2

The close reason is called "exact duplicate" and the banner says:

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.

But it seems like often things get marked as duplicates even if it's not literally the same question.

What's going on here? How close do things actually have to be to be duplicates?

(Note: this is intended to be a more clear description of the policy that's evolved on the site over the years than given in the answers to this older question.

1

We generally close B as a duplicate of A if conceptually, everything that B asks is also asked by A, i.e. (good) answers to A are guaranteed to completely and clearly answer B.

That "completely and clearly" part is really important. It's not enough for questions to just have the same bottom-line answer ("it's safe!"). It's the whole of the potential answers: the possibilities, the reasoning, everything. If there's something big you need to do in order to answer B, and A doesn't ask for it, they're not duplicates.

That is, when we say "this question has been asked before..." we mean the concept of the question, not the literal question. If they're conceptually asking the same thing, they're duplicates, even if they don't look the same on the surface.

Some non-exact cases that are generally still duplicates:

  • B is a sub-question of A, i.e. A answers more than just B, but still comprehensively covers B.
  • B is the same as A, but with added or changed details that don't make any difference to the answer.

Some cases where questions may not be duplicates:

  • A doesn't cover everything that B covers, i.e. only some of ways to answer to B would be covered by A.
  • A is so general that it doesn't clearly answer B. If the OP would have to really dig through to understand, or there might be details missing, it's not a duplicate.
  • B has new details that make A not really apply. This isn't carte blanche to ask endless variations, of course (we'd try to generalize if that happened), but not all situations can be covered by a single question.
  • A does answer B, but substantial explanation is needed to demonstrate why it actually applies.

Note: I'm happy to discuss and tweak this, add examples, and so on. However, my goal is to broadly document our current general approach, not to make any new rules. Anything contentious is probably best left for case-by-case considerations.

  • I'm not very happy with this answer, because I see it as describing a neccessary condition only, but not a sufficient one. I'm sure we've had situations where two very different questions happened to be solvable with a similar approach, and people started suggesting that they should be closed as duplicates. In my eyes, they are not duplicates at all, even though the answer is "the same". I will need to find a few examples before posting this as an answer though. – rumtscho Feb 7 '16 at 18:36
  • @rumtscho Is that similar to the issue of "close as dupe based on question" and "close as dupe based on answer". I think the former is good, the latter is bad. – Catija Feb 7 '16 at 18:48
  • @rumtscho Again, "completely and clearly". If the questions are very different, even if there exists a common solution, completely addressing them involves different answers. In my experience, this is what we end up doing - we have closed an awful lot of duplicates where we knew based on the question that the answers would all be the same, even if the questions weren't identical. – Cascabel Feb 7 '16 at 22:01
  • @Catija I think there's a subtlety there that people get caught up on sometimes. I agree that we don't close as dupe based on existing answers. But closing based on the question is often basically the same as closing based on what it takes to answer it, i.e. the set of potential good answers. If you know that every possible way to answer one question also answers the other, and vice versa, then they're probably duplicates - and although you're thinking about answers, it's still based on the question. – Cascabel Feb 7 '16 at 22:05
  • @rumtscho I edited to try to emphasize that if the questions are truly different, even if there's a common solution, they're not duplicates. – Cascabel Feb 7 '16 at 22:19
  • Can y'all talk some sense into the SF&F folks? Because they close totally unrelated questions as duplicates all the time because one of the answers to the old question happened to mention something in passing that sorta-kinda answers the new question, if you squint your eyes and tilt your head. – Marti Feb 10 '16 at 16:39

You must log in to answer this question.

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