Lately, I've seen a bunch of poor answers flagged as "not an answer" (NAA) in the flag review list.
Such flags seem to be controversial, even among reviewers. For example, at the moment:
this answer has two "very low quality" flags, one "not an answer" flag, one automatic "low answer quality score" flag and three dispute flags, while
this answer has one "very low quality" flag, one "not an answer" flag, one automatic "low answer quality score" flag and five dispute ("invalid flags") flags.
Thus, I'd like to request the community's opinion on this, and hopefully establish some kind of clear guideline on the appropriate use of these flags.
I assume that we all agree that "answers" that don't even try or claim to solve the challenge in any way, such as comments, new challenges or xkcd comics posted as answers, qualify for the NAA flag. However, the answers I mentioned earlier are somewhat less obvious cases — they seem to have been intended as solutions to the challenge, even if they miss that target pretty badly
Is flagging as "not an answer" appropriate for answers that:
solve the wrong challenge, as in the first example above (where the task is to check whether a number is a power of two, while the flagged answer checks whether it is even), or
fail to meet the criteria for the type of challenge, as in the second example above, which is a non-golfed solution to a code-golf challenge?
Just to be clear, I'm not talking about answers that merely solve the challenge poorly or fail to work in some edge cases, as long as they seem at least potentially salvageable. What I'm talking about are answers where no amount of editing or bugfixing (short of throwing the answer away and starting from scratch) could possibly turn them into valid solutions to the challenge. Should such "answers" be flagged as "not an answer", or should we just downvote them?