Consider the following chain of questions, presented in chronological order:

The second was marked a duplicate of the first, and the third a duplicate of the second. Now the fourth is being voted closed as a duplicate of the third.

I agree the fourth is a duplicate of the third, as both ask about gravitational redshift. The first, however, asks about cosmological redshift. These are quite different, and indeed the asker for the latest questions points out that the cries of "Noether doesn't apply" in response to the earliest question are just wrong in this newest context.

The link is the second question, and to be honest I don't know whether it is about gravitational or cosmological redshift.

This seems to be a general flaw in the system, but I'm not sure what the fix is. Two options come to mind:

  1. Just ignore this condition. If we close A as a duplicate of B and B as a duplicate of C, then even if A is not really a duplicate of C, they'll be related enough that the person can piece together an answer on their own. Besides, B usually accumulated B-specific answers before being closed, and these probably answer A.
  2. Restrict ourselves to only marking exact duplicates (in concept if not in words). (I feel the word "exact" used to be here in this context, but I can no longer find it.) I have anecdotally noticed over the years that as we experienced users see more and more questions, we are more likely to see similarities to other questions, and are perhaps a bit quick with declaring questions to be duplicates.

What do people think? Is there something wrong here? What can we do to fix it? See also Liberal definition of duplicate questions and the health of physics stack exchange.

  • 5
    $\begingroup$ I think people are too quick in declaring a question to be a duplicate. I also think they should look at the answers on the alleged duplicate. If those answers are incorrect, they shouldn't close the new question. Instead they should give a correct answer. What I dislike is seeing somebody vote to close a question when they've answered the alleged duplicate incorrectly. That doesn't do the health of physics stack exchange any good at all. I recommend you try to answer the latest question to appreciate this correct/incorrect issue. $\endgroup$ Commented Jan 14, 2016 at 22:50
  • $\begingroup$ Isn't the first option what we're presently doing? $\endgroup$
    – Kyle Kanos
    Commented Jan 15, 2016 at 2:12
  • $\begingroup$ @JohnDuffield that sounds like it should be an answer $\endgroup$
    – David Z
    Commented Jan 15, 2016 at 18:01
  • 1
    $\begingroup$ @David Z : maybe, but I note the new question is now closed and my answer downvoted, so I don't think I'll bother. $\endgroup$ Commented Jan 18, 2016 at 13:36

1 Answer 1


The idea behind closing one question (A) as a duplicate of another (B) is that any valid answer to A would also be a valid answer, or part of a valid answer, to B. I don't mean that you should be able to copy an answer from A and paste it into the answer box on B, but more that the core concepts, or "answering ideas", of the answer-post to A would also be "answering ideas" in an answer-post to B. (I hope my invented terms of "answering idea" and "answer-post" are clear enough.)

Anyway, in this situation, you might as well post the answer to B instead of A. It helps everyone when the knowledge is centralized in one place as much as possible. (Of course, this assumes that B is still narrow enough that it can reasonably be addressed in a single answer.) If there is one question that we can treat as canonical on a certain subject, one which many other questions can be marked as duplicates of, that often helps.

Admittedly, as I post this I haven't yet read the specific questions being referenced, but in general I would say this: the chain of duplicates should be broken at the point where a marked duplicate question (i.e. a question that has been closed as a duplicate) is asking something that the target of its duplicate link is not, and thus the criterion I listed in bold above doesn't apply. If there is no such point in the chain, then there's no problem.

That's the theory, anyway. In practice, sometimes a question (A) is a duplicate of a few other questions (B,C,...), in the sense that each concept contained in a valid answer-post to A is also an "answering idea" to one of the other questions, B or C or so on, but perhaps no single one of B,C,... entirely incorporates A. In this case, we can't always put all of B,C,... in the duplicate link list - in other words, it's not always possible to close as a duplicate of multiple questions. It requires at least as many (non-binding) close votes as there are target duplicates (B,C...). So sometimes extra duplicates will be indicated in comments or something, and we should take that into account.

Also, questions are a bit of a special case, and beyond the scope of this question.


You must log in to answer this question.