17
$\begingroup$

I had a decent answer that received upvotes to a question that was later closed as a duplicate. Since the answer was well received, I thought it would be good to post it to the original question.

I had assumed that this was the desired behavior so that the original question would collect the good answers. So I was surprised that my upvoted answer to the duplicate was deleted.

@rob I would rather keep my points and have my answer to the original question be the deleted one, however I don’t understand this policy. If we wish to collect good answers in the original question then shouldn’t we not penalize users for doing so?

Could someone clarify this policy and/or revisit it? Personally, I think it is counterproductive if one of the goals of the duplicate question mechanism is to collect good answers in the future.

$\endgroup$

2 Answers 2

Reset to default
5
$\begingroup$

Totally fair question; I wondered if I'd end up discussing it here.

In general, if a verbatim-identical answer goes in more than one place, then something is screwy with the duplicate system.

My judgement was that the new duplicate question doesn't add much value over the older one, and so would probably be purged by the roomba after a while. A good answer like yours can save a closed question from being cleaned up. But you found a better home for your answer. Having two copies of a good answer seems like a way to game the reputation system.

Here is a proposal. If you think the two questions are distinct enough to both warrant answers, let's use the mechanisms we have to get the community to re-open the closed question: edit it so that the differences are clear and put it into the reopen queue. When the question is reopened, we can move your answer back from the older one to the newer one. But I'd prefer not to keep the new duplicate in the archives forever over a difference of two upvotes.

I'm happy to discuss other ideas.

$\endgroup$
17
  • 1
    $\begingroup$ I don’t think that the questions were distinct. I thought it was indeed a complete duplicate, and because it was a complete duplicate my answer didn’t need any revision to answer the original. I agree with the decision to close as a duplicate and don’t see a benefit of keeping both questions. I don’t want to game the system but I also don’t want the system to game me. Personally, I see no harm in leaving both answers. The closed thread is invisible and will not gain additional reputation, so I don’t see it as gaming. Leaving it merely allows me to keep the reputation earned. $\endgroup$ Commented Apr 26, 2020 at 0:24
  • 3
    $\begingroup$ @Dale if they're duplicates (which you seem to agree with) they're duplicates. Double-posting of information is precisely what closure of duplicates is there to avoid. Your new answer will get to the score of the old one soon enough. That's your rep right there =). $\endgroup$ Commented Apr 26, 2020 at 0:34
  • 2
    $\begingroup$ @Emilio Pisanty I have no objection to the closure of the duplicate. I object to the deletion of my upvoted answer therein. That is a separate thing, and seems both unnecessary and counterproductive. I don’t see how that action supports the goals of the site. Although you do seem to be right about it getting upvotes at the new place, which surprised me since it is an older question with an accepted answer $\endgroup$ Commented Apr 26, 2020 at 2:17
  • 2
    $\begingroup$ @Dale At the time your answer was removed, it stood at +3/-0 votes, and the new answer had no votes up or down. Now the new answer stands at +3/-0, so the effect on your reputation is currently the same as if the two questions had been merged. You are coming up on the 20k reputation milestone and we are discussing a transient reputation change of 0.2%. With respect, I feel like your concern would carry more weight if more reputation were at stake, but I don't see the urgency here. Maybe I am missing something? I invite you to explain. $\endgroup$ Commented Apr 26, 2020 at 2:46
  • 1
    $\begingroup$ @rob I am no longer concerned about the reputation in this instance. But I do question the policy. I see it as discouraging a behavior that I presume we want to encourage. Do you not see the discouraging effect? Perhaps I am wrong in presuming we want to encourage people to post upvoted answers to original questions as I did. If so, what do we want to encourage? What should I have done? $\endgroup$ Commented Apr 26, 2020 at 2:47
  • 2
    $\begingroup$ @Dale I look at it from a different direction -- if two questions are exactly the same, and an answer on each is exactly the same, this is a bad thing. Let's say X weeks from now, the answer on the still-open question is improved/revised. Do you want to go back and edit the one on the closed question too? Won't it be confusing to have different answers to identical questions? It introduces opportunities for people to be confused about what the real answer is, and this hinders the goal of the site -- we want to produce high quality Q&A. $\endgroup$ Commented Apr 26, 2020 at 2:48
  • $\begingroup$ @tpg2114 the deleted duplicate question is not relevant. Few people can see it, and those that can have the link to the original. The goal, as I understand it, is to have one definitive original question with all of the best answers identified and sorted. What happens in deleted questions stays in deleted questions. $\endgroup$ Commented Apr 26, 2020 at 2:55
  • 2
    $\begingroup$ @Dale And if your answer wasn't deleted, the question wouldn't be deleted (which is what rob was talking about with the roomba). The closed question would have stayed there, and somebody could come along and post yet a new one that says "Question A is a duplicate of B, but they both have upvoted answers that are different... which is right?" We've seen questions like that before. But, by deleting your answer, now the duplicate question can be roomba'd and deleted for good -- so there's no confusing differences that can even crop up between them. $\endgroup$ Commented Apr 26, 2020 at 3:01
  • 1
    $\begingroup$ It's important to remember that "closed" is not "deleted." The question that was closed because it was a duplicate would be fully visible to everybody, and if your answer there was not removed, your answer would also be visible to everybody. Once the duplicate is deleted properly, then only >10k rep users could see it. I suppose in theory, rob could have deleted the question -- but that would delete your answer too. And it would also mean OP couldn't easily edit it to make it distinct and reopened. If it was edited to be distinct, then your answer wouldn't apply anyway. $\endgroup$ Commented Apr 26, 2020 at 3:05
  • 1
    $\begingroup$ In the end, deleting the answer makes the most sense -- it let's the closed duplicate get automatically deleted if OP doesn't edit it to make it distinct. So no chance for confusion about the proper answer if things diverge later. And if OP were to make it distinct, then your answer wouldn't apply anymore (probably) and could be downvoted or lead to confusion. This really was the most logical way to ensure minimal confusion and automatic "garbage collection" of abandoned things... I guess my comments are long enough, I probably should have just made an answer. $\endgroup$ Commented Apr 26, 2020 at 3:08
  • $\begingroup$ @tpg2114 AFAIK the purpose of the reputation system to encourage behaviors that the community values. Here we apparently value both collecting answers in the original question and roomba-ing old questions. But the required action costs reputation. Should I have done something different? $\endgroup$ Commented Apr 26, 2020 at 3:17
  • 1
    $\begingroup$ @Dale Just because something increases reputation doesn't mean it's good for the community. In general, that's how it works, but the system isn't perfect, and when there is tension it's people who make decisions. If you had left your answer on the closed question, it would have stayed there, visible, probably for the life of the site, and been find-able by people who searched. That outcome would have been fine. You moved the answer to an identical question with higher visibility. That outcome is also fine. But I think it's not fine to have exactly the same answer in two different places. $\endgroup$ Commented Apr 26, 2020 at 3:25
  • $\begingroup$ @rob OK, so it sounds like I may have had a mistaken impression that there was a clear preference for moving the answer to the original question. I will keep that in mind in the future. $\endgroup$ Commented Apr 26, 2020 at 3:31
  • 3
    $\begingroup$ @Dale The "proper" solution would be to ask for the answers on the closed question to be merged into the open one. The end result would be that your answer on the closed question disappeared and reappeared on the still-open one. However, our tools are not great -- merges aren't always done correctly -- and if you wanted it merged, we probably would ask you to just post it there yourself and delete the one on the closed question. The difference between the merging and self-posting-and-deleting option is that you lose 30 rep by the latter, until the answer on the open question is upvoted. $\endgroup$ Commented Apr 26, 2020 at 3:32
  • 2
    $\begingroup$ If we were talking about a really high value answer, maybe we'd work through the merger and possible bugs in how its implemented and whatever. It would be worth it. But for 30 rep on an experienced user, it's not worth the chance of borking the system (in my opinion). $\endgroup$ Commented Apr 26, 2020 at 3:33
2
$\begingroup$

I've been doing a pretty intense programming sprint for the past few weeks, and my directive is to hack in a complicated feature into a big legacy code. The directive is to make it work, quickly, and don't worry about polishing it up or making it fit existing standards.

Of course, if you've ever written a lot of code, you know that doing the same thing in a lot of different places the same way introduces a whole lot of extra effort. If I found I did something wrong in one of these duplicated code sections, I have to remember to change all of them. And if I forget one or two or many, I will think my corrections didn't work and spend a lot more time trying to figure things out...

I bring this up because it's a useful analogy and not because I've spent the past 2 weeks trying to fix something when it turns out I forgot to put a change in all places... yeah, let's stick with that story.

If I took the extra time up front to make sure duplicate code is put into a single place and referenced where needed, then I just have to fix one thing. And I'd be a lot less frustrated trying to make progress.

Duplicate questions with duplicate answers have the same potential for regressions. One question (Question B) gets closed as a duplicate of another (Question A) -- this is the system working to "refactor" things into a single place. But duplicate answers are not directly linked to each other. If an answer on Question A is updated, revised, improved... that does not change the form of the answer on Question B.

This introduces a regression. A user can arrive at Question B and see the answer as well as the duplicate banner, then go to Question A and see a similar-but-different answer from the same user. Maybe they check time stamps and put together what might have happened, but that's pretty unlikely. So instead, they ask yet another question (Question C) that says:

I read Question A and Question B, and they are identical, yet their answers are slightly different. Which one is the true answer?

And now we've made a mess of things. For all we know, Question C collects a different set of answers that are distinct from what is on A or B and now we've failed at creating a definitive, high-quality question-and-answer site. Instead, we have different answers to the same question spread across multiple places.

Instead of letting that happen, deleting the answer on Question B makes it clean. Question B is closed with a link to Question A, and Question A contains all of the answers. In addition, as rob discussed in his answer, deleting the answer means Question B can be edited to make it distinct or it will be removed over time because it is abandoned -- again, making the site cleaner.

So what is the preferred course of action? We want all of the answers on Question A and any duplicates of A should not be considered source material. There is a moderator tool that can merge questions together. This would have taken the answers from Question B and moved them to Question A and preserved rep gained/lost for the people who answered the question. The way to ask for a merger is to A) Vote to close as a duplicate and wait until that is done, and B) Flag the question and explain why it should be merged with the duplicate.

That sounds pretty great, so why don't we do it that often? Well, the advice I often see from other moderators and community managers is that merging is messy. If something goes wrong, it's not trivial to unwind the mess and get things back to the way they were.

Furthermore, merging the question, especially if it happens right after it is closed as a duplicate, prevents OP from editing their question to make it distinct. They would instead go ask a new question that they think is distinct. Minor burden perhaps, but the people who ask 100% identical questions to others on the site tend to be new, whereas people who would know to ask for a merger tend to be established -- I would rather see OP learn to edit their questions than to get into a habit of re-asking the same thing when their original is closed.

All that is to say, for easy-to-copy answers on identical questions, if somebody asked for a merger right after the question was closed as a duplicate, I would probably just suggest they post their answer on Question A and delete it themselves from Question B. Sure, they lose a little rep on Question B -- but posting the answer on Question A bumps it to the front page and if the answer is good, they should get the rep back (and possibly more). And it's a safe, clean way to do it where if something goes wrong, we don't need CM/database developers to fix it.

And it prevents divergence leading to regressions and follow up questions about why there are different answers.

The ideal use case I can see is when we have Question A with many great answers, and then Question B is posted and although Question B is a duplicate of Question A, let's say that Question B is a much better question. We want the best questions with the best answers, and so in that case it might be worth closing A as a duplicate of B, even though B is newer. Then it might be worth merging the answers from A into B. But this is a pretty rare thing and we'll cross that bridge when we get there.

$\endgroup$
1
  • 1
    $\begingroup$ For what it's worth, I have only ever used the merge tool on substantially identical questions, not ordinary duplicates. For instance sometimes a question will double-post by accident, or sometimes a question that starts on another site will get re-asked here and also get migrated here by the other community. These are quite rare; it's been a couple years since I did one. Organic duplicate questions almost always have enough personality that an answer tailored for one would be out of place at the other. $\endgroup$ Commented Apr 26, 2020 at 21:20

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.