In my opinion, one has to distinguish between seeing an _answer_ as

* _an answer to the question_,

and

* _a solution to the problem_.

When I'm reading a question that I **know the answer** to (i.e., I can provide the solution to the problem) I **write an answer**.

However, when I'm reading a question for that I **might have seen/found a/the problem/flaw** (i.e., I neither _know_ that for sure nor did I test my idea/feeling) I **post a comment**.

I don't intend to _do whatever I want_, as @Chip described it—this is just my personal intuitive way of responding to the question under the given circumstances.

Oftentimes, however, the actual solution is either just a one-line (or even one-word) answer, and when posting this as a comment, one does not yet know this is _the_ answer.

Also, suppose an existing answer does lack only a tiny bit to be complete (i.e., the problem is not yet solved). In that case, I would choose to comment this answer instead of copying the whole code, then add/edit the single missing/wrong line, and provide this (complete) answer. No matter if the particular (incomplete) answer was written just a couple minutes ago or a decade ago, I don't feel right about posting an answer in that case.

Concluding, I don't see a problem with writing a _maybe-answer_ as a comment, and then turn it into an actual answer if it **is** the answer to the question. But that's just my gut feeling.