20

Apologies if this counts as "bumping" my question as previously asked on Language. It's only just dawned on me that quite likely the SO internal search engine is the same across all sites, so others may be interested here.

Here's the link. To summarise - similar questions come up repeatedly. Often they get closed, with a link to the earlier instance of that question. But it's often hard to locate that earlier instance (because exact wording may vary quite a bit, among other reasons). So I asked for an enhancement along the lines of...

Increment a count against existing question when a new one is closed as "exact duplicate" thereof, and make that count available as a "sort results" key in searches.

Apart from making them easier to find once the links build up, Questions which start to accumulate high "dupcount" values would attract more views, and would probably get progressively tweaked to maximise Answer quality on what would in effect be the "showcase" answers for questions which in my opinion ought to be considered more important, since people keep asking them.

Improve this question
15
  • 1
    I'm not sure this would work exactly as presented, but +1 anyways for an interesting concept!
    – Pops
    Aug 25, 2011 at 21:54
  • @Popular Demand: Well don't just leave me hanging, please! I obviously think it would, or I wouldn't suggest it. Why do you think it wouldn't?
    – FumbleFingers
    Aug 25, 2011 at 21:57
  • I actually asked a very similar question recently, but it didn't attract much interest. Perhaps these proposals could be merged?
    – Chris Frederick
    Aug 25, 2011 at 21:57
  • I'm having a hard time putting it into words. I guess it's that the duplicate system doesn't quite work ideally as it is, so using it as a metric just doesn't feel like a good idea to me? Would people eventually unfairly pay less attention to low-dupe-count posts? I'm still trying to think of a way to build off of this idea.
    – Pops
    Aug 25, 2011 at 22:01
  • Related: meta.stackexchange.com/questions/77103/…
    – Pekka
    Aug 25, 2011 at 22:34
  • Also related: meta.stackexchange.com/questions/90496/… . I think this is a commonly repeated question!
    – Andrew Grimm
    Aug 25, 2011 at 23:09
  • @Popular Demand: On Language, people pay less attention to old questions. Most activity is on recent minutes/hours, maybe the last day or two. Stuff that's weeks/months old just sits there. A lot of people do identify an original of a duplicate, but some get missed - or at least hang around for quite a while and people waste effort rekeying similar answers. Sometimes better, sometimes not. People will "pay attention" to what they want. So what? Equality is for people, not questions. But I'm mainly asking for a tool for me, a gain for the site is a secondary (though worthwhile) issue.
    – FumbleFingers
    Aug 25, 2011 at 23:50
  • @Eat more Twisters: No idea of the etiquette here, but I've just copied my comment against Chris's Question on yours as well, linking back to mine here. I think maybe the overall context may vary across different SO sites, and different users will want different things anyway. I think my idea is simpler though - plus yours didn't get actionned, obviously, so maybe mine mine fare better because it won't change the UI much (I know nothing's trivial to implement, but I can't believe it's that difficult).
    – FumbleFingers
    Aug 26, 2011 at 0:14
  • @Andrew Grimm: All the above. I don't understand GraceNote's point about "links", but whatever they are, they don't do it for me. I'm only interested in incrementing a "beauty contest score" each time an old question survives when a new one is closed as a duplicate because of it.
    – FumbleFingers
    Aug 26, 2011 at 0:31
  • I can't think of an issue/bug tracker that doesn't do this, seems like a good idea. Also, duplicates can sometimes have useful info so it would be helpful for the user to be able to see them.
    – studgeek
    Aug 26, 2011 at 0:46
  • @studgeek: I'm not asking for an all-singing all-dancing set of cross-links. Hopefully SO is designed to be more scaleable than any bug tracker I ever heard of, so one must be realistic. Just a count would make a huge difference in many ways. More functionality would always be nice to have in reserve, but I don't want/need it now.
    – FumbleFingers
    Aug 26, 2011 at 1:00
  • I totally support your suggestion @FumbleFingers. I actually think it would be great if they did a combination of your suggestion and mine (but would be happy to see any one of them implemented)
    – Pekka
    Aug 26, 2011 at 6:41
  • @Eat more Twisters: I certainly wouldn't object if the extra features you seek were to be provided as well, though to me it looks like 90% of the effort for only 10% of the functional value. Certainly on Language, vote counts aren't very significant at the moment. Most questions have quite an ephemeral existence, in that they're virtually ignored after a day or two. Thus many answers tend to remain "unpolished" because they're not significantly scrutinised and re-evaluated over time. People just ask again, often getting another "half-answer" before being closed in favour of the original.
    – FumbleFingers
    Aug 26, 2011 at 14:26
  • @FumbleFingers To do what you are suggesting still requires one of the people marking as duplicate to indicate what question is being duplicated. Since SO will have the information, I don't see the downside of indicating that on the duplicate/new question (its certainly not a technical hurdle). And it will let users tracking the duplicate/new question to the right answers (on the duplicated/old question). It also will also give them the context to unvote the duplicate if appropriate. I agree only a count on the duplicated/old question would be fine.
    – studgeek
    Dec 2, 2011 at 18:33
  • @studgeek: I don't see what you're getting at. My issues are firstly that it's hard for the first person to find a specific earlier duplicate, and secondly that the whole system offers no way to easily trawl through hoary old chestnuts to make sure they really are up to snuff. Any "new" duplicate automatically gets a comment linking to the earlier one as soon as someone flags it, so I don't see any problem there.
    – FumbleFingers
    Dec 2, 2011 at 19:03

2 Answers 2

Reset to default
4

For info, we do now make this data available when selecting a duplicate (i.e. the "close as duplicate of" popup); see I want to know how duplicious a question is, and if it is the best example of such

It is not currently available separately (i.e. for use as a sort); I'm also a little dubious how much real use that sort would see. It might be possible to get this from SEDE, though.

Improve this answer
4
  • I don't really see the relevance of your first sentence. So far as I know, "close as a duplicate" just shows you any existing questions already chosen by others who voted to close before you. Useless if you are the first voting to close, and potentially unhelpful if not. It encourages you to choose the same earlier question if you don't have anything to help you find the one you really want, which is exactly what triggered me to ask for this feature in the first place.
    – FumbleFingers
    Sep 1, 2011 at 12:27
  • ... also, you're dubious how much real use it would be to sort on the proposed new field. The fact is on ELU we do get quite a few questions coming up again and again. Often they're just slight variations on a theme. IMHO having this count would encourage us to find and polish up a single comprehensive version of the best Answer(s) in one place. Without it, the overall quality of our "answer database" gradually goes down over time, rather than improving.
    – FumbleFingers
    Sep 1, 2011 at 12:32
  • 2
    @Fumble this is still an area we are evolving; there is some merit in the earlier proposal, so that is implemented now as a stop-gap while we find ways of getting the desired step-change in accuracy. Like you, we also want it to be simple and awesome - but it isn't quite as easy as just listing all answers by "dupliciousnessnessness"
    – Marc Gravell
    Sep 1, 2011 at 12:43
  • Okay, I'm satisfied with your answer now. I'm not on the design team, obviously, so it's not my call how a solution to this problem actually gets implemented. I'm happy just to know it's an area which is being addressed - your original answer implies the existing system already does part of what I wanted, which I couldn't in all honesty agree with, but I can go with your last comment.
    – FumbleFingers
    Sep 1, 2011 at 15:12
4

I totally agree that it would be helpful to leverage the two-way relationship between a question and its duplicates (the duplicator and duplicatee?). Technically speaking, the original question has a one-to-many relationship with its duplicates. SO prominently displays one side of this relationship—from each duplicate question to the original—but I don't see why it can't display the other—from the original to its duplicates.

If nothing else, this "exact duplicate" count that you propose could be used for queries with the StackExchange Data Explorer. Alternatively, a list of duplicates could be displayed underneath the original question. Anything would be better than keeping a list by hand.

Improve this answer
1
  • I've not seen StackExchange Data Explorer before. Looks interesting, thanks. There are probably loads of things that could be done - but a bi-directional link of "duplicator" and "duplicatee" sounds like hard work to me, so I wouldn't hold my breath. Ask for a "little" thing, and we might just get it!
    – FumbleFingers
    Aug 26, 2011 at 0:37

You must log in to answer this question.

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