5

Earlier today, I was pointed towards a bunch of question that were duplicated one of the other. But while reviewing the stuff linked to the question I noticed that most of those questions don't have the enough elements to qualify as exact duplicate of the question they were linked to. This is rather preoccupating since duplicates are the hardest to review/evaluate:

If it's a duplicate we need to read both the questions - the one we are closing and the one we are targeting - to ensure that we select a proper duplicate target.

Most of our queue is (was?) composed of duplicates, it makes harder for reviewers to evaluate the questions as duplicates as the question they are reviewing is too bad/unclear to make the reviewer feel sure what is being done, and then either: do not review or robo-review and just agree with the previous.

How to solve? If there is a possibility that the questions is already asked or you found something similar, but the question is unclear, flag/vote as unclear, and leave a comment of a possibly related question. This will help reviewers greatly and in the meantime can help OP to say "Yes" that was what I was asking or "No" and clarify the question.

Remember, the duplicates should have the same or more quality that the question they are being linked to.

7
  • 3
    I agree we should use "unclear" more often rather than point at sort of duplicates.
    – don.joey
    Jan 2, 2014 at 21:10
  • do you mean if the question is un-nessisary close the dupe and point (if needed) to the better original q/a page? Jan 9, 2014 at 0:47
  • 2
    I'm saying that if you could not write an answer in the current state of the question you shouldn't flag it as duplicated either.
    – Braiam
    Jan 9, 2014 at 1:01
  • 1
    -1 why not just say that then? the way you have your question worded is really noisy and, ironicly, unclear
    – virtualxtc
    Feb 17, 2014 at 11:40
  • @virtualxtc "I noticed that most of those questions don't have the enough elements to qualify as exact duplicate of the question they were linked to" where is unclear? Why you don't edit it to change the wording if it was unclear for you?
    – Braiam
    Feb 17, 2014 at 14:28
  • @Braiam the edit feature is grayed out.
    – virtualxtc
    Feb 17, 2014 at 14:32
  • The first two paragraphs are extremly hard to understand. In the 3rd paragraph - who is "our"? what "queue"? There's also a lot of grammar issues though out. I'd recommend deleting everything and replacing it with: "Don't flag questions as duplicate if you could not write an answer to the question in its current state!"
    – virtualxtc
    Feb 17, 2014 at 14:37

1 Answer 1

5

The Real Solution

The solution to bad posts that have no value is for them to be deleted after a sufficient time has passed after they're closed. For posts that clearly have no value at all, that can be a very short time.

Duplicates are actually deleted automatically in accordance with the same rules that other posts are deleted automatically: if they're unanswered without positive score, the Commmunity bot will delete them after a while. (Of course, we can and should be willing to delete posts manually when we know they aren't doing anybody any good.) So if we have long term duplicates that you consider low quality, that should be discussed in detail.

It's not good for the site to have valueless posts of any kind. Closing questions with a close reason other than duplicate doesn't address the issue of the site being littered with low quality closed questions. If there is any real issue here, it seems to be it must be that.

This shouldn't be a problem for reviewers unless reviewers are reviewing improperly. (And if so, that is the real problem.)

When a post is in the close queue, the job of the reviewer is not to decide on the narrow question of whether or not previous close reasons were correct. Instead, the reviewer's task is to look at the post and determine--as independently as possible from other voters--whether or not it meets any of the criteria for closure. (Of course there is also, in borderline cases, the task of weighing all the reasons one can think of for closing a post against all those for keeping it open. But usually, we just look and see if the post belongs open on the site.)

If reviewers have to slow down to think more about posts, good. Anything that achieves that is a good thing.

If a post has close votes (or flags) for closure as a duplicate, the reviewer should consider the master question, other questions they're aware of (sometimes searching is necessary), and also all the other close reasons.

If a post clearly fits another close reason, reviewers can vote to close it for that other reason. It's neither necessary nor beneficial for reviewers to deliberately make any effort to vote in accordance with the way others have already voted.

Consider that it is a deliberate design decision to only show reviewers if and how many close votes a post already has, after they have decided they think they want to close it and clicked Close.

Reviewers who prefer to select other close reasons are particularly unburdened by this issue.

Of course, in the case of duplicates, it's often necessary to examine the proposed master question, and for posts that are not properly commented ("possible duplicate ..."), that may involve clicking Close to view them. But if you know a post qualifies for closure based on another close reason, considering whether or not it's really a duplicate as proposed is entirely optional.

I don't agree that it's always better to pick close reasons besides duplicate whenever possible. I'll avoid discussing that particular issue here since I know you're making a new meta question to discuss that. But if that is true, it's especially clear that the reviewer confusion issue is not a real problem: If you strongly prefer to close for reasons other than duplicate, then you'd only be considering closing as a duplicate if you've already decided no other close reason applies. But the posts you're talking about are all posts where some other close reason does apply...

7
  • "the job of the reviewer is not to decide on the narrow question of whether or not previous close reasons were correct" and here we go to the part where reviewers just agree with the previous reasons. You cannot trust the reviewers to do their job correctly.
    – Braiam
    Jan 12, 2014 at 15:11
  • 1
    Also, I proposed a real solution. Read "How to solve?" paragraph
    – Braiam
    Jan 12, 2014 at 15:21
  • @Braiam Whether or not we can trust reviewers to do their job correctly, we can at least avoid making it especially easy for reviewers to do their job incorrectly. In your post you object to how it isn't easy for reviewers to "evaluate the questions as duplicates...". But as I've demonstrated, that seems only to be a significant issue if they're doing their job wrong. We shouldn't try to make it easy for people to review incorrectly. Furthermore, to the extent that easy means robo reviewing and making reviewing easy at all is not a good goal (as then hard means paying attention). Jan 13, 2014 at 8:38
  • Ok, lets put it this way. I prefer robo-reviewers closing as Unclear, rather than pointing OP to a post that will not help and/or possibly could be counterproductive/destructive in OP's case. Duplicate is super special in SE. It means: hey your question is good, on topic and understandable but has already been asked here [link]. If it's not an exact duplicate for any reason as lack of information or having another more fitting close reason.
    – Braiam
    Jan 13, 2014 at 12:27
  • Duplicates are questions that in normal circumstances you can answer, the answer is within the scope of the site and could be answered with a couple of paragraph. If it doesn't fit those conditions then it's not a duplicated. (I had to reopen questions that I answered that were closed to a sort of I-catch-all duplicate when obviously none of the answers apply to OP)
    – Braiam
    Jan 13, 2014 at 12:28
  • @Braiam I don't see anything like "hey your question is good" on that page; besides the opinion of one person I respect (you), I've never seen any respectable source even suggest that duplicates have to be good-quality posts. Furthermore, you are simply wrong to cite that blog post as a source for the claim that duplicates must be exact: please read "2. Accidental duplicates" on that page! Duplicates do not mean and have never meant what you're saying. Jan 13, 2014 at 14:52
  • @Braiam From our chat, my understanding is that you planned to post a new meta question to discuss and get others' opinions on whether or not "duplicate" is the close reason of last resort that you claim it to be. That's why I've deliberately avoided covering that in my post here. For example, the main point I made in chat is not covered in my meta-answer here. Do you still plan to post that meta question? If not, I'll considering expanding this post. Jan 13, 2014 at 15:08

You must log in to answer this question.

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