Let's define the Socratic method in the context of StackOverflow as: a heavy use of questions, in answers, to teach people how to answer the question themselves (example).

As far as I know, this goes against the Q&A format as we all know it, as the Socratic method provides that the person, receiving the socratic answer, should keep questioning himself/herself favoring a debate, which is not what answers are there for.

Quoting from the about page:

This site is all about getting answers. It's not a discussion forum. There's no chit-chat.

Should this behavior be discouraged in answers?

  • 12
    It certainly should be encouraged for homework questions (those that the OP has indicated are homework). – Oded Apr 23 '13 at 11:10
  • 4
    The Socratic method may not make sense for answers but it does for comments in that posting questions as comments to a post may help the poster to clarify and refine their question or answer by editing it. – N.N. Apr 23 '13 at 11:22
  • 5
    The example you have linked to does provide the answer, it just doesn't state it explicitly. Once the OP follows the instructions in the answer, the solution becomes obvious. – Asad Saeeduddin Apr 23 '13 at 12:07
  • 4
    I am not going to vote for any answer to this question that doesn't employ the socratic method. – Pekka Sep 25 '13 at 2:37

This should be allowed.

...as long as the "answer" doesn't require a response from the asker in order to lead a reader to the solution.

While not appropriate in many (perhaps most) instances, rhetorical questions can be an excellent way to demonstrate faulty reasoning and teach folks how to solve similar problems on their own. However, this is not a medium for discussion - those providing answers must show their hand by answering their own question (or at least hinting at the expected answer) in order for their answer to be an effective tool for education. The example provided does this - I would consider it a perfectly valid answer for this reason.

  • Why not just answer the question tho? Instead of talking to the OP like s/he's a child How many bits wide is an int? You seem to think it's three bits wide. Certainly not correct! I wouldn't call an answer an answer if all it's saying is "Think about it" – TsSkTo Sep 25 '13 at 8:16
  • 1
    @TsSkTo: If you think the answer is unhelpful, then down-vote it. I don't think the format is inherently harmful though - whenever possible, it is beneficial to teach folks how to fish rather than just handing them the catch they seek. – Shog9 Sep 25 '13 at 16:42

Rhetorical questions should be encouraged in comments, but discouraged in answers.

There are several reasons for this. First and foremost, a question does not satisfy any common sense interpretation of the term "answer". It's a response, yes. It may be helpful, yes. It may allow someone to derive an answer, yes. But an answer it's not.

Second, we have an established guideline/culture for using comments to express questions and for using comments as a mechanism to allow users to answer questions for themselves. Using the comment mechanism for rhetorical questions is only natural in this context.

Third, I contend that a fundamental attribute of the StackExchange "brand" is a vehicle for getting answers to questions quickly and efficiently, either through Google search (for existing questions/answers) or by a real time response. Allowing/encouraging answers in the form of questions is at odds with that attribute.

Finally, we have a flag category of "Not an answer" whose name and explanation is an ideal fit for rhetorical-questions-as-answers given this answer. If you accept/encourage rhetorical questions as answers, then we have to add "Not an answer" to the rogues gallery of Stack Exchange phrases with gaping exceptions that everyone is/was supposed to remember (e.g. "Off topic", "Not a real question", etc.)

  • 4
  • 1
    "we have an established guideline/culture for using comments to express questions and for using comments as a mechanism to allow users to answer questions for themselves." wait, what? No, no no no no NO! Questions and solutions buried in hard-to-search discussions is the problem that Stack Overflow was created to solve, not provide yet another venue for! We've tried to encourage folks to post their own solutions as answers - twice! – Shog9 Sep 25 '13 at 2:09
  • 3
    For the rest of this, I gotta ask what you consider more important: useful content that helps folks to learn, or slavish devotion to the harshest possible interpretation of every rule without regard to its intended purpose. And then suggest you read this: meta.stackexchange.com/questions/141210/… – Shog9 Sep 25 '13 at 2:12
  • @Shog9 As with your earlier "why so?" question, I'm not sure what your "no, no, no" comment is referring to. That thousands of questions get resolved solely as a result of questions expressed in comments is a fact of life as is the fact that many experienced SO uses are consciously choosing the comment format over the answer format for their interaction with the OP, particularly for questions that are of marginal long term significance (e.g. simple question answered by any basic documentation, duplicate question, ill-formed question, question about logic error in presented code). – Peter Alfvin Sep 25 '13 at 12:32
  • @shog9 I believe it's also true that there is "official" SO documentation which indicates that if you want to ask the OP a question, you should generally do that in comments, although I see that "When should I use comments?" is not answered in the FAQ on comments. – Peter Alfvin Sep 25 '13 at 12:34
  • @Shog9 As for your last question about which I consider more important, please forgive me if I don't play along with the false dichotomy you implied. In this any other situations, I'm saying that: a) language matters and we should continually strive to keep SE's use of language in line with the everyday usage to the extent possible, b) a question does not meet the everyday interpretation of the word "answer", and c) questions as answers are not maximally "helpful" or in line with SE's goal. – Peter Alfvin Sep 25 '13 at 12:41
  • @Shog9 Examples are pervasive as I said, but note this very recent and very typical case of a user solving his own debugging problem as an indirect result of questions (in this case from me). Admittedly, these were non-rhetorical questions on my part, but I contend the distinction is typically not significant. In this case I was just working back from the failure symptom asking "why", the standard problem solving approach. stackoverflow.com/questions/19008878/… – Peter Alfvin Sep 25 '13 at 16:18
  • @Peter, I'll make this simple: the goal here is to get useful answers to real questions. When you're evaluating a post - or a comment, or an edit - the first question you should ask and answer is, "does this work to achieve our goal?" If the answer is "no", then it's time to worry about flag or close reasons or whatever. If you skip step #1, whatever you do next is bound for trouble. – Shog9 Sep 25 '13 at 16:39
  • @shog9 I think that's a great rule for flagging except that the system is currently designed (for some reason) to use flagging as a vote-to-close mechanism. Since closing and vote-to-close is an acceptable, constructive way to give feedback to users about how they can improve their question, then I don't think that's a case of the question "not working to achieve our goal". Similarly, I honestly believe that questions-as-answers, like salvageable-but-not-meeting-criteria questions, don't support our goal as is, but support our goal in the sense that can be improved to meet our standards. – Peter Alfvin Sep 25 '13 at 16:58
  • You're making this overly complicated. 1) Find a post. 2) Is it well-written, helpful, constructive? GOTO #6. 3) Can you make it so? Do it, then GOTO #6. 4) Does it need to be removed? Flag or vote, GOTO #6. 5) Down-vote. 6) Sip your tea. – Shog9 Sep 25 '13 at 17:04
  • @shog9 Huh? Sorry, but that seems neither simple nor realistic. You seem to have completely skipped the step that involves giving feedback to the OP intended to help the OP improve their post to meet the question or answer criteria, unless you intended to cover that by "make it so", which seems like a real stretch. Further, you seem to imply that flags should only be used for removal, when (as I've said repeatedly) they are the mechanism <3k users have for the vote-to-close process (which is not about removal in many cases). – Peter Alfvin Sep 25 '13 at 17:12
  • With the possible exception of duplicates, vote-to-close is always about removal. It does other things, but if you don't think a post should be removed in its current state, you should not be voting to close it, or asking anyone else to close it for you. Anyway, this discussion wasn't originally about closing and has become distractingly tangential - if you want to continue, you can find me in chat - otherwise, I think we're talking past each other. – Shog9 Sep 25 '13 at 17:15

Yes, it should be discouraged altogether in answers. You have a problem, someone provides an answer. However, if further discussion ensues in the comments then the answerer and questioner can continue their conversation into chat.

You must log in to answer this question.

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