What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 131 Stack Exchange communities.

I just came across a question where the question asker slightly modified his question but was not allowed to! Here's the revision history.

It was pointed out to the asker that one of his code segment doesn't compile and it makes no sense to ask difference between a compiling code and non-compiling code. I believe, this was the reason that question received 4 downvotes. (Otherwise, it's not such a bad question and wouldn't have received 4 downvotes IMO).

Asker edited the question saying:

Removed the first one which gave a compiler error. (My bad!)

Asker's edit was rolled back with the reason:

I've rolled it back because all answers reference both example lines of code; so they should both remain, regardless of their correctness.

It's not that the asker changed the entire content of the question, he just modified an offending which part which accumulates downvotes for him and he did in quick time too (3-4 minutes).

Now the broader question, is it ok to not allow an asker change his/her question even if it changes the original context slightly and may make some (or part of) answers irrelevant?

I have always thought it's the answerer's responsibility to edit/delete the answer if the question has been modified and his answer became irrelevant. What's the consensus on this?

share|improve this question
3  
Actually, it took him 14 minutes, not 3-4, and at the time of the edit, all three answers to the question had already been posted. –  animuson Feb 11 '13 at 0:33
1  
@KingsIndian: You can always mouse-over any time on SO sites to see the exact timestamp –  Nicol Bolas Feb 11 '13 at 0:36
1  
On a related matter, unregistered users often lose the ability to edit their previous posts when their cookies get lost. Merging such duplicate accounts is one of the on-going housekeeping tasks for moderators. –  dmckee Feb 11 '13 at 1:11
1  
The problem with that edit is that it changed the question entirely, which invalidated all the existing answers. Originally the question was asking why one line of code worked and the other didn't, and it got changed to erase the 1st line of code, and instead asked how the 2nd line of code worked. Generally once a question has answers, we discourage users from editing their question to change it to a new question and invalidating the existing answers. The current edit is better - it leaves both lines of code but clarifies that he already knows the 1st is a compile error and wants to know why. –  Rachel Feb 11 '13 at 13:48

3 Answers 3

up vote 2 down vote accepted

Answers were already based on the merits of the original question. Editing it out like that would cause confusion to someone just passing by/didn't have time to look at the revision history.

My approach has always been to allow for clean-up of the original question, but not to change it in any way. Removing that line of code does change the question.

If the OP has another question, then they can ask it. If they want to modify their question, then whether or not it's similar to the one they just asked must be taken into consideration, too - as to avoid it being closed as a dupe if they did make another question.

share|improve this answer

I have always thought it's the answerer's responsibility to edit/delete the answer if the question has been modified and his answer became irrelevant. What's the consensus on this?

I could not disagree more on this statement. The OP asks a question, and someone takes the time to look at the question, figure out the problem, and craft an answer, then the OP should not do anything to cause that person "harm" (through downvotes).

Generally, I think the latitude the OP has in editing the question depends on the answers.

  • If there are no answers, then the OP should have a significant amount of latitude to edit.
  • If there are answers and the edit, while changing the context of the question, does not invalidate any existing answers, then the edit is ok.
  • If the edit invalidates existing answers either by removing the offending code, correcting the error, or changing the context of the question, then the edit should be rolled back.

The basic rule of thumb I use is if the edit will cause existing answers to attract downvotes for not answering the question, then it is unfair to the original answerers and the edit should not be allowed. Permitting the OP to change an question that will attract downvotes for an answer is unfair to the original answerers who tried to help the OP, so they should not be held responsible for the OP's decision to change the question.

Where this guideline causes a problem is when the question is changed and no one sees it, and another answer comes in addressing the new/updated question. At this point, rolling back will invalidate the new answer. Here I think the guidelines do cause a problem and need to be handled on a case-by-case basis. Maybe the original and updated questions can co-exist and merged* so both answers are valid. In other cases, it might take some kind person (or the asker of the question) to ping the owner of the now wrong answer to let them know the question was changed and their answer is no longer valid.

*- not merged in the sense that duplicate questions get merged, but the original question is edited back into the current question

share|improve this answer

Now the broader question, is it ok to not allow an asker change his/her question even if it changes the original context slightly and may make some (or part of) answers irrelevant?

In general, no. If editing your question makes existing answer invalid, there are three options:

  • The question wasn't clear, it has been misinterpreted or it contains an obvious typo.

    In this case, I think it's OK to edit it. If the question is ambiguous, users should askl for clarification before answering.

  • The edit isn't necessary to solve the OP's problem.

    In this case, leave the question as it is. Purposely invalidating existing answers is somehow rude.

  • The edit is necessary to solve the OP's problem.

    For the same reason as in the previous bullet, it's better to ask a new question.

See also: Exit strategies for “chameleon questions”

share|improve this answer

You must log in to answer this question.