4
$\begingroup$

Today I found something strange, a question that is marked as a duplicate of another question that is closed as off-topic (while IMHO it was not off-topic, but that is a meta off-topic comment in here).

Q1: Effect Size/Mean Squared Error from Linear Mixed-Model in R

Q2: Partial $\eta^2$ for repeated measures ANOVA (with R's car package)

This kind of "duplicates" make no sense: if the question is a duplicate of an off-topic question, then it is also off-topic so should be closed on those grounds. On the other hand, marking questions as duplicates of unanswered questions also makes no sense, as it does not provide any help to those who asked.

This is of course closely linked to another discussion on meta: What should be our policy on duplicate answers?

What do you think?

IMHO, it makes sense to mark questions as duplicates if there exists (1) a real duplicate question, (2) with a high quality answer because if there is a duplicate with one-sentence answer, then it is no help to redirect to it (generally, not always).

$\endgroup$
  • 1
    $\begingroup$ In this case the linked duplicate was cross-posted on SO - the cross-posting is linked from CV in the comments, and someone seeking this information can find a good answer on SO. So "it does not provide any help to those who asked" isn't quite true, though it's a somewhat roundabout route! $\endgroup$ – Silverfish Dec 17 '14 at 13:22
  • 1
    $\begingroup$ But there is also a general issue here: marking as a duplicate of a question with low-quality answers. It happens. $\endgroup$ – Tim Dec 17 '14 at 13:25
  • $\begingroup$ My remark was on this specific case rather than on the general issue - hence a "comment" not an "answer". This is certainly a topic worthy of discussion, though the particular example in the title/body of this Q introduces some side-issues e.g. cross-posting. $\endgroup$ – Silverfish Dec 17 '14 at 13:31
  • 2
    $\begingroup$ Tim, yes it is problematic to refer someone to a duplicate question whose answers are of low quality. (I hate to do this.) But the problem is not with the duplication: it lies with the poor answers! The solution is to improve the answers. I confess that in such cases, since I usually have neither the time nor the ability to improve the answers personally, I look for some way (however small) in which the two questions differ. I write a comment pointing that out and emphasizing that difference--and leave the new question open. $\endgroup$ – whuber Dec 17 '14 at 16:25
5
$\begingroup$

The implicit assumption behind the issues raised here seems to be that closing is a form of removing, hiding, or destroying something we do not want. I would like to suggest that closing, on the contrary, is an act that adds information and value to our site. To understand this, bear in mind that the threads associated with closed questions remain visible and searchable.

Pointing to an off-topic as a duplicate is rare, but it is not inconsistent. Although questions can officially be closed for only one reason, often several apply. Moreover, off-topic threads frequently collect comments (and, on occasion, answers). Linking to such threads is informative.

Closing duplicates of unanswered questions makes perfect sense. When the same question appears in two or more places, then for the purpose of deciding whether they are duplicates it does not matter whether neither has an answer. We want to link them so that all the answers (and, hopefully) comments appear together in a conveniently read, easily searched way.

The proposed solution--do not mark unanswered questions as duplicates--would only cause unanswerable questions to proliferate and degrade our site. As far as I can see, the only reasonable alternative is to delete duplicate questions. It is rare for moderators to do that, because duplicates are almost never phrased exactly the same way: they may contain different key words, for instance. In many cases it might not be evident to a neophyte that two questions are really duplicates, because recognizing that might require understanding the meanings of related technical words or of making a simple mathematical transformation. The act of linking a question to a duplicate embodies such an insight and itself creates meaning. This adds value to our site, whether or not either question currently has an answer.


There is a problem, however. It lies with the system's stock phrase, automatically applied when a duplicate is closed:

This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.

This obviously is not true when the duplicate does not have an answer--but there is no way anyone here on CV can modify this text to make it more accurate.

$\endgroup$
  • 2
    $\begingroup$ Just for disambiguation, non moderator users can't flag questions as duplicates of unanswered questions (source). This post is another source supporting your arguments. $\endgroup$ – Andre Silva Dec 17 '14 at 16:05
  • $\begingroup$ It is not only the text: if I am not mistaken, the system does not allow to mark a question a duplicate of another question, if that second question does not have any upvoted answers (NB: it does not work even if there is an answer, but it is not upvoted). $\endgroup$ – amoeba Dec 17 '14 at 16:06
  • 1
    $\begingroup$ Thank you, @Andre. Also related is the SE team's explanation at meta.stackexchange.com/questions/166707 (which IMHO misses the point, as explained by a highly-upvoted answer there). $\endgroup$ – whuber Dec 17 '14 at 16:22
  • $\begingroup$ With closing unanswered there is one problem: a recent question is closed and linked as duplicate to old question. So you can say it is linked to something that is already somewhere "deep" and forgotten. Closed question disappears with no answer. (I don't know if it was clear what I wanted to point out?) $\endgroup$ – Tim Dec 17 '14 at 16:25
  • $\begingroup$ Old questions are not forgotten, Tim, nor are they buried deeply. They are available through searches--and, in fact, the original poster of the new question should already have found the old question (everybody searches the site before asking a question, right?). Any action that modifies that thread will put it right at the top of the queue of active questions. Several badges (Revival and Necromancer) exist to encourage people to answer old questions. In effect, activating an old question can be a better way of getting answers than re-asking it! $\endgroup$ – whuber Dec 17 '14 at 16:30
  • 1
    $\begingroup$ @whuber Well, yes an no. If there is a question that hangs on the front page, I argue that there is a greater chance that someone will answer it, then if you have deliberately to search for the question to answer. Also consider that if someone asked the question long before then he may be not interested in the answer any more, so as answerer you could ask yourself if there is a point in answering this kind of question. $\endgroup$ – Tim Dec 18 '14 at 11:08
  • $\begingroup$ Yes, there is a point to answering any question: that's what this site is all about! We would become a much worse place for everybody if we kept duplicating old questions on the (speculative) basis that the old proposer might no longer be interested in the answers. $\endgroup$ – whuber Dec 18 '14 at 16:57
  • 1
    $\begingroup$ I agree that there is a point. I am rather taking about how this could end up in practice. I argue that the longer the question is unanswered, the lower is the probability that it will be answered soon (we could even do a survival analysis of unanswered questions ;) ). $\endgroup$ – Tim Dec 19 '14 at 9:21
  • 2
    $\begingroup$ Are you aware, Tim, of the "Community bot" that promotes upvoted questions without accepted answers to the top of the queue every 30 days? $\endgroup$ – whuber Dec 19 '14 at 16:35
  • $\begingroup$ @whuber, I almost never see community-promoted questions answered because of such promotion. It seems to me that people tend to actively ignore them (at least personally, I usually do). I wonder if the available data allows to look into that (e.g. what % of community-promoted questions get answered in the 24 hours after being promoted). $\endgroup$ – amoeba Dec 21 '14 at 22:43
  • 1
    $\begingroup$ @amoeba I beg to differ: I see many such questions upvoted, commented on, and answered. (I personally have answered quite a few older questions that came to my attention this way. I have also closed quite a few of them, too!) You might shift your point of view a little and view the Community questions as easy opportunities for a "Revival" or "Necromancer" badge :-). (BTW, I don't think we have access to records of when a community promotion occurred: those events would have to be inferred from a database query that attempts to reproduce the promotion algorithm itself.) $\endgroup$ – whuber Dec 21 '14 at 23:23
  • $\begingroup$ I try to look at Community questions whenever I can, and some of them do carry answers worthy of an upvote (cc/ @amoeba). $\endgroup$ – Andre Silva Jan 5 '15 at 23:20

You must log in to answer this question.

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