The accepted answer feature comes from Stack Overflow, where it serves a useful purpose. By accepting an answer, the questioner confirms that the answer solved the original problem. This makes sense when the questioner has a programming problem - if implementing the answer allowed the program to work properly then it objectively worked.
A site like Science Fiction & Fantasy, though, is very different in this regard. The questions are theoretical, and the answers can not be fed into a computer program to test their efficacy. As such (with the exception of story-identification) it is a useless, even negative, feature. The person asking the question is in no better position than anyone else to judge the correctness of the answer; on the contrary, the questioner is often in a worse position, since questions are often asked when someone has a lack of knowledge about the topic.
Across the Stack Exchange network, the accepted answer is displayed before all other answers (unless it is a self-answer). A question on the network-wide Meta site three years ago asked if this could be changed per site. There has been no response so far, and it does not seem that this feature will change anytime soon. That leaves us with the possibility that the questioner can accept a bad or wrong answer and cause it to appear before better, correct answers.
However, this problem is not engendered by the duplicate process. This is a potential problem on any question, and has to be dealt with accordingly. If an accepted answer is incorrect you should downvote it, post a comment explaining why it is wrong and post a better answer. Beyond that there is not much you can do. Users who visit the page will continue to see the wrong answer first (and perhaps only, if they don't continue scrolling down the page). Hopefully, readers can figure out that an accepted answer does not become better than a higher voted answer, because an individual user awarded it a checkmark.
As it relates to duplicates, then, an answer's status as accepted should be ignored. Whatever criteria are used to determine the direction of closure should continue to be used regardless of which answers are accepted or not. The acceptance of an answer carries no weight or meaning.
Additionally, to whatever extent feasible, it is not the job (or privilege) of close-voters to determine an answer's correctness. When users reach 3,000 reputation and gain the ability to vote to close questions, or earn a Gold Tag Badge and gain the ability to single-handedly close questions, their upvotes and downvotes are still only worth 1 apiece. Those privileges do not necessarily mean that those wielding them have more expertise in any particular question. If the argument falls back to relying on objective post score (to eliminate close-voter subjectivity), we return to the fact that the Stack Exchange network recognizes the right of the individual questioner to supersede that.