As Gaming grows, we're seeing a big increase in duplicate questions from new users. This puts us in a bind, because while we certainly want to close the duplicates, doing so can have a negative impact on whether those users choose to continue with the site. It seems like everyone would be better off if we could prevent those duplicate questions before they were even asked.

The "Similar Questions" feature aims to do this, but because it is passive, a new user to the site is unlikely to notice it and/or understand its implications. Could we add something that if there is a question that is very likely a duplicate of the one about to be asked, new users would get a popup indicating that we think there question has already been asked, and linking them to the duplicate instead? (This would obviously be dependent on how well we can detect duplicates automatically. Utilizing the tags on the question would help immensely for sites like Gaming)

| improve this question | | | | |
  • 1
    +1 Google Mapmaker does this already, very useful feature. Make sure it's fine-tuned to show a few relevant results, not a whole bunch of them--which everyone will just skip. – Manishearth Mar 28 '12 at 15:30
  • 1
    +1 - Given the lesser focus on code and more on text (with potentially more specific keywords) I could see this working much better on gaming.se than it would on SO. – cdeszaq Mar 28 '12 at 15:47
  • 2
    The "Similar questions" functionality isn't very good until the user has typed in the body of the question. It is unlikely you could get a reliable match until the user finished at least part of the body. How would you expect this to work? Would the popup appear when they hit "submit?" – user102937 Mar 28 '12 at 15:47
  • @Robert Yes, I was figuring this would happen when they clicked Submit. At the point we have the whole text and all the tags, giving us the best opportunity to correctly spot a duplicate. I know it isn't the friendliest thing to have a popup at that point (after they've done all their typing), but I think it is better than letting them ask the question and then closing it immediately. – bwarner Mar 28 '12 at 16:35
  • @RobertHarvey - The current "Similar questions" functionality could be made to update as the user is typing in a more "in your face" fashion if there is a particularly strong match, or perhaps the "visual strength" of a suggestion could increase as the user types and the match gets stronger. – cdeszaq Mar 28 '12 at 16:40

This puts us in a bind, because while we certainly want to close the duplicates, doing so can have a negative impact on whether those users choose to continue with the site.

Why in the world would closing questions as a duplicate have a negative impact on whether users want to continue? Their question has already been answered before they even had to ask it. Talk about awesome. It's like your site can read their mind. They get [nearly] immediate answers to their question that have already been vetted by the community without even having to wait. I fail to see how this is anything but a positive experience.

Besides, if you really think the fact that a question is closed as a duplicate is somehow an off-putting experience, then how is an "aggressive duplicate prevention" system going to be any less off-putting?

The pop-up that they'd see when trying to submit the question would be not only annoying and increase user friction, but unless it was impossible to dismiss (which is a really bad idea for imperfect algorithms, and this is the kind of thing you just can't get to be perfect), wouldn't keep them from submitting the possibly duplicate question anyway. And in that case, you'd just have to go through the routine of voting to close it as a duplicate anyway.

It's much better to have an actual human read the question, understand it, and propose a duplicate. Then to have other actual humans (4 of them) do the same thing.

| improve this answer | | | | |
  • 6
    Closing is a negative experience, regardless of how it is spun. From their point of view, the new user has had their hand slapped by the moderators for doing something wrong. A user actually has to participate in the site for some time before they "get it" and realize that having a question get closed is not a bad thing. – Jason Berkan Mar 28 '12 at 16:57
  • 2
    @Jason: That seems non-responsive to the issues I raise in my answer. First of all, as you say, new users don't have a baseline by which to measure what is positive or negative. There's no particular reason to assume that being pointed to a question that already contains the answers to their question is a bad thing. It's not a "hand slap" unless you tell them it's a hand slap. Second, if it is a hand slap, then why in the world would it be any less negative for a message to come up that says, in effect, "You cannot post this question as it's a duplicate of another question already asked"? – Cody Gray Mar 28 '12 at 16:59
  • 3
    I don't know if the proposed solution is good or not. I was just taking issue with the "closing is not negative" statement. While true, I have found that is not the way new users think. I make (or made, when I had the time) a conscious effort to ensure all new users on Gaming that have a question closed get a good comment explaining why it is closed and what they should do next. – Jason Berkan Mar 28 '12 at 17:05
  • 3
    @JasonBerkan: if a user cannot understand the importance of Closing as a Duplicate, you do not want them participating on your Q&A site. They likely have mistaken it for a forum. – user7116 Mar 28 '12 at 17:27
  • 4
    @sixlettervariables - Sorry, but that is far too rude of an attitude for me. I want new users on the site - there are large sections of the gaming community that are not represented on Gaming SE. Slapping their hand and then being rude to them is certainly going to send them away, ne'er to return. – Jason Berkan Mar 28 '12 at 17:33
  • 2
    @JasonBerkan: some people aren't actually looking for SE, and that's Ok. It does not make sense to suspend the rules because new people might not understand them (or misunderstand it as a negative). – user7116 Mar 28 '12 at 17:35
  • 1
    @sixlettervariables - I'm not certain anyone is talking about suspending any rules - duplicates are duplicates and I VTC them. bwarner just wants a better automatic way to hit new users in the face with a duplicate before it gets posted (and then closed). – Jason Berkan Mar 28 '12 at 17:38
  • 1
    @JasonBerkan: they will still ask their duplicate question (just phrase it differently), or worse, a non-duplicate will be blocked. Humans closing as a duplicate is the only accurate approach. (perhaps I should rephrase "suspending rules" to "automating intentionally manual processes") – user7116 Mar 28 '12 at 17:42

If a question is very nearly exactly duplicate, and the canonical question is upvoted and has a selected answer, then close it as a dupe. This ensures searchers are pointed to a great answer.

Otherwise, leave it alone. Who cares if it's kind of sort of a dupe? Besides, you should stop worrying and love dupes.

| improve this answer | | | | |

The assistance should always be for people trying to get answers. I recommend making suggestions a bit more apparent when there is a high match in the algorithm. The users can happily ignore them (as they can currently), but by appealing to their sense that there may already be an answer, you can increase compliance voluntarily and reduce the amount of dupes while not punishing diligent users.

Here is the current dialog:

Current New Question Dialog

We get the 'similar questions' on the right, but they are always there, always changing, and a user will likely notice them or not, with no difference between whether the similar question is likely a dupe, or just possibly. Perhaps we could add separate boxes that appear when the algorithm detect a particularly strong possibility of duplicate to give the user a more visual hint that they may find their answer elsewhere. Something like this:

New and Improved Question Dialog

My CSS skills were lacking, but ideally these dialogs would be more noticeable (like the red notices that pop up when you vote up comments too fast, etc.). If we allowed mouseovers to pop up the question and the highest voted and/or accepted answers, the user could decide if it's a dupe without even changing pages.

| improve this answer | | | | |

Currently, the SO logic for handling duplicates is this:

  1. OP prepares question
  2. Automatically give OP suggestions, based on question context
  3. OP ignores everything, posts eye-popping exact duplicate
  4. Viewer flags the question as possible duplicate
  5. Five users are required mark the question as duplicate

I disagree with the popup idea, but I think the logic could make better use of the calculated score for duplicates.

What if the logic was the following:

  1. OP prepares question
  2. Automatically give OP suggestions, based on question context
  3. If the question is ranked as a highly possible duplicate and the OP has practically no rep, don't open it immediately and add it to a queue for one or two 3k+ rep users to check, also including the Possible Duplicates.
  4. After the question is checked and confirmed not to be a duplicate (which will be highly impossible), it is normally opened.

This will have the following advantages:

  1. Take weight off the "Close Votes" queue (currently at 85.9k, raising about 2k per month)
  2. Prevent duplicates without offending the user at all (I understand that they shouldn't be offended, but sometimes they are...)
  3. Much easier environment to maintain, since most unhealthy questions will close before they even open
  4. Absolutely no rep inflation on questions that are exact duplicates.
  5. OP may be forced to try to research and format next questions accordingly

Of course, we all understand the disadvantage of expected user-interaction to open a question, but it should only happen in occasions that there is high chance that this question is a duplicate.

Moreover, the current user-interaction and review speed actually scares me, as I have seen answers being flagged and marked as duplicates in less than 2-3 minutes.

| improve this answer | | | | |
  • The first post (and late answer) queue doesn't hide those posts until they are reviewed. The post are posted normally, and simply added to the review queue as well. Other than being in that queue, they are no different than any other post. – Servy Oct 22 '13 at 21:14
  • @Servy, thanks for the edit. I just fixed a couple of spelling errors. As to first posts and late answers, yes, I know that they are already open, but I thought I didn't need to point it more, since it is now (thanks to you) boldly saying "don't open it immediately". The similarity is about being reviewed. – mavrosxristoforos Oct 22 '13 at 21:17
  • but it says, "similar to the first post queue". That's not similar to the first post queue at all. Just remove that entirely. – Servy Oct 22 '13 at 21:19
  • Alright, did that now. – mavrosxristoforos Oct 22 '13 at 21:20
  • This is a great way to piss off the best types of new users. They checked on google, checked on the SE site, couldn't find an exact post (although there are similar ones that they refer to in their question), and asked their own question as a result. And how do we reward their research? By assuming that they're asking a duplicate and tossing it in a queue before it sees the light of day! We need to trust new users to be better at reviewing whether their question is a dupe than an algorithm is. If the chance is absurdly high, at most we should create a message that lets them know. – jmac Oct 23 '13 at 1:34
  • It could start off with ridiculously common duplicates, like php mail not working. I don't know how the current algorithm checks for duplicates, but I'm sure that it does have some kind of scoring. I'm also sure that you can think of many ways that this could work out, and stop the actual duplicates, and not a good question. Even if it does stop a good question once in a while, reviewers will get it up before we know it. – mavrosxristoforos Oct 23 '13 at 3:04

You must log in to answer this question.

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