# Course of Action in case you realise the answer you post needs more.

I recently answered a question. But minutes after posting the answer, I realised that I missed out a constraint. I edited informing any reader that this particular constraint is missing.

Now I tried to work upon this missing constraint, but haven't been able to find a way to fit it in. Now in future if I find how to fit the constraint in, I will obviously edit and add that info.

But till then the answer remains incomplete. But I believe it still holds some value or hint towards finding the actual answer to the question.

However, I understand that some won't share the same opinion.

I am looking for opinions for what should be done in this case. Should I delete the answer or keep it up? What is the ideal and SE-ethical course of action here?

• Just put a line at the top, maybe in capitals, explaining this is a work in progress and should not yet be regarded as a full answer. Think about it, the way to understand the strength of a theorem is to see what happens when the hypotheses are weakened. – Will Jagy May 29 at 18:34
• You could delete the answer until a further constraint is found. At that time, you can edit your post, refine your answer, and undelete it. Also, it it the OP's problem to further refine the question with restraints, not yours. – amWhy May 29 at 20:50
• Also, @WillJagy encouraging answers that are merely "works in progress" is not an ideal scenario on math.se. – amWhy May 29 at 20:56
• @amWhy I ask this because I have seen several high rep users answer as : "HINT:.. (Basic framework for the question)......", so is it wrong to to something like this? – Saket Gurjar May 29 at 21:00
• Hints should only be given when you know the answer. You state here that you do not know that answer. – amWhy May 29 at 21:01
• I would go a step farther, and suggest that hints really shouldn't be given as answers, but be given as comments. I am aware that the community disagrees with me, and I respect the community consensus insofar as I will not vote to delete hint answers on that basis alone, but MathSE is meant to be a repository of knowledge, not half-baked almost-knowledge. If you know the answer, post it. If not, delete it. There is room for partial answers, but they should be clearly noted as such. – Xander Henderson May 29 at 23:54
• In any event, there is nuance. Comments are a little small for conveying proper nuance. – Xander Henderson May 29 at 23:58
• Xander, You're a Doctor now!?!?!?!? Congratulations!!! – JonathanZ supports MonicaC May 30 at 0:16
• @JonathanZsupportsMonicaC As of 3 pm today, yes. Thank you. I'll likely revert the name change in a day or two, but for now, I'm going to act like an arrogant jerk. – Xander Henderson May 30 at 1:32
• I happened to pass by and now I can't stop myself: Congratulations @Dr.XanderHenderson !!!! – sai-kartik May 30 at 9:10

I'm going to expand my comment into an answer. As I see it, there are several kinds of answers which have been brought up in the discussion, either explicitly or implicitly:

1. There are wrong answers, which the answerer realizes are wrong only after posting. These answers should be deleted unless the wrongness is indicative of something interesting, and the answer can be edited to highlight that interesting thing. Honestly, this seems like a rare event, hence wrong answers should, under most circumstances, be deleted. If the answerer resolves the issue later, they can always fix the error and undelete the answer.

2. There are incomplete answers, which the answerer realizes are incomplete, but which cannot be completed by the answerer. It could be that the incompleteness results only from the ignorance of the answerer, or it could be that question does not really lend itself to a complete answer (e.g. it sits at the edge of research mathematics).

Such answers (or parent questions) might be handled in a number of ways, depending on the nature of the answer (and question). If the issue really is one of cutting-edge knowledge, migration to MathOverflow might be appropriate. If the problem is that the answerer just doesn't know how to answer the question, then deleting the answer is likely appropriate (again, such an answer can be undeleted if and when the answerer resolves the incompleteness). If the incomplete answer is enlightening, even if it doesn't completely answer the question, then it is appropriate to leave the answer alone.

Because there is such a range of possible reasons for the incompleteness of an answer, it is best to consider such things on a case-by-case basis. When in doubt, ask in meta, Constructive Feedback, or (possibly) CURED. In any case, an incomplete answer which is known to be incomplete should be marked as such in an obvious manner.

3. There are hint answers. The community consensus seems to be that hints can be appropriate in cases when the answerer knows that the hint will lead directly to an answer. A hint answer should provide the asker with a clear path to their desired result, and should not consist of a some general advice about a technique which might work in most cases. Good hints are hard to write, thus I would generally suggest that one avoid this kind of answer.

• I'm a little worried by your policy that "these answerers should be deleted"! I hope I'm not on your target list. Have you become Dr. Evil? $\ddot\smile$ – Calum Gilhooley Jun 5 at 16:30
• Sometimes a user explicitly requests just a hint, not a full answer. What then? Do we give a full answer anyway, or just give a hint as requested, or should such questios be disallowed? – bof Jun 6 at 9:25
• @bof quoting myself: There are hint answers. The community consensus seems to be that hints can be appropriate in cases when the answerer knows that the hint will lead directly to an answer. A hint answer should provide the asker with a clear path to their desired result, and should not consist of a some general advice about a technique which might work in most cases. Good hints are hard to write, thus I would generally suggest that one avoid this kind of answer. – Xander Henderson Jun 6 at 13:05
• Of course, it might be worth noting that this is a compromise position. My personal opinion is that the original asker is only one of many interested parties. Math SE is meant to be a repository of knowledge, with the questions being asked having broad interest to many potential readers. In general, I think that requests for hints are too narrow to really be appropriate here. As such, I believe that hint answers are totally off-topic, as well. However, the community consensus is that requests for hints are fine, and therefore hint answers are fine, too. – Xander Henderson Jun 6 at 13:16
• But, even in that context, a hint should be something which the answerer knows is going to lead to an answer, not just a vague suggestion for a course of action. – Xander Henderson Jun 6 at 13:16
• If a hint is very narrowly keyed to the situation of the person asking the question, then it verges on being "personal advice", and so is probably off-topic. But most hints have a degree of objectivity. Textbooks often contain hints; and what's good enough for a textbook should be good enough for Maths.SE. (Shouldn't it? Or is this evidence that I have always misunderstood what Maths.SE is for?) By the way, the amusing typo on the fifth line of your answer is still there! I think you should keep it. $\ddot\smile$ – Calum Gilhooley Jun 6 at 14:34
• @CalumGilhooley I don't disagree with your assessment. Personally, I think that hint answers are not good answers. Unfortunately (from my perspective), the community consensus is that hint answers are fine. In that context, however, I think that it is not unreasonable to insist that hints lead directly to an answer. Writing a good hint is hard, and the hints in textbooks are typically very well considered (and the textbook author knows precisely how the hint is going to lead to a solution). I see no reason to expect less here. – Xander Henderson Jun 6 at 15:15
• I agree with your last three sentences. I've seen a few "hints" posted in Maths.SE that just hadn't been thought through. Posting a "hint" shouldn't be an excuse for sloppiness (although I understand the temptation all too well!). – Calum Gilhooley Jun 6 at 15:29
• Does that mean that the exact same question may appear twice, once asking for a complete answer and once asking for "just a hint", and neither is a duplicate of the other? – bof Jun 6 at 22:20
• @bof I'll go out on a limb and say yes - but I'm slightly fearful that I'm about to vanish in a puff of logic! – Calum Gilhooley Jun 7 at 0:23
• @bof This topic seems to revolve around the question of whether or not incomplete answers are acceptable. My personal feeling is that incomplete answers are almost never okay, including hints. I believe that there is a very little bit of room for answers which are wrong, but wrong in an interesting way. That is my opinion. However, I see nothing in any of that which suggests that two questions, one asking for a hint and one asking for a solution, are not duplicates of each other. Two such questions are definitely duplicates (in my opinion). – Xander Henderson Jun 7 at 2:28

@SaketGurjar: I don't know if this depends on your reputation level, but I just discovered that when I deleted an answer, when I go back to the page it shows me my previous answer, marked as "This is deleted", and there's an option for me to undelete it.

This was a total surprise to me! So even though you "delete" you answer, your work is not lost. Though I don't know for how long the deleted answer stays around and is eligible for un-deletion.

• As I understand it, deleted answers generally live forever, and are visible to users with enough "reputation" points, 10,000 or something like that. – bof Jun 6 at 9:29
• One reason for this feature is that (on occasion), an answerer knows how to fix their erroneous answer, but it is a substantial fix and will take a while to execute. So as not to leave a misleading answer up on the page, one tactic is to delete the answer, fix it, and then undelete it. In the meantime, the answer is invisible to all except those who (as @bof points out) have sufficient reputation, and can therefore be expected to understand what's going on. – Brian Tung Jun 12 at 4:36