15
$\begingroup$

Many things changed when MathOverflow joined the Stack Exchange Network a year ago. While most changes were quickly absorbed a few are more subtle and might have been missed by some. In particular, closing as duplicate has changed in a way that is somewhat incompatible with historical usage.

The main change is that it is now impossible to close a question as a duplicate if the original has no answer. Historically, it was encouraged to mark as duplicate all repeat questions, regardless whether the original had an answer or not. This is compatible with mathematical tradition, where one wants to preserve the authorship of a question, especially an open question! Unfortunately, this isn't a relevant tradition for many other network sites.

Here are the meta.SE posts where the changes were originally discussed:

Keep in mind that duplicates are actually useful as they generally improve search results. On the other hand, it is necessary to close duplicates to avoid fragmentation of answers.

$\endgroup$
  • $\begingroup$ I agree to what you write -- but what is the purpose of your post? -- Finding enough people to convince SE to allow again closing questions as duplicates of unanswered questions? $\endgroup$ – Stefan Kohl Aug 1 '14 at 17:37
  • 2
    $\begingroup$ @StefanKohl: Mostly, I just want to let people know that this is the current deal. $\endgroup$ – François G. Dorais Aug 1 '14 at 17:39
  • $\begingroup$ Unless a duplicate asks the same question in a substantially different way, I think in the present situation deletion is probably the way to go -- otherwise one would open the doors for plagiarism. $\endgroup$ – Stefan Kohl Aug 1 '14 at 17:49
  • 8
    $\begingroup$ @StefanKohl The moderators are in regular contact with us when it comes to things .. well .. not quite fitting as well as they once did prior to coming over to the Stack Exchange 2.0 platform. This post is precisely what I need for the impetus behind a proposed change. It allows me to (1) show that people care about it and (2) a clear explanation of why, and will likely predicate a feature specification that I'll bring to our team. It's .. as much a discussion as it is an artifact that something might have gone sideways for MO after the upgrade. $\endgroup$ – Tim Post Aug 1 '14 at 19:03
  • $\begingroup$ @StefanKohl What did prevent this form of "plagiarism" before? Put differently I do not see a goal of such behavior that would be prevented by the old or envisioned dupeclosure mechanism. $\endgroup$ – user9072 Aug 2 '14 at 15:10
  • $\begingroup$ @StefanKohl I'd say a comment stating it was asked before and linking there makes this sufficiently obvious too. And if one wants one could even edit it in. But MO anyway not being for "original" questions this seems somewhat of a corner case to me regardless. $\endgroup$ – user9072 Aug 2 '14 at 21:31
13
$\begingroup$

The reasoning is pretty simple, closing a question as a duplicate of something asked in the past that didn't get an answer either is .. rude, and often self-defeating. It basically says "That person has no chance of getting an answer, so we'll make sure you don't, either"

This makes sense nearly universally. But this is a research level site, and I realize that introduces a whole new set of dynamics.

I'm going to look into it. It might be possible to add a caveat based on the age of the proposed duplicate. Sometimes, things need fresh attention, or simply perhaps a question asked in a more provocative and interesting way in order to attract the person that might know the answer. But I can see the desire to preserve the chain.

For now, my response is yes - I can see it's a problem here and why, and I'll see what I can do. Updates, of course, coming as they develop.

$\endgroup$
  • $\begingroup$ Tim, I made a feature request on meta.MSE, a year ago, that a duplicate of relatively new questions be allowed (say things not older than a week). This allows to handle unregistered accounts which keep popping up and asking the same questions, or unregistered account on two sites asking the same question and one is migrated to the other. $\endgroup$ – Asaf Karagila Aug 2 '14 at 2:24
  • 4
    $\begingroup$ I don't agree with the interpretation in the first paragraph. Closing the question as a duplicate means that if that person eventually gets an answer, you get it too. The current system ensures (or strongly encourages) that only one of you may get an answer. $\endgroup$ – Emil Jeřábek supports Monica Aug 2 '14 at 10:04
  • 3
    $\begingroup$ It seems to me that the current system is designed for stack overflow, which get hunderds of questions, and the askers are only interested in the answer for a couple of hours or maybe days, so an older unanswered question is effectively dead. On MO, it is fairly common that questions only get answered after many months. $\endgroup$ – Emil Jeřábek supports Monica Aug 2 '14 at 11:04
  • 1
    $\begingroup$ @EmilJeřábek It's definitely a difference between technical, and research-level sites. Stack Overflow can see 15 incarnations of the same question in a single day. On high-scale sites, closing an unanswered question as a dupe of another unanswered question does almost guarantee that neither question gets answered, especially if the dupe is very old and unlikely to churn again. It's not .. a perfect system by any means (and something we're working on). Fortunately, for Math Overflow, it's very much simplified. Just allow it, because it's expected. I'll see what I can come up with :) $\endgroup$ – Tim Post Aug 2 '14 at 15:04
  • $\begingroup$ @EmilJeřábek re old questions getting answered, the revival badge measure this rather well (first answer after 30+ days, and answer getting score +2): Now, MO has about 1.5k revival badges for 52k questions and SO has about 135k for 7700k questions. Thus while slightly more frequent on MO it seems really pretty much the same on MO and SO. $\endgroup$ – user9072 Aug 2 '14 at 15:18
  • 1
    $\begingroup$ @quid: Alright, however what really matters is not how many questions are revived, but the expectation of users asking the questions. If a question is reasked or answered with a month delay on a technical site like SO, chances are the original poster no longer cares, hence it make sense for the system to focus solely on the user experience of the user asking the duplicate question (or users that might get interested in the question in future). It does not work that way on MO. $\endgroup$ – Emil Jeřábek supports Monica Aug 2 '14 at 18:01
  • $\begingroup$ @EmilJeřábek I know neither SO nor the habits of programmers well enough to judge what is typical there. What I do know however is that it is not unusual for a mathematician's questions especially one asked on MO to become obsolete quickly (too). Anyway, personally, I (also) prefer if dupe-closure of unanswered questions were again possible. Still, on the one hand I was genuinely curious if what you said was refelected in data, and on the other hand, I somewhat dislike sweeping statements how MO is so different, while really it is not that much. Hence my comment. $\endgroup$ – user9072 Aug 2 '14 at 21:26
3
$\begingroup$

One, perhaps silly, approach would be to put an "answer" on the original question (thereby bumping it on the front page) stating that this question has been re-asked (and give a link). Then the new one could be closed as a duplicate, thus linking back to the original question. The spurious answer could be moved into a comment to maintain the link to the newer version of the question.

However, if there are moves afoot in SE headquarters, then this may be moot.

$\endgroup$
  • 3
    $\begingroup$ The new question can only be closed as a duplicate if the old question has an upvoted or accepted answer, thus this requires two people (someone must get the hint and upvote the spurious "answer"). And since conversion to a comment needs a moderator, who can close anything as a duplicate without restrictions anyway, one can as well avoid all this and just flag the duplicate for moderator attention. $\endgroup$ – Emil Jeřábek supports Monica Aug 2 '14 at 14:23
  • 1
    $\begingroup$ This procedure seems too complicate to be workable, and I think bumping the old question better should be avoided. Why not just close the new with costum reason: "This question should be close as the same question was already asked, see [...]" In addition one could edit the link into OP to be still closer to what dupe-closure does. (Personally I am not even sure there is any need for a closure here; the above is under the assumption one wants to close.) $\endgroup$ – user9072 Aug 2 '14 at 15:07
  • $\begingroup$ Fair enough. The system wasn't quite as simple as I'd imagined. $\endgroup$ – David Roberts Aug 2 '14 at 22:25
1
$\begingroup$

What about the option to merge the original with the duplicate? If the original has no answer then the merge should be pretty straight-forward. The new question could have the wording of the original and the wording of the re-ask, perhaps under separate headers. The merge could also port over comments from the original thread. I don't know if the new platform allows merging but I remember that the old one did and hopefully that functionality is still present in 2.0.

$\endgroup$
  • $\begingroup$ Only moderators can merge questions. $\endgroup$ – François G. Dorais Aug 5 '14 at 13:03
  • 1
    $\begingroup$ Right, but it seemed like other solutions also required flagging for moderator attention. So the effort from the community would be no larger if my suggestion were carried out, but the output in terms of consolidation might be better. $\endgroup$ – David White Aug 6 '14 at 0:00

You must log in to answer this question.

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