There are hordes of dubiously labeled duplicates on this site. Many tie questions with titles that are so diverse that there is no way a person looking for one the Qs would ever search for the wording in the other. And on the whole, we spend a fair bit of effort doing a basically bad job of managing them. What is the point?

It is to be expected to find an answer already posted somewhere on ELU. If Q1 asks what rule applies here and A1 says you use this rule because ..., then it's reasonable that there will be many questions that A1 applies to that are not duplicates. The current duplicate notification bubble states "This question already has an answer here:". This is a terrible wording because it appears to sanction closing a Q just because there is an answer somewhere else. This notification should only be used on Qs that are not duplicates.

I propose a strong test for duplicate closure - 1. Unless replacing the Q's title with the alternative title improves the Q, don't close it as a dup. Just point to the alternative as a related Q, and treat A's in the alternative as fair game in answering this Q. (Which might mean a new feature of two be added to make this fair to all.) And 2. The alternative must actually have an authoritative answer. Both 1. and 2. must be true to close, otherwise, the OP should be prompted to consider whether they are dups, and the OP's opinion should be respected.

  1. What impact would such a strong test have on the site?
  2. Would new features be desired to implement it, such as crediting someone whose answer to a different question was referred to?
  3. If this is too restrictive, why is it too restrictive? How do duplicate closures benefit the site and why would a tight policy not attain those benefits.
  4. Can we please change the wording in the duplicate closure bubble so it isn't about answers, but about questions?

ADDED: Laurel's answer linked to an existing blog on the subject - https://stackoverflow.blog/2010/11/16/dr-strangedupe-or-how-i-learned-to-stop-worrying-and-love-duplication/

This seems to cover the same issue and does answer one of my questions. The persistence of related questions that may actually have identical answers isn't a problem.

It also points out the putoffish structure that is created when referring a commonly asked question to a funny question that contains a canonical answer. I mentioned this in one of my comments to Laurel's answer as well.

  • 2
    Examples of poor matchings, complete with explanations, please.
    – Mari-Lou A
    Oct 30 '17 at 0:17
  • 1
    @Mari-LouA Well, the one that spurred this question just got edited, and so did it's dupe target, plus three others got tag edits, and there has been a flurry of reopen votes. So it's not going to be as easy as it sounds. I'm interested in the policy here. I don't think you need an example to answer 1. through 4. I'd prefer not to point to a specific question because it isn't about the asker or the answerer, it's about why we take an answer-based approach to deciding if a question is a duplicate, and what would happen if we just quit doing that.
    – Phil Sweet
    Oct 30 '17 at 1:22
  • If the new question is clearer then close the old as a duplicate of the new. Or if neither question is good, write a new canonical question for both. Oct 30 '17 at 5:18
  • Then please rephrase your opening line There are hordes of dubiously labeled duplicates on this site it's hyperbolic. If there are hordes then finding two or three examples is easy. You're talking about one or two questions being mislabelled, not two or more dozens. You don't need to implement a new test if, generally speaking, users are closing questions correctly.
    – Mari-Lou A
    Oct 30 '17 at 7:04
  • In my 6 months experience on ELU, most questions that got closed as duplicates and did not get reopened are actually genuine duplicates of the original question @Phil Sweet, and the 'answer based approach' is the exception. However I completely agree that a basic condition for a question to be closed as a duplicate should be, as you say, "the alternative must actually have an authoritative answer." When one of my questions got closed as a duplicate of a very old question with just 1 accepted but not authoritative answer, I was advised by 3 senior members to place a bounty on that question. Oct 30 '17 at 10:31
  1. Unless replacing the Q's title with the alternative title improves the Q, don't close it as a dup.

I don't think 1 is a good test. Take for example "My wife and I's seafood collaboration dinner". This is a title that would only fit one question on the site, the one it is attached to. This question, however, is used as a canonical duplicate target for many questions including How do you make the possessive form with "He and I"-style subjects? and "Each and every one of you's opinions". (At this point, I would be against changing the title of the seafood question, because it's memorable as it is and I'd never find it if it changed. For some other questions, it would be beneficial if somebody made edits to their titles, bodies, or tags to make them easier to find or more clear or just plain old better.)

Many [duplicates] tie questions with titles that are so diverse that there is no way a person looking for one the Qs would ever search for the wording in the other.

This is actually one of the biggest benefits of having dupes. They are signposts to people who search for X so they can find the question that answers their question but doesn't mention X at all. This is mentioned in the blog post Dr. Strangedupe: Or, How I Learned to Stop Worrying And Love Duplication:

One thing I want to be clear about, though, is that duplication is not necessarily bad. Quite the contrary — some duplication is desirable. There’s often benefit to having multiple subtle variants of a question around, as people tend to ask and search using completely different words, and the better our coverage, the better odds people can find the answer they’re looking for.

I could be interpreting this far too literally though. Your case would be improved by citing specific examples where what you suggest would work.

  1. The alternative must actually have an authoritative answer.

To some extent, this is already a feature. It is not possible to close if the duplicate doesn't have either an accepted answer or an upvoted answer (except under certain conditions, as described here). This is one of the reasons it's essential to downvote.

On the other hand when the system fails and a question is eligible as a dupe target despite the answer(s) being very bad, I am somewhat torn between closing and not. I think I've gone both ways: sometimes I've voted to close, others I just leave a comment.

What to do

You already have a lot of tools at your disposal to combat wrongful dupes:

  • Vote to reopen
  • Edit to clarify why it is not a dupe
    • Under certain circumstances this sends the post to the reopen queue
  • Comment to clarify why it's not a dupe.
  • Bring it to meta and give everyone your argument for why the question should be reopened
    • People don't do this enough I think

All these tools currently exist and can be pretty effective.

  • 2
    Using My wife and I's seafood collaboration dinner as a canonical anything deserves 40 lashes with wet nori. That's exactly what needs to be avoided. Nothing can be expected to lead to that question directly. And anything that gets redirected there will also feature a rebuke for not first checking to see if this crazy title contained an answer. Answers don't make a question canonical, the question makes the question canonical. Any index of canonical answers lurking in funny questions should not impact the surface structure of the site, which is organised by questions, not answers.
    – Phil Sweet
    Oct 30 '17 at 0:29
  • Regarding the second para. One problem here is that the 'signposts' are accompanied by a putoffish display of notes, edits, system generated remarks, and comments even if the question gets returned as a hit far more often than the one with a canonical answer. Why run everyone through this gaunlet. Secondly, it amounts to paying a premium to have the canonical answer appear on the second page of a Google search instead of the first. (I think that is a reasonable effect of this policy, but I'm not a search engine guru).
    – Phil Sweet
    Oct 30 '17 at 0:55
  • Thanks for explaining the restrictions that are in place and listing the current options. All I can say is that they seem to be inadequate on the whole. Can ELU gin up a page like the SE page that you linked to?
    – Phil Sweet
    Oct 30 '17 at 1:05
  • I agree with @Phil that the "my wife and I's seafood" question is greatly overused as a duplicate target. I made a Meta post about it earlier: Can we switch to a better canonical question about using pronouns in compound possessives?
    – herisson
    Oct 30 '17 at 1:18
  • I agree with @Laurel that making it easier to search for frequent questions with good answers would be a great help; better search would solve some of the problem with closing duplicates, and make it easier for users to find answers on the site.
    – Xanne
    Oct 30 '17 at 6:12
  • @Xanne there is already something in place which helps users with linking new questions with older questions. It's the frequent button. The question Is there a correct gender-neutral singular pronoun? has been used to close maybe a 100 questions. Not all the questions linked have been closed as duplicates.
    – Mari-Lou A
    Oct 30 '17 at 7:13
  • @PhilSweet: For non-logged in users, there are automatic redirects from posts marked as duplicates to the duplicate target in at least some cases (it doesn't happen if the closed posts have answers of their own). See on Meta SE Automatically redirect anonymous user from unanswered duplicate question to corresponding answered version. So I'm not sure what you mean by "second page of a Google search instead of the first"
    – herisson
    Oct 30 '17 at 7:18
  • @Mari-LouA I hadn't thought of the "frequent" button as something to help users in this way. I've looked at these questions but never seen a recommendation to a new user to try it, although I will start doing so, and using it more for finding good dupes. The "help" page is on the whole a list of what not to do rather than help finding answers on the site.
    – Xanne
    Oct 30 '17 at 7:28
  • Right now there's a question english.stackexchange.com/questions/416298/…; and I know we have good answers (e.g., Lawler) on this; and I can't find the right one from the search box or the frequency list. If I did what I think would be really helpful, I'd go to Purdue OWL, i.e., outside the site that wants to become the source of all good answers, and summarize & quote from there. That's a site failure, isn't it?
    – Xanne
    Oct 30 '17 at 7:56
  • 1
    @PhilSweet There is a difference between a canonical question and a canonical answer (which is what the duplicate message points to: "This question already has an answer...").
    – Andrew Leach Mod
    Oct 30 '17 at 9:58
  • Do you mean that when a question is closed as a duplicate, the message 'this is an exact duplicate of an earlier question' points users to a canonical question and the alternative message 'this question already has an answer here' points users to a canonical answer @Andrew Leach? It is a very sensible arrangement and extremely effective as long as it leads to a proper answer to OP's question, but how is it done? Oct 30 '17 at 10:36
  • @EnglishStudent I don't know. If you can find a concrete example or two of each, that might make a good Meta question. I don't recall seeing such a question before. Even as a mod, I don't know whether I can influence which message the system uses.
    – Andrew Leach Mod
    Oct 30 '17 at 10:39
  • 1
    Thanks @Andrew Leach. I looked at a number of recently closed questions at english.stackexchange.com/… and found that both the messages I referred to earlier are used on every closed as duplicate question. Right below the title it says "this question already has an answer here" and points the reader to the original question. At the end of the body of the question comes the 'close message': Marked as duplicate by user 1, user 2, user 3, user 4, and user 5 xx hours ago: This question was marked as an exact duplicate of an existing question. Oct 30 '17 at 10:46
  • 1
    (contd.) @Andrew Leach, interestingly, instead of 'this is an exact duplicate' some of those closed-as-duplicate questions carry this alternative close message: "marked as duplicate by user 1, user 2, user 3, user 4, user 5: This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question." What is the difference? I shall try to frame it as a new meta question. Oct 30 '17 at 10:56

You must log in to answer this question.

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