I've been seeing a lot of questions marked as duplicated. Now, I understand there's a rationale about questions being asked again and again.

But there are many cases where one of the following things happen:

  • Questions marked as duplicated even though the original has no accepted answer
  • Questions marked as duplicated even though the original is extremely old and outdated
  • Questions marked as duplicated even though the original has a very debatable answer (when not completely wrong).

About the last point: in many cases, when someone doesn't know something, it means s/he doesn't know something. Litwrally. So how will s/he know the answer is correct? (unless it's a well documented answer). So, "accepted answer" shouldn't be a really important metric. As a matter of fact there was an accepted answer that literally says this:

Secondly - Accepted answers don't really mean much more than "the person who asked the question says this solved their problem", it doesn't necessarily mean that answer is the correct one. An answer with 50 votes vs an accepted answer with 3 votes should tell you that the 50 vote one is more likely to be of use.

Is there a way to clean up things a bit? For example, if any of the above applies, or if the original is 5+ years old it shouldn't be marked as duplicated. Furthermore, the original should be revised to see if it's OK or if it should be deleted

  • I would be interested to know how many such questions exist, but I am not sure how easily it will be to get this kind of data so we can do a quick review and see how many questions fall into each category that you have mentioned.
    – Michael Lai Mod
    Dec 23, 2020 at 20:41
  • @MichaelLai Here's an example that comes from you ux.stackexchange.com/questions/134321/… . And maybe I'm a bit biased in some way, but almost any answer marked as duplicated falls into any of the 3 categories, easily 50-60% of them, probably higher
    – Devin
    Dec 23, 2020 at 20:45
  • I'd have to go through the change history of both questions, but as it current stands, the original question that I posted 10 months ago does have an answer (which I haven't accepted because it was worded as a suggestion and references cited are from commercial or private site rather than somewhere like W3C). The question marked as duplicate is worded very generically ("Can you name this UI component?"), so keeping both in this instance doesn't seem to make sense. You are welcomed to post your answer to the original question or I can reconsider the merit of the duplicate question. Thanks.
    – Michael Lai Mod
    Dec 23, 2020 at 20:56
  • 1
    @MichaelLai it wasn't a stab at you or anything, I just pointed it out to you because it included you. As I said, there are many similar cases. I also mentioned that having an answer doesn't necessarily mean it's correct. Furthermore, I said they should be revised, duplicate or not. A good bunch of "correct" answers are arguable, duplicated ot not. Anyways, don't take it personal, I just thought an example of your own would be more illustrative FOR YOU. I could have examples for any person with 50+ answers, which probably includes me
    – Devin
    Dec 28, 2020 at 16:53
  • You've got to have a pretty thick skin to be a moderator (I think), so I hope I didn't come across as being defensive or anything :D If anything I don't feel like we get enough engagement with the community, so I am always very happy when there are questions on Meta and there is robust discussion on issues that can help shape the community. Your efforts in improving UXSE is definitely appreciated! :) I'd love to go through and clean up a few more things as well (while it is quieter during the holiday period) so please feel free to call it out.
    – Michael Lai Mod
    Dec 29, 2020 at 0:50


You must log in to answer this question.

Browse other questions tagged .