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

Disclaimer: I only use SO from time to time, so this is only based on a "light" view of it.

Also, this is not a question about "should we close as a duplicate?". The answer is obviously yes. This question is more about the feasibility of managing all duplicates, and the implications of that.


SO generates a lot of traffic. Questions on the front page stay for about 15-30 minutes max, during which most of them get around 3 answers. It is hard to answer a question without checking new questions very often, since they get answers so fast.

It works well this way; most questions (apart from exotic ones) get correct answers, very fast.

Many questions are duplicates, but only a few seem to be "regular" or obvious ones. In most cases, it will take longer for someone to find a duplicate question, and vote for it, than for someone else to give a correct answer in the meantime.

This is a question only about Stack Overflow, because the other trilogy sites don't have such a large volume of users. On the other sites, typically Super User, duplicates are found, and closed quite fast, most of the time before an answer falls on them. But this is possible only because of the much smaller volume of users and questions.

So the question is: are these duplicates a problem which is out of control when these sites are gaining such volume?

Is it something to really worry about, or does it work fine this way? People can come and get an answer very quickly, without prior searching, just by asking. For most questions, there will always be people to answer. So from the point of view of someone coming to ask, it is not an issue. For obvious reasons, it is also not a problem for people answering, they get reputation this way.

It may be an issue from a storage point of view, since clearly duplication means more redundant data. But more importantly, it can be an issue to find an answer without asking a question yourself. With many duplicates, multiple useful answers become distributed over different questions, and it becomes harder to find most of them.

Extra question following on from these thoughts: should we be worried that the volume of users is finally pushing the site to be much easier to get duplicate answers from, than to search from?

share|improve this question
    
To be honest, if they searched for the answer first (as they should) Google would have already shown the SO answer on top. –  Ivo Flipse Feb 23 '10 at 14:12
    
Questions can stay on the front page for much longer, during off-peak hours. –  Brad Gilbert Feb 23 '10 at 15:46
    
@Ivo: why would they have to? I would say the vast majority of the questions asked on SO could be avoided with a google search. Think if everyone did that. The website would be virtually empty and it would lose the (according to Jeff) 90% of our users, the ones that did google but found an answer on another website because no one asked that question here. If I come up with a question never asked before I ask it even if I could find an answer to google to expand our knowledge base. –  Andreas Bonini Feb 23 '10 at 15:56
5  
@Koper: just as there is a difference between asking your question on SO and asking your question on SO once every hour, there's a difference between finding an answer somewhere on The Internet and finding an answer on the site where you plan to ask your question. Nothing wrong with wanting a Stack Overflow answer to your question. –  Shog9 Feb 23 '10 at 16:35

5 Answers 5

up vote 10 down vote accepted

Should we be concerned? Yes. Here's an example:

How can I achieve jQuery animation queuing?

This is a fairly common question. I found 3 duplicates within minutes of it being asked, and posted them in comments. Note that the first two links I gave have well over 1000 views, hinting at their popularity...

The question was answered, and not closed. And... The answer is reasonably good - I might even recommend it in the future. But since it's attached to yet another question on the same topic, it adds still more overhead to the process of finding an answer for anyone who goes looking for one. If the answer had been posted to (or merged into) one of the previous questions, then it could be viewed much more easily by future readers.

This has the potential to get completely out of hand... Good answers can get less attention than they might have otherwise, mediocre answers can get accepted, and direct comparisons between alternate solutions become more difficult.

But asking people to not answer questions is futile. I think Lance has the right idea here: we need more merging.

share|improve this answer
    
Mostly agree here. The only thing I worry about is questions that support very different wording. I've taken to commenting on those with "new" phrasing to suggest not deleting the duplicate in order to maintain a highly searchable tree of pointers back to the canonical version of the question. I don't know how well that is working, though. –  dmckee Feb 23 '10 at 19:06

I think this is a potentially undesirable result of the reputation system. Would it be possible to add a new feature to the reputation system and answer system where a user could choose to answer a question OR below the text box with an answer, provide the link to a duplicate question. Then if that "answer" get's upvoted, they would receive 15 instead of 10 reputation points. This would theoretically encourage regular users to help moderate duplicates more carefully.

Just a shot in the dark. I find SO an extremely fascinating psychology case in results based on different kinds of encouragement.

share|improve this answer
    
Very good idea. Without such an incentive quality of site will be going down. I don't get all these people who don't see a problem with dupes in which we all drown. –  Piotr Dobrogost Sep 30 '12 at 19:32

I think closing duplicates is fine, but we shouldn't stress about people answering them. As Koper pointed out (and the logic that's always been touted), it just gives us more ways to find the relevant question in a search.

It would be nice if there was more active merging of answers to questions (on real exact duplicates). I think for a high enough level of rep (15k?) there should be the ability to migrate answers from one question to another.

I've put in a feature-request for this.

share|improve this answer
    
@Shog9, thanks for the edit. –  Lance Roberts Feb 23 '10 at 17:25

are these duplicates a problem which is out of control when these sites are gaining such volume?

No, duplicates are not out of control, nor is the volume difficult to manage. We have over 2,500 people that can vote to close for duplicate, and for the most part once one of them finds a dupe, a question is closed very quickly.

share|improve this answer
1  
Yes, once "one of them finds a dupe". Once this is done, anyone able to vote, seeing the "close(1)" will participate to close it. But before that, unless the duplicate is obvious (referred in "related questions"), there will be answers posted. –  Gnoupi Feb 23 '10 at 13:56
1  
About the only friction you get now are when high rep users post the quick and dirty answer without searching for a dupe - others chide them for this, though IMHO it doesn't matter in the long run and the animosity is misdirected. –  Adam Davis Feb 23 '10 at 13:57
8  
The problem is that few of that 2500 bother to search, and quite a few seem disinclined to vote to close. –  nb69307 Feb 23 '10 at 13:57
1  
At the end of the day, the site is to help people with questions get answers. If someone posts an answer before someone else finds the dupe, then the OP gets the info they need - a win, even if the question is ultimately closed as a dupe. If the reverse is true, they win - they go to the dupe and get their answer. –  Adam Davis Feb 23 '10 at 13:58
1  
If it's a dupe that receives answers, but the dupe is never found, then it's too bad only because the existing answer might have better answers. But in terms of filling out the google index with searchable programming questions, it's still a net win. It's going to help more people. –  Adam Davis Feb 23 '10 at 14:00
2  
Personally, most of the times that I initiate a vote to close, it's because I remember the original question. The other case is when it must be a duplicate - cases when I can't imagine it hasn't been asked before. In that case, I search. –  John Saunders Feb 23 '10 at 15:58

With many duplicates, multiple useful answers become distributed over different questions, and it becomes harder to find most of them.

I think it becomes easier. If 5 people ask 5 questions in completely different ways it's much more likely for user #6 to find at least one of them in the dropdown under the title.

Anyway, I agree with you: most of the time giving an answer is faster than finding a duplicate and trying to close it.

share|improve this answer
3  
And if questions 1 and 4 have great answers while 2, 3, and 5 have lousy answers, then users have a better chance of finding a lousy answer to their question instead of being directed to a good one. –  Shog9 Feb 23 '10 at 16:52

You must log in to answer this question.

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