5

An interesting discussion about duplicate closures happened in this question's comments section:

enter image description here

Personally, I'm in the "duplicate means duplicate question" camp. However, since it appears two veteran users are of a different opinion, I thought it would be best to put the question to the wider community.

What do you think? When is it appropriate to vote to close a question as a duplicate? Should the two questions be asking the same, or is it enough if the one's answers answer the other?

  • as sort of a prequel, consider editing this to add similar discussion that took place about half month ago: six comments starting from this one: "Strictly speaking, answers don't make a post a duplicate target..." etc – gnat Nov 10 '16 at 20:04
  • 7
    @gnat: I think this site would really benefit if whenever you are tempted to vote for a dupe, take a deep breath, count to ten, and ask yourself for a moment "will it be clear for others that this question is a dupe". And when the answer is "unsure", please, don't vote for closing, just leave a "related, maybe a dupe" link. That would probably avoid a lot of destructive discussions here. – Doc Brown Nov 10 '16 at 22:21
  • @DocBrown I almost bought your comment until I noticed that you have an answer in the question we discuss and might be simply afraid of getting it compete with brilliant canonical in case if duplicate is confirmed. Got to be tempting to squeeze a little bit of easy rep in your neck of woods instead of trying to beat or improve canonical post; Hans probably has a point when he wrote about "voting force" driving the changes that degrade original idea of a great Q&A site – gnat Nov 10 '16 at 23:38
  • 6
    @gnat: Really, really irrelevant. – Robert Harvey Nov 10 '16 at 23:41
  • 2
    @gnat: I can assure you, my comment was not motivated by that question in special. Will you "buy" my comment if I delete my answer from that question? Say "yes" and I will be happy to delete it. – Doc Brown Nov 11 '16 at 9:04
6

The guidance for duplicate questions is very clear and specific. In a nutshell:

If you're going to close a user's question as a duplicate, it has to be a real duplicate. For example, if a user asks, "What does the IP address 128.0.1.1/24 mean?" it's OK to close that as a duplicate of a more general question like "What do IP addresses of the form a.b.c.d/e mean?" But it's not OK to close it as a duplicate of a twenty-seven page guide to netmasks. That's the moral equivalent of saying "RTFM."

Nor is it OK to close a question as a duplicate of a seemingly unrelated question that happens to provide an answer that fits the new question. That's not a duplicate. It's just coincidence. It's not helpful to point people at some random post and say "You're answer is over here somewhere."

  • If it's not useful to point people to where they can find the answer to their question, then why do we have duplicates in the first place? Why do we even have answers in the first place if pointing someone to the answer to their question is something that "is not helpful". – Servy Nov 10 '16 at 21:47
  • 5
    Because "Why the f*ck did you point me to a question about Tahiti when I asked about Hawaii? I know they're both surrounded by water, but damn." – Robert Harvey Nov 10 '16 at 21:52
  • 2
    "Because finding the area of an island is done the same way whether you're on Tahiti or Hawaii?" If there's an underlying question in which one island is given where it being another doesn't change the answer, then the questions can be marked as duplicates, despite not being the same question. Of course, if the question is asking something for which it being about Tahiti vs. Hawaii would change the answer, then of course they don't have the same answer, and they're not duplicates. – Servy Nov 10 '16 at 21:56
  • @Servy: The proper way to do that is to stand up a canonical question about islands in general, and close as a dupe of that, or modify the original question so that it has more general utility (if, in fact, the two OPs are actually asking about the same general principles). – Robert Harvey Nov 10 '16 at 21:57
  • 4
    @Servy: let me tell you I am backing up Robert here. Questions should be closed as dupes if the other question is either an immediate dupe, or a more general version of the same question. If answers just fit by coincidence, that is IMHO not enough for a closing reason. – Doc Brown Nov 10 '16 at 22:23
  • 3
    @DocBrown Again, the tautological definition of, "it's a duplicate if it's a duplicate" is not a useful metric. It provides no means for someone to figure out what to do in any situation. Your fear of "coincidental answers" is unfounded; I challenge you to find even one example of entirely unrelated questions that have nothing to do with each other that just coincidentally happen to have the same quality answer (by SE's standards of quality for an answer). You'll only ever find questions with the same answer when the questions are similar enough that you'd want them to be closed. – Servy Nov 10 '16 at 22:29
  • 2
  • 1
    @Servy: "You'll only ever find questions with the same answer when the questions are similar enough that you'd want them to be closed" - that actually sounds to me you agree that questions should to be closed only if they are "similar enough" to be a clear dupe. – Doc Brown Nov 10 '16 at 22:37
  • 1
    @DocBrown Yes, but I'm phrasing the actual guideline that I would explain to others in such a way that they have a clear and understandable metric to follow, rather than an entirely ambiguous statement of "similar enough" that is unclear and highly subjective. By defining "similar enough" to be when a good answer to one answers the other you have provided a means for anyone to determine when questions are "similar enough" and when they're different enough that they should have their own answer and not be closed. – Servy Nov 10 '16 at 22:42
  • 1
    @Servy: if answers to one question are serving well to another question can be at least as similar subjective as if one question is asking the same as another one. Actually, I think both should match - the question should be the same, or a little bit more general, and the answers to the dupe should be good answers to the new question. And the primary "metrics" I favour here is common sense. – Doc Brown Nov 11 '16 at 9:36
  • 1
    @Servy: that is why you need at least 3000 rep for voting to close. – Doc Brown Nov 11 '16 at 14:12
  • 1
    @Servy: I did not say "no criteria", don't twist my words. Quite the opposite, I guess I expect stronger criteria than you for a legitimate close - the questions should match, the answers should fit, and I expect people should apply some common sense. – Doc Brown Nov 11 '16 at 16:14
  • 1
    @DocBrown Again, you have not actually listed any criteria. You're recursively defining your terms. When you say questions should be closed when they're "a match" and you provide no definition of what a match is, then you are telling people to just do what they want. You have provided no useful means for someone to determine when to close a question. Given your "criteria", gnat's close vote is appropriate, because his common sense tells him that the questions are duplicates, and you're telling he he should only ever consider his common sense, and nothing else. – Servy Nov 11 '16 at 16:24
  • 1
    @Servy: this gets boring, you ignore what I am saying, twisting my words, ignore that I actually agree to you that also answers must be a good fit, and I don't like your tone. Sorry, I have better things to do. – Doc Brown Nov 12 '16 at 20:31
  • 1
    @DocBrown How am I ignoring what you're saying? You're ignoring what I'm saying. All you've ever done is say that duplicates should be closed when they're duplicates, blatantly refused to provide any criteria for when questions actually are duplicates, and any time I've pointed out that you've not actually said anything meaningful you've resorted to attacking me while ignoring all of my arguments. You haven't said anything that I could ignore. – Servy Nov 14 '16 at 0:10
3

My understanding is the same as the one that Servy presents in the comments captured above.

Question A is a duplicate of Question B if the answers to Question B answer Question A.

That said, there are some other considerations.

First, Question B may answer parts of, but not the entirety of, Question A. In that case, Question B should be linked to in the comments of Question A. This also causes it to show up in the sidebar of Question B. If necessary, answers can also link to one or more answers of Question A as supporting information.

Second, the answer to Question B cannot be hidden in one or more answers to Question A. The answer should be clearly obvious by reading the answers to Question B. If the answer is hidden away in a link to an external resource or a small aside, then Question B is not a duplicate at all.

Third, if both questions have answers, the quality of those answers should be used when considering duplication. Question A may be older, but if Question B attracted sufficiently high quality answers, then Question A should be marked as a duplicate of Question B.

See:

The Wikipedia of Long Tail Programming Questions:

If you're going to close a user's question as a duplicate, it has to be a real duplicate. For example, if a user asks, "What does the IP address 128.0.1.1/24 mean?" it's OK to close that as a duplicate of a more general question like "What do IP addresses of the form a.b.c.d/e mean?" But it's not OK to close it as a duplicate of a twenty-seven page guide to netmasks.

Handling Duplicate Questions discusses three kinds of duplicates. Two are easily closable as duplicates (the "cut-and-paste duplicate questions" and "accidental duplicates"). The third, the "borderline duplicate" covers the case that I describe above.


I would like to generally speak about voting to close as a duplicate, as well. One person's vote to close is a single opinion, one that just happens to leave an autogenerated comment that includes a link to the question they picked as a duplicate. That autogenerated comment does not need to be flagged as non-constructive, rude, or offensive. However, you can flag if Also, you do not need to engage with people over a single close vote. I'd ask that no one votes blindly - if someone has suggested a question as a duplicate, read both questions and, if you feel they are duplicates, vote with them. If you don't feel that way, then don't vote to close as a duplicate. If the question has several close votes that you feel are inappropriate, then you should open a here on Meta.

  • 1
    @RobertHarvey Done, with two links to the SO blog. Is that sufficient? Otherwise, I can keep going to find some Meta.SE posts. – Thomas Owens Nov 10 '16 at 20:46
  • 2
    Gnat's proposed duplicate (the one that started this conversation) doesn't fit any of those three kinds of duplicates, nor do any of those three kinds discuss closing a question as a duplicate because "you can find your answer over here." – Robert Harvey Nov 10 '16 at 20:48
  • 1
    @RobertHarvey I'm not speaking to the validity of that close-as-duplicate vote. This question is a generic discussion of what a duplicate is, not a question about that specific question and if it is a duplicate. If it doesn't fit the definition(s) above, then as other people review the question, they should see that and choose to not vote to close. Eventually, the close vote will expire. – Thomas Owens Nov 10 '16 at 20:50
  • 2
    @RobertHarvey You mean the quotation that describes how a question can be closed as a duplicate of a question that isn't actually asking the same thing as it? Your citation provides a counter-example to your claim. – Servy Nov 10 '16 at 20:55
  • 2
    @Servy: I'm not really in the mood to rehash canonical posts here, and the constellation of problems that accompany them. You know what I mean, and I know that you know what I mean. Canonical posts are something that we can all agree on; they have no relevance to gnat's proposed close vote. – Robert Harvey Nov 10 '16 at 20:57
  • 2
    @RobertHarvey So given you do understand that questions that aren't actually asking exactly the same questions can potentially be duplicates, why continue to say otherwise? I've only ever said that I felt that the one close vote in question wasn't appropriate; that has never been an issue. – Servy Nov 10 '16 at 21:03
  • 2
    @Servy: The sticking point is that you and gnat seem to think that you can map a new question to some existing answer on some other question without considering what the original question is asking first. You can't. I never said that the questions have to be exact duplicates, I said that you have to map questions to questions, not question to answers. Nor am I averse to considering the quality and relevance of the original answers, but the question to question mapping has to succeed first. – Robert Harvey Nov 10 '16 at 21:06
  • 4
    You see, you're still missing the point. Vote to close as duplicate was never designed to eliminate duplicate answers, only duplicate questions. A trivial, but hopefully useful example: The two questions "What color should I paint my car" and "what color should I paint my mailbox" are not duplicates, even if the answer is "green," and even if the detail in both questions points to green being the correct answer in both cases. – Robert Harvey Nov 10 '16 at 21:13
  • 4
    @RobertHarvey A post explaining why it's a good idea to paint a mailbox green wouldn't make a good answer for why you should paint your car green. The points wouldn't apply, making it a poor answer to the question. The criteria you'd consider when determining the color of your mailbox and your house are different, thus the answers to the two questions are different, even if the resulting color chosen is the same. The differences in the question are germane to the answer, so the questions are not duplicates. – Servy Nov 10 '16 at 21:17
  • 3
    Of course, an exemplary answer to a question asking what color to paint your house if you have an odd numbered street address is going to be an exemplary answer to a question asking what color to paint your house if you live on an even numbered street address. The questions are different, but since the differences in the question aren't relevant to the answer they are duplicates. – Servy Nov 10 '16 at 21:19
  • 2
    @RobertHarvey What is so complicated about it that you don't understand? It's your position that makes no sense at all. "Questions are duplicates when they're duplicates, and they're not duplicates when they're not duplicates." You've defined your term recursively. You're basically defining questions as being duplicates whenever you say they are, which isn't a useful criteria for anyone but you to determine if a question is a duplicate. – Servy Nov 10 '16 at 21:39
  • 1
    @Servy: Let's try this again. 1. Is the question a duplicate by any reasonable definition? No? STOP. Yes? 2. Does the dupe target satisfactorily answer the duplicate question? No? STOP. Yes? Its a duplicate. Simple as that. – Robert Harvey Nov 10 '16 at 21:42
  • 1
    @servy: Any assessment of duplication must first evaluate the questions, without regard for the answers. Does that make better sense to you? This isn't a tautology; it's simply how this works. I think you're being pedantic about the wording I'm using; again, you know what I mean, and I know that you know what I mean. If you're in any way involving the answers in your duplication checks without first checking the questions alone to see if they can reasonably be interpreted as duplicate, you're doing it wrong. – Robert Harvey Nov 10 '16 at 21:46
  • 2
    @RobertHarvey That's one example. It's not a definition. Any acceptable answer to the latter question is going to be an acceptable answer to the former question. Note you didn't need to actually see the answers to determine that. If you can provide an acceptable criteria for determining if questions are duplicates that never even considers what their answers might be, great; let's hear it. You also haven't provided any reason why it's important that one be prohibited from considering what the acceptable answers for a question are in determining if they're duplicates. Why shouldn't you? – Servy Nov 10 '16 at 21:59
  • 3
    @RobertHarvey It seems you really don't get it. Using answers as a guidelines is the best way I've ever seen to explain it to someone without a lengthy conversation going over lots of nuance. If I just tell someone, "close questions as a duplicate when you think they're close enough to being the same" I know they won't actually understand (correctly) what questions to close. If I tell them, "if copy-pasting the answer to the question you're considering closing would make an acceptable answer, then you can vote to close as a duplicate" that they'll have a pretty good idea of what to do. – Servy Nov 10 '16 at 22:53

You must log in to answer this question.

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