Yesterday, for the second time in the past three months, I answered a question, only to get a response from the OP saying, in effect,

This answer was correct when you posted it.  I’ve changed the question since then; so your answer is now wrong.  Please answer my edited question.

What is the appropriate reaction to this situation?

  • Roll over and …
    • just ignore the question and the OP (“You’re dead to me.”), or
    • answer the edited question,


  • Tell the OP that it’s poor form / bad etiquette / against the rules to change a question so drastically as to invalidate answers that were correct when posted (citing what reference?), and encourage / request the OP to …
    • partially rollback the question, so it shows a Part I and Part II, or
    • fully rollback the question, and ask a new question, or
    • edit the question myself (doing one of the above)?

P.S. In the more recent case, the OP accepted my answer and then unaccepted it (even though no new answer(s) had been added).

  • 1
    Have originally written an answer that was so complete and so convergant , that anything else they ask or edit it to along the similar lines of the question are still answered :-) Or take that opportunity that the question was changed some to finish it up along that same idea. Under the idea that if anyone else comes along finding it in search it is useful to them also. – Psycogeek Jul 15 '15 at 23:07
  • 4
    You might also mention that asking questions here is free. They should probably just ask the new question as a new question. Tell them they'll get more points for it anyway and that should be incentive enough to fix it. – krowe Jul 21 '15 at 23:47

For me, it comes down to how drastically the OP edited their question.

Since the "spectrum" of possible edit-changeiness (how much the content of the question was edited) is infinite in a continuum, I'll break it down into three rough categories:

  • A little tweak in the question that (one would hope) requires just a little tweak or added paragraph to your answer.

In this case, I'd suggest:

Roll over and answer the edited question.

  • Significant additions or modifications that change the question fundamentally while remaining vaguely on the "same topic", requiring a complete rewrite of your answer or adding many paragraphs to your answer to accommodate their revised question.

In this case, I'd suggest:

encourage / request the OP to partially rollback the question, so it shows a Part I and Part II

and then

  • If they comply:

    Roll over and answer the edited question.

  • If they refuse or ignore your request:

    Roll over and just ignore the question and the OP (“You’re dead to me.”)

  • The question after their edit is on a completely different topic. The original question is now GONE and what is there now has not even a tangential relation to the original question. OR The original question remains, but the edited "new question" is not even tangentially related to the original.

In this case, I'd suggest:

encourage / request the OP to fully rollback the question, and ask a new question.

  • If they comply, your answer will stand on its own, and you won't have to do anything (except, if you wish, answer their new question).

  • If they refuse or ignore your request, flag it for a moderator.

I wouldn't personally have the temerity to just edit their question in such a fundamental way as to change it back to the original, because that could start an edit war. But you're free to try to roll it back once and that's not against the rules. Once. If they roll back your roll back, to prevent an edit war, flag the question or ignore it (either one is fine).


Allquixotic's answer is good, but I would look at his middle case a little differently. These tend to fall in two categories:

  1. The original question was not worded well. The edit clarifies what was intended and is consistent with the original wording potentially having meant that.

  2. After getting some input, the OP decides he really should have asked a slightly different question. Or, it was a multi-part question and having gotten the needed information for one part, the OP drops that part to refocus the question on the remainder.

In the first category, that's the breaks. It's up to you whether you want to modify your answer or walk away. If you walk away, your original answer should be dealt with appropriately. If it isn't really an answer to the revised question, delete it. Or if a lot of work went into it, ask a new question and use it for a self-answer.

I would treat the second category like allquixotic's third case. If the OP has changed the question in a way that isn't simply clarification of what was originally meant, and it renders existing answers irrelevant, that makes it a different question.

People contribute questions to SU's knowledgebase in exchange for potentially getting an answer themselves. Once a question has been asked and has received answers, it assumes a life of its own, and the OP doesn't have carte blanche to do whatever they want with it. It defeats the site's purpose to allow the OP to replace the question with a different question. The OP is free to ask as many variations or new questions as he wants, but as new questions.

My opinion is that in the latter situation, the question should be rolled back and the OP encouraged to ask the revised question as a new one. This can be initiated in a comment, but if the OP isn't cooperative, it should be bumped up to a moderator for resolution. It probably is not a good idea for an answerer to be the one to roll the question back.


Based on this question and other answers, I would suggest the following rules:

  1. Editing one’s own question once it has an answer should be subject to approval, the same as editing someone else’s question.
  2. Unaccepting an answer an answer once it was accepted should require an explanation and be subject to approval.
  • 2
    Interesting concepts, but …  Are you familiar with the term “non-starter”?  How about “lead balloon”?  Let me start with your second proposal.  The idea that a downvote should require an explanation has been proposed many times and has repeatedly been shot down with extreme prejudice.  The idea that an unaccept should require an explanation would meet a similar fate.  … (Cont’d) – Scott Jul 24 '15 at 0:44
  • 1
    (Cont’d) …  Also, there have been suggestions that moderators, or a community vote, should be allowed to forcibly “accept” an answer to a question that has no accepted answer — specifically, in the case where the OP has commented “this answer worked”, but not accepted the answer.  The response has been an adamant assertion that the checkmark is the OP’s castle, and nobody else should be allowed to touch it.  I expect any proposal to restrict the OP’s ability to unaccept an answer would similarly be denied.  … (Cont’d) – Scott Jul 24 '15 at 0:45
  • 1
    (Cont’d) …  Backing up to your first idea — as stated, it’s too restrictive.  It happens all the time that an answer (or answers) get posted, and then the OP says, “No, that’s not what I meant; here, let me edit the question to clarify what I meant.”  Requiring such edits to be reviewed and approved would put an undue burden on the process.  My case — “Yes, that is what I meant; here, let me replace the question (which you answered correctly) with a different question.” — is rare.  But thanks for contributing your thoughts. – Scott Jul 24 '15 at 0:45

You must log in to answer this question.

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