Sometimes I see an interesting and accepted answer that includes useful ideas on nuanced topics, but that (I think) ultimately has flaws that makes it wrong. I hope to learn the proper etiquette for handling such answers. I personally do not like downvoting if I think an answer contains some useful ideas, though I am comfortable commenting when I think there are flaws.
I see a couple of options. One is to open a bounty on the original question and specifically writing in the bounty description box that the current accepted answer is incorrect and giving detailed reasons why, and requesting an answer that resolves the incorrect features of the accepted answer. I am unsure if that that might be an abuse of the bounty system, especially as I'm unsure which bounty reason to use in that case.
Another option is to open a new question, following largely the steps above. That is, the new question would ask the same question as the original question, but with some additional paragraphs pointing out why the accepted answer to the original question was incorrect and specifically requesting an answer that does not make the same errors.
Another potential option would be to write an answer to the original question, but the trouble is, I often don't know the answer to the original question. I only have reasons without much evidence why I believe the accepted answer is incorrect.
I am likely missing other important ways to address incorrect answers to other peoples' questions, so please let me know of other methods.
As an example to make my claim concrete, Andrew's answer discussing the black hole information paradox is what I would call "includes useful ideas in answering a subtle question, but is ultimately incorrect." As I commented on the answer, the answer incorrectly assumes that the Hilbert space dimension of a set of degrees of freedom is the sum of the individual Hilbert space dimensions rather than their product, making the resulting claims suspect. There doesn't appear to be a simple way to remedy the answer, and this is not an area of physics in which I would feel comfortable making my own answer to the original question.