I have found a few cases when I try to answer someone's question based on my best understanding of what they are asking. Sometimes, I have made some assumptions based on their vague descriptions, or just based on speculation as to the source of the problem. Occasionally, these answers are based on wrong assumptions (shocking, I know). What I have found in some of these cases is that I get downvoted for these reasonable-yet-not-applicable answers. This has driven me to provide all my answers which are based on any kind of assumption as comments first. This works okay, but I've seen some discussion here on meta which seems to discourage posting answers as comments. So which is it?

  • 10
    You win some, you lose some. Psychic debugging has the potential for great rep gains.
    – user7116
    Commented Apr 2, 2012 at 19:36
  • Related: meta.stackexchange.com/questions/15798/…
    – Shog9
    Commented Apr 2, 2012 at 19:58
  • Just a few minutes ago I gave someone an upvote to counteract a downvote, because they gave a perfectly correct answer to the question but wasn't a solution to the actual problem. Commented Oct 17, 2012 at 2:09

3 Answers 3


Don't post speculative answers.

Posting answers as comments isn't what comments are for, if you aren't sure what the question is about, post a comment and ask for clarifications. There is a ton of unanswered questions you could be answering instead of trying to guess what the OP means and providing wrong answers when you fail to do so.

  • 2
    And then the person who guesses right in the meantime gets all the glory. In a perfect world this would be the right answer, but that's not where we live. Commented Oct 17, 2012 at 2:06
  • @MarkRansom Glory? Meh.
    – yannis
    Commented Oct 17, 2012 at 2:11

First off: there's no possible way you can do this "safely" if you're guessing. Accept that. If you can't live with the possibility of being wrong, follow Yannis's advice and stick to answering less ambiguous questions.

If you do want to gamble, I have some recommendations that can help to make this a bit less stressful for everyone involved:

  1. Edit the question!

    If there are two ways it could be understood, edit it to leave one. Explain what you're doing in the revision summary. Then leave a comment for the OP explaining what you've done and why, and advising him to roll it back and edit to clarify if you were wrong. This shows the author what he should have done originally, makes the question easier to find for future readers with the same problem, and helps to avoid issues caused by two answerers interpreting the same question in completely different ways and effectively answering two unrelated questions.

  2. Be willing and ready to delete your answer

    Like I said, it's a gamble. In spite of your best efforts, you might be completely wrong. Know when to fold 'em... And since we're listening to Kenny here,

  3. Know when to walk away, know when to run

    Ideally, the asker will take a hint from your edit and clarify his question. But sometimes, the lack of clarity indicates severe confusion, which he'll try to resolve by engaging you in a lengthy discussion in comments... Or worse yet, he'll keep changing his question each time you solve one problem to present another one. Unless you've nothing better to do than spend the rest of the day teaching him to program one-on-one, you'll need to extricate yourself from this situation. Even if you do have the time, you'll want to take it to chat.

  • 5
    I like this answer in general, but is #1 really okay? I can see this as a good option when some context cues can push your interpretation in one direction or another, but if it's truly ambiguous, this seems somewhat counterproductive. Commented Apr 2, 2012 at 20:13
  • 1
    I do wonder about option 2. Since we are attempting to build a comprehensive knowledge base for technical problems, presumably someone could find the original question through a google search and find value in my answer, because although my answer did not correctly assume the conditions for the particular OP, it might be close enough to be useful for other people with similar issues. Is that not worth considering? Commented Apr 2, 2012 at 21:00
  • @Chris: if you're not comfortable editing in that situation, you probably shouldn't be answering either. If you want to anyway, you'll need to accept that you're playing an even risker game.
    – Shog9
    Commented Apr 2, 2012 at 21:07
  • @Jake: sure. There are plenty of perfectly good questions that can have multiple, good, non-identical answers. As sixlettervariables said in a comment, there's potential in these questions. It's still a gamble. Only you can decide how long you want to stick it out. You might win big, and find that while your answer didn't help the OP it ended up being quite beneficial to future readers - or you might find the down-votes multiplying as more and more new readers find it unhelpful. I can't tell you when to fold... (-3 is probably a decent rule of thumb; at least there's a badge for that)
    – Shog9
    Commented Apr 2, 2012 at 21:16
  • @Shog9: I don't spend a lot of time answering questions, so I can't argue from a position of great experience in this kind of situation. You say if you aren't comfortable editing, you shouldn't really be considering answering. I agree with you, but mostly in the case where you have an AWOL OP who doesn't seem to be working to reduce that ambiguity him/herself based on the early comments. I think editing the question to tip it in one direction is a good idea even if for no other reason than to generate better answers to something useful. But with an active OP, waiting it out might be better. Commented Apr 2, 2012 at 22:16

You should request clarification from the poster in comments, and build an appropriate answer based on that feedback.

If the question so vague that it requires guesses on your part to answer, it's not a good question yet. Make the person who asks provide the required detail before you attempt to answer.

Note that this doesn't always work. If the person doing the asking doesn't provide sufficient details to make his question answerable, just walk away (possibly casting a close vote as you go).

You must log in to answer this question.

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