Possible Duplicate:
Answering a Question Vs. Solving a Problem

People don’t always say what they mean.

I’ve run into cases that go something like this. The question was quite clear on what it was asking, and I answered it (correctly as far as I know) according to what the question was actually asking. But then someone else posts an answer which speculates that the asker might have actually meant something else and answers that question.

That’s fair enough of course: the asker probably asked their question because they don’t fully understand the problem they’re having, so it may be genuinely difficult to describe the problem. But what happens then is that the speculative answer gets upvoted and accepted, despite not answering the question as it is stated. Even if the question doesn’t describe the asker’s actual problem, someone else finding the question and actually wanting to know the answer will find a non-answer.

What would be a good course of action in these cases? I can think of several, but they all have downsides:

  • Edit the question so that it asks the question that the accepted answer is answering. This may be seen as too intrusive: it actually changes the meaning of the question.
  • Post a comment asking the asker to edit their question appropriately. This is futile. Even after knowing the solution they may still not be able to phrase the problem, and of course they are not interested in doing so because they’ve already got what they wanted.
  • Post comments on the question and the accepted answer pointing out that the answer is wrong and everyone should upvote my answer. I think the problems with this are obvious: it is more likely to encourage downvotes than upvotes.

Discuss away!

  • 1
    I know this is a duplicate. I can't find it, but I know there is. Aug 18 '10 at 15:31
  • 1
    The question we just closed this as a duplicate of is itself closed as a duplicate of meta.stackexchange.com/questions/8891/…. We should close this last one as a duplicate of this very question, to "close the circle". :D Aug 18 '10 at 23:39
  • This is weird. My post is an approximate duplicate of “Answering a Question Vs. Solving a Problem”. That, in turn, is an approximate duplicate of “Is “Don’t do it” a valid answer?”. Taken together, however, my post is not at all a duplicate of “Is “Don’t do it” a valid answer?”.
    – Timwi
    Aug 19 '10 at 2:28
  • The other question which is impenetrably long was closed as a duplicate of a question which this question is not a duplicate of. This one should stay open, or an alternate duplicate found.
    – devinb
    Aug 20 '10 at 15:18

Four part solution:

  1. Answer question as stated, so that the Q & A combo makes sense to people looking at the thread in the future. You may include a PS with speculation on what the asker could have actually meant including a suggestion on how to better phrase the question.
  2. Watch another answer that doesn't answer the question as is get up voted beyond all reason.
  3. Face palm
  4. Repeat 3 as needed

You should always answer the question as stated. Regardless of whether or not this particular user meant exactly what they said, other users who land on that question from a search will.

Each answer has three audiences.

  1. The user who asked.
  2. Users who have the same question as the user who asked.
  3. Users who pass by.

You should be focusing on 1 and 2, you should be ignoring 3.

As noted though, people do not always know what they want. For that reason you should answer both the question as asked, and then question they really meant.

In order to accomplish XYZ, you will need to ABC. If the reason you need to do this is because of (some large cause), then the resolution to that problem is Foo and Bar.

This answer will help the maximum number of people.

  • 1
    Fine in theory, but in practice this doesn’t work. For one, the answer to the question-as-stated could already be longer than a screen. But secondly, I can’t answer what they really meant unless I know what they really meant, and if I try to speculate about what they really meant, I would often come up with ~10 possible interpretations. Should my answer really be 15 screens long?
    – Timwi
    Aug 18 '10 at 14:44
  • @Timwi, make your answers a little shorter. And your question started on the premise of people speculating (correctly) about the author's actual intention. This is something which comes with experience. If you only know their question, the only answer that. However, if it is a situation you encountered before and have extra knowledge, then you should include that. Also, keep your answers as short as possible.
    – devinb
    Aug 18 '10 at 15:01
  1. Answer as you see fit. If you suspect the asker is asking the wrong question because he doesn't know how to ask the right one, then summarize the right one in your answer and then answer it. If you're wrong, expect down-votes (but do that anyway).

  2. When your answer is accepted, edit the question to reflect what the asker meant to ask. Otherwise, folks will have a hard time finding your awesome answer in the future.

    This should be common sense, like moving the milk back into the fridge when you find it has been placed in a cupboard - no one is going to look for it in the cupboard, and if you sit, paralyzed in fear of making a mistake, the milk spoils and no one wants to drink it.

  • I like the suggestion of editing to clarify once an answer has been chosen
    – going
    Aug 18 '10 at 22:46
  • +1 for awesome spoiled milk simile.
    – Jaydles
    Aug 19 '10 at 0:05
  • 1
    @Shoq9 - #2 Violates the Q & A must match! big time, since there's probably other answers besides yours. If you edit the question, then all the other old answers (which answered the original Q) will not be answering the written question anymore. Aug 19 '10 at 1:03
  • @Peter: well, first off you're assuming there are other answers, and that they actually manage to answer either question. But yeah, it's a gamble - it's a gamble either way, which is why I've stated before that if you're not able to accept failure you're better off staying clear of ambiguous questions and users with a track record of asking something other than what they're actually after. If you want to be diplomatic about it, then leave enough of the original question for it to be recognizable... But all else being equal, giving deference to the accepted answer seems appropriate.
    – Shog9
    Aug 19 '10 at 1:19
  • BTW: this is something of a special-case of the Chameleon Question. See that discussion for some ways of extracting yourself from a bad situation... (and it is a bad situation - to return to milk-in-a-cupboard, it would be much better if folks would remain focused enough to just put it back in the fridge...)
    – Shog9
    Aug 19 '10 at 1:21

What I've done on occasion is first answer the question directly, and then answer the question that should have been asked.

To answer your question, it's X. But if I were you, I would strongly consider doing Y instead.


@Jon already answered your question (+1), but it sounds what you really meant to ask is How do I spline my reticulations? And the answer is …

  • That assumes that I know what the question is that should have been asked, but I’m referring to a case where the question doesn’t even hint at it and seems like a clear question about a specific thing.
    – Timwi
    Aug 18 '10 at 14:47
  • 1
    Now, why would you choose Jon, of all names, to use in your example?
    – Pops
    Aug 18 '10 at 19:27
  • 1
    @Popular Because I don't know how to spell Ngu Soon Hui. Aug 19 '10 at 0:38
  • that's... very clearly a lie. Why, you don't even have the edit icon on that comment!
    – Pops
    Aug 19 '10 at 2:46

The aggressive solution I've seen a couple of times is a downvote to the question, with a comment of "Your question is unclear and doesn't seem to accurately reflect what you want to know. If you remedy this, I'll remove the downvote."

Absolutely objectively speaking, the downvote EXISTS to signify that the question is "unclear or not useful". Adding a comment to indicate how that person could improve their question is quite polite, and even gives you a chance to remove the perceived 'sting' of a downvote.

It's likely only appropriate in a case where the asker downvoted a perfectly correct answer, simply because it solved the problem as asked vs. the question meant. It may never be a good option - but I wanted to document it here as something that exists, if nothing else. ;)

  • I am not talking about questions that are unclear or questions where it is obvious that it is asking the wrong thing. I’m talking about cases where the question is actually clear, and there is no way for me to know that it is the “wrong” question.
    – Timwi
    Aug 18 '10 at 14:46
  • 1
    I don't think its a matter of whether the question is clear or not, but rather the person not understanding enough about the scope of the question to ask it in a manner that will get him what he wants. As for your remedy, I think its far from effective, esp. for questions where the person is clearly bad in English and couldn't express himself properly. Even in cases where the problem is hilariously badly phrased (like this stackoverflow.com/questions/3492662/…) I can't bring myself to downvote it.
    – Yi Jiang
    Aug 18 '10 at 14:49
  • @Kara, are you simply reporting an observation or are you endorsing this solution? (Normally, I wouldn't ask, but since you've got that big ol' rhombus after your name, I want to be extra-clear.)
    – Pops
    Aug 18 '10 at 19:29
  • @Popular Demand Endorsing anything becomes political, stupid rhombus. ;) I think it's effective, but heavy-handed. I would absolutely not do it personally, but I'm going to edit the long-winded interpretation of it up above. Aug 18 '10 at 19:55
  • @Kara - The reasons to downvote a question is that it is unclear or not useful. Asking a question that isn't what you mean is not necessarily either. A clear question, even if it's not what the asker means, will probably be very useful to many people. It seems like doing what you describe is misusing the down vote feature. Aug 18 '10 at 20:35
  • @Yi Jiang Do you believe that a downvote is ever appropriate? It's not actually meant to be a personal attack, simply a method of communication. Aug 18 '10 at 20:35
  • @Peter Ajtai I don't understand - if you ask a question that does not convey what you intended, isn't that the definition of 'unclear'? (I think these are great distinctions to discuss, btw) Aug 18 '10 at 20:36
  • @Kara - Not at all. "How do I know if a number is odd?" remains a perfectly clear and useful question whether I actually meant prime, even, or whatever. You can't vote on whether a question is clear and useful to the asker, since you can't read minds; therefore, each question must be considered on its own merits. Aug 18 '10 at 20:39
  • 1
    For your example, I agree completely. However, if the asker accepts an answer that does not solve the question as asked, the question is wrong and needs to be edited. Thought now it seems like we're thinking of entirely different types of questions. Aug 18 '10 at 21:00
  • @Kara - I thought this is why there's accepted answer vs up voted answers. If the accepted answer doesn't answer the written question, it is instructive to everyone, and the bad example can teach how to write better questions. If all the questions are retroactively edited, important and useful information about the Q&A process is lost..... imo Aug 18 '10 at 23:31
  • @Peter: remember though, the primary purpose isn't to teach about Q&A, it's to answer questions. My primary use-case - and, i'd wager the primary use of SO for most readers - is reading answers to questions that show up on Google... for this to work, the Q & A must match!
    – Shog9
    Aug 19 '10 at 0:52
  • @Shoq9 - Precisely! Which is why you must answer the question asked not the question maybe meant. Which is also why, you shouldn't go back and edit the question after a series of answers has been given, since it would make the old answers NOT match the question asked.... Eventhough, now the new edited question would match the original meaning of the asker............ It's a little confusing to explain in words, but major editing of questions after answers have been given is bad. Aug 19 '10 at 0:56
  • @Peter: You're still confusing process with end-result. It'd be great if we could somehow force people to ask the question they actually want answered. And answer the question asked. But since we can do neither, the latter actually becomes a reasonable response to the former: which leaves only the problem of turning a mismatched Q&A into a matched Q&A.
    – Shog9
    Aug 19 '10 at 1:23

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