3

That may not the best title, but it describes the situation.

Over the past few days two questions have been posted:

“Very not fair” vs “not very fair” which is correct

and

What is the difference between “you're so not cool” & “you're not so cool”?

These questions appear to be very similar, and in fact it's been suggested that the second be Closed as a duplicate of the first.

My question is:

How is the OP of the second question supposed to be able to find the first question?

I realise that semantics play a big part in many of the questions asked and searching for similar questions is not straightforward.

The feature when asking a question

Questions that may already have your answer

is useful and has helped me on occasion, but could it be more useful?

By only looking at the titles of the two questions, algorithmically it is not obvious how to link them together since the keywords are mutually exclusive.

It seems to me that if either the search for "already asked questions" could be improved or if there was a better way to ask a question (to help the search engine) it might lead to better "already asked questions" showing up and possibly lessen duplicate questions being posted.

Any thoughts on how to go about this, or is it too hard a problem due to the nature of natural language?

BTW, there were no "already asked questions" which seemed to apply to this one, apologies if this has been hashed out already.

  • On another note, all this fluff about tags, question titles etc. is only to make searching easier. Right now, sadly, ELL is a tag salad, and although question titles have improved there are still too many that basically say "Is this correct?" "What this means?" and so on. I do not see enough meta interest in doing some huge editing, and that's what stops me from wrecking havoc. :) – M.A.R. Apr 18 '16 at 17:56
6

The scope of ELL is considerable: the vocabulary is huge, most of the common words have a plethora of (sometimes subtly different) meanings, and the ways of expressing a particular idea are manifold. Apart from very simple questions, finding a duplicate is like looking for the proverbial needle in a haystack.

The onus should not primarily be on the OP to find duplicate questions, and it's not likely in the foreseeable future that The System would be able to spot duplicates. If it reaches that stage, The System could probably answer the questions!

The responsibility therefore rests with the answerers to use their greater experience of the language, their knowledge of the system and their memory to find the needle in the haystack.

Marking a question as a possible duplicate should not be regarded as a stigma for the OP, and we should be quick to do so in order to avoid wasted effort on the part of other answerers.

Once the possible duplicate is flagged, there is no particular urgency. Other askers and answerers can look at the posited duplicate and decide whether it a) answers the same question b) does the job well. If yes to both, they can move toward ratifying the closure. If in doubt, they can leave the question open and see if a better answer appears.

Whether the question is closed or not, the link still remains in place, and searchers who find one answer but are not satisfied with it can easily move on to a related answer.

One thing that would increase the effectiveness of the system would be to show reverse links: if a question is posted as a possible duplicate, and either has or gets answers, the original question could link back to the 'possible duplicate with answers' question.

  • A reverse link or something like it is a great idea!. Taking it a step further, "grouping" the questions and making them somehow "related" may be more to the point. "Tags" don't really do this as they are too broad but a step in that direction. Related questions could be implemented so that if there are two questions, marked as related, they would be "grouped", then any other question marked as related to either previous question would join that "group" then anyone would be able to see all the related questions in one place. (cont) – Peter Apr 15 '16 at 18:56
  • A question could also belong to more than one "group", but it would still be up to the answerers to flag the relationship. – Peter Apr 15 '16 at 18:57
  • 3
    @Peter there is already a "reverse link" in the sidebar. If you want to link questions together, just post a comment with the link, and it will show up in the sidebar. As soon as a question has a proposed duplicate, a comment is posted with a link to it. If the comment doesn't get deleted, they will be linked regardless of whether one gets closed. Example: ell.stackexchange.com/questions/58870/… – ColleenV Apr 15 '16 at 20:31
6

There's a point I'd like to add that hasn't been raised in the other (excellent) answers to this question: duplicate questions are actually a good thing, provided the duplicate question meets our usual quality standards.

I can't find the post, but somewhere out in the metas there's an answer from our SE overlords (I think it was Jeff Atwood, but perhaps I misremember) saying that duplicates are very much our friend, and closing questions as duplicate is an awesome thing; both for the network and the asker. If the original question had an answer, the asker gets help right away! That's pretty great for them. But better yet, it's now easier for future visitors to find the answer to their question, because maybe they would have phrased it like the person who wrote the duplicate question, not the original. The more good duplicates you have of a question, the more varied search results trying to find the same answer are going to point to ELL... And get them to the right answer by clicking one more link.

So duplicates can be good, if they're written well, and you shouldn't be upset to find your question closed as a duplicate! It's a good thing, and you get an answer even more quickly than you thought :)

  • 2
    I'm pretty sure the Jeff Atwood blog post you're talking about is the same one linked in my answer. Dr. Strangedupe: Or, How I Learned to Stop Worrying And Love Duplication – ColleenV Apr 16 '16 at 20:49
  • 1
    I agree with this so much that after due consideration I've actually downvoted the question - because it strongly implies that duplicates are a bad thing, and that we should be encouraging new users to make more effort to find any earlier dup before posting. What would be useful though is to make sure every question title reasonably reflects the issue being queried, since that would make it easier for more experienced users to find those dups, if they exist. – FumbleFingers Apr 20 '16 at 18:43
  • 1
    @FumbleFingers Absolutely! There are two (somewhat-competing) parts to making duplicates the most useful: 1) preserving the word usage of the querent, and 2) making the title clear and understandable, to promote dupe-finding/closing, and also to increase the frequency which the dupe appears in search results. Both have the same goal, and on most sites I'd say #1 most often takes precedence. On ELL, #2 is more likely to be an issue. I'd say the best practice is to preserve searchable keywords in the OP, while editing to ensure the title is grammatical/understandable. – WendiKidd Apr 28 '16 at 5:04
  • Exactly how to strike a balance between the two is an individual judgement call. But yes, please do edit! Editing is our friend (especially in regard to titles). :) – WendiKidd Apr 28 '16 at 5:04
3

I disagree with your premise that a poster should avoid posting duplicate questions, and that duplicate questions that are difficult to determine are duplicates are something we should try to reduce. I think we need to increase our efforts to find the same or similar questions and link them together as duplicates, or put a "related" link in the comments so that similar questions show up on the sidebar.

Finding a duplicate question that isn't a nearly exact duplicate is not something that is reasonable to ask a learner to do. I don't even think it's reasonable to ask a learner to tag their question sometimes, because the entire reason they're asking the question is that they don't know. Now, there will always be those folks that don't search before asking, but even if we made their duplicate easier to find, we wouldn't prevent them from posting their question.

We had a discussion about duplicates under "a" vs "the" to introduce a company, and I linked Dr. Strangedupe: Or, How I Learned to Stop Worrying And Love Duplication (which is linked from the help topic on duplicates) to try to explain that some duplicates are good for the site, and that closing as a duplicate doesn't mean the question is "bad" or shouldn't have been asked.

I think that some folks are reluctant to close questions as a duplicate, because the way things are phrased makes folks think the one that didn't get closed is somehow more definitive or better. I don't claim to understand all of the thresholds and heuristics, but if the question and or answers are up-voted in the one that is closed as a duplicate, all closing does is prevent new answers to it - new answers would be directed to the question that is still open. An example is Certain kinds of fish or certain kinds of fishes? which has been around since it was closed in 2013.

Tangentially related discussions for a deep dive into the duplicate question topic area:


Duplicate can be considered as an insult. I think we need to use a better title
Should we close questions that are exact duplicates of open ELU questions?
Correct motivation for learner to ask a (possible) duplicate question?
Marking older questions as duplicates

  • Thanks for the additional links! I do think that "duplicate question" is not the greatest phrasing, maybe "related question"? The discussion points out the finer definitions of "duplicate" but I don't think we can expect an ELL'er to reach into all of the discussions. I'm not saying the OP shouldn't post their "duplicate" question, just that if they post, then it's Closed as Duplicate, it may not be the best experience for the OP, also if previous questions similar to the OP's are revealed to them, it may save some heartache. (cont) – Peter Apr 15 '16 at 19:05
  • Also, I wouldn't think the OP would be able to find the duplicate, but if they can be guided in a better way, it may help them as well. I've always found that the more a new topic is discussed, the easier it is to formulate an understanding of it. A "grouping" (see comment for (at)JavaLatte) might help with this. As things stand, the "duplicate is only a one-to-one relationship. – Peter Apr 15 '16 at 19:09
  • @Peter I think that you're ignoring the other tools that SE has to group questions together. There's the sidebar linked/related questions and tags. It's not feasible IMO to have one canonical question that you're only one click away from no matter how you asked your question, especially when we have a language barrier. If we were better about tags, our questions would be naturally grouped without having to mark them as duplicate. Also, if you want learners to get better suggestions of related questions before they post, encourage them to write better titles. – ColleenV Apr 15 '16 at 20:43

You must log in to answer this question.

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