We've had some discussion recently about the quality of the questions asked on the site, and one point that has been raised is that many simplistic questions are asked repeatedly, generating a large number of duplicates.

When I see a question I think is a dupe, I tend to do a quick site search to see if I can find a good candidate to offer up for a "Close as Dupe" vote, but this can be time-consuming and I'm sure that I'm not always picking the best option.

Given this, would it make sense to have a curated list (perhaps managed by the moderators) that would essentially be an FAQ list of some of the most common questions with pointers to existing threads on the site that answer these questions that we could use as a first point of reference when we see potential duplicates? It seems to me that if we had such a list handy, we could close these duplicated more quickly and more accurately, and redirect people to the best existing threads more regularly.

  • That's a pretty good idea, but you don't need a moderator to curate it. There is such a feature built-in - the "Frequently Asked Questions" list - security.stackexchange.com/questions?sort=frequent - but if you feel it isn't picking up the canonicals youd like, feel free to go ahead and start collecting them. In fact, go ahead and use this question right here! ;-)
    – AviD Mod
    Aug 3 '14 at 10:48
  • 1
    That's a pretty neat feature. I guess it's a combination of number of dupes linked, views, and upvotes...so upvoting good questions and answers and linking dupes to the best answers should reinforce the ordering of the list? Hopefully it also draws on this to suggest existing entries to people when asking questions.
    – Andy Boura
    Aug 4 '14 at 13:53
  • @AndyBoura I had not gotten so far as to thinking about automating it, but more a feature based on tribal knowledge. However, if it could be automated in a manner similar to what you describe, that would certainly be very cool.
    – Xander
    Aug 4 '14 at 13:55

In my experience, having a big list of canonical questions isn't very manageable. Tagging questions as or some such (the way meta questions are tagged ) isn't very nice either because each question then stands individually. It's better to have questions organized by topic.

Fortunately, Stack Exchange already has a way to classify questions by topics: with tags. And tags have a place for canonical questions: the tag wiki.

If you think a thread covers an important aspect of a topic, and in particular if it's a good potential target for duplicate closure, add it to a list in the tag wiki for the tag that best describes the topic.

You can build up lists of existing canonical question by looking at the most-upvoted questions in a tag. Don't blindly take the most-upvoted questions though: sometimes the questions cover overlapping questions and are redundant, sometimes the questions are popular trivia of little general interest, etc. The votes are a hint but are not what intrinsically makes a question canonical. Also the canonical question lists should only list questions with a good answer.

On Unix & Linux, I try to maintain lists of common questions in tag wikis (but these lists aren't nearly complete or ubiquitous). I've standardized on “Further reading” as a title for the canonical questions section — perhaps not the best choice. You can see examples in bash, shell, ssh, grep, … I've even created a few on this site: man-in-the-middle, obscurity, … I invite others to follow suit.

  • TIL what the tag wiki is for. Who woulda thunk it? That all makes sense, and I think that's a fine solution for the problem.
    – Xander
    Aug 4 '14 at 17:07

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .