When should comments be used to determine a dupe story-id?

Question

The general policy for closing story-ids as dupes was determined to be that they're only a dupe if the OP accepts an answer.

Later, that answer was edited (based on a comment side discussion) to include allowing comments from the OP to count.

That commenting portion was not something I personally upvoted the original answer for. So, I have questions about when commenting should be accepted:

• When the OP comments on their own question, but there's no answer?
• When the OP comments on their own question, but not one of the answers?
• When the OP comments on an answer?

This came up recently because of a situation where a comment was left on a brand new question asking if it was a certain story, with the intent to close it as a dupe if they replied "Yes.", and no intent to have an answer posted first.

The alternative being someone posting an answer, seeing if the OP accepts, and failing that relying upon a comment (more as a last resort, because of drive-by askers). This is how I believed we'd been doing things all along, but I'm wondering if we should be doing comment first/only instead.

Answer 1

Confirmation by OP comment should be enough to close as duplicate

First, closing when the question has an accepted answer is not controversial: that’s the substance of the original answer.

Further, let’s note that closing based on comments is, in a sense, our current policy, as indicated in the linked answer. It’s true that it was edited in after most of the the voting was done (and probably should have been put as a separate question), but the answer has nonetheless retained its high score after that was edited in. Further, closure of questions based on OP comment acceptance has been going on at least since that edit (six months) with a minimum of controversy (i.e., no “reopen this question” posts on meta after a story-identification question was closed due to an OP confirming the answer to be correct in a comment).

Now, why do we wait for acceptance before closing story-identification questions? What makes them special (and what likely motivated the other answer in the first place)? It’s probably worth going to the second-most-upvoted answer on that page, which explains why we shouldn’t close without OP confirmation:

If we’re wrong, we’ve just annoyed the OP.

And further:

At best, we should try to guess what the story is – but only the OP can confirm. We shouldn’t take that control out of their hands.

We want to make sure that the OP has a chance to confirm that a story is actually the one they remember. We’re not just trying to get an answer out there, or we’d close without acceptance.

If the OP confirms in a comment that the story is really the one that they remember, that’s at least as good as an acceptance. There’s always the (admittedly quite small) risk that a new user does not quite understand what clicking that checkmark means. But if they leave a comment, we can be reasonably sure of what the correct answer is. It doesn’t matter where they leave a comment: on the question, on the answer, or anywhere else. That tells us what the answer is, and that’s always been enough to close as duplicate.

And one other thing can’t be emphasized enough: there are a lot of people who come back, see the right answer, say “Wow, this is right, thanks!” and never return (and never accept an answer). Being able to close as duplicate in this case only makes sense: these are users that want to accept, but just didn’t know how.

In a similar vein, a self-answer is also sufficiently definitive confirmation that the OP has really found the work they’re looking for. We’ve closed in a number of those cases too, again pretty non-controversially.

Whether to comment or answer should be a matter of preference

With all this in mind, I don’t think we should be enforcing a policy of only answering and never commenting, or only commenting and never answering. Either an accepted answer or an OP comment is enough to close. While some questions, particularly those that are asked over and over with pretty similar descriptions, aren’t going to benefit from numerous copies of what’s basically the same answer, there’s always the possibility that a new answer might be better written, more comprehensive, or useful to a different segment of searchers than the existing ones. Both situations are sufficiently common that we shouldn’t say that one is always preferable.

Answer 2

The spirit of the existing policy is to dupe-close ID questions if and only if the OPs have confirmed that both answers are the same.

Even if it's "obvious" to every reader that an answer is correct, we don't close the question without some form of feedback from the OP. Conversely, if the OP has accepted an answer, we close the question without checking whether the answer actually matches every detail described in the question. The policy is to judge by OP confirmation and nothing else. It's a very clear policy, and it's worked well so far.

Now does it really matter what specific form that OP confirmation takes? Whether it's the pressing of a green tick button or the submission of a comment to say "this is it, thank you so much" - both express the same idea, even if they look different on the screen and are registered differently by SE's software. In fact, if anything, a comment is more clear, since a checkmark could in theory just mean "this isn't the story I was thinking of, but it's the answer I found most useful".

Let's not apply policy mindlessly, but instead think about the reasons for the policy and why it was created. Let's also remember that the existence of answers isn't supposed to have much effect on whether or not a question is a dupe. Even our story-ID dupe policy is based on the questions being the same because they're looking for the same story, not on the answers being the same. If an OP confirms in a comment that yes, the story they read was indeed Asimov's The Last Question, then their question is a duplicate of this one because we know both are looking for the same story, whether or not the new question has an answer.

Answer 3

To continue this as a discussion (and not a comment chain):

One of our initial topics on the subject had a well-supported discussion that story-id should be an exception to dupe anyway:

Why would VTC as duplicate be appropriate for a story identification question?

This initial questions top post had little in the way of clearly stating "Yes, story-id questions with the same answer should be closed as duplicates." It's more of a pondering of how the system works. The next highest answer is for excluding duplicates.

The follow-up question then became about closing when there's no acceptance. The rest of that has already been discussed.

So, to me, the spirit of the policy has been to close as duplicate as a final measure, when a correct answer has been given. Instead, it seems to have turned into this quest to close all possibly story-ids as duplicates. Especially since there was (possibly is) a fairly strong resistance to closing story-id questions as dupe.

---

I'm still under the belief that meta consensus is not just which answer gets the highest votes, but about the ideas that are generated from the discussion. So, my interpretation of what we should do has been influenced by that: not merely looking at the highest answer. (As a tangent, I don't feel that all of our discussions that are now turned into policies were intended to become the be-all-end-all policy at the time, but have since become practically immutable and forget about the other opinions expressed at that time.)