What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 122 Stack Exchange communities.

This question (currently deleted (visible only to 10kers)) is a great example of a duplicate question that should not be deleted. [Note: this question was later merged into this one, which resulted in the original being hard-deleted. This issue with merging was later fixed ]

The titles of the duplicates are completely different. Having duplicates around greatly aids search hits on the site.

We need to keep questions like this around for the benefit of the community.

Both questions have around 4K views. They are both relevant. I cannot believe that a question with over 3K views is being deleted.

Edit
Search for python continue line and python line break in google. Both answers are top hits. They should both be open.

share|improve this question
15  
+1 Fully agreed. Helps stops future duplicates too... –  Dan McGrath Dec 8 '09 at 17:54
5  
We're already through this, aren't we? meta.stackoverflow.com/questions/230/… –  Ladybug Killer Dec 8 '09 at 18:48
1  
Well, John, that question is asking about etiquette. I am trying to make a bit more of a stand. –  jjnguy Dec 8 '09 at 18:58
    
Have yet to use the option, but noted. –  OMG Ponies Dec 8 '09 at 20:38
1  
Why do you want to flood Google with essentially the same results instead of keeping one definitive question and answer for that specific problem? –  Gnome Mar 24 '10 at 21:17
10  
@Gnome, You don't get the same results with different searches. But, with different searches, you get relevant results if the two questions are left undeleted on the site. –  jjnguy Mar 24 '10 at 21:44
2  
@Gnome, also, Flooding may be a bit strong of a word...The internet is huge, and the results from SO make up a tiny fraction of google's search results. –  jjnguy Mar 25 '10 at 3:11
2  
I'm upvoting this. Deleting duplicate questions — just deleting any user-contributed content without notifying the user — is a terrible idea. BTW, the page seems to have now been so thoroughly deleted (how?) that it's not visible even to 10kers. –  ShreevatsaR Sep 21 '10 at 5:22
    
@Shree, wow. Yeah. I wonder why that was deleted... –  jjnguy Sep 21 '10 at 12:14
2  
I'm guessing it was merged before the improved merging capability was finished. –  mmyers Sep 27 '10 at 13:08
    
@mmy, yeah. That makes sense. –  jjnguy Sep 27 '10 at 13:09
    
Shouldn't this question be tagged as feature-request? –  Anderson Green Nov 8 '12 at 2:33
    
@AndersonGreen no, this was directed at the community. –  jjnguy Nov 8 '12 at 4:10
1  
So somebody undelete it! We have definitely reached a consensus –  Doorknob Apr 9 '13 at 20:24
    
@Doorknob that question no longer exists. Not sure what happened to it. –  jjnguy Apr 9 '13 at 21:25
show 3 more comments

11 Answers

up vote 90 down vote accepted

I agree. As far as I'm concerned, only off-topic or unintelligible questions should be deleted; duplicates are quite useful to keep around.

share|improve this answer
5  
Especially if they have links to other similar questions. :D –  Gordon Gustafson Dec 8 '09 at 22:16
3  
what about unintelligible duplicates all by the same user like these: stackoverflow.com/questions/8519944/… and stackoverflow.com/questions/8529423/… and stackoverflow.com/questions/8526636/… ? –  noah1989 Dec 16 '11 at 10:20
1  
@noah1989, those are pretty much worthless and two of them have been removed. Thanks for bringing them up. –  mmyers Dec 16 '11 at 15:54
1  
Wow, +72 for basically saying "I agree" :P –  Doorknob Aug 14 '13 at 13:45
1  
What about duplicates that don't offer anything new in their topic or content that doesn't already exist in another duplicate? –  crush Jan 31 at 20:14
    
@crush same as unintelligible, delete them. –  Braiam Mar 17 at 13:15
add comment

This seems to be Jeff's intention as well:

There’s often benefit to having multiple subtle variants of a question around, 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.

And some time later:

I wish people wouldn't delete questions with good answers. You're destroying the useful contributions of your peers!

share|improve this answer
add comment

It's a tough call, especially with older questions. Duplicates are great when they work to funnel traffic back to an authoritative answer, but some older dups lack links to the questions they duplicate, or contain many good answers. In the former case, the question is actually counter-productive to searching (rather than acting as a sign-post, it behaves as a distraction), while in the latter case users must read both questions to access all relevant answers.

And then there are duplicates of questions that should never have been asked on SO in the first place. Whatever your feelings are on questions like "Best Programmer Joke", there's really no need to have multiple instances of those cluttering up the site and encouraging more new users to mimic them...

So current my strategy is this:

  1. Leave duplicate questions that are significantly different in wording.
  2. Delete duplicate questions that:
    • Don't have links to the original and don't have good answers, or
    • Aren't valid questions to begin with.
  3. Request merging of valid, effectively-identical questions where both questions have good answers.

Unfortunately, merging is also broken. It leaves no signpost, not even a deleted one. Migrated questions at least leave the question itself, locked, for future searches to find... a strange discrepancy indeed, considering most migrated questions are completely off-topic and don't particularly benefit the site when they appear in search results. I strongly encourage everyone reading this to vote for the When merging questions, leave the duplicate one as a placeholder suggestion (and the answer about automatic redirects as well... ); if this were implemented, I would have no qualms about asking moderators to merge questions like the one referenced here.

Update: Since it was implemented, please try to merge duplicates that get good answers whenever possible.

share|improve this answer
    
I would rather see that links to the original get added for duplicate questions that don't have links to the original. –  alexanderpas Sep 7 '10 at 18:43
    
Probably worth noting that when I originally wrote this answer, dup-closing and merging were pretty primitive compared to today. Doesn't change the guidance, but does make deleting these less likely. –  Shog9 Mar 24 at 13:58
add comment

The reason we want to keep both of these questions is because they have key words in them that will assist future searchers.

This is handy, because oftentimes you don't know the proper terminology, especially when you're dealing with a subject that you're unfamiliar with.

A new programmer might not know the phrase "line continuation" -- if they did, they would probably just google "python line continuation" and get an answer right away. But because this is a place where people can ask questions and get responses from real live people, we can ask open-ended questions, or describe something that we're not sure there's a proper term for.

So if they are identical -- completely, totally, same-hash-result identical, then yes, of course, delete them. If they are two different paths to the same answer, keep them.

share|improve this answer
7  
I'm reminded of the time there were two questions on the SO front page, one asking "What does X mean?" and one asking "What do you call Y?", and they answered each other. If my memory were better, I'd tell you what X and Y were. –  David Thornley Dec 9 '09 at 15:24
add comment

I think if you can, by editing the title in the original, coalesce two questions into one, then deleting is still ok. Not every duplicate needs to be kept and with some manipulation you can make the original easier to find and remove the need for the duplicate. I don't know that the referenced question is a particularly good example of that, though changing the title of the first to:

How can I do a line break or continuation of code in Python?

might suffice to make the second question unneeded. The reason I say that, is that the question is relatively simple and the answers nearly identical in content.

In some cases, however, the question might not be worth keeping, but the answers in the duplicate may be. In that case, merging questions might be a better option and I would flag for moderator attention if that were the case.

Note, too, that there is a risk with duplicates -- the search may find the question with an answer that is not as good as the original, in fact, there's no guarantee that it won't be wrong. A question closed as a duplicate may not get the attention it needs to have a correct answer emerge as the best. Merging questions, where feasible, could avoid this.

share|improve this answer
    
Maybe if we looked at it as "migrating answers" and not "merging questions"? Keep the duplicate question - closed as a dup with link to the canonical question - and just move the answers to the open question? –  AnonJr Dec 8 '09 at 21:02
    
That sounds like a reasonable way to merge them IF you can't change the title to match search queries that would also have returned the other one. If you can massage the title -- or the titles already turn up in the same queries -- then I don't see any point in keeping both questions. –  tvanfosson Dec 8 '09 at 21:28
add comment

I am one of those who flags duplicate questions for moderator attention, asking that the answers get merged onto the original question. Keeping duplicates around means that the responses are distributed across several questions, which hampers those who actually do perform searches on the site first before asking a potential duplicate.

However, perhaps we can handle duplicates better, perhaps by saving the text of the duplicate question (rather than discarding it, as now) so as to preserve the alternate phrasing. However, I would prefer that the moderator who performs the merge performs the reconcilation by hand, by editing the original question as appropriate so as to include the alternate wording.

share|improve this answer
4  
If merging would be improved, so that the original question was still preserved in some way to assist in searches, merging would be the solution for all duplicates. But before that there needs to be some improvement in merging questions, see also meta.stackoverflow.com/questions/16437/… –  sth Dec 8 '09 at 20:09
add comment

At the risk of incurring everyone's wrath, I'll admit that I was one of the people who voted to delete that. I'll explain my thought process for the sake of discussion...

Although I understand the stance put forth by Jeff about the usefulness of duplicates (i.e. improving the coverage and search results), I can't help but wonder if a year-old, sparsely-tagged, closed duplicate question is really going to help at all. Consider that even though the original question and the duplicate have been around for over a year, we still had this question appear just 11 days ago (and I'm fairly certain I've seen even more copies of it show up before).

If I go to the search box and type the title of the older duplicate verbatim, I still get the newer duplicate showing up first in the list. The newer duplicate also shows up first if I search "python line break". Maybe this is a failing of the SO search box, which a lot of users will probably use instead of the Google search route. Or maybe this is intended, since the older duplicate is so old.

In short, I must admit that I have doubts about keeping duplicates around:

  • How well do they really redirect traffic to the originals?
  • How many duplicates are really necessary? Do we really just need a couple?
  • Should older duplicates be progressively purged since newer duplicates may show up first in searches anyway, negating the need for the older ones?

EDIT #1:

I'll concede the point for this particular question, since it does have a lot of views and a good rank on Google. However, I still don't like the idea of having content distributed across multiple questions, and would still lean in the direction of a merge at least. Perhaps just the answers from the duplicate can be merged into the original, leaving the duplicate question as merely a routing point to the answers?

Incidentally, I just noticed that 10k+ users can vote for delete/undelete as many times as they want (unlike close/reopen votes). That could make for some ugly delete/undelete wars (like the one the question seems to be undergoing now). But don't worry... I won't be voting to delete that one again.

EDIT #2:

I just wanted to address the assertion that a high ranking on Google means we should keep a duplicate around. Seeing as how popular SO has gotten, getting a high rank on Google almost seems trivial. Almost any SO question could probably achieve a good Google ranking, so a duplicate with a good Google ranking isn't necessarily so special.

Take, for example, this MATLAB question I answered earlier today. Now, MATLAB is hardly a popular tag on SO (currently just 1,003 questions out of 578,286 total), and the question above currently has only 39 views, 4 answers, and nothing voted higher than 4. And yet it is already fifth on the list when I search Google for the phrase "MATLAB duplicate rows", and it's only fifth because four links to The MathWorks (the makers of MATLAB) come in before it!

I'm not saying it's a bad question. Many MATLAB users would probably find it quite useful and informative. But if an underdog like that can rocket up Google's search result list, is good Google ranking really a legitimate reason to keep old duplicates?

share|improve this answer
1  
what was the viewcount on the question? If the viewcount is low and stays low, then people aren't arriving there. That is data supporting your position. (I haven't actually looked in this case, but I'm just saying, the data is there.) –  Jeff Atwood Dec 8 '09 at 18:58
    
Both view counts are above 3K. –  jjnguy Dec 8 '09 at 19:00
1  
But how do you know when those views took place, or that any of them translated into people going to the original? –  gnostradamus Dec 8 '09 at 19:05
2  
Both questions are top hits on google. They are both valuable to have around. –  jjnguy Dec 8 '09 at 19:12
    
Regarding the multiple delete votes, see meta.stackexchange.com/questions/9431/…. –  mmyers Dec 8 '09 at 20:40
    
@gnovice The reason I felt google rank was important was because people don't ask the same question in the same ways. And, since they were both showing up in google searches, they were both helpful to googlers. –  jjnguy Mar 24 '10 at 21:42
add comment

Just as an example: I just noticed a question on Super User that was closed as a duplicate 12 hours after it was asked, but meanwhile has received 20,648 views in about 2.5 years. The older duplicate was viewed 6,665 times.

If this duplicate would have been deleted, then I guess we would have seen more duplicates posted, and maybe people would not have found their answer at all.

However: I find it odd that not all viewers clicked through to the other question. This might be caused by the answer that was posted to the duplicate, but I'd still be curious what the other answers would be.

share|improve this answer
add comment

Everybody can help with the searching side of this issue

If the title of the duplicate question was added as a comment to the original question, that would help with the searching issue. The alternate wording of the same question would appear on the same page and possibly come up as a search result.

Suggestion for site programmers

If two questions are really exact duplicates, no reason why the answers could not be merged on one page (with respective votes in tact), and the duplicate title added on as some sort of alternative/secondary title for searching purposes.

share|improve this answer
add comment

To post a counter argument to this (though I agree with the points made). If duplicates were deleted then the problem with lots of questions getting lots of points would be less prevalent. (As the rep from upvotes would be lost on the next recalc.)

Maybe that could have deterred the "halfing the points for questions" change. (Or at least if dupes did not gain rep. it may have been able to make the change less needful.)

share|improve this answer
1  
I don't have a problem with dupes gaining rep, as long as they're not posted in the same words as one of the others. Using different wording to express the same thing is helpful for Googlers. –  mmyers Mar 24 '10 at 20:36
    
@mmyers, it is fine to leave them there for the googlers, but if dupe questions were not gaining rep for the questioner then there would not be a draw to post them just to get rep (for those that abuse). This would stop a lot of the question spamming. (I think I will post this as a separate question.) –  Vaccano Mar 24 '10 at 20:54
    
Found it already posted here: meta.stackexchange.com/questions/2038/…. Added to that question. –  Vaccano Mar 24 '10 at 21:03
add comment

I actually started following the recommendation of this question. Notorious as I am for deleting, I actually stopped deleting duplicate questions.

A little while later, I changed that to delete duplicate questions where the titles seemed sufficiently similar to the original. It made sense for the "extra" titles to stay around, as pointers to the original.

After a while, I got tired of that exercise. After all, the answers weren't getting merged together, so they were going to be effectively lost.

I'm now back to deleting duplicates along with everything else. When we get a better process for merging duplicates, perhaps adding a table of alternate titles to help search engines, then I'll go back to not deleting them.

share|improve this answer
    
FYI: That's terrible, and 4 people hate your idea. –  bjb568 Mar 13 at 6:31
add comment

You must log in to answer this question.

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