involves a non-mainstream theory (and isn't just the result of the answerer making a mistake or getting confused/mislead in some way), then it should be deleted for being non-mainstream.