826

At present if you see a question that may be a duplicate and has an easy answer you can post the answers or post a link to the duplicate question. (Or the few hi-rep users can vote to close it as duplicate, think of normal users here, not the people that read Meta)

It is better for Stack Overflow if a link to the duplicate is posted as a comment, or the user votes to close the question as a duplicate. However, the user gets more rep if he/she posts the easy answer quickly.

I'm thinking of something along the lines of:

  • First person to post a comment that points to possible duplicate gets some rep if the question is closed as a duplicate and at least 2 of the other closers chose that same dupe question.

  • Whenever someone votes to close as a duplicate, a “possible duplicate” comment should be posted with a link to the other question, if there is not already a link to it in a comments. (The system now does this)


As Ether said, The removal of reputation earned from answering a question that is later closed as a duplicate would at least remove the negative incentive, but is that enough to correct this behaviour?

Is a “duplicate finder” badge part of the solution?

  • A bronze badge for being the first person to vote to close a question as a duplicate with the same “duplicate of” question chosen as most closers choose.
  • A silver badge for doing the same 25 times.

See also Dr. Strangedupe: Or, How I Learned to Stop Worrying And Love Duplication

Improve tools for closing as duplicate, would also help.


Now that the duplicate can be marked by the OP as helpful, we have another trigger to consider for awarding badges and/or rep.

26
  • 67
    I know you were just throwing ideas out there, but I think this might work better as a badge instead of giving rep for it.
    – Jon Seigel
    Jan 29, 2010 at 18:53
  • 62
    Not sure if these suggestions will work, but right now the incentives encourage rather than discourage duplicates, so I would like to see something done. Jan 29, 2010 at 18:58
  • 77
    I've seen people (mostly, moderate rep users: 1-5k) post duplicate links as "answers". This should be discouraged IMHO. I've no practical solution at the moment though.
    – mmx
    Jan 29, 2010 at 19:49
  • 11
    @Jon, the advantage of giving rep is that new users see it on the page of how to get rep; I don't know if badges motive anyone. However there is no reason not to have a badge(s). Jan 29, 2010 at 20:39
  • I tried this, it got nowhere meta.stackexchange.com/questions/9686/…
    – waffles
    Jan 29, 2010 at 22:10
  • 1
    In the case of duplicate questions, I usually say "as was said in so-and-so's answer on this similar question [link], check this, do that, and record the other. I, too am frustrated at the repeats...
    – gWaldo
    Sep 28, 2010 at 15:19
  • What about an incentive for intentionally writing a question that could collapse a collection of closely-related (exact duplicate or not) questions? The idea of curating a smaller set of more "canonical" questions and answers for themes that repeat themselves constantly would make the site easier to use. The original questions could be linked (or merged, as appropriate) to a distinguished Q+A, for which the curator receives additional rep (maybe by counting incoming links from other questions?), perhaps.
    – andersoj
    Nov 5, 2010 at 3:51
  • 18
    A hundrer upvotes and no official comment from the powers-that-be? That seems odd.
    – MPelletier
    Nov 28, 2010 at 15:26
  • 2
    See Jeff Atwood's insightful comment on the effects of this on participation, at blog.stackoverflow.com/2011/01/…. Jan 7, 2011 at 23:29
  • @Reno: Isn't that obsolete due to the new flagging system? I don't have vote power, so I just flag dupes.
    – John
    Mar 5, 2011 at 23:58
  • @John New flagging system? Do you mean by clicking It needs moderator attention -> other ?. I'm sorry i don't know. Anyway the HQ chat room helps close dupes faster.
    – Reno
    Mar 7, 2011 at 7:27
  • @Reno: For me, I can click "It doesn't belong here" and it lists the five close reasons. I can flag a question with any one of them. Note: "It doesn't belong here" doesn't show up on answers, only questions.
    – John
    Mar 7, 2011 at 15:29
  • 1
    @Reno: See here for more info on the new flagging system.
    – John
    Mar 7, 2011 at 15:36
  • 1
    @Mehrdad: I've started a posse to clean those up. Over 600 have been removed so far.
    – hammar
    Aug 30, 2011 at 19:11
  • 3
    … have good intentions and want to help people, they are currently encouraged by the system to copy 90% of another answer and change a few minor things rather than closing as duplicate and letting the asker figure out how to apply the knowledge from the parent post themselves. This also applies to a lot of other close reasons, basically everything except "unclear". Jan 16, 2020 at 17:23

25 Answers 25

413
+750

I'd also like to see the removal of reputation earned from answering a question that is later closed as a duplicate. Currently there is little incentive to post an answer on the original question (or merely post a link to the original), rather than re-answering the question. It's tiresome seeing the same answers again and again, and seeing this behaviour awarded with reputation.

18
  • 59
    This is one of the best ideas I've heard for this problem in a long time. However, at The Great Rep Recalc (if it ever happens), I can imagine a sudden surge in Meta questions from relative newbies wondering where their hard-earned Internet dollars went. Still think this is a great idea, though.
    – John Rudy
    Jan 29, 2010 at 20:46
  • 7
    This was wrong. See Jeff's comment below. <strike>The reputation gained for answers to closed questions does get removed eventually, when reputation is recalculated.</strike> Jan 29, 2010 at 23:16
  • 8
    @Æther: We just need to get Jeff to do global recalcs more often than once in the history of the site. :) Jan 29, 2010 at 23:23
  • 1
    @Bill: aye, or schedule targeted recalcs on users if they are affected by various things like being involved in a closed or deleted question.
    – Ether
    Jan 29, 2010 at 23:56
  • 1
    @Bill: what I find interesting is that an upvote on a post that has already been closed still results in a rep gain. No wonder so many people lose tons of rep on a recalc...
    – Ether
    Feb 1, 2010 at 23:39
  • 70
    @bill that is not correct -- reputation is only removed on recalc from deleted posts. We still need some duplicate questions around to cover users who have the uncanniest ability to ask the same exact question using zero words in common. Feb 2, 2010 at 4:26
  • 8
    @Jeff: except that's rarely the cause: it's far far more probable that they either failed to search, or the search algorithm failed them (seems like more the former than the latter, these days).
    – Ether
    Feb 2, 2010 at 4:32
  • 8
    @Ether: if you think the users search first you are delusional :P If you think that they should then I probably agree but the reality is that they don't, so what they should do is irrelevant. Feb 7, 2010 at 19:34
  • 1
    Jeff is against this, so it ain't gonna happen: blog.stackoverflow.com/2011/01/… Jan 6, 2011 at 7:05
  • 9
    @Andrew: well it wouldn't be the first time that Jeff is wrong :)
    – Ether
    Jan 7, 2011 at 18:30
  • 5
    I'm so tempted to flag this as 'Not an Answer' :P
    – Doorknob
    Feb 5, 2013 at 14:00
  • 7
    This would effectively solve this problem. How many times would a user have to see the message "Reputation removed because you answered a question which turned out to be a duplicate. Click here for more details." before they started searching for duplicates BEFORE posting an answer?
    – lnafziger
    Jul 12, 2013 at 15:52
  • 2
    I would be in favour of this, provided that there is an increase in merging of answers and that any answers merged into the original question get to keep their reputation. This means we don't penalise a really useful answer simply because an (obscure) duplicate wasn't noticed immediately. Apr 30, 2014 at 13:40
  • 2
    I would only do this if the question is closed shortly after being asked. I have dupeclosed questions asked like 5 years before. At that point sometimes the older one has been less popular for some reason, so I close it as a dupe of the newer. That would be wrong if the first answerers lost their rep. And closing the newer but more popular one would be fighting against the community.
    – Oriol
    Jun 12, 2016 at 22:40
  • 1
    PLEASE yes. I've seen users with high enough rep who seem to wait for a duplicate, pounce with low-quality answers on a low-quality question, harvest rep, do nothing as question is duplicate-marked . . . and repeat. This is a biggie. Jul 7, 2018 at 0:28
275
+650

I'm reviving this thread because I'm surprised there's no badge offered for this. Looking for duplicates requires more effort than other types of close vote and we're always saying that badges should be encouraging positive behaviour on the sites.

I propose the following badges:

Seeker - First question closed as a duplicate of another question you found.

Scout - 50 questions closed as a duplicate of another question you found.

Reconnoiterer - 200 questions closed as a duplicate of another question you found.

The underlying criteria would require at least 3 of the total close votes to concur with the found duplicate. Reconnoiterer could possibly be a repeatable to encourage continued behaviour. I can't see any "gaming" issues arising from this, because it requires agreement from other members of the community.

Clarification: "found" equates to finding a duplicate and casting a vote to close on that duplicate, or flagged for review with the appropriate duplicate chosen.

24
  • 1
    How would you propose tracking who found a dupe? It would be easy if you also had close vote privileges, but what if someone without close rep -- not saying "3k" because of the betas -- found one and submitted it via a mod flag, or something?
    – Pops
    Sep 28, 2010 at 14:13
  • 2
    @Popular: I guess the wording could be a little better (I was trying to make it consistent with the current set of badges *wink*). This would apply only to user who cast the first close vote identifying that particular duplicate. So, to clarify, User A votes to close a question with Dupe A. User B comes along and votes with Dupe B instead, then 2 other users agree with him and 1 agrees with User A. User B's dupe has 3 votes, User A's has 2, so only User B's count towards these badges is increased.
    – Andy E
    Sep 28, 2010 at 14:26
  • 3
    I like your idea, but I'm not sure I can get on board with badges only for people with close-vote rep, especially if they're doing something that anyone can do (finding dupes). Closing dupes... well, that's its own reward.
    – Pops
    Sep 28, 2010 at 14:49
  • 4
    @Pop: There's a few other badges that require rep. Copy Editor and Strunk & White require editing privileges, which is only 1000 less than closing privileges. Hell, Legendary requires a minimum of 30,000 reputation ;-)
    – Andy E
    Sep 28, 2010 at 15:05
  • 1
    @Andy, yeah... not a huge fan of Copy Editor/Strunk & White either. As a former copy editor, I think they reward a desperately needed behavior, but it does bother me that you can't even start working towards them for a while. Legendary, at least, you can start working on on day one... but it's really its own animal.
    – Pops
    Sep 28, 2010 at 15:22
  • 35
    I fully support this, I'm constantly hunting dupes and I find very depressing to see that some users prefer answering obvious dupes to get some easy rep instead of closing, diluting existing (good) stuff. The system really needs to encourage the "right" behavior with some incentive. Sep 29, 2010 at 5:41
  • 4
    It seems to me that the easiest way to link the duplicate to the "finder" is to expand the options under "flag" to include "duplicate" with a required field for the URL of the duplicate.
    – ale
    Nov 4, 2010 at 0:22
  • 1
    I like what jdv come up with with looking in comments for the first link to the duplicate if it predate the first close vote. Nov 23, 2010 at 17:19
  • 1
    I prefer: Scout/Sharpshooter/EagleEye
    – yhw42
    Nov 24, 2010 at 13:53
  • 4
    @yhw42: the names aren't intended to be set in stone, and I expect that if these badges were implemented the team would think up their own names if they didn't like these ones. However, Sharpshooter implies a standard of accuracy rather than the ability to find things. Bloodhound might be a better alternative.
    – Andy E
    Nov 24, 2010 at 14:26
  • 4
    "I can't see any "gaming" issues arising from this, because it requires agreement from other members of the community." I can. You could easily create a new account or even several new accounts and start posting duplicates and then with your real account you will be the first one to report them as duplicates.
    – Erik B
    May 2, 2011 at 16:22
  • 2
    @Erik: something like that could be picked up on fairly quickly by moderators (same IP posting question as first to vote; same person first to vote on all the OP's closed questions, etc). I would say there are more complex anti-gaming mechanisms in place on SO already.
    – Andy E
    May 3, 2011 at 10:19
  • 1
    @Andy, It doesn't really matter. All I'm saying is that it's gameable and nearly impossible to detect by mechanisms and/or moderators. Users that are actively searching for dupes would have the same usage patterns as the ones who are gaming the system to get these badges. However, I wouldn't consider it worth my time to create 200 accounts from 200 IP-addresses and post a duplicate question from each of these accounts just to earn three badges.
    – Erik B
    May 3, 2011 at 11:23
  • 1
    I absolutely agree, we need some badges (at least) to give an incentive for closing as duplicate. Apr 27, 2012 at 17:23
  • 1
    "Reconnoiterer" => "Master Locator"? Apr 29, 2014 at 7:53
93

I'm wondering if it doesn't make sense to lower the reputation threshold to vote to close as a duplicate. Unlike other votes to close, there is already a bit of a check on this; you have to actually point out which question this is a duplicate of. That discourages (at least a little) people from doing it casually. So unlike "subjective and argumentative," where it's really a matter of opinion, and makes sense to restrict to higher reputation users, it would be good to have more participants in fighting duplicate questions.

Regarding your second point, the site already marks questions as possible duplicates after a certain number of votes. Perhaps it could do so (less intrusively, maybe) after the first vote?

2
  • 23
    or maybe a lower rep limit needed for the first person to vote to close as a duplicate, so low rep users don't just "jump on the band wagen", but can still help in the process of controling duplicates Apr 29, 2010 at 8:50
  • 1
    I agree the rep should be lowered, but not just for the first user to vote... that seems to be just asking for the first disagreeing person to vote it down. But I'm surprised that voting as a duplicate requires such high rep - surely if I can already flag the moderator.... I can also vote for duplicate.
    – Taryn East
    Jun 7, 2011 at 17:54
53

Send us a package of cookies for every 100 duplicates we vote as duplicate.

No rep gaming issues, no badge gaming issues, no problem!

0
52

I am all for rewarding closevotes (not only finding duplicates). If reputation

is a rough measurement of how much the community trusts you; it is earned by convincing your peers that you know what you’re talking about.

it is illogical that only answering increases reputation but closevoting doesnt. If I supply a closevote and people follow it, I have effectively convinced them that I know what I am talking about, too - namely that a particular question should rather be closed than answered.

Moreover, even if a question is closed, it will usually have received some answers before. The reputation gained on these remains unless the question is deleted. Deletion rarely happens due to the 20k requirement. But if it remains it implies they gained trust, while those that closevoted dont gain anything at all, although they did the right thing and those that answered did the discouraged thing.

So if reputation really is a measure of community trust, closevoting should gain reputation as well. This is true for all the additional privileges gained. If we want more people to use them, we have to reward them for it. Currently, the only reason why a 3k+ user would closevote is because s/he understands the necessity of it. Unfortunately, there is even a lot of high rep users who dont see that and repwhore on whatever they can answer. Are those really the people we trust?

Regarding the problem of incorrect closing I am quite sure that this is only a minor issue. I've seen far more correctly closed questions than reopened ones. The few false positives could be further mitigated by allowing to uv/dv on the votes instead, e.g. cast a closevote and have people vote it up or down until a threshold is reached. +5 will close, -5 will fade the vote. And allow people to revoke their vote please.

Regarding rewarding and badges, I dont see why we cannot have both. Someone who continuely does her/his community duty and closevotes should be rewarded for it with a badge or two. Badges are not enough to encourage closevoting though, because once you got the badges, you dont have an incentive to closevote anymore. So we also need some reputation gain.

4
  • 17
    I think it should only be the first person to cash the "close as duplidate" vote, but the rep should only be given when the question is closed. The first person had to do the real work in finding the duplicate, the rest of the close votes take very little work. Nov 14, 2011 at 13:03
  • Vote to delete is a 10k privilege.
    – jscs
    Nov 14, 2011 at 16:44
  • 7
    @Josh true, but you need 20k to delete immediately. All others have to wait for two days, which makes deletion happen rarely because no one heads back to a closed question.
    – Gordon
    Nov 14, 2011 at 16:57
  • I like your reasoning about getting rid of the rep earned by people who answer questions that are later closed as dupes. But FTR I see a steady trickle of questions closed as dupes when really they're only superficially similar. Apr 20, 2013 at 17:19
22

Another possible solution might be to separate the problems of flagging duplicates and voting to close the duplicates. What if anyone (or almost anyone) could flag a duplicate using the functionality presently built into "close as duplicate?" Then "close as duplicate" would be enabled for higher reputation users once the first duplicate has been flagged.

1
  • I was thinking along these lines, but with an incentive as well. However I wish to be able to see the list of possible duplicates on a question before it gets closed. Jan 29, 2010 at 20:43
20

This is a terrible idea. There, let me be the first to put in a negative answer.

I have had to reopen some questions because not all SOpedians are as smart as the next bloke. They see some similar words, and think it must be a duplicate.

There also seems to be some moths that are driven to brackets around Close(n) and see their call of civic duty to blindly follow where others have gone before.

Similar to my advocating against this: Should the sportsmanship badge be awarded multiple times?, I'd like some hard numbers for % of questions actually closed as duplicates.

Let's say theoretically

0.1% closed as duplicates
0.12% failed attempts to close
0.016% reopened
no badges

I reckon you will see the close/reopen buckets both increase significantly (more towards close since it is incentive-ised) and failed attempts go through. My reasoning is simple - human nature. Just as people will cast votes for opposing answers for a week and never again for and vote solidly for another 17 days* after for (and also thereafter rarely vote again).

*If you don't believe me, just go to https://stackoverflow.com/users?tab=voters They stick out like a sore thumb with votes an exact multiple of 30.
Disclosure: I'm guilty too, but I actually continue voting as much because I do really read that many questions!

2
  • 2
    Therefore the electorate badge need to be redefined so it needs high voting over a long timer, e.g. count only 10 votes on each day. +1 for some good points Mar 4, 2011 at 9:30
  • 3
    I would suggest heavy penalty for wrongly closing a question as a duplicate. Something that exceeds reward by two or three times. This will make the users careful closing as duplicates
    – One Face
    Feb 13, 2015 at 8:46
14

Not sure this is a great idea. On the surface it sounds interesting, but often times people simply skim the question and then vote to close not realizing that there are differences in the questions. It then is much harder to get the question reopened and basically an uphill battle.

2
  • maybe if more people votes to close as duplicate, there could be a requiment for more votes to close??? Jan 29, 2010 at 20:41
  • 6
    The incentive will only be for the first person that finds the dupilcate, so the other 4 votes to close will be as now Jan 29, 2010 at 20:42
14

The trick is to find a system that only rewards work. My proposal:

  • When (say) 5 people have identified a question x as a duplicate of question y, award rep to those 5 people and close the question.
  • Don't reveal which questions have been suggested as duplicates before a question is closed.

The 2nd point is important to avoid triggering a close-as-dupe epidemic. It means that people individually have to do the work (searching for dupes) to get the rep. You have to be fast (otherwise 5 other people will get there first), and you have to be right (pick the dupe that others picked).

Inspired by: The ESP Game from Games With A Purpose (from about 22:30 onwards, though it's all fascinating stuff!)

Thoughts? Are there ways that this system could be abused?


UPDATE 2/5/2012: Eliminating Cheating

Jeff's comment raises a valid concern that people could easily game this system by voting to close a question as a dupe and then writing a comment mentioning the other dupe; others will then vote to close as the same question. I can see this happening because there's positive feedback at work here -- both the original comment-writer and subsequent close-voters stand to make some quick rep. But I think I now have a solution to that:

To prevent people "dropping hints" about dupe-parents in comments/answers, you could simply forbid people from both identifying a question as a dupe and commenting on or answering that same question -- i.e. as soon as you do one, you can't do the other(s).

Think about it: if you see a question and immediately recognise it as a dupe of something else, there's no need to comment on/answer it. The only scenario I can think of where this might be too hard-line is when someone uses a comment to ask for clarification, and only recognises the question as a dupe after receiving that clarification, but this would be pretty rare. In the unlikely event that this turned out to be a common problem, the system could just be changed to waive the restriction for users above $SOME_THRESHOLD_REP.

14
  • 8
    I think the second point here would result in very few questions actually being closed as a dupe. A lot of genuine dupes are closed as a direct result of a link to the found duplicate being displayed in the comments. This would probably result in duplicates for niche tags almost never successfully reaching those 5 votes.
    – Andy E
    Nov 24, 2010 at 12:20
  • @Andy E's head: That is a factor, but wouldn't you agree that it would do a good job on the most common dupes? Which (by virtue of being the most common dupes) are the ones we're most interested in eliminating? Also this system offers rep as an incentive, something that's not present in the current system. Nov 24, 2010 at 12:30
  • 3
    The problem is we allready have so many duplicates, that everyone could choose a different quesion as the "parent" Nov 24, 2010 at 12:36
  • 1
    @Ian Ringrose: I agree that's a problem. What if there was a standard way to choose which parent is preferable, e.g. the oldest or the one which has already had the most dupes assigned as "children"? ("Number of children" could be a per-question property -- it would definitely help "canonicalise" questions I think.) Nov 24, 2010 at 12:43
  • 3
    @Ian Ringrose: Another possibility would be to reveal all suggested parents if the number of suggestions gets to a certain number without any clear winner appearing. In this case no rep would be awarded, but community-minded folks or moderators would have a useful selection of dupes to choose from and could close as dupe "traditionally". Nov 24, 2010 at 12:49
  • 2
    Just had another idea: When X is marked as a dupe of Y, the question Y gets a +1. Rationale: (1) A question that is asked many separate times is an important question; and (2) it makes it easier for people to identify which of several roughly-equal candidates they should nominate as the dupe-parent -- just nominate the highest-scoring one. This will fold the "number of children" aspect I mentioned above into the dupe-parent-selection decision without having to separately track this quantity (which would be messy and confusing). Jan 26, 2011 at 5:46
  • 1
    This is a great idea, except it's easy to cheat: just post a comment containing the identified dupe and 5 users can paste it in. I don't know how you could prevent that or if it is even possible to prevent. Mar 4, 2011 at 9:58
  • 1
    @Jeff: See what you mean. You can't prevent side-channel communication completely (e.g. if someone was determined enough, they could set up a separate site just to game it) but you can raise the "effort barrier" to the point where most people will just not bother. How about this: only award points for dupe nominations whose question digit strings (e.g. this one's is "37466") do not appear in any comment. If someone does put such a digit-string in a comment, you could pop up a warning box ("Are you sure? No points will be awarded if the question is closed as dupe!") to pre-empt rage. Mar 5, 2011 at 3:29
  • @JeffAtwood: To really prevent people "dropping hints" about dupe-parents in comments/answers, you could simply forbid people from both identifying a question as a dupe and commenting on or answering that question -- i.e. as soon as you do one, you can't do the other(s). The only scenario I can think of where this might be too hard-line is when someone uses a comment to ask for clarification, and only recognises the question as a dupe after receiving that clarification, but this would be pretty rare. (And you could even waive this restriction for users above $THRESHOLD_REP.) May 2, 2012 at 4:27
  • 4
    Your suggestion will actually lead to less duplicates being closed. I don't automatically know a Q is a duplicate when I see it, nor do I immediately start searching to see if it is. Duplicates get closed because one person notices a dupe, links to it, and then others check to see if it actually is, and then vote to close. With your suggestion, you would need five people to independently recognize that something is a dupe. This will cause less questions to be closed as dupes. May 25, 2012 at 12:25
  • @HodofHod: I agree that's a factor, but I doubt it's the overriding one. One of the reasons why you don't immediately start searching to see if a question is a dupe under the current system is because there's no reward for doing so -- and what I'm proposing would change that. I think my way is also less likely to result in the "piling on" effect that we sometimes see (i.e. when the question's not really a dupe but gets closed-as-dupe anyway, because overly-officious people don't bother to read the other question carefully). May 26, 2012 at 13:54
  • 2
    Do not forbid to comment and flag for the same question. Just do not give award points for flagging, if the user also add a public comment for this question or referring to this question. I personally prefer to add comment to link questions together even if moderators would not agree, that it is an exact duplicate. Apr 19, 2013 at 22:44
  • @MichaelFreidgeim: Good idea -- we want the least restrictive measures that prevent cheating. However I think it's necessary to prevent others from getting rep in this situation too, since without this it becomes tempting for others to quickly hit "Close as dupe" without thinking it through. Apr 20, 2013 at 17:02
  • Agreed. After 2 questions are linked together, reporting them as duplication should NOT be rewarded. It's too easy :). Apr 20, 2013 at 19:09
13

You might be interested in my script Duplicate Question Suggestion Boxes which makes this more easy.

enter image description here

An example of its effectiveness, where otherwise the link would be hidden somewhere in the side bar:

enter image description here

Of course, we can't enable this for all users.

0
9

The key to any community-based, "Duplicate Task Force" is opening and simplifying the process.

First, provide 1k+ rep users with a "flag as duplicate" option, which will allow for a single link to be provided. This makes it easy and painless to report.

Display the flag and the duplicates found to the asker (and others? Debatable.). This makes the process open, and will help the asker.

Then, facilitate moderators' (or 10k+ rep users') job by providing them with a list of duplicate-flagged questions. Allow sorting by number of duplicates found. This makes the process manageable.

Finally, provide the users who flagged the question with one duplicate or another (it's not a game to get the right one) with a small incentive. Lock down rewarding after question has been closed as duplicate. That makes finding duplicates rewarding.

3
  • 4
    this is possible now with blog.stackoverflow.com/2011/01/improved-flagging Jan 24, 2011 at 4:38
  • @JeffAtwood: Alright! And thanks for finding this itty bitty suggestion and replying. How will the new improved-flagging help simplify warnings of duplicates? Besides there Newgrounds-style whistle level, which is pretty neat by the way.
    – MPelletier
    Jan 24, 2011 at 11:46
  • 1
    another option possible duplicate of could've been added!
    – Harish
    Jan 25, 2011 at 12:05
9

For new users, few questions are duplicate. Both new answerers and new askers get frustrated when they are trying to do the right thing, but it gets closed as a duplicate. Their initial reaction is not: "Thanks for pointing me to the right question", but: "Why are you rudely interrupting my conversation?"

The irony is that the old users enjoyed the very same thing they are denying to the new users. With the very same questions, and the very same answers: the only difference is the timestamp on the question.

Even more irony is that most high-reputation users got high reputation by answering questions instantly. The fact that they can answer instantly is a strong indicator that the question has been asked before, and is on the top of their tongue. In other words: most high reputation users are very good at answering duplicates.

At the end of the day, the fact that only a small number of duplicates gets closed means its not a big problem. But until duplicates are closed in a socially acceptable way, I think it would be a very bad idea to reward or even encourage more duplicate closes.

2
  • The questions should merge somehow
    – joojaa
    May 29, 2015 at 5:17
  • 3
    Almost all the statements that are presented as facts are dubious at best. Mar 14, 2016 at 5:37
7

We should just give a badge for 'Closing as Duplicate' enough times (when it's the confirmed close reason). I do not want to disincentivize good answers even to duplicate questions. All questions deserve good answers.

1
  • 8
    And sometimes the duplicate status is not at all obvious. I've searched for and failed to find duplicates, then answered the question, only to then have somebody find a dupe using different search terms that I hadn't thought of. That doesn't reduce the value of my answer. Jul 12, 2013 at 16:07
6

I think this is a great idea. Especially the "duplicate finder" badge. It has been argued that this may require a feature to mark duplicates, but that may not be needed. Anyone who has sufficient privileges to add comments can add the link to the duplicate in a comment. If the question is closed, the Stack Overflow engine can harvest the URL, and award a badge to the user who was the first to link the question with the duplicate, albeit through a comment, or through a close vote.

Some of the other answers argue that questions are closed as duplicates, where the question is not actually a duplicate. This is true. And this may happen more often if duplicate finding is encouraged.

But this is not a problem.

If the person asking the question disagrees, she can always compose a better documented question, referring to the other questions, and point out why her question is not a duplicate. I don't expect this to happen more than a few times in every hundred duplicates, so we should not worry that this will flood the site with more duplicates.

5

I know its tiresome to see duplicate questions but I am guilty of this as it is sometimes hard to find the duplicate in the sea of questions. I wanted to ask a question about deleting my own posted question, I did a search for this and came up with 50 pages of questions with the same search parameters. I looked at the first few pages...but this got me thinking that sometimes users post duplicates because they need the answer but don't want to search through the sea of questions as this would take more time than posting a question. I like the idea of rewarding the finding of a duplicate to the first person who finds it then it gets put up to vote and if enough votes agree then it gets merged into the old question answer.

5

One way to incentivize finding duplicate questions would be to make it easier to find them. A tab could be added to the review section for "Possible Duplicates".

When you type in the title for a new question, you see "Questions that may already have your answer". The functionality that creates that list could be adapted.

New questions that meet a certain threshold of commonality with existing questions could be shown along with the list of "Questions that may already have your answer" that met the threshold.

2
  • There's already a panel suggesting possible duplicates when you're typing a question. Mar 14, 2016 at 5:25
  • @ivan_pozdeev Yes, I said that in my 2nd paragraph "When you type in the title for a new question, you see "Questions that may already have your answer". The functionality that creates that list could be adapted."
    – codewaggle
    Mar 14, 2016 at 22:10
5

I think that a combined approach would be the best solution.

Andy E's badge suggestion looks promising, but I think it would be much more effective when combined with removing rep earned by answering dups as Ester suggested and possibly even awarding some small amount of rep for successfully closed dups like Gordon suggested.

Above and beyond these measures I suggest:

  • Making the Related Posts column visible in the First Posts Review queue, so that obvious duplicates stand out and can hopefully be caught as a part of the review process.

  • Adding "asking too many duplicate questions" to the list of reasons that one could be question banned, even if those duplicates have not been deleted.

Note:
Apparently there's no way to find out whether asking an excessive amount of duplicate questions is currently taken into account in the auto-ban equation. If they are please disregard.

5
  • I'm pretty sure it is already taken into account, but there's no way to confirm.
    – user206222
    Jul 10, 2013 at 17:56
  • @EmrakultheAeonsTorn I know deleted questions are, but I couldn't find any indication that duplicates were.
    – apaul
    Jul 10, 2013 at 17:58
  • There won't be; the algorithm is a closely guarded secret.
    – user206222
    Jul 10, 2013 at 17:58
  • I know, but when looking at meta.stackexchange.com/questions/86997/… duplication doesn't seem to be mentioned as a possible reason for a ban
    – apaul
    Jul 10, 2013 at 18:00
  • I would be surprised if it were ignored, since it detracts from the site, but I'm not going to preclude the option that it isn't.
    – user206222
    Jul 10, 2013 at 18:01
5

I have some ideas for gamifying the process instead of just giving rep incentives and would like to know where the potential holes are for promoting exploitation of badges/reputation. Each one of these suggestions have the goals of:

  1. Reduce existing duplicates that have not yet been identified
  2. Reduce rep-hunting against low-hanging dupe-fruit
  3. Identify users who answer duplicates

I don't like the idea of awarding reputation points for these kinds of things, but I also know that reputation is a valuable metric to many users, so many of these potential suggestions are more geared towards badges than rep.

Tackling #1

New Badges

Potential Name: Death by a thousand cuts || Hirudotherapist
Purpose: A hard-to-achieve badge to encourage dupe-hunting
Tiers: 1; metal-grade, flexible (probably gold)
Unlocked with: 1,000 questions closed as duplicates by actions of the user
Rationale: Whether they're normal users or moderators, this is something that can be worked towards over time to better the site(s).

Potential Name: Justice never sleeps || Mold-Hunter
Purpose: Cleaning out the old dupes that snuck past us already
Tiers: 3, threshold values flexible
Bronze Tier 1 Unlocked with: 1 question closed as duplicate by actions of the user for a question older than 6 months
Silver Tier 2 Unlocked with: 25 questions closed as duplicate by actions of the user for a question older than 6 months
Gold Tier 3 Unlocked with: 500 questions closed as duplicate by actions of the user for a question older than 6 months
Rationale: Even old dupes can hurt. If we encourage badge-hunters to start from the back and work their way up, it could have the nice effect of not overlapping badge-hunters with users just trying to do the right thing as they come across them.

My hope with these two badges is that anyone looking to do badge-hunting with these will try to tackle both the backlog of dupes while not completely ignoring incoming dupes.

Also, with the thresholds being so high, the more people that participate will make it harder to achieve them, making these eventually a sort of status-symbol badge for those who threw themselves to the task, first.

Tackling #2

I'm not sure what all is in place for #2, so these are less concrete gamifications than conceptual musings.

Rep Changes

Are we already taking away earned reputation from upvoted answers to questions closed as duplicates? If not, I think we should revisit the idea. This is a very straight-forward way to stop #2, but I admit that it is quite rep-invasive. Maybe a 24-48 hour window for losing rep while an alternative addresses potential dupes in that window.

The alternative might be to freeze rep-acquisition for all answers as soon as the question is flagged as a potential duplicate; this would be invisible to most users until actually marked as a duplicate, if I understand it correctly. Upvotes would still "count" but the user getting the upvote only gets the rep if the flag ends up not being accurate.

Another potential dupe penalty if not already enforced is decrementing the tag's badge score increase when the question is closed as a dupe. I'm pretty sure that only applies to accepted answers, but the psychological price would still be there.

With or without rep changes, I think next badge can might help with #2

New Badge

I'm a big fan of this next badge, as it rewards casual and hardcore reviewers with different tiers with timing and quantity instead of just quantity. I think it's a potential candidate for a new set of badges across SE regardless of how you feel about the rest of my suggestions.

Potential Name: Quick-Draw || The Fastest Gun in the West
Purpose: Reduce new duplicates, encourage review-queues, encourage experienced/knowledgeable users to find existing answers instead of re-answering them
Tiers: 3, threshold values flexible
Bronze Tier Unlocked with: 1 question closed as duplicate by actions of the user for a question less than 6 hours old
Silver Tier Unlocked with: 100 questions closed as duplicate by actions of the user for a question less than 3 hours old
Gold Tier Unlocked with: 1,000 questions closed as duplicate by actions of the user for a question less than 1 hour old

Tackling #3

New Badge

Potential Name: Good Intentions || Paving the Road to Hell || Doh!
Purpose: Identify users capitalizing on answering dupes while not exactly calling them dunces, rep-moochers, or menaces
Tiers: 1, a new one each time it occurs
Bronze Tier Unlocked with: Had an accepted (and/or up-voted) answer for a question that was closed as a duplicate
Rationale: An accepted answer to even a duplicate question added value somewhere, but not the kind of value that SE wants to encourage. The more badges a person has of this, the more that it will stick out by sheer count compared to other badges for people that make a habit of doing this kind of rep-hunting against low-hanging fruit likely already addressed in another question


I'd like to point out or at least disclose that I have a bad habit of doing #3; I never do it on purpose, but knowing that I would be getting something like this would encourage me to look for existing answers even more thoroughly instead of blindly answering something that I know off-hand.


Some might be asking why the emphasis on people contributing to dupes and dupe-hunting instead of penalizing the people writing the dupes in the first place. My rationale is this: getting downvoted and your question closed is fairly demoralizing as it is. It has happened to me several times, and each time it does I get more encouraged to research further before asking. I really don't think this aspect of tackling dupes can be improved by much, as the Similar Question mechanic when writing out a new question I feel works fairly well, at least for me.

2
  • what does "by the user" mean? the user voting? or does that require the author to agree with the closure (dupe hammer closures would therefore bypass that)
    – Kevin B
    Jul 6, 2018 at 23:10
  • @KevinB It means whatever the user does directly contributed to it being closed as a duplicate, be that directly closing/dupe-hammer, flagging, or other means. Jul 6, 2018 at 23:11
4

We could implement this in much the same way as the vote to close system. A user could submit an "answer" that is somehow marking this question as a duplicate and all other users could see it, and respectively vote if it really is a duplicate. 5 votes closes the question, and maybe gives -5 rep to the asker (seems fair to me if they can't search).

1
  • 1
    Not all duplicates are the result of failing to use teh google. Jan 6, 2011 at 7:08
4

Count me in on the idea to reward for finding duplicates.

Though I suggest to instead reward with reputation rather than a badge (rationale follows):

  • Upon closing a question as a duplicate, award N rep (see below for considerations regarding the amount) to the user who first suggested the duplicate
  • The rep is retracted if the question is later reopened
  • Does not apply to own questions
  • Does not apply to closures that aren't subject to peer review (by mods/golden badge holders)
  • Discussable: what to do with the normal accept mark:
    • count as well
    • forbid/delete (the duplicate takes precedence)
    • retract the rep for the duplicate if present (normal acceptance mark takes precedence)

Rationale

  • The user who found the duplicate essentially answered the question
  • This will eliminate the current ill motivation to re-answer: re-answering (incorrect/suboptimal course of action) yields reputation while finding a duplicate (correct course of action) doesn't
  • The incentive to game the system is limited by the fact the reward is limited to just the amount awarded at closure

Downsides

  • the OP does not participate in the award, thus it violates the principle that "acceptance means it has helped the OP the most"
    • not necessarily: if the OP did not protest the marking, this means they are satisfied with the "answer"
  • allows users to conspire by intentionally asking duplicates, akin to upvoting rings (this is nothing new but it means the algorithms will have to be adjusted)
  • allows rep-whores to spam "looks similar" suggestions hoping that voters overlook the differences. This would also be cluttering the already overloaded Close review queue (again, statistical analysis will have to be updated to detect and stop this pattern).
  • escalates (as it exists already) a flavor of the Fastest Gun in the West Problem - once a question is closed, it's nign-impossible to re-open and re-close it with another reason - e.g. if a more fitting suggestion arrives.
    • Even the official position is that if a question is closed, and shall stay closed, it should not be reopened even if the reason currently stated is not the most appropriate one (anymore).
    • this may be solved by treating duplicate suggestions as answers to some degree: e.g. a preferred one could be selected by an OP. I'm not the first one to get this idea: Show duplicate suggestions as answers .

Considerations regarding the amount

  • Finding a duplicate is a noticeable to significant amount of work, but still less than composing an own answer.
    • On the other hand, the closure cannot bring further rep from upvotes.
    • Still, answers to duplicates bring little rep as a rule
  • The reward amount shall make duplicate searching be seen as an easier course of action than re-answering reward/work ratio-wise. Still not much enough to justify aggressive "partial-duplicate" spamming hoping that voters will overlook the differences.

If not 15, maybe 12 rep will be seen adequate (note the word "seen"). 10 appears too little already - answering a question straight is more than just "being useful".

3
  • 15 seems like a bit much... edits are only 2... I have a difficult time seeing marking something as a dupe being as valuable as having an answer accepted.
    – Catija
    Mar 14, 2016 at 4:31
  • 1
    @Catija the basic idea is that doing research on the question's subject is a noticeable amount of work. What can I do if the research's result turned out to be that there's already an answer (more specifically, there's already an answer on the same SE site)? Sure, referring to an already-written answer is not quite the same amount of work as composing one's own, so the specific amount to award is fully discussable. Mar 14, 2016 at 4:38
  • 2
    @Catija on the other hand, 1)rewarding less will not eliminate the motivation to re-answer; 2)the reward is already limited by not allowing to get any further rep from upvotes. Mar 14, 2016 at 4:47
3

This answer is more for consideration purposes if/when implementing this feature-request:

  1. The binding vote-to-close (and re-open) for gold tag-badge holders should be considered here. For now questions closed as a duplicate (or re-opened after being closed) may be excluded from any consideration for badge-worthiness.

    There is distinct difference between how Stack Overflow is hammered with close votes in comparison to other network sites. So this may be a non-issue to start with. After all, tag gold badge users are considered trusted in that "field".

  2. Another thing to consider is how mods are affected by finding (obvious) duplicates. As they possess the ultimate dupe-hammer, they may never be able to attain these badges.

6
  • why should such questions be excluded? If I go and dupehammer enough questions to get this badge or rep or whatever the prize is, but they should never have been closed, flag me. I think people with gold tag badges have proven themselves pretty trustworthy, and MOAR CLOSING AS DUPE is exactly what we want from them. May 17, 2014 at 15:40
  • @KateGregory: I do mention "for now"... I know that Stack Overflow is a different beast compared to the other networks. There closures are a problem since people just seem to pile any old question on the community.
    – Werner
    May 17, 2014 at 16:28
  • Yes, you mention "for now" but you don't mention why. You seem to think it's obvious since you say "without a doubt" - but I don't know why you think they should be excluded, and I am asking you to explain it. May 17, 2014 at 16:38
  • @KateGregory: I've lightened my opinion with a mild caveat.
    – Werner
    May 17, 2014 at 16:43
  • 2
    I am sorry that there are people in the world who say "why do you want X?" when they mean "I don't think you should want X." If I was one of those people, then "for now" and "maybe" would mollify me. I am not. You are free to want and support whatever you want. Perhaps I will join you in wanting that if I have any idea why it is good. I am asking for your reasons because I can't think of any and if some exist, I want to learn them. I am not asking you to back away from your opinion, but to share more of it with me for my enlightenment. May 17, 2014 at 16:45
  • @KateGregory: No worries. I know there are robo-reviewers out there. And I was concerned about that. Non-Stack Overflow networks don't get review audits (or at least I've never seen one on TeX - LaTeX where I'm active). I was just concerned that someone with a dupe-hammer that doesn't have the sense to properly review get to just run wild... Of course, this may be a very cynical view on things, since the community will step in to correct this behaviour. I was erring on the side of caution.
    – Werner
    May 17, 2014 at 16:49
3

I'd rather give badges if the duplicate was marked by the OP as helpful – as you probably know, when you suggest a duplicate, the OP can mark it as helpful or editing the post to explain why it is different. This makes sure that people won't suggest harmful duplicates as they wouldn't be rewarded anyway. I'd rather not give rep.

I suggest the badges:

bronze – 1 duplicate accepted by the OP

silver – 25 duplicates accepted by the OP

gold – 100 duplicates accepted by the OP

Badge names are all over the post. I don't have an original suggestion for them.

1
  • It does not make me sure, because badge-driven robo-reviewing is a well-known problem. Nobody gets rep for reviews, and yet...
    – user259867
    Aug 2, 2015 at 16:33
2

I was looking at Show duplicate suggestions as answers . I think if the candidate duplicate is offered as an answer, then the person that posted this should get rep credit for people that upvote this "duplicate" answer. Possibly the finder gets 2 points of rep per upvote, and the author of the original answer gets 3 points of rep per upvote. This should be a strong incentive for identifying a duplicate.

Once a question has been determined to be a duplicate, all the provided answers should be deleted, leaving only the duplicate answer candidates. That would remove the rep gained from providing a new answer to a duplicate question. This should be a strong disincentive for answering a question that is likely duplicate.

0

Maybe you could add "flag as possible duplicate" as a top-level option in the flag-moderator box? (with a requirement for a link).

That makes it more obvious promoting it happening more, and opens it up to lower-rep people to flag as possible duplicate, without directly voting for duplicate.

2
  • 2
    That's a good idea that already exists! "Flag" -> "Doesn't belong here" -> "Exact Duplicate".
    – yhw42
    Jun 7, 2011 at 18:16
  • 5
    Thanks for letting me know. That is extremely non-obvious to me. "doesn't belong here" smells to me of offensive language or spam... I would not click on it unless I found a particularly noisome question. Especially given that this thread is about promoting duplicate-post flagging, I'd consider adding "duplicate" as a top-level flag option to make it more obvious.
    – Taryn East
    Jun 7, 2011 at 19:17
-2

I'm very much in favor of awarding rep to those who post links to previous answers to the duplicate question. I'm less in favor of giving rewards to simply flagging questions as a duplicate.

0

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