3
$\begingroup$

In a couple of recent situations I have found myself at odds with the duplicate question policy of the site. The most recent case was this question Proof of Ohm's Law [duplicate] that was marked as a duplicate of two other questions which not only were asking different things but also none of the answers really addressed the OP's question. I actually found a good answer for the question in this arXiv article but I could not answer because the question was already marked as duplicate (I would need to read the article very carefully because I find it a bit complicated).

What I want to note with this is that the duplicate policy (as it is now) has several flaws that reduce the effectivity of the site as a place to share knowledge:

  • Some questions that ask different things are marked as duplicates if the prior question has an answer that may vaguely be related to the latter question. This is clearly a problem because when one answers a question, if you do it right, you have to take care to address the specifics of the question. Redirecting a question to one that is only somewhat related will very likely undermine the nuances of it, and it may not even be a proper answer for the question.

  • Physics evolves and hundreds of articles are published every day. For example if I ask about cosmic censorship in an unspecific way, the question would be immediately marked as duplicate. No one would be able to mention the existence of the $C^{0}$ stability idea, which is a rather new topic. The result is that even if there is new physics and new ideas, these cannot be shared as effectively because something related has been asked before.

  • There is no specific set of guidelines that render a question a duplicate. This makes the decision of marking a question as duplicate a personal decision of the mod, if you are unlucky enough to find an intransigent mod (which are a thing in PSE) before you get a good answer then you will never get an answer. The question boils down to: What constitutes a duplicate question? Is it an exact & paste? Is it asking the same thing but with other words? Is it asking for something previously answered but with a different focus? Is it asking for something already answered but in more detail? Is it asking for something that has not been asked before but someone already answered in another question? Is it asking for something related to an answer to an existing question? ─The problem is that for me only the first two would be duplicates whereas for some users, all of these consititute duplicates.

  • There is no formal mechanism to appeal to a duplicate question. You can flag, delete, edit and report questions with a single button, but there is no real way to appeal if your question (or one you like) gets marked as duplicate, unless you are a mod or have the gold badge, which most users are not and do not.

I think the solution could be to make a more formal process of appeal (resembling the review queues) where the user is allowed to make a case for his question if he wants. And should the appeal be rejected, a good reason should be given (at least a better reason than "it's duplicate because I say so" which is what is basically implied when the only comment is "possible duplicate URL", and any debate is dismissed or ignored).

$\endgroup$
15
  • 13
    $\begingroup$ You are mistaken, there is an easy and "formal" way to appeal a duplicate closure (like any other closure): Edit the question, and explain how it is different from the duplicates. The first edit after closure automatically places the question in the reopen review queue. A single reopen vote (from any user with more than 3k reputation) will also place it there. $\endgroup$ – ACuriousMind♦ May 13 '19 at 21:21
  • 4
    $\begingroup$ In regards to your second point, there's no prohibition against answering old questions. If you think some question has had a new development, just add a new answer. $\endgroup$ – knzhou May 13 '19 at 22:58
  • 4
    $\begingroup$ A majority of questions closed as a duplicate are closed by a consensus of five regular users instead of a unilateral moderator closure (source: this SEDE query - though admittedly the balance is less direct than for off-topic closures). Getting caught up in the "intransigent mod" paradigm largely ignores the way most closures go down. $\endgroup$ – Emilio Pisanty May 13 '19 at 22:59
  • 1
    $\begingroup$ As for the specific example that you mention - frankly, in this situation, if you have a clear mental picture of a suitable question and answer where the answer provides substantial new content and the question clearly delineates the existing Q&As on the site, sets itself apart from them, and motivates the extensions, then I would encourage you to post a full Q&A with both. It will serve the site far better than the existing (thinly researched, and frankly pretty lazy) duplicate that you linked to, particularly given OP's lack of engagement. $\endgroup$ – Emilio Pisanty May 13 '19 at 23:09
  • $\begingroup$ @ACuriousMind it seems to me that such a system does not work very well. For one, the users with more than 3k reputation are less than 0.3% of the PSE users. Second, there is no change in the user interface to follow a "reopen ticket" (or however you wish to call it) and finally, there is no feedback on why the question was not accepted after the appeal (i.e., no counter-arguments to the reasons you believe your question should be reopened). If it were a court it would be the equivalent of you presenting an alibi but getting convicted anyway without even hearing a veredict. $\endgroup$ – user137661 May 14 '19 at 4:16
  • 1
    $\begingroup$ @knzhou how do you suppose anyone could answer a question that they never see? Unless you dig for old questions that is not going to happen because "that has been asked before" and there will therefore be no new questions on that topic. $\endgroup$ – user137661 May 14 '19 at 4:20
  • $\begingroup$ @EmilioPisanty on most of the closed questions I have seen I usually see only the name of one mod or maybe two. I assumed that means that such a thing was decided unilaterally but maybe it only uses the name of the mod that checked it. I am not versed in the subtleties of the SE interface. On the other hand, I will try to make a Q&A version with the material I posted, but that will probably take some time. $\endgroup$ – user137661 May 14 '19 at 4:26
  • 3
    $\begingroup$ I think you've got a good point Salvador. IMHO people here are too quick to close an interesting challenging question as a duplicate. Then when you look at the duplicate, you see that it's ancient, it isn't really a duplicate, the answers are unsatisfactory, and some of the names are familiar. $\endgroup$ – John Duffield May 14 '19 at 7:25
  • 2
    $\begingroup$ "the users with more than 3k reputation are less than 0.3% of the PSE users" I'm not sure if that's a correct number, but it's still a completely redundant point because (1) Everyone who has enough rep to cast close votes has enough rep to cast reopen votes (2) A huge number of 3k rep users are not active at all, and a number of them who are active don't participate in moderation like closing and reopening. $\endgroup$ – user191954 May 14 '19 at 7:25
  • $\begingroup$ @Rishi it is about 0.26% in number of existing accounts. As you can see I was answering an argument by ACuriousMind who seems to think that the current system works. If you assume that low rep and high rep users are similarly active (although admittedly this is a big interpretative leap) then it would be a meaningful fraction of the number of users that can use the mechanism described by ACuriousMind above. $\endgroup$ – user137661 May 14 '19 at 14:17
  • $\begingroup$ "how do you suppose anyone could answer a question that they never see?" I thought you were complaining about new questions being closed as duplicates of old ones. Then it's easy to find the old one, because there's a link to it right in the closing text. $\endgroup$ – knzhou May 14 '19 at 14:22
  • $\begingroup$ @knzhou the question feed shows only active questions. Questions that are marked as duplicate will disappear from the feed in just a few hours so they will be much harder for anyone to find just as much as their old counterparts. $\endgroup$ – user137661 May 14 '19 at 14:38
  • 5
    $\begingroup$ @SalvadorVillarreal I thought you were complaining about seeing questions but then being unable to answer them. If you see a question, you can either answer it, or if it's been closed as a duplicate, answer the original question, so what is the problem? $\endgroup$ – knzhou May 14 '19 at 14:41
  • 4
    $\begingroup$ @SalvadorVillarreal Also, all questions disappear from the global feed quickly, so I'm really not sure what you're specifically complaining about. If you want to avoid missing questions about a subject you like, you can just look at the feed for that individual tag, which will move much more slowly. $\endgroup$ – knzhou May 14 '19 at 14:43
  • 2
    $\begingroup$ "the users with more than 3k reputation are less than 0.3% of the PSE users" - you're off by an order of magnitude. Users at 3k+ rep are about 2% to 3% of the active user population if you define the latter as having posted within the past 6 or 12 months -- and it rises to 6% if you shorten the cutoff to 1 month. (See this SEDE query for the data.) $\endgroup$ – Emilio Pisanty May 16 '19 at 10:48
1
$\begingroup$

For clarity, let me respond point by point:

  • Some questions that ask different things are marked as duplicates if the prior question has an answer that may vaguely be related to the latter question. This is clearly a problem because when one answers a question, if you do it right, you have to take care to address the specifics of the question. Redirecting a question to one that is only somewhat related will very likely undermine the nuances of it, and it may not even be a proper answer for the question.

Our duplicate policy is that questions which are asking the same thing should be marked as duplicates of each other. Questions which are asking different things should not be. If people are marking questions as duplicates in a way that contravenes the policy, I'd agree that is something that should be improved, but it's not a strike against the policy itself.

  • Physics evolves and hundreds of articles are published every day. For example if I ask about cosmic censorship in an unspecific way, the question would be immediately marked as duplicate. No one would be able to mention the existence of the C0 stability idea, which is a rather new topic. The result is that even if there is new physics and new ideas, these cannot be shared as effectively because something related has been asked before.

Again, questions should be marked as duplicates if they are asking the same thing, and not otherwise. I don't see how the fact that hundreds of articles are published every day has anything to do with that, and I don't understand the reasoning by which to conclude that the duplicate policy prevents people from asking about C0 stability.

If you edit the question to clarify, maybe I can give more information.

  • There is no specific set of guidelines that render a question a duplicate. This makes the decision of marking a question as duplicate a personal decision of the mod, if you are unlucky enough to find an intransigent mod (which are a thing in PSE) before you get a good answer then you will never get an answer.

Well, again, the "specific" guideline is that questions which are asking the same thing should be marked as duplicates, and those which aren't asking the same thing should not be. You may want something more specific, but there are many cases (including, I would argue, this one) where we've found it harmful to try to make the guidelines too specific, because we can't anticipate all the actual situations that arise.

The question boils down to: What constitutes a duplicate question? Is it an exact & paste? Is it asking the same thing but with other words? Is it asking for something previously answered but with a different focus? Is it asking for something already answered but in more detail? Is it asking for something that has not been asked before but someone already answered in another question? Is it asking for something related to an answer to an existing question? ─The problem is that for me only the first two would be duplicates whereas for some users, all of these consititute duplicates.

I believe your instinct is correct, that only the first two of these are necessarily duplicates.

  • There is no formal mechanism to appeal to a duplicate question. You can flag, delete, edit and report questions with a single button, but there is no real way to appeal if your question (or one you like) gets marked as duplicate, unless you are a mod or have the gold badge, which most users are not and do not.

As mentioned by some people in the comments, that's simply not true. To quote ACuriousMind:

You are mistaken, there is an easy and "formal" way to appeal a duplicate closure (like any other closure): Edit the question, and explain how it is different from the duplicates. The first edit after closure automatically places the question in the reopen review queue. A single reopen vote (from any user with more than 3k reputation) will also place it there.

$\endgroup$
4
  • $\begingroup$ " I don't see how the fact that hundreds of articles are published every day has anything to do with that, and I don't understand the reasoning by which to conclude that the duplicate policy prevents people from asking about C0 stability." What I meant is that you can even have an exact copy of a question, but what the answers entail can be very different (broader) if it is asked at a later time. $\endgroup$ – user137661 May 16 '19 at 15:41
  • $\begingroup$ And the reason I believe the "formal" process is not very good (although it is true it doesn't exist) is, as I explained in the other comment sections, is that even if you appeal and you make a case for your question, the request can be rejected without even sending a notification, much less with an explanation for such a rejection (I know this for a fact because I have experienced it before). $\endgroup$ – user137661 May 16 '19 at 15:45
  • 1
    $\begingroup$ but what the answers entail can be very different (broader) if it is asked at a later time true, but you can always answer the older question with newer information, so I'm not sure the point you're making here. $\endgroup$ – Kyle Kanos May 16 '19 at 18:06
  • $\begingroup$ Most questions that are closed as duplicates are not on "hot" topics on which there is evolving literature. Quite to the contrary, most duplicates or near-duplicates are on fairly mundane topics since by simple common sense questions on these topics are more likely to be asked more than once whereas more specialized questions on very recent topics have narrower appeal. $\endgroup$ – ZeroTheHero May 17 '19 at 2:36

You must log in to answer this question.