74

Scenario:

An answer is posted that is substantially the same as an earlier posted one, and its existence adds no (more) value.

Suggestion

A specific new "duplicate" action, not a flag, like the "close as duplicate" action for a question, to delete the answer as a duplicate of another "original" answer that is selected/identified as part of the action so a comment/message can be added to the duplicate (eg "Deleted as duplicate of link-to-original, with the guidelines that answer:

  • was posted after the original (perhaps by at least say 1 minute)
  • has content is a subset/derivative/copy the original (ie adds no more value)
  • possibly, has fewer than "n" upvotes (but probably drop this criterion)

Same kind of requirements as delete votes to take effect of actually deleting the answer.

Motivation:

  • reduce/discourage plagiarism
  • reduce answer clutter
  • eliminate moderator flags raised for this reason
  • encourage the community to take ownership of keeping their sites "cleaner"

Edit:

I think it would be OK for users to keep reputation earned from answers deleted as duplicates. The answer is not being deleted because it is outside quality guidelines, but rather "merged" with another "same" answer.

  • 8
    Yes! This would be so useful for questions like this. – Mysticial Nov 7 '14 at 23:22
  • 1
    This will be rarely useful on many sites, while causing various issues. At the very least, this should be something that is only activated for some sites. – quid Nov 7 '14 at 23:32
  • 6
    has fewer than "n" upvotes No. – bjb568 Nov 8 '14 at 2:42
  • 3
    @bjb568 good idea. It would only take a couple of enthusiastic newbies trying to win their Vox Populi badge to upset the applecart. Question updated. – Bohemian Nov 8 '14 at 2:51
  • 7
    I'd say up the minimum time difference to be way more than a minute, perhaps 10 or even 30 - loads of people answer similarly with simple questions when playing FGITW without anything untoward going on. – Flexo Nov 8 '14 at 3:24
  • 1
    @Flexo OK nothing untoward, but why not clean up all but the quickest gun if the slower ones have the "same" answer? – Bohemian Nov 8 '14 at 4:14
  • 4
    You end up penalising people who's only crime was trying to help and spending slightly longer waiting for loading/compiling/typing. Penalising fair answers is counter productive and it's not like having 3 FGITW answers kicking around is actively harmful to begin with. – Flexo Nov 8 '14 at 4:23
  • 1
    I submit answers through mobile web, it takes longer to type and it doesn't show any indication that another answer has been posted for that question. After I post my answer, I see another answer posted 5 minutes ago. Consider all aspects that all users don't access SE via the same means. – Infinite Recursion Nov 8 '14 at 15:01
  • 8
    @Flexo and what you end up is penalising web search visitors, forcing them read same System.Text.Encoding.UTF8.GetString over and over and over again. Letting dupe answers proliferate is a step toward old-fashioned "forums", and quite a substantial one I think – gnat Nov 8 '14 at 17:41
  • 6
    1 minute? Uh...no! I don't want my valuable answer killed because I didn't synchronize myself properly with someone else without intentionally plagiarizing. – Christian Rau Nov 8 '14 at 23:26
  • 1
    +1, nice concept.... but later shouldn't be your only criterion for marking an answer as duplicate -- what if the later answer is a higher quality, e.g. brings no newer information to the table, but explains it better? – Shokhet Nov 9 '14 at 18:13
  • 2
    @Shokhet then criterion #2 adds no more value would not be met, so it shouldn't be deleted. Although it would be preferable if the canonical answer was edited with the new information ;if a simple addendum) rather than posting a new answer. – Bohemian Nov 9 '14 at 22:11
28
+100

I absolutely agree with the intent. Sometimes I see answers posted 1, 2, 3 years after the initial influx of answers and these late answer merely repeat what was already said. I tend to see them as attempts to farm for rep. "Surely if I add my own answer which repeats the one that has already gotten all those upvotes, I'm also going to get upvotes when people visit the question. After all, my answer is right, as evidenced by the earlier answer."

This being said...

How Much Difference is Enough?

The problem is how people will determine the "adds no more value" criterion. There are those posts that most people would immediately recognize as adding nothing. However, sometimes someone will add an answer that is basically proposing a solution that is the same as an earlier one but is explained differently. The problem is how different is different enough that the new answer is adding value?

For instance, I tend to see things more abstractedly than the average reader. If an early answers explains the solution using stack and a later answer uses graphs, but it is the same solution, it is unlikely that I will see the later answer as being different enough form the earlier one. However, I know from discussions on the Metas that some would say that there is value in explaining things differently. Some people will understand one explanation more readily than the other. I do not dispute this, but again the issue is what constitutes different enough? Because surely there are explanations that don't in fact explain things differently.

I note that the question here mentions that moderators are already handling flags to deal with such answers. However, I would suggest that users are now not actively looking for such answers to flag whereas if there is a formal flag for it, they will be actively looking for such answers to flag. So the number of flags cast if the proposal goes through should be higher than the number of flags currently cast to delete duplicate answers. However, I realize that if a new flag is introduced, it will be first handled by the community and will get to moderators only in the case where the community cannot handle it. So perhaps when everything is said and done the moderators will be burdened as much as they are now but will only handle contentious cases and this would be a net benefit. At any rate, my concern is with an outcome in which the moderators are more burdened than they are now.

Note here I'm not arguing that the mere fact that there is likely to be a substantial subjective component in the decision to flag or not is enough to make the proposal a no-go. It is merely an issue we should mitigate. The "very low quality" flag is an example of a flag that is consistently misunderstood, at least initially. People use it to flag answers that are full of typos or provide a solution without an explanation. Yes, these make the answer "very low quality" in as anybody would understand the term generally speaking but it does not qualify the answers for the "very low quality" flag.

A Departure from Other Answer Flags

The answer flags we already have typically do not require knowledge of the technologies involved. My go-to example is "Thank! Alice's answer really helped me." I don't need to know C# or SQL or any other technologies. When it comes to duplicate answers, except in the most egregious cases (e.g. cut-and-paste), I need to be able to understand the technology used. In contexts where ++i and i++ would appear to do the same thing, there may be substantial implications that are known only to those who are working with the language. (Yes, I'm looking at you, C++.) Or an answer that looks like it is a different solution is in fact not proposing anything different. An example would be "How do I check that two jQuery objects ($el1, $el2) that I know are collections of only one element refer to the same DOM element?" Answer 1:

$el1[0] === $el2[0]

Answer 2:

$el1.get(0) === $el2.get(0)

In my view the 2nd answer adds nothing. AFAIK, with an index of 0 and in an jQuery object that contains at least one element, the .get(0) and [0] will return the same thing, always. Someone who does not know the technology will most likely not know this and might decide that there is in fact a difference, because, well, they look different. And in the off-chance that someone would actually argue that there is a functional difference I'm ignoring, then this person would have to know jQuery, which is precisely the point I'm making: handling such tags will often require knowing the technology used.

Again, the point here is not that this consequence kills the proposal but, you know, caveat emptor. We have to be aware of the changes this will cause. The queue these flags end up in will most likely end up being filtered by tag by those who review.

  • 3
    wait a minute. Closing duplicate questions often "require knowledge of the technologies involved" doesn't it. And this is considered considered okay. Probably fundamental difference is that closed questions remain visible for everyone so that community can correct mistakes when these happen. For answer deletion, this just ain't so and this is really really big difference. To at least partially account for that, maybe dupe-deleted answers should remain visible to their authors, and maybe authors should have an option to unilaterally undelete... – gnat Nov 8 '14 at 19:45
  • 1
    ...no, an option to undelete and visibility to authors is probably not sufficiently reliable to catch up with safety net established for closed questions. One way I can think of to resolve such a difference is some kind of "folding". Answers are deleted as usual but "dupe target" answer gets an additional link, "duplicate answers" available to anyone and allowing to view texts of deleted answers... and probably also allowing to flag and vote undelete - now that would likely be about as safe as question closure – gnat Nov 8 '14 at 20:01
  • 2
    Right, closing quite often require knowledge of the techs involved. That the new answer flag would require such knowledge is not a deal killer but something that should be discussed before the proposal is acted upon. If the flags for duplicate answer end up in the LQP queue, for instance, people may start filtering by tag, which I'm thinking is not the case now mostly because there's no point to such filtering. And the ideas you are mentioning are exactly the kind of thinking we must perform before moving forward with the proposal. – Louis Nov 8 '14 at 20:21
  • 1
    If the knowledge of the technologies is really an issue, we could enable the flag only to those who has receive a given ammount of reputation from answers to similar tagged questions. – SimonBesner Feb 8 '18 at 15:34
23

Duplicate answers don't fit well with the stated Stack Exchange mission "to make the Internet better". Site visitors have to waste time and effort trying to figure which one is more (?) correct and what is the purpose of repeating same answer over and over again. This leads to frustration and disappointment.

From this perspective, deleting duplicate answers serves essentially the same purpose as closing duplicate questions: it spares site visitors from the burden of looking in multiple places to find answers to their questions. That's certainly beneficial.

Now, ask yourself a question, what if decision about duplicate is mistaken?

  • If we talk about closed duplicate questions, the answer is easy. There is a fairly solid safety net built to correct possible mistakes. Duplicate questions are visible to anyone and conveniently linked from dupe target. Anyone (anyone!) can find and look at these and if they believe that closure is a mistake, they have multiple ways to challenge it and get it corrected: meta, flags, reopen votes and reopen queue, edits (some of which even automatically push a question into reopen queue), chat...

  • With deleted answers, our options to correct mistakes are not even close. That's really bad; given that answers are generally considered more important than questions, it would be logical to expect comparable level protection against possible mistakes wouldn't it.

Now, it looks like we have conflicting requirements.

If we keep duplicate answers visible to question readers, this degrades quality of content. But if we delete, it becomes way too difficult to correct mistakes which also degrades quality of content and, which is maybe even worse, goes against the goal to "maximize the happiness and enjoyment of answerers".

Is there a way to reconcile, to meet both requirements? Hard to tell.

One approach I can think of partially resembles the way how system handles duplicate questions. Answers are technically deleted as usual so that question readers see "clean" content. But along with this, dupe target answer gets an additional link: "duplicate answers". Anyone can click that link, get to a separate page and review content of deleted duplicate answers... and probably also flag and (for same users who are granted a privilege of voting to delete duplicates) vote undelete. That would hopefully be about as safe as duplicate question closure.

(Attentive readers may notice that answer is essentially built upon my prior comments here: 1, 2, 3).

  • There are answers so blatantly duplicative, that deleting them is a very straightforward decision. However, since moderators aren't supposed to have technical knowledge of the answer in question, how about giving more powers to uses with high rep on the relevant tags, such as nominating for deletion? @CodyGrey proposed something similar. – Dan Dascalescu Jun 30 at 4:29
12

I like this idea, but I think you should only be allowed to vote against an answer which has not been edited for either 24 or 48 hours. In various "fastest gun in the west" cases, simple answers will be posted within several minutes of each other, and the answerer ought to be given a chance to expand upon it.

The other issue, which is not as simple to address is if the earlier answer is the less informative duplicate. What could happen in such a case, I am unsure.

  • If there are rapid concurrent answers, then guideline 1 "at least 1 minute later" would not be met (that period to be tweaked of course). If the earlier answer is "less informative" then the guideline 2 "adds no more value" would not be met. – Bohemian Nov 8 '14 at 0:26
  • @Bohemian what if at the time of posting the second answer is more informative, but then later the first is expanded and the second becomes less informative in comparison? (How does one even know if it happened under 5 minutes?) – quid Nov 9 '14 at 0:34
  • 1
    @quid only guidelines. Like closing questions as dupes, deleting answers as dupes is up to the community to figure it out. – Bohemian Nov 9 '14 at 2:01
  • Let's start by solving the problem of duplicate answers posted at least a month after the original. – Dan Dascalescu Jun 30 at 4:30
1

I happens a bit where two people will be writing an answer at the same time. When you click done you see a duplicate that was posted while you were working on your answer.

I always just delete my answer as the other was posted first.

On many occasions I will see duplicates of mine posted later and will comment this is duplicate of a an answer posted earlier. About 1/4 of they time they refuse.

I just let it go but I think I will start flagging moderators.

Another thing that happens is people will copy paste to add something. Unless adding something substantial I think you should comment on the existing answer. Or edit the existing answer.

Questions should look for the answer posted first and give it the check. I don't say anything to the OP in that case.

  • Are you saying that 75% of those you notify of posting a duplicate answer acknowledge that? Wow! I've had a 0% success rate. – Dan Dascalescu Jun 30 at 4:31

You must log in to answer this question.

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