In which cases is it appropriate to accept a partial answer?

For example, if an answer gives the state of the research on the area, that does not completely answer the question, but says also that what remains open is hard, is it appropriate to accept it?

Remark: it could be relevant to have two levels of acceptance: partial and complete, with easily distinguishable colors.

  • 15
    $\begingroup$ Sure, why not? Especially if it comes from an expert, this may be the best currently available answer. Use your best judgment, and keep in mind that an answer can be unaccepted if a better answer comes down the road (in which case one should leave a courteous comment explaining why the answer was unaccepted). $\endgroup$ – Todd Trimble Sep 4 '14 at 11:38
  • 2
    $\begingroup$ I agree with you with the fact that it can be unaccepted. Perhaps it's suitable to wait (for example) a week. One problem could be that some people consulting a list of posts and seeing that this one has an accepted answer (whereas it is just partial), do not open it. It could be relevant to have two levels of acceptation: partial and complete, with two different colors easily distinguishable. $\endgroup$ – Sebastien Palcoux Sep 4 '14 at 12:07
  • $\begingroup$ Yes, waiting a while for answers to come in is part of exercising one's good judgment. (I'm not too worried about people not opening posts that have accepted answers; probably they were not very interested in the topic if they do not open it.) $\endgroup$ – Todd Trimble Sep 4 '14 at 12:20
  • 5
    $\begingroup$ I'd also find it good to be able to distinguish between "answered completely and unconditionally" and "answered up to open problems". -- More concretely, so far I have 3 questions which this applies to: mathoverflow.net/questions/138106, mathoverflow.net/questions/118523 and mathoverflow.net/questions/168109. $\endgroup$ – Stefan Kohl Sep 4 '14 at 12:30
  • 1
    $\begingroup$ @ToddTrimble, I think the question should not be "why not?", but "why to accept?". In my opinion, accepting an answer means closing the question ("I don't have this question anymore") --- the op says that he is fully satisfied with the answer and is not interested in any further answers (otherwise, what would be the purpose of the accept button? --- you may always leave a comment that you have found someone's answer the best and add "thank you"). (cont...) $\endgroup$ – Michal R. Przybylek Sep 4 '14 at 21:43
  • 1
    $\begingroup$ (...cont) On a few occasions I think I saw a situation when someone accepted a non-optimal (or even wrong) answer, but I did not post my answer, because I did not want to intrude the op, who (by checking the accept button) clearly indicated that he is happy with the provided answer --- why on earth should I try to destroy someone's happiness? $\endgroup$ – Michal R. Przybylek Sep 4 '14 at 21:44
  • 2
    $\begingroup$ @StefanKohl, I do not think so. I have just said that, in my opinion, someone should accept an answer if he is fully satisfied with the answer. If it turns out that the answer to his question depends on an open conjecture and the op is interested in the conjecture (whether it is true, false, or neither) then it may be logical to not accept the answer unless the conjecture is settled, or another answer to his question is found that does not depend on the conjecture. It may be also logical to accept the answer and ask another question about the conjecture. It is all to the op and his needs. $\endgroup$ – Michal R. Przybylek Sep 4 '14 at 22:20
  • 3
    $\begingroup$ @MichalR.Przybylek There's an answer of mine that you once accepted, although it was later shown to have a mistake in it. So I guess you should probably unaccept it according to your criteria. But I think "accept" can actually have several meanings, such as "provisionally, this seems to be the best one can say (but this may change if more information comes in)", or "this didn't answer every last thing, but it was extremely helpful and put me on the right track", etc. Historically, acceptance has been used these ways. I think there's a lot of latitude and it's hard to give a hard and fast rule. $\endgroup$ – Todd Trimble Sep 4 '14 at 23:31
  • 1
    $\begingroup$ @ToddTrimble, not at all. Once, I was interested in getting an answer to my question, because I lacked essential tools to approach it. Then I got one answer that I thought was correct, so I accepted it. As I said in the comments, I learnt much from your answer, and then I was not interested in getting the correct answer to my question from anyone but myself anymore (I generally prefer thinking over reading). (cont...) $\endgroup$ – Michal R. Przybylek Sep 5 '14 at 8:00
  • 2
    $\begingroup$ (...cont) Put it another way --- getting an answer to an interesting question is like paying for the knowledge by the question --- you lose your question permanently (so you can not enjoy your time on thinking about the question anymore), but get something in the exchange. On the other hand, getting an interesting wrong answer is a win-win situation: you still have your interesting question, and at the same time you could learn something new :-) $\endgroup$ – Michal R. Przybylek Sep 5 '14 at 8:01
  • 1
    $\begingroup$ However, I think all of the above is a bit off topic. The question should be: "what is the meaning of the acceptance button". In my opinion it is: "I do not have this question anymore" and it seems the SE policy for questions with accepted answers justify this opinion (notice that MO is not for writing about interesting things if no-one asks for them). If I understand the other opinion, it is something like: "Thank you, your answer was really helpful". Is it correct? $\endgroup$ – Michal R. Przybylek Sep 5 '14 at 8:15
  • $\begingroup$ @MichalR.Przybylek It's an excellent question, and the proposals are interesting. I still think it is reasonable and appropriate to accept an expert answer that the question is likely very hard (in effect accepting the likely reality of a situation), and still have the question be alive. But mine is just one man's opinion, and it would be good to hear from others. $\endgroup$ – Todd Trimble Sep 5 '14 at 11:47
  • $\begingroup$ @ToddTrimble, I think we are in violent agreement when it comes to most parts. So let me try to focus, once again, on the differences --- to answer the question "should I accept an answer?", in my opinion, one should first know "what is the meaning of accepting an answer (i.e. clicking on the acceptance button) --- both in technical and social terms (it would be grate if it was a single answer compatible with both aspects)?". $\endgroup$ – Michal R. Przybylek Sep 5 '14 at 12:58
  • $\begingroup$ My point is that the button is a technical tool that modifies the chances of getting better answers, and as such should not be used for purely social purposes. $\endgroup$ – Michal R. Przybylek Sep 5 '14 at 13:29
  • $\begingroup$ @MichalR.Przybylek: don't you think that the creation of two levels of acceptance is a suitable solution for all the opinions? $\endgroup$ – Sebastien Palcoux Sep 5 '14 at 13:38

One should accept an answer if one is satisfied with the provided answer. In my opinion, in many practical cases, if one gets a satisfying partial answer, but for some other reasons is also interested in the full answer, then it is better to ask a separate follow-up question, than to leave the existing question open (after all, the op has gotten what he wanted).

You have not told what you want to achieve with "two levels of acceptance", so let me just suggest some work-arounds that you may find helpful:

  • if by a partial acceptance you want to say that you have found the answer the best, just say it in a comment,

  • if by a partial acceptance you want to reward the author with points, just do it manually,

  • if by partially accepting an answer you want to make it easier to find your question among other questions having partially accepted answers, then create a tag "partially-answered" and attach it to the question.

  • $\begingroup$ The creation of the tag "partially-answered" is an interesting idea. I'm not sure that it is really usable: for example some questions has already the maximum number of tags. Nevertheless, I'm agree for trying to use this tag. $\endgroup$ – Sebastien Palcoux Sep 6 '14 at 18:50
  • $\begingroup$ I've just created this tag. $\endgroup$ – Sebastien Palcoux Sep 6 '14 at 18:53
  • 4
    $\begingroup$ @SébastienPalcoux: I think "partially-answered" would qualify as a meta tag. -- At least as I understand, the use of meta tags is rather not encouraged. $\endgroup$ – Stefan Kohl Sep 6 '14 at 20:03
  • $\begingroup$ @SébastienPalcoux, there are a lot of problems with the current tag system. A smart person should perhaps create a proposal for hierarchical tags. I usually do not spend much time on MO and restrict my attention to category theory, and, sometimes, to logic. So, for example, when someone uses tag "monoidal-category" without tag "category-theory" it is very likely that I will miss his question. $\endgroup$ – Michal R. Przybylek Sep 6 '14 at 20:07
  • $\begingroup$ @StefanKohl: if this meta tag is not accepted by the community, then the problem is still open. Some meta tags are well-accepted: open-problem, big-list, reference-request, soft-question... $\endgroup$ – Sebastien Palcoux Sep 6 '14 at 20:26
  • 4
    $\begingroup$ Dear @Sébastien Palcoux, there is a big difference between the tags you mention and the tag 'partially-answered'. The tags 'open-problem', 'big-list', 'reference-request', 'soft-question' give some information on the type or subject of the question, while 'partially-answered' attempts to describe the present status of the thread. This is traditionally not the role tags play. In fact, I cannot think of a single tag on the main site whose purpose is to describe the status of the thread. $\endgroup$ – Ricardo Andrade Sep 7 '14 at 19:13
  • 1
    $\begingroup$ @RicardoAndrade: you're right, except for 'open-problem' which is quite close to 'paritally-answered', in fact 'open-problem' is described by: << If it turns out that a problem is equivalent to a known open problem, then the open-problem tag is added. >> Anyway, my initial proposal was to create two levels of acceptance, but such a tag is a simpler solution, so if you (and the community) disagree, then I'm ok for deleting this tag, but then are you agree with my initial proposal? Else, what do you propose? $\endgroup$ – Sebastien Palcoux Sep 7 '14 at 19:42
  • 2
    $\begingroup$ @SébastienPalcoux for the "open-problem" : it is first around since a very long time when these things were not that clear, second some (me for example) try/ied to change its use but things are moving slowly, third it is still a difference to indicate that something is about a (well-known) open problem as opposed to it being just some question that is not yet fully answered on this site. In any case please do not tag any further question with this partiall answered tag for now. Finally, I for one do not really like some of the other metaish tags either but it is hard to get rid of them now. $\endgroup$ – user9072 Sep 9 '14 at 22:23
  • $\begingroup$ @quid: what do you propose instead? $\endgroup$ – Sebastien Palcoux Sep 9 '14 at 22:33
  • $\begingroup$ @SébastienPalcoux I am sorry, I have no proposal at the moment. Still, the tagging system should not be used in this way. If you feel you must do something right now edit the title of the question something like that. (I do not propose to do this it is just relatively less harmful.) $\endgroup$ – user9072 Sep 9 '14 at 22:37
  • $\begingroup$ @quid: if you want to delete this tag, you can (I believe I can't do it that myself). It's an interesting idea to put "partially-answered" in the title (or in the body), because then we can find all such posts by using the advanced research option: title:"partially-answered" (or body:"partially-answered"). In fact it would be simpler to put "partially-answered" in a comment, unfortunately there is no advanced research option for the comments, isn't it? $\endgroup$ – Sebastien Palcoux Sep 10 '14 at 10:28
  • $\begingroup$ @SébastienPalcoux it is best to let a mod deal with it (only they can remove it without bumping the questions again; FGD mentioned in chat he will comment later). Yes, to edit a note in the body could actually be a good idea. One could use the "keyword" partially answererd (or something linke this) and at the same time add a paragraph what it is precisely that is still missing. This seems appropriate to me. re comments: AFAIK they are not searchable by design The abstract idea here is that everything that is actually important should be an edit to the question or an answer or a new answer. $\endgroup$ – user9072 Sep 10 '14 at 12:02
  • 3
    $\begingroup$ @SébastienPalcoux and others. I deleted the new tag you created. I don't think the basic idea for the tag was bad, but such tags are not the solution. At best the new tag would be a crutch, it's much preferable to have a genuine solution to an actual problem. $\endgroup$ – François G. Dorais Sep 12 '14 at 3:40
  • $\begingroup$ @FrançoisG.Dorais: What's your genuine solution? What do you think about my initial idea of two levels of acceptance? or the idea to put "partially-answered" in the body (with some explanations) for finding the post by using the advanced research option body:"partially-answered". $\endgroup$ – Sebastien Palcoux Sep 12 '14 at 9:17
  • 1
    $\begingroup$ @SébastienPalcoux: I think the best solution is to ask a follow-up question and link the two. Anything else would have lesser visibility. $\endgroup$ – François G. Dorais Sep 12 '14 at 14:20

You must log in to answer this question.

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