272

And the reputation gained should count towards their ability to close the questions themselves single-handedly. Otherwise this site will continue to be the endless stream of extremely low quality content.

I can spot a few people who care for the quality. Alas, they gain almost no reputation points. At the same time, people who care for the reputation points only writing hasty answers to off-topic or duplicated questions always get rewarded. This is an economical matter. Given the reputation points is a "currency", the population will always opt for the action that will give them a reward. It means Stack Overflow is intentionally driven towards being an endless stream of low quality content. It should be changed towards quality, not quantity.

The irony is, finding a duplicate often takes incomparably more effort than writing an answer. I myself is often tempted, "What the heck, what I am wasting my precious life for? Why not to write that primitive answer to a primitive question and move on?". There is not even a ready-made link to Google search to ease the task, so you have to type site:stackoverflow.com manually and then paste the question title in the address bar. While it can be easily done with a single click, if only Stack Overflow did care for people who care for Stack Overflow.

51
  • 5
    @JeanneDark and you can do the same with answering if you really wanted (meaning, socketpuppets). The positives of this outweigh the negatives. BTW, if anyone is going to use socketpuppets, chances are they'd do it for answers because the rep amount you gain is a lot more. You could have the dupe reward be the same as an edit (+2) if you were really concerned.
    – Script47
    Commented Jul 18, 2019 at 8:20
  • 5
    Yes, yes you can. Stack overflow staff is paranoid about people who would possbily amass enormous reputation value by some sort of fraud. The problem is, such a fortune cannot be converted into anything useful. So may be it is not that dangerous as it feared.
    – Your Common Sense
    Commented Jul 18, 2019 at 8:23
  • 76
    or probably we disable the reputation of answers in case the question is closed as duplicate. So the question/answers behave like wiki posts. Doing so you will earn nothing answering trivial duplicate (you will probably gain some reputation only if you get upvoted/accepted before the closure)
    – Temani Afif
    Commented Jul 18, 2019 at 8:26
  • 10
    For any given feature that provides rep, there's a potential to abuse it. A high-rep user could also go on an incorrect duplicate closure spree to gain rep, or as already mentioned, use socks to get more rep. If a feature like this is implemented, there needs to be a system in place to avoid abuse as well, especially close-spree abuse. Sockpuppet abuse doesn't do too much damage in comparison
    – Zoe - Save the data dump Mod
    Commented Jul 18, 2019 at 9:07
  • 57
    @YourCommonSense I've seen users with hundreds of thousands of reputation with multiple hammers who post answers to duplicates so low-quality it's almost worth nuking off the site instantly. One of those actually has over 800k reputation, and 22 hammers.
    – Zoe - Save the data dump Mod
    Commented Jul 18, 2019 at 9:15
  • 7
    There are less than 10 users with that much rep, @Zoe. If we exclude Jon Skeet and a SE employee, only 6 users. Now it's only a matter of counting hammers. Which is a phrase I didn't expect to use today. --- Too easy, I shouldn't have started from the bottom.
    – yivi
    Commented Jul 18, 2019 at 9:22
  • 29
    @Zoe to make it clear: I don't give a damn if some reputation billionaire won't be closing duplicates. I only care for knowledgeable and conscientious newcomers who don't have a chance to get reputation points in the situation when most questions already has answers.
    – Your Common Sense
    Commented Jul 18, 2019 at 9:47
  • 9
    I think this would be a great idea if done well, but I also think it will never, ever ever be implemented or even considered. Good luck.
    – Magisch
    Commented Jul 18, 2019 at 9:49
  • 5
    Reward for finding duplicates or for having a question closed as a dupe that someone has found? Only the latter would make sense, but the reward seems rather irrelevant, based on my personal experience with ~50% of all dupe flags simply ageing away ..
    – shmee
    Commented Jul 18, 2019 at 9:50
  • 5
    Related: Reward for close voters finding appropriate duplicates?
    – Jeanne Dark
    Commented Jul 18, 2019 at 9:58
  • 4
    @shmee I'd rather say finding should he awarded, because that is the work done. Dupe hammering can be done in seconds, and there are enough gold badgers around that could ... (at least under the js tag)
    – Jonas Wilms
    Commented Jul 18, 2019 at 10:14
  • 7
    We reward answers with reputation based on votes. We don't award reputation for answering. We award it for answers that others agree with. If we were to award reputation for identifying duplicates, at a minimum it should only apply when those potential duplicates have been somehow validated. Duplicates marked by hammers are often wrong. So very, very, off-the-rails. what-are-you-thinking wrong. We shouldn't award reputation for unilaterally marking a duplicate any more than for just answering a question. There must be careful review and validation, even more than now.
    – Scott Hannen
    Commented Jul 18, 2019 at 14:19
  • 4
    While I'm interested in topic in general <non welcoming text>this is really low quality post - exact duplicate of multiple discussions on "reward searching duplicate" topic, does not have any concrete proposal, does not explain how already discussed problems should be solved, does not even list known issues... </non welcoming text>. Positive thing is new set of people have chance to rehash the same concerns discussed earlier...
    – Alexei Levenkov
    Commented Jul 18, 2019 at 17:06
  • 9
    @Zoe I had two high rep users reopen questions that were closed as dupes just so they could answer them. :/ The answers didn't offer anything that wasn't in the canonical question already. The internet points offered by gamification is just too difficult to pass up. I like the goal of this question, but I don't think it will come close to solving the problem.
    – John Conde
    Commented Jul 19, 2019 at 23:18
  • 6
    SO is never going to do this. Their revenue is coming from users writing bad questions like this (and having people write bad answers to them). It causes more traffic on the network, which translates to more ad money and more people buying their services. Rewarding people for finding duplicates will cause these questions to be closed and deleted more (which will cause some users to ask and answer less, as a pose to their question being answered).
    – S.S. Anne
    Commented Jul 20, 2019 at 13:20

9 Answers 9

Reset to default
78

The irony is, finding a dupe often takes incomparable more effort than writing an answer.

It's not the irony; it's the core of the problem. People want answers and they want them quick. Even for answerers, if they know the answer, it's easier to write them down repeatedly than searching on the site for duplicates.

Of course, you could give rep for finding duplicates, say for example +20 rep for every proposed duplicate that also gets confirmed. Maybe it increases the amount of questions that get closed as duplicates later on, but probably it would not change anything about questions first getting answered before anyone has found a dupe.

The value for the questioner is quite low by then. And I don't expect people to go hunting for dupes even for rep. If rep is a currency, it lost its value for me a long time ago.

Every technical solution for making finding dupes easier is welcome. I set my hopes in NLP-ML support over the coming years preventing dupe question from being asked in the first place, but that won't come soon, I guess.

This means that I totally agree with this idea and propose a fixed rep gain for every successful dupe closing, maybe even badges. But I also think that this will not solve the problem really. It's better than nothing though.

8
  • 3
    We can also punish (as -5 or something) user whom their question was closed as duplicate. It also don't solve the first timer or rare user but it may help as well
    – dWinder
    Commented Jul 18, 2019 at 11:07
  • 55
    @dWinder punishing users for duplicates isn't a good idea, especially if it's used abusively. But even then, duplicates aren't meant as an insult to the author ("haha, you failed to research") in a way that's worth punishing them over. Downvotes are already punishment enough, especially for new users.
    – Zoe - Save the data dump Mod
    Commented Jul 18, 2019 at 11:28
  • 29
    @dWinder we should not punish the asker because it's not easy to know that a duplicate exist. It would be also hard to identify trivial questions where duplicate is clear and other questions where the duplicate is not easy to find. Besides duplicates question are good because the redirect to the target. The core issue is the duplicate answers not questions
    – Temani Afif
    Commented Jul 18, 2019 at 11:28
  • 5
    @TemaniAfif Some duplicate questions are not easily identifiable, and some duplicate questions are useful signposts to the questions they duplicate, but lots of duplicate questions are very poorly researched and do have duplicates that are readily accessible, and most duplicates are not useful signposts. Sure there are rare gems that offer a radically different approach to solving a given problem that's also frequently the way others also approach the problem, but most duplicates are less discoverable than the questions they duplicate.
    – Servy
    Commented Jul 18, 2019 at 13:25
  • 3
    @TemaniAfif Indeed, it's often impossible to recognize that a question is a duplicate without already knowing the answer to the question. Some connections are only obvious in retrospect.
    – user149341
    Commented Jul 19, 2019 at 23:07
  • 1
    If the user answering dupes receive a notification that one of the questions he answered was marked as dupe/off-topic, could that have any effect on the behavior of these users?
    – Berriel
    Commented Jul 20, 2019 at 23:37
  • "And I don't expect people to go hunting for dupes even for rep." You'd be surprised.
    – Knu
    Commented Jul 21, 2019 at 0:27
  • A question marked as dupe should offer no net positive rep to the asker, even if it is upvoted. It can still offer net negative rep. Even if the asker accepts one of the answers. We could also consider eliminating any rep gain for someone who answers a question that is eventually marked as a dupe. I don't really like the idea of awarding rep for dupe-finding, especially not to dupe-hammer holders.
    – Robert Crovella
    Commented Jul 21, 2019 at 0:58
43

I would say at least answering duplicates should not be awarded, as suggested in the comment "or probably we disable the reputation of answers in case the question is closed as duplicate." by Temani Afif.

The reputation should be retracted only for answers which are older than the question which is used in the dupe closing, to prevent unexpected reputation change caused by "good old questions with valuable answers which get closed as duplicates of newer ones" (as commented by Martin Backasch).

I think users often know (or suspect) that the question they are answering is likely to be a duplicate. Removing part of the reward could help.

13
  • 2
    it's also the easiest to implement. You close as dupe --> all posts are transformed to wiki, that's all. If it's reopened for any reason, we put them back as normal posts.
    – Temani Afif
    Commented Jul 18, 2019 at 11:24
  • 16
    This would create a strong incentive of people who farm reputation that also have gold badges (quite a lot of them, considering farming reputation is one way to get a gold badge) to unilaterally reopen questions on which they have an answer after they get closed.
    – Magisch
    Commented Jul 18, 2019 at 11:48
  • 4
    @Magisch we can easily detect such bad behavior and flag it to moderator. If you do it 1 time or 2 it would be ok, but massively reopening questions you answered is easily visible to us. (I personnaly keep an eye on all the question I closed and get back again to the reopened ones to see what happened)
    – Temani Afif
    Commented Jul 18, 2019 at 11:53
  • 4
    @TemaniAfif I'm unconvinced that we have the tooling and opportunity to deal with large scale misconduct (this can filter down all the way to people with as little as 3000 reputation). There would need to be a systemic barrier in place or automated measures to monitor such behavior for this to be a viable strategy.
    – Magisch
    Commented Jul 18, 2019 at 12:23
  • 5
    Although I agree with the proposal, I much more after rewarding the dupe-finders than punishing dupe-writers. There are a lot of people who deserve to have some reputation but being honest and thinking globally they forfeit it for sake of the quality. It is my long desire to reward people for the quality of their contribution, not quantity.
    – Your Common Sense
    Commented Jul 18, 2019 at 13:35
  • 6
    I support this suggestion. Pretty fed up with (often high rep) users who answer questions that should be closed.
    – Modus Tollens
    Commented Jul 18, 2019 at 13:52
  • @TemaniAfif: This may be easy to implement, but in my is eyes it is only one side of the medal. We have to consider there are good old questions with valuable answers which get closed as duplicates of newer ones.
    – Martin Backasch
    Commented Jul 18, 2019 at 14:49
  • 2
    @MartinBackasch that's why I said that they become wiki after the closure so all the Rep gained before will remain.
    – Temani Afif
    Commented Jul 18, 2019 at 14:50
  • @TemaniAfif: Okay, that's one fact I did not know (yet). I thought converting an answer to an wiki answer will lead to reputation lost.
    – Martin Backasch
    Commented Jul 18, 2019 at 14:52
  • " good old questions with valuable answers which get closed as duplicates of newer one" I did not think about this, and this is really an important point.
    – Suma
    Commented Jul 18, 2019 at 15:00
  • @Suma this will change nothing as I already said in my comment. The keyword is wiki posts so you will loss nothing, we will simply block future reputation
    – Temani Afif
    Commented Jul 18, 2019 at 15:05
  • 6
    I think when answering an obvious duplicate, one should not get the award even in the time window before the question is closed as a dupe.
    – Suma
    Commented Jul 18, 2019 at 15:09
  • 1
    High rep doesn't get you a gold tag badge. In fact what does earn you a gold tag badge strikes me as bizarre. I am routinely puzzled by having a C++ gold badge when Raymond Chen doesn't. That's just <expletive deleted> up.
    – user4581301
    Commented Jul 20, 2019 at 5:26
28

Answering obvious duplicates definitely needs to be discouraged more, but I'm not so sure rewarding dupe-finders is necessary. It could help, but it could also cause other issues, like users racing to get the first duplicate-vote in without making 100% sure that the dupe target is correct (this already happens often enough currently with hammer-wielders).

Many users already vote to close duplicate and off-topic questions to keep the site clean despite the fact that said actions don't benefit them personally in any way. At least on the more active tags, a question which is obviously a duplicate is already very likely to get closed. (It may not happen as soon as we would like if a hammer-wielder doesn't see it immediately, but that's another issue.) The problem is that, as OP says:

people who care for the reputation points only writing hasty answers to off topic or duplicated questions always get rewarded

If users gain no reputation for answering duplicate/off-topic questions, and if they know when seeing such a question appear that it will probably be closed, they will be much less likely to put in the effort of trying to put together an answer before the closure comes. That alone would probably solve much of the problem. Dupes are already found often and (relatively) easily, even though it doesn't benefit the finders doing the cleanup; I'm doubtful if rewarding them would improve the situation much.

There is at least one caveat, though - having some answered duplicates can occasionally be useful. We don't want duplicate questions to be answered over and over like they are currently, but a problem can have many possible variants, for which a few question-answer pairs might be nice, "to cover all possible search terms and common permutations of the question."

1
  • I would say that seeing the ones answering duplicate/off-topic questions is a gentle reward for those who are finding them. I would feel myself rewarded just by knowing that finding the duplicates helps to keep the economy fairer.
    – Berriel
    Commented Jul 20, 2019 at 23:17
15

I would not want to receive rep for dupe closing questions because then one could quite easily claim that I'm doing it for the rep and dismiss my actions. I'm also not a fan of people having their "reward" removed when the question they answered is dupe closed... I'd rather we instead find some middle ground where we can get questions dupe closed quickly and taken off the first page... But hey, that already exists! People need to just get over the implied negativity that such actions have and take action.

The problem I have is with questions that are dupes that aren't easily closable because they, and all of their dupes, are so poorly worded/researched that it's essentially impossible to find a good dupe to close them with. No AI or incentive is going to fix that problem. There will always be users who can't write/speak effectively about the subject they are writing about. Several of these cases under the tags I frequent are wholesale solved with very good canonicals, but we need more of them. I was hoping SO Docs would be the solution to this problem, but it went in an entirely different direction.

10
  • 3
    "I'm also not a fan of people having their "reward" removed when the question they answered is dupe closed" - at least the ones who vote to close and answer should have their reward removed.
    – Modus Tollens
    Commented Jul 18, 2019 at 15:32
  • i mean... that's just semantics at that point, yes, some people are motivated by rep, but some of the worst (best?) offenders have so much rep that they can care less about losing a little rep. They aren't doing it for the rep. A direct answer to the OP and then having the OP say "Thanks! This works!" is far more satisfying than closing a question as a dupe to some people. the rep is a side effect.
    – Kevin B
    Commented Jul 18, 2019 at 15:37
  • What else could be the reason? It's more work and more time consuming to post a complete answer again and again. Note that those questions are often the ones where the OP never reacts in any way, too.
    – Modus Tollens
    Commented Jul 18, 2019 at 15:38
  • 1
    Wouldn't removing rep from your answer if you also dupe close disincentive dupe closing? It wouldn't disincentive the answer, because you still get the feel good out of helping someone.
    – Kevin B
    Commented Jul 18, 2019 at 15:40
  • 5
    I mean, look at me for example. I have 87k rep. I throw rep away every day casting downvotes on answers that deserve it. it's cost me thousands of rep, if not tens of thousands of rep. But that doesn't stop me from doing it. rep is meaningless after a certain point, a point that I'd hazard a guess most people who regularly dupe close are far past. I wouldn't want new users coming in and seeing that they can earn 2k rep from casting dupe votes, we already see the consequences of that in editing and the review queues.
    – Kevin B
    Commented Jul 18, 2019 at 16:01
  • 2
    I know, I have far less and do it regularly, too. That doesn't bother me too much. What bothers me is the behavior of answering what should be closed. I feel like a janitor who likes the place clean but has users stomping through freshly swept rooms with muddy shoes because they don't care or have other motives. That's frustrating, that's what bothers me. They should know better. And I don't have the power to stop them or the means to reason with them. And I don't even get paid :D
    – Modus Tollens
    Commented Jul 18, 2019 at 16:05
  • I agree, it is frustrating, but you have to understand it from their perspective to get past that. They're not doing it to spite you, they're doing it because they enjoy it. it makes them and the op happy, regardless of whether or not the question ends up closed, as long as they helped someone it's a win. Removing rep, closing the question, downvoting it, none of that changes it. (though, it does make me feel better to have it end up closed.)
    – Kevin B
    Commented Jul 18, 2019 at 16:07
  • 1
    But it isn't a win when you look at the whole picture. Their behaviour is harmful to the site. Question is - how can we make things better? How can we discourage them?
    – Modus Tollens
    Commented Jul 18, 2019 at 16:08
  • I don't think you can discourage them. Removing rep won't do it, what other thing can we do to discourage it? we certainly aren't going to suspend them for helping people. There's literally nothing that will discourage it, while dupe close voters are regularly discouraged from doing so.
    – Kevin B
    Commented Jul 18, 2019 at 16:27
  • Though... one could argue that because you can earn 2k rep from editing, more people are comfortable editing, even after they're way past earning rep for doing so simply because they were introduced to it with a positive light. Maybe there's some other way we could introduce close voting/flagging and down voting more positively? Maybe rep is the answer, but rep isn't the only lever we have.
    – Kevin B
    Commented Jul 18, 2019 at 16:53
-3

I much more after rewarding the dupe-finders than punishing dupe-writers. There are a lot of people who deserve to have some reputation but being honest and thinking globally they forfeit it for sake of the quality. It is my long desire to reward people for the quality of their contribution, not quantity. --YCS (OP)

I agree, but there might also be a new quality vs quantity issue if we create incentives for quick-and-dirty links, as brought up by CertainPerformance referring to racing/FGITW and by vaer-k:

I think rewarding marking dupes will incentivize even less careful reviewing of duplication than we already have, resulting in a deluge of dupe marks, where the marker has barely even taken any time (or perhaps even no time at all) to consider whether the question even truly represents a duplicate or instead merely covers similar territory. I think this will actually have the effect of increasing the workload of reviewers, as the many, many more questions marked as dupe will need to be reviewed by 5 more reviewers as a result.

Anything above +2 rep (that new users get for edits) could be expected to lead to every single question getting marked as an (almost always erroneous) dupe, harming the signal-to-noise ratio in the dupe links.

My suggested variation on the OP's idea: As a user, I would like to offer the reward as a "dupe bounty" and allocate it to whoever found the best link, similar to a normal bounty. Then both I and the bounty hunters would have incentives aligned with high quality links. (I would take it seriously since I'd be forgoing scarce rep; and they would take it seriously, since they would know the reward would be given based on another user's judgment of quality.)

Regarding details, maybe:

  • can create a bounty as soon as a question is posted
  • 5-25 points, with higher costs mandatory at higher rep
  • offered for up to 4 days, with minimum duration of 1 day
  • each user can have at most 3 open at once
  • if the question is dupe-closed during the bounty period and is new (created within the last 30 days?), all its Q&A posts immediately become community wiki (CW), per @TemaniAfif's suggestion in several comments in this thread
  • the bountier can also manually assign the reward to an answerer

I include the last point since a good answer might win the bountier over, for example, by providing and explaining links to related questions.

YCS had this counterpoint to @Temani's CW suggestion:

@TemaniAfif this is my life's dream. But it will spoil the "gamification", and will never be implemented. So the only feature that possibly could be accepted is not one that strips a person of the hard gained precious points but one that awards them.

My understanding is that CW does not strip points, but only blocks future growth, and if we are looking at a 4-day window before CW goes into effect as I suggested, the extra attention to the Q&A would probably benefit answerers who get in before it is closed more than it would hurt them. (I don't have any strong desire to hurt or prevent the answerers, but I am open to being convinced.)

I'm sure there are other problems with the particulars above (like... isn't it still vulnerable to FGITW as soon as the bounty is posted? what if I don't want to wait 4 days for the active bounty to end before contributing my rep to the reward pile? what if it's already marked as a dupe, but I'd still like to offer a reward and trigger CW?), but anyway I do think we would want some way of ensuring quality of dupe links is maintained if the OP's suggested rewards are attached to them. Right now, the links are very high quality. It would be a shame to see the site flooded with less meaningful links.

And I do think some version of the OP's suggestion is a constructive step even if AI might some day save us or whatever.

2
  • A bounty program for finding duplicates is just not feasible. Bounties would have to be created and where to create them or where not? This will add more work into finding duplicates. Also, I don’t think that users are willing to give their rep in bounties just for finding dupes
    – weegee
    Commented Jul 23, 2019 at 6:13
  • @weegee I would be willing to, and do know how I’d use it, but yeah, there might be few others who think the same way.
    – Frank
    Commented Jul 23, 2019 at 10:21
-4

I disagree with this proposal for a few reasons.

  1. While some people answer duplicates knowingly, sometimes(most times) it's done unknowingly.

  2. I have come to appreciate that what defines a question as duplicate on the site is somewhat ambiguous and most often opinion based. One could easily tag a question as a possible duplicate yet the original poster and/or other users will disagree with the same.

  3. On rewarding duplicate finders: This will create a mess because you'll have more people looking to find duplicates and other uncalled for "arguments" for those whose primary goal here is the reputation.

  4. On punishing duplicate answerers: This relates to point one. Not every question is an obvious duplicate and even if it were, "punishing" a user for answering a duplicate is not the best of solutions. Take the recursive duplicates for instance, these questions are useful because they have different titles, different wording and may be answered in a slightly different/ more updated manner.

Bottom Line: Leave the system as is.

9
  • 1
    these questions are useful because they have different titles : the issue is that most of them have the same title, the same answers, even the same person answering --> those are the problematic duplicates we don't want.
    – Temani Afif
    Commented Jul 20, 2019 at 13:44
  • I follow the r tag mostly. I'll give an example using two of the most common questions. The first relates to non standard evaluation within the tidyverse. This has been answered so many times in an almost similar way but even if you mark it as duplicate it won't be closed(and rightfully so) because what was true yesterday is not true today. I also think some users will be biased towards more recent answers. The second is to do with reshaping data, pretty common but questions still arise each day that one could argue are duplicates but they often have a bit of uniqueness.
    – NelsonGon
    Commented Jul 20, 2019 at 13:48
  • 2
    I am following the css tag and already closed almost 2000 questions as duplicate and I can show you a lot of the same question answrered again and again and again with the same answer and the same users. Even in the future the answer will not change because we have canonical duplicate that are updated with new answers. Having 2 or 3 duplicate question is not bad but having 100 duplicate questions dealing with the same issue is not good.
    – Temani Afif
    Commented Jul 20, 2019 at 13:54
  • 4
    one could argue are duplicates but they often have a bit of uniqueness --> all the questions are unique but can be reduced to the same issue and closed with the canonical duplicate but most of people are lazy reading duplicate question and only want answer to their question. They don't want to do the effort to read the duplicate question and identify the main issue and then be able to fix their own using the duplicate target.
    – Temani Afif
    Commented Jul 20, 2019 at 13:57
  • Fair points, I still think maybe rather than punishing the answerer, people should just take more effort to let them know that it's a duplicate and ask them to delete their answer. I honestly don't see a way to solve this issue that is not "harmful" to contributors in the long run.
    – NelsonGon
    Commented Jul 20, 2019 at 14:03
  • Same, I don't agree with the point of punishing users for answering dupe questions, it at least helps the questioner. I don't care if my answer is to a dupe question, I care about helping the one I have answered to. I only search if it is a duplicate AFTER I have answered them.
    – Syed M. Sannan
    Commented Aug 16, 2022 at 9:08
  • @Stranger This site isn't a helpdesk. Your approach of providing a custom answer to every question doesn't scale and disperses equivalent knowledge. Whoever expects a tailored answer every time they ask are by default asking for too much. On the other hand, setting up duplicate targets provides a quick reference to solving a common problem without the issues described.
    – E_net4
    Commented Aug 16, 2022 at 9:29
  • @E_net4thecommentflagger and what about the answers to duplicates that actually solve a problem differently and in some cases more efficiently? or what about the ones that have modern practices than the original ones? sometimes dupes are different and useful.
    – Syed M. Sannan
    Commented Aug 16, 2022 at 11:55
  • @Stranger Such answers likely can and should be posted in the canonical question instead.
    – E_net4
    Commented Aug 16, 2022 at 12:12
-5

To implement something like this, we have to strike on some existing problems regarding answers

As mentioned by @Tim Post, answers sure get old and the methods inside them get deprecated or no longer maintained. FRs to actually "Mark answers as deprecated" were posted but they got no response back from people. If a method is deprecated then questions about that can arise which are not dupes of the actual question and help future visitors. These are not classified as dupes.

Before implementing this FR, the FRs that are behind these should be implemented before for a better result of what we actually want. Like Remove the stigma of duplicates and to mark answers as deprecated.

As soon as this FR gets implemented, users will race to get their dupe flags handled. This will also storm the review queue and make it hard for people there to approve on most of these flags.

More bad flags will arise too and we don't have enough people to review back there. Which should be our main concern.

-9

There's no way mere human beings can keep up with it at this scale. "Artificial Intelligence" is such a buzzword these days; I hate even using the term. But predictive analysis can give people some super powers in this space.

We've already reduced the friction for anyone with demonstrated knowledge in a given tag to mark a duplicate. Some folks are .... less accurate ... than others, but arguments around the efficacy of the system tend to center on the wrong target being selected, not so much that questions weren't a duplicate to begin with.

If we look more at those that have demonstrated knowledge in a tag and help them coordinate their efforts better, we may actually get ahead of the problem. If we train more exclusively on their history in marking duplicates, we can get a strong signal that something is likely to be a duplicate. This gives us many more options in UX, including more definite language for the user saying you already have an answer to this, the work is done for you, go take a look at this other question.

But sometimes it is necessary to answer the question if it's not obvious to the asker how it applies to their scenario. And there's no harm in helping them see this by showing them how it applies to what they were searching. I'd also caution you to not always assume the motive for answering is just to collect rep, or 'economic' as you put it -- people that have very recent memories of struggling as a beginner have a different perspective than you might.

I don't think it's a question of a sum of rewards. I don't think adding incentive for finding duplicates (though, recognizing effort more isn't off the table) is going to put a scratch in this. I think we're just at a scale where expecting humans to do this is lunacy unless we have machines capable of sorting through thousands of possible targets nearly instantly helping them.

Also there is the problem of information deprecation we must solve as almost a dependency to this. An answer from 2008 might not be helpful to the same question asked in 2020.

Anyway, if we had any real confidence that more incentives to find duplicates would be more helpful than possibly harmful, we would have explored it by now. I hear what you're saying, but we still have to spend more time inventing things that don't currently exist before we solve this problem. I'm very confident that we will. What's equally hard is helping people get to a place where it doesn't stress them as much while we figure out exactly how we're going to get there.

10
  • 23
    Sure maybe humans can't close every question that's been answered, but lots of questions are obvious duplicates that are easily closed quickly, and very often even those questions aren't closed, despite the duplicates being found by simply copy-pasting the title of the question into a search engine and looking at the first result, and yet there is virtually always numerous people falling over themselves to answer these frequently asked obvious duplicates. Just because perfection is impossible and certain cases are hard doesn't mean you should ignore the easily solved problems.
    – Servy
    Commented Jul 18, 2019 at 15:09
  • 4
    I'm with @Servy on this. Often it is pretty clear from the title and a quick glance at the question that a duplicate has to exist, because the question is so simple. Just an hour ago, it happened again, a very simple question got 4 duplicate close votes pretty quick, but managed to get two answers by established users before it could be closed. I wouldn't say anything if this was a rare occurrence or the questions were complicated and it would be real effort to find a duplicate, but this happens regularly on very easy questions where it's pretty unlikely those users didn't realize the dupe
    – Modus Tollens
    Commented Jul 18, 2019 at 15:27
  • 19
    I also feel obligated to point out that it's super demoralizing when, on the rare occasion a very obvious and low quality duplicate gets closed quickly, it gets low quality answers posted after the question has already been closed, because SO allows questions to be answered even after they're closed. To me that's just inexcusable, and is easily fixed by SO with minimal development effort.
    – Servy
    Commented Jul 18, 2019 at 15:30
  • 14
    how about we at least try to move forward, at least with simpler solution that doesn't involve AI (but instead leverages human factor)? Could: "Possible duplicate of.." be given prominence for answering users? Just a little, safe and simple step forward, how about it?
    – gnat
    Commented Jul 18, 2019 at 16:12
  • 1
    @gnat I like that idea!
    – Modus Tollens
    Commented Jul 18, 2019 at 16:18
  • 12
    This is another argument for a beginner Stack Overflow where the rules are more relaxed around duplicates, as well as everything else. You effectively already have duplicate Stack Overflows in other languages, except this time the language changes to "beginner"
    – Jeff Atwood
    Commented Jul 18, 2019 at 22:12
  • 9
    @JeffAtwood: Who would ever visit such a site? Only the beginners would go there. The blind leading the blind, as they say.
    – Robert Harvey
    Commented Jul 18, 2019 at 23:07
  • 3
    @RobertHarvey I actually think it'd be the opposite. You could probably find some answerers willing to go there because hey, it's easy rep and fame for answering easy questions. But the challenge would be getting enough new askers to go ask on the 'novice SO' instead of the 'professional SO staffed by experts' and many people have an inflated idea of their proficiency. This SO has a reputation for getting you answers to questions, a new SO would have to earn it first before considering it. Then again, a site like ELL seems to have worked out, so maybe in practice it wouldn't be so bad.
    – Troyen
    Commented Jul 18, 2019 at 23:41
  • 3
    But there'd need to be a lot of effort put into making sure "beginning SO" doesn't just become "dumping ground for anything regular SO doesn't want to handle".
    – Troyen
    Commented Jul 18, 2019 at 23:41
  • 1
    @JeffAtwood I don't disagree with you. But threads are lighter than forks, although prone to blowing off more toes. Ultimately, well, we'll have to see. I'm still here because I still see a very clear path to things finding their groove again, so to say. If I didn't, I wouldn't be.
    – user50049
    Commented Jul 19, 2019 at 4:27
-26

I think we should learn to stop worrying and love duplication. Rewarding marking dupes sets up a dangerous pattern of incentivizing rampant dupe flags for questions whose similarity is barely, if at all, recognizable. Multiple variations of the same question are healthy for the system, as Jeff Atwood outlines here, provided they otherwise properly adhere to site guidelines.

Furthermore, voting to close on dupes can cull high-quality content in favor of pre-existing low quality content. For example, see this question, which contains a quality title, a detailed and well-researched question, including multiple reproducible examples with descriptions of the expectations and actual results, and a high-quality correct solution with links to documentation. It is marked as a duplicate to this question, which contains a poor question title and very poor information in the question, so poor that the asker had to return to edit in more low-quality content as a weak attempt to add clarity to the question. The asker of question #2 even explicitly states in a comment that the answer to question #1 is not the solution to his question, question #2.

Yet the first high-quality question is deemed to be closed as a duplicate to the pre-existing low-quality question? I think this example highlights a fundamental problem with the way duplicates are handled on SO, one that goes beyond the issue raised here in this meta-question. Additionally, in this case, this example barely constitutes a duplicate, and even if it does, it represents a borderline duplicate, which provides a "benefit [...], as people tend to ask and search using completely different words, and the better our coverage, the better odds our fellow programmers can find the answer they’re looking for."

I think rewarding marking dupes will incentivize even less careful reviewing of duplication than we already have, resulting in a deluge of dupe marks, where the marker has barely even taken any time (or perhaps even no time at all) to consider whether the question even truly represents a duplicate or instead merely covers similar territory. I think this will actually have the effect of increasing the workload of reviewers, as the many, many more questions marked as dupe will need to be reviewed by 5 more reviewers as a result.

16
  • 7
    Voting on meta is different. On feature requests or discussions it can mean disagreement.
    – Modus Tollens
    Commented Jul 18, 2019 at 16:17
  • 6
    @ModusTollens Thanks for explaining. SO should update the tooltip over the up/down arrows to indicate that.
    – vaer-k
    Commented Jul 18, 2019 at 16:19
  • 1
    I mean... i see nothing wrong with the dupe target. The question is straight to the point, and it has a high quality answer. :shrug: If people come across your question, they'll see your answer and then the dupe link bringing them to another answer. It won't auto-redirect because it has an answer.
    – Kevin B
    Commented Jul 18, 2019 at 16:21
  • @KevinB I'm not surprised that there are differing perspectives about this particular anecdotal example, but the larger point I'm making is that new content marked as dupes can be higher quality than pre-existing content. Can you agree that that could happen? Would you also agree that losing the new high-quality content to a dupe mark would be a detriment to the site?
    – vaer-k
    Commented Jul 18, 2019 at 16:25
  • 2
    That can and does happen, yes, and we have the tools to fix that. You can even draw attention to specific cases via meta or chat.
    – Kevin B
    Commented Jul 18, 2019 at 16:25
  • What tools are those?
    – vaer-k
    Commented Jul 18, 2019 at 16:25
  • 1
    Questions can be reopened and reclosed. A dupe badger can do it single-handedly.
    – Kevin B
    Commented Jul 18, 2019 at 16:26
  • This is a possible scenario, but the should forget the aspiration to be the source of all answers for programming questions, becoming a mere forum
    – Your Common Sense
    Commented Jul 18, 2019 at 17:00
  • 3
    If you feel the question that yours is duped to has a bad title or poor information in the body, you have recourse for that: edit them! Make it better for everybody! We're not just here for askers, after all, but all the people that come after them.
    – fbueckert
    Commented Jul 18, 2019 at 17:33
  • 2
    You're missing the "(some)" from the title of that blog post... Jeff was quite clear about the fact that "[h]aving dozens and dozens of variations of the same question is clearly bad" and encouraged us to "please continue to close and merge duplicates as you see fit".
    – Heretic Monkey
    Commented Jul 18, 2019 at 20:19
  • @HereticMonkey I assumed any reasonable person would realize it's obvious that dozens upon dozens of variations of the same question is clearly bad, so I left that part out. And since I never made the argument that we should just accept all possible duplicates, I'm not quite sure what you're arguing here.
    – vaer-k
    Commented Jul 18, 2019 at 20:21
  • It would be great to hear the thoughts of even just one of the people who disagree with my opinion here, rather than just downvotes. I'd love to learn more about how the community understands these issues.
    – vaer-k
    Commented Jul 18, 2019 at 20:26
  • 3
    @vaer-k I immediately downvoted when I read the first line because I disagreed so strongly (even though you were quoting). Then I kept reading and un-did my downvote. Your points are good and valid, but are a bit of a side issue; I believe it has been discussed elsewhere what the proper protocol for merging a newer, higher quality duplicate into a lower quality old question (although it's certainly not a solved issue!). My guess for the downvotes are gut reactions to the first line, and people tend to vote kind of emotionally like that on Meta. A lot of us also just really hate dupes :P
    – HFBrowning
    Commented Jul 18, 2019 at 23:38
  • As a follow up, I think if you had moved the last paragraph to the top, and then kept the second and third paragraph (discarding the first completely), you probably would have gotten fewer downvotes. Welcome to Meta by the way :)
    – HFBrowning
    Commented Jul 18, 2019 at 23:42
  • 1
    @vaer-k actually, I misspoke; according to this, the questions have to be exact duplicates to be merged, which I guess is why I don't see it very often (I didn't know that). Thus your points are even more valid.
    – HFBrowning
    Commented Jul 18, 2019 at 23:49

You must log in to answer this question.

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