What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 128 Stack Exchange communities.

Duplicate questions are a drain on the trilogy community -- not necessarily a big drain, but a drain. Either you have to be a duplicate cop, or you find yourself answering a question only to have it closed as a duplicate.

I suggest these changes, the goal being to encourage people to search before posting:

  1. Even before the question is closed, list the links to questions people have voted as duplicates for everyone (not just people with close rights) to see, including the questioner. Remove these if the question survives for a day (or change them to "related questions"). (Edit: I just went to post this as its own feature request, separate from the rest of this suggestion, because it has utility beyond the thrust of this idea. But someone already has [and quite some time ago]. If you have an opinion about this point separate from the rest, I recommend going and voting on / commenting on that question.)

  2. If a question is closed as a duplicate, ding the questioner's reputation by a smallish, but real, amount; say -10. This can be avoided if the questioner sees people are voting to close it, looks at the questions they're pointing to, and deletes the question themselves. (Edit: Farseeker and others have pointed out that there should be a threshold so this doesn't hit newbies. I'm all in favor, that's a very good idea. It can be based either on rep, or on number of questions closed as duplicates. I prefer the latter -- Farseeker's suggestion was three, which seems reasonable.)

  3. Obviously, if this isn't done already, reverse any rep gained from up-votes on the duplicate question. (Edit: Gnome points out that this is currently done, but only if the question ultimately gets deleted, which sets the bar very high indeed. I submit that it being closed is sufficient.)

Obviously, I did search before posting this! And reviewed the "Related Questions" list carefully, but to my surprise I didn't see this suggestion made.

share|improve this question
Related to your "related" quandry: meta.stackexchange.com/questions/34460/… –  Mark Henderson Mar 22 '10 at 8:31
Also related, but again not quite the same: meta.stackexchange.com/questions/4303/… –  T.J. Crowder Mar 22 '10 at 8:37
+1 The line between duplicated questions and related questions is a fine one - I would like to see that argument expanded on (but that would mean that this could well be a duplicate comment). –  amelvin Mar 22 '10 at 11:41
see also meta.stackexchange.com/questions/37466/… –  Ian Ringrose Mar 22 '10 at 13:38
add comment

7 Answers 7

I would also want to penalise the people who answer questions that get closed as duplicates. Some of them appear to be doing this quite cynically, and are in effect gaming the system. I suggest -10 rep.

share|improve this answer
If you're referring to me (Jimmy) @ stackoverflow.com/questions/2491177, after I saw your comment, I was the second close vote and I voted up your comment. I'm not going to delete my accepted answer, because the suggested dupe doesn't even mention bitbucket, Not that I disagree with your suggestion, just being clear I had no intention of gaming. –  Jimmy Mar 22 '10 at 10:57
@Neil: I disagree. If someone takes the time to properly answer a question, I don't think we should penalize them. I'm a bit surprised, actually, that they lose the rep if the question is ultimately deleted (see Gnome's comments...just about anywhere in this question, really. :-) ) That's surely penalization enough; they've wasted their time. –  T.J. Crowder Mar 22 '10 at 11:03
@T.J. That penalty is only incurred if the question actually gets closed. Lots of things that are blatantly dupes don't. –  nb69307 Mar 22 '10 at 11:37
Not sure if expecting people to check whether a question is a duplicate before they answer would work - wouldn't it stop fast answers being produced? –  amelvin Mar 22 '10 at 11:45
@Neil: I've seen this discussed on meta before (while reading archives, effectively, from around last Summer), and a few people made the point that it's considered polite to provide answers that might help OPs asking duplicate questions (legitimate answers, not comments). If you point out to someone blatantly gaming that if once the question gets closed+deleted, they will lose any gains, don't you think that could be enough to encourage desirable behavior from them instead? –  Gnome Mar 22 '10 at 12:00
add comment

You're acting like all duplicates are a bad thing.

If a question is hard to locate (perhaps because of multiple different sets of terminology), then a duplicate that uses different terminology to express the same thing, and comes with a link is a good thing. It increases the visibility of the question that has the answers. Several times I've found solutions to problems I have had by first finding a closed question and then following the duplicate link.

Unless you can find a way to differentiate between the 15 millionth repeat of a simple question or a alternate phrasing of an existing question I don't think there should be any penalisation.

Add to that the issue that you don't want to discourage question asking. Surely it's better to have a few duplicate questions now and again (that are quickly closed by the community - so is it really a big problem) than discourage people from asking them in the first place.

share|improve this answer
+1 - I started an answer trying to say this, but it didn't read right or as well as this one. –  ChrisF Mar 22 '10 at 9:28
Most of the duplicates I've seen have been a complete waste of time, very few of them have increased the "question surface area" (as Farseeker put it). You're assuming the questioners have searched but failed to find; I don't think that's usually the case. But you're right, it does happen. Perhaps the cost should kick in if the question is deleted, rather than simply if it's closed. (blog.stackoverflow.com/2009/04/…) That sets the bar very high indeed. And actually, perhaps that's a whole different way to go, not duplicates-specific. –  T.J. Crowder Mar 22 '10 at 9:50
@TJ: That's what happens currently (during a recalc you "lose" any rep for posts deleted since your last recalc), in case you missed that part of my answer. What advantage do you see in tacking on another -10? Don't make deleting posts about rep punishments, keep it about improving the site. –  Gnome Mar 22 '10 at 10:16
@Gnome: I do understand your point, and did when you made it originally in your answer. I just disagree. (No need to shout, btw! :-) ) –  T.J. Crowder Mar 22 '10 at 10:59
+1 You're right to say that discouraging questions - especially from new users - is a bad thing. –  amelvin Mar 22 '10 at 11:43
There is no shortage of questions at all. And all users of web resources like SO should learn to lurk a bit before posting. –  nb69307 Mar 22 '10 at 11:50
add comment

There should not be a cost, but remember there is currently no gain (eventually):

Questions and answers sometimes get deleted; votes on deleted posts do not count toward your reputation. [blog]

There will always be a need for questions and putting up barriers to asking is not smart. Such punishment will make anyone think twice before asking anything, and while that might be a good thing from a pedogogical standpoint, it will hurt the site. The keys is merely don't reward mildly undesired behavior (too much).

Not to mention dealing with potential abuse and gaming of any system that removes reputation—some people love to screw with others.

share|improve this answer
The goal isn't punishment, but incentive to search. I think if the cost is smallish, it's not a barrier. Regarding gaming the system: I don't think this changes anything in that regard. The only gaming would be to vote to close questions that aren't actually duplicates, which the SO community will catch and which will come to moderator attention if done persistently. –  T.J. Crowder Mar 22 '10 at 9:02
@TJ: Punishment is incentive, it's negative incentive. You only have 402 rep on meta. Imagine you stood a chance to lose just 40 rep if this got dupe-closed, that's not a lot, right? Just 4x answer upvotes. But it's still 10% of your current rep, and some dupes are really hard to find if you haven't seen/read it before. Intermittent reinforcement works best, and that's what you get from upvotes and rep.</armchair-psych> –  Gnome Mar 22 '10 at 9:36
@Gnome: My suggestion is -10 per duplicate closed, not -10 per close vote. And note that with my suggestion, if the questioner is actively involved in the question, they'll likely be able to avoid even that (and get their answer) by seeing and following the links to the questions people have said were duplicates and either A) Saying why they're not, or B) Deleting their duplicate themselves. –  T.J. Crowder Mar 22 '10 at 9:55
@TJ: If the punishment is so small that it doesn't matter for anyone but still prevents some good questions from being asked (because they want to avoid stigma, etc.), what's the point? Take Simon's approach, why are dupes bad? Aren't all/enough of the obvious dupes being dealt with appropriately? As I point out here, any rep for dupe-closed-deleted questions is "taken away" on the next recalc (and everyone's getting recalced today). –  Gnome Mar 22 '10 at 10:12
@Gnome: Thanks for clarifying the closed/deleted thing (my take away from your earlier link was closed, not deleted). I understand what you're saying, I just don't agree with it. –  T.J. Crowder Mar 22 '10 at 11:00
+1 - There will always be a need for questions and putting up barriers to asking is not smart - agreed. –  amelvin Mar 22 '10 at 11:46
Gnome, I'm curious, which triology+plus site(s) are you active on? The OpenID you're using here doesn't have a match on SO, SF, or SU. Looking for "gnome" on some of the plus sites (doctype.com, etc.) didn't turn up any obvious matches, but it's an imperfect method. It's just that you clearly have a StackExchange interest, being so active here on meta... –  T.J. Crowder Mar 23 '10 at 7:58
@TJC: I have cleverly hidden a reference to my SO account on my meta profile. –  Gnome Mar 23 '10 at 12:29
@Gnome: LOL, I'm so used to seeing StackOverflow profiles that link back to themselves, I just read right past it, missing the minor detail that this is meta. Why not link up? –  T.J. Crowder Mar 23 '10 at 13:02
@TJC: As I got more active on meta I embraced the---how to put it?--- unique unicorn-and-waffle aspect, and thought I might want to keep the accounts separate. –  Gnome Mar 23 '10 at 13:05
add comment

Hmm, I'm wary about #2. I can see how it's a problem, but the main issue that I have with it is that some dupes are impossible to find. When one person uses the term "PSU" and someone else uses the term "Power Supply", with no cross-over in terms, you'll never find the question. And if you can't find it, it's not fair to penalise the OP.

When someone creates a duplicate question, and the question is closed as dupe with a link back to the original, you're greatly increasing the surface area of the question, which means that next time someone comes along you can find the question easilly, regardless of which search term you use.

Secondly, the SOFU search leaves a lot to be desired. I've given up on it and just use google site:serverfault.com search. But you really can't expect a novice user to know or try this.

If this were to be considered, I would really strongly suggest it have thresholds in place. Maybe after the 3rd dupe or some other metric (that would need to be determined by studying the number of 'close as dupe' questions for each user).

With #3 - totally agree

share|improve this answer
I like your threshold idea, although in some ways I think a threshold already exists: You have to have gained at least 10 reputation before you can lose it, and if you've done that, you've done some kicking around the site already and should know about duplicates. But I still like your idea. Also agreed about searching: I do both, on-site using tags and then via Google. –  T.J. Crowder Mar 22 '10 at 8:39
That's a good point. If they've got rep to loose then the've already been active on the site, and a -10 to a 1 rep user (i.e. a first time offender - I use that word very lightly) isn't going to penalise them... Hmm... –  Mark Henderson Mar 22 '10 at 8:41
add comment

I think this really is an issue. Especially superuser (if I remember it right, there was a big discussion about this a few weeks ago) has big problems with duplicates, and a massive trend of experienced users not to vote to close them, but to answer them.

  1. Totally seconded

  2. would not be necessary IMO if the loss of reputation from the question's closure would be calculated immediately - I am pretty active on SO and I didn't know you actually lose the reputation gained from closed questions at re-calc.

  3. See 2. - if you feel the effect of answering to a duplicate straight away, you are going to have a lot less incentive to answer it.

Moderators would have to be able to reverse such close votes, but with no negative effect kicking in, the possibilities to screw with people (as feared in the comments) would be much reduced.

This would prevent questions from being answered of which people know deep down that is must be a duplicate, and make those people search for the duplicate instead. Maybe this would be best paired with a (very small) incentive on finding and voting duplicates.

share|improve this answer
Re #1 -- turns out someone's already recommended this, and quite some time ago: meta.stackexchange.com/questions/1383/… As you seem to agree that it's a good idea, suggest moseying over and voting. ;-) –  T.J. Crowder Mar 22 '10 at 10:11
@Pekka: Closed doesn't matter for a recalc: deleted matters. The obvious solution to the problem you describe is to get active 10k users who go through and delete closed dupes, get regular (e.g. biweekly or monthly) recalcs, etc. But I'm not active on SU. –  Gnome Mar 22 '10 at 10:18
@T.J. Done. @Gnome Ah, I see, so there is a difference. Something would really have to be done about that IMO. –  Pëkka Mar 22 '10 at 10:27
@Gnome: Thanks for clarifying the closed/deleted thing. –  T.J. Crowder Mar 22 '10 at 10:59
add comment

How about just halve the reputation gain upon duplicate closed Q/A on recalc?

share|improve this answer
I thought that all of the reputation gain for closed questions was discarded on recalc. At least, that's Gnome's claim here meta.stackexchange.com/questions/2038/… although he doesn't cite a source. I couldn't immediately find one, but have no reason to doubt him. –  T.J. Crowder Mar 22 '10 at 10:16
@Gnome: Thanks for clarifying that. –  T.J. Crowder Mar 22 '10 at 11:00
@TJ, sure thing, but will delete since I updated it into my answer. (I like clean comments sometimes :P) –  Gnome Mar 22 '10 at 11:10
add comment

I think @gnome (+1) is right when he talks about barriers to question entry, just noted down the rep of every person asking a question on the first two pages of SO right now :

85 - 28 - 1 - 58 - 16 - 868 - 1 - 394 - 8 - 12.5K - 64 - 814 - 704 - 1 - 18 - 175 - 161 - 868 - 814 - 321 - 147 - 29 - 1 - 1493 - 126 - 457 - 1 - 22

So in this sample the majority of questions are being asked by relative newcomers to the site - the people least likely to 'know' that they should be searching for duplicates in the first place. If this is a representative sample (and based on what I see day-to-day I think it is) duplicate question sanctions will penalise newcomers and this will surely impact on the flow of questions?

Instead if duplicates are a real evil punish those that indulge in them when they are experienced enough to know better rather than when it is more likely to be an honest mistake due to site inexperience.

So why not only penalise duplicators who have the rep to edit questions as by that point they really should know better.

I already think that the line between related questions and duplicate questions is fairly blurred and this measure would make SO even more expert-friendly (as opposed to user-friendly).

share|improve this answer
"Instead if duplicates are a real evil punish those that indulge in them when they are experienced enough to know better rather than when it is more likely to be an honest mistake due to site inexperience." Completely agreed, and this is a theme that Farseeker called out early on as well. I think the threshold idea is a very good one. –  T.J. Crowder Mar 22 '10 at 11:02
I don't agree: Newbie or not, not taking the time to search before asking is a behaviour that I want to discourage. –  rds Feb 18 '13 at 9:33
add comment

You must log in to answer this question.

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