What's going on …

There is a trend that is very easy to explain.

  • A question is asked.
  • It gets a few answers, and because it's the first of its kind (i.e. topic-wise), it receives a lot of popularity
  • Votes trickle in, but at some point they stop, because the question isn't "new" anymore.
  • The question and its answers stay like this.

Now, sooner or later, this happens:

  • A new question is asked. It covers the same topics as the older one.
  • It gets a few answers, which are now state-of-the-art, gets discussion, comments, etc.
  • Somebody looks for a duplicate and finds our previous question.
  • Close votes trickle in, sometimes very fast, sometimes not.
  • The new question is closed eventually.

I won't give any specific example here, because if you've used the site, you'll know that this is more or less the same for every duplicate closed. Some are worse, some are better.

What the problem with this is …

In the following "duplicate" is referring to the question linked to as the "exact duplicate".

  • The duplicate might be absolutely outdated. This means that either technology has changed, software is no longer updated, new features have been added, etc.

  • You can't really fix a duplicate. If the above is true, it is very hard to "fix" such a state. As the duplicate is linked to quite often, it will probably have accumulated a large number of votes and views. At the same time, its answers might not reflect the current state of the art anymore (for reasons given above).

    In order to fix something like this, you'd have to go through the whole question/answer pair and see if you can salvage it. Even worse, you might even want to close the duplicate as "not constructive" or "off topic", or "too localized", if it happens to fall under any of these categories. You would then end up in a state where the "new" question is locked (in some sense), and the old one too.

  • You can't really add new information. Well, you certainly can. But if you post a great answer to a question that's a year old – or if you've finally found the solution to a big problem –, you'll start off with a score of 0, among many other low-quality answers like "I kinda had the same problem, and fixed it by …", or "This is a great solution: <insert link here>".

    How are you ever going to make it to the top again? You won't have the ability to gather all these votes that would allow you to outscore the other answers.

  • By closing the new question, you stop the process of getting new information. After a question's been closed as a duplicate, you eliminate the possibility for users to add new information to an old problem. This is not to say that they couldn't just post at the duplicate, but for the reasons in my previous points, this is not always useful.

    Also, from my experience, I haven't really often seen users posting their answers at the duplicate, even when they are able to see the "possible duplicate of … " link.

Should we do anything?

Duplication is not necessarily a bad thing. Every once in a while, we need to come up with something new. I'm out of ideas how to approach this inherent problem of Stack sites.

What do you think we can do? How can we teach people to vote more on duplicates, sort out the crap that's accumulated, edit old posts, et cetera? It would take a lot of involvement that's much more time-consuming than just posting a new answer, not worrying about the fact that the question you're answering is a duplicate.

share
1  
The moderators need to hit that merge button more often. –  Daniel Beck Oct 3 '11 at 8:18
1  
Works for some cases. For others, if the OP or the answers failed to generalize the problem statement, it would lead to confusion though. –  slhck Oct 3 '11 at 8:20
2  
The question then should be edited to apply more broadly, if necessary. That's a given. For many questions, it probably won't make much of a difference. It needs to be decided case by case anyway. –  Daniel Beck Oct 3 '11 at 8:23
    
At some point a person might analise how the format or practices can turn people off so badly, that they Dont even show up, to ask or answer questions. My vote would be to continue the Dumping process. I would rather have my question dropped down into "stupid endlessly repeated User" section :-) As oppoesed to be in the "We already know that shut-up" section. :-) –  Psycogeek Oct 12 '11 at 7:49
add comment

2 Answers

One option: Vote to close the original question instead. If its information is no longer current, and the newer question and answers to it have been posted anyway, make the newer question the "original". Could single, still applicable answers be moved to the newer question? Or will merging always merge into the older topic?

This invites participation on the site without being told "this topic from mid '09 covers that" — even though its information is out of date and the user has no way of making it appear on the front page again (without posting a "not an answer" or hijacking an old topic, as is happening sometimes with bounties). Having a few of these topics on the front page, instead of buried in the archives, will also lead to new, good answers that wouldn't have been posted otherwise.

OTOH, this will invite creation of duplicates. Users might post answers for those questions even more than now, since the topic might become the often-linked-to new canonical question on the topic. Some pretty much solved topics (backups *cough*) will stay on the front page forever, without actually getting new answers.


Feature Request: There should be a way to deprecate answers that are no longer useful, something which counts as a topic modifying action, pushing it to the front page.

They might have been useful at the time, but things change. I don't want to have them deleted, and make them disappear. Deleting once perfectly valid answers that, as a kind of warning, have a date attached to them, punishes actively participating users, as the reputation gained is lost. In addition, someone still might be using that out-of-support ancient software version* for which the answer still applies (and newer ones don't). Just tag it as This is probably no longer useful unless you know better, make it a grayish text color and don't count as answer with respect to the Unanswered tab (similar to deleted answers, but still visible to everyone).

*I'm not referring to the Windows 8 Preview discussion here. More like "Our organization has rather slow update cycles, so we're still using Windows XP / Red Hat Enterprise Linux 3"

As a specific example, consider this question. The user tried to use instructions for resetting the password that were at least two years old by the time his OS was released. With this feature, we could mark answers as deprecated on this site, as an additional indicator that you should really know what you're doing before doing what the answer says.

share
    
I kinda like the idea of "reverse" duplicates. This would however require a bit of moderation overhead, as not all look for the old duplicate and close it, unless you specifically tell them to (or 10k users / moderators do it). –  slhck Oct 3 '11 at 17:03
    
Also, as you said, there's not always a need for those. Who gets to judge if an old duplicate is outdated? In the end, maybe it's about educating people not to post at the newly asked question but focus their energy on cleaning up the duplicate itself. Then again, you can't get them to do that either. Either way, it's a very hard problem to solve, I guess. –  slhck Oct 3 '11 at 17:04
    
@slhck In some ways, it needs the deprecation feature to make reverse duplicates work. Or at least, they'd work better. It could work similar to off topic destinations: a voter decides which question to keep, and the majority wins. Of course, this requires specific support for this feature on the site, and will generate negative responses ("But MY QUESTION is the original!!"). Another, "cheaper" option would be to bump the old question once there's a newly closed duplicate of it, just to get some interest. But "manual" reverse duplicates is the only thing I can think of without new features. –  Daniel Beck Oct 3 '11 at 17:25
    
Well, I always try to fix what I can, whenever I VTC a duplicate, but the impact of bumping it is marginal I'm afraid. –  slhck Oct 3 '11 at 17:39
    
The invitation of creating duplicates is something we don't really want, I've only read in a slant way but you can only do this if the new question has already generated more useful content or the old question is simply not relevant anymore. As for the feature request I'm pretty sure it won't happen given that we can't vote to accept stuff either, and who is wanting to go through all the old content like that? As you describe it it doesn't really fit in the system, so I doubt if anyone else besides those that read about it will use it... –  Tom Wijsman Oct 3 '11 at 21:31
1  
I think the suggestion for a "deprecated" or "superseded" close reason is a very good idea actually, and might be well received. Suggest it in a separate post here or Meta Stack Exchange. –  nhinkle Oct 3 '11 at 23:46
1  
+1 for the deprecated answer feature. its so simple but a great idea –  Joe Taylor Oct 4 '11 at 10:50
add comment

There are several different states for this situation:

  • The original question is still relevant:

    If the original question is understandable, has high-quality answers, and the answers are still relevant and useful, then the new question should be closed as a duplicate. If there are any useful answers with new content on the new duplicate, then the questions can be merged.

  • The original question is outdated:

    Even if the original question was high-quality at some point, if the answers are no longer relevant, or if there are better solutions now, you have a few options:

    1. Close the original question as a duplicate of the new question, but reference it in the new question. The historical relevance lock reason can also be applied here.
    2. Try to update the content in the old question, then merge the new question into it. This should only be done if there are a manageable number of answers. This way, you bring in new perspective while retaining potentially useful older answers.
       
  • The original question is cluttered:

    Sometimes the original question may still be a relevant, reasonable question with decent answers, but it's got so many answers that it's hard to wade through everything. Adding answers with new information is practically impossible because they'll never be seen.

    Use the information in questions like this to create a new answer, then close the old one as a duplicate of the new one. If it's such a popular question that it's got an unmanageable number of answers, it might be worth writing a Community-FAQ on the subject to try to create a canonical answer.

  • They're actually different questions:

    Sometimes, "duplicates" are different in some way, and it might be valuable to have multiple questions on the subject. In this case, it's OK to leave both. Perhaps put a note on each one to the effect of "see this related question", but they can coexist.

    For example, you might want to have a question about backing up Windows 7 and a question about backing up Windows XP. You could say that they're duplicates, but technology changed a lot between XP and 7. Even though the XP one is outdated by now, there are still a lot of XP users out there, so you don't want to close it - it might still be useful for someone. In this case, it makes good sense to keep both questions open and separate, but put a reference between them.


When in doubt about a specific question, you can always ask a question here on meta, or discuss it with a moderator (and other users!) in the Ask a Super User Moderator Chat Room. In general, I personally think we should be promoting high quality new information, so long as we're not duplicating something that's already good and current.

share
    
Wow, I somehow must have missed the notification for this answer. That sums up everything pretty well. I wouldn't do anything without community support anyway, so I guess the best thing to do would be to bring it up here. –  slhck Oct 5 '11 at 20:33
add comment

You must log in to answer this question.

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