In this question I forgot to add that I want the space to be sufficiently large.
One user wrote a (correct) answer pointing out a trivial example on $2$ points, answering the question.
It was my mistake to unintentionally leave out the condition that would make the problem interesting.
Question. How to proceed? Just edit the question and include the additional condition, rendering the correct (but trivial) answer obsolete? Or put the problem with the new condition into a completely new thread?
Let me copy the discussion in the comments of Dominic's linked question. The pre-history of this discussion was Dominic's question posted March 6 6:02, 2018 (all times are GMT+1), followed by Francesco's answer [b], and Dominic's edit [c].
[b]: Answer posted by Francesco Polizzi (in its original form): Take the discrete space with two points. It is clearly homogeneous and Hausdorff, and its only self-maps are the two constant maps, the identity and the involution exchanging the two points. - answered Mar 6 at 8:55 Francesco Polizzi.
[c]: Edit by Dominic of the question Mar 6 at 10:19, adding the assumption that the space is infinite and then after the comment [d] below: "EDIT. I forgot to add "infinite" in the original question." edited Mar 6 at 10:33
(the next steps are comments to the question)
[d] Well, you should not change the question once there is already an answer. In this way a perfectly correct answer becomes a wrong one. Please write an edit, explaining the modifications you made. – Francesco Polizzi Mar 6 at 10:25
[e] I disagree with you. Even if this corresponds to the original formulation, this does not deserve an answer, but rather an comment saying that the question should be reformulated. If this were the intended question, it would be closed as off-topic. – YCor Mar 6 at 13:06
[f] @YCor: of course you can disagree. At any rate, this was the question, it was unanswered, open for more than one hour and it also had un upvote. Sometimes it happens in life that people miss trivial counterexamples, this is not my fault. – Francesco Polizzi Mar 6 at 13:21
[g] That said, I still think that modifying a question with an answer, making a right answer a wrong one is a bad practice, regardless if someone finds such an answer interesting or not. – Francesco Polizzi Mar 6 at 13:22
Now let me comment further. I think (1) it's bad practice to modify a question after it's been replied, because it's a lack of respect towards the person who did the answer, who possibly did some efforts to handle the question. In a sense, it reflects that the OP does not consider the efforts made by the person who wrote the answer as worthwhile.
But it turns out that this judgement can be appropriate, when the answer consists of a trivial counterexample or just a link. (In my own experience, a case with possibly a little effort by the replier, and where the question was subsequently modified, is this answer.)
Next, I think and it's been discussed at many places, that (2) it's bad practice to post trivial answers to trivial questions (one reason among other being that it forbids the OP to delete the question; one other being to discourage homework questions - anyway it's not the point here). If a question has a trivial answer, it should be closed, and the trivial answer should be mentioned in the comments. If one can give to the OP the benefit of the doubt that there's a missing hypothesis, then it's better post a comment addressing this, rather than immediately voting to close.
If we had followed Francesco's request ([d] above) in the current case, what would have happened would be:
Sometimes an interesting question is spoiled because the poster forgets an important detail, or explains it ambiguously. Sometimes there is no mistake, but it simply happens that the first answers or comments make it clear that the truly interesting question is just a bit different. What it usually happens, and I find perfectly correct, is that the poster apologizes in comments, and decides, maybe after suggestions by other users, whether it is worth posting a new modified question, with a link to the previous one, or just making an edit (complete with date and explanations) in order to update the original question. Both ways are good to me; which is more convenient, really depends on the case. Of course, I agree with Francesco that changing the question without any explanation and without leaving any trace does not add to the clarity, even when there is absolutely no bad intention, like in the present case. Talking to my personal experience, let me add that sometimes I know in advance that a poster (especially newcomers) will mix up the question, and I copy their exact words in my answer, as a precaution. Also, sometimes the first wrongly stated question actually helped (and forced) me to understand better the right answer to the successive right question.
In my opinion the best way to handle this is usually to accept the trivial answer and ask a new version question containing the correction. My arguments:
If you didn't think of the trivial example then it's likely that other people won't either. A unique service that MO provides is that you can google trivial questions and see trivial answers, whereas such trivialities are usually filtered out of published papers and textbooks. This has saved me a lot of time and pain in the past.
Noticing simple examples often requires a lot of serious intuition and expertise in a subject, so it's generally not the case that the answerer is getting undeserved reputation.
The main reason to modify the question instead of accepting the trivial answer is to avoid polluting your question list with what turned out to be a slightly silly question. But taking that hit carries significant possible upsides for others.
Of course my arguments don't apply if the error is so trivial that it requires no real expertise to spot it - e.g. forgetting to exclude zero in a nontrivial question about prime factorizations. Those issues should be settled in the comments. But I could see a reasonable person (myself included) spending a bunch of time trying to prove the statement in your original question before running into the two point counter-example, so I would have kept it around for posterity and asked a new question.
