What is the proper way to correct trivial mistakes in answers? Correctness of even small details is desirable, but especially important when programming. Examples:

Case A:

Use the FooBra library.

This is a typo, as everyone know's it's FooBar.

Case B:

Use FooBar's size method.

You believe another, conceptually identical, method (in the method-of-a-class sense) is more appropriate. Must not be different ways to get the same final state, as that would be a conceptually different answer. You might be only mostly sure and somewhat asking for clarification. ('size' is in the answer, but you believe 'length' is "what he meant".)

In no particular order, these are what I think are the important aspects:

  • correctness: of the answer, but also of the answer's author's intent
    • e.g. the author honestly believes exactly what they said is correct, instead of being close to what you believe is correct with a typo as the difference
  • credit: given where due
  • cruft: keep the answer clean
  • repetition: avoid nearly identical answers
    • e.g. worst case: 1,000 word post copied because of one typo
  • disputes: avoided

Only non-community-wikis are relevant here, as part of why wikis work is such corrections can be edited easily, and the vibe I get from SO is that even more liberal corrections are acceptable in CWs than what I'm targeting.

  • 1
    DROP TABLE STACKOVERFLOW; 'it's the only way to be sure. – Adam Davis Nov 17 '09 at 23:50
  • 5
  • 5
    I dunno, FooBra sounds like an interesting library to me. – Ether Jan 2 '10 at 18:46

I would say that obvious syntactic errors can be corrected by anybody (with enough rep to edit the answer). Things that look okay on the surface but might need further clarification (ie. your length/size example) can be handled through a comment on the answer.


Generally, correcting typos is (or should be) considered completely valid. The only time I might hesitate is if I notice that my edit will be the one to push a non-CW answer into CW. Then I'd probably comment to encourage the original author to edit. (OTOH, if an answer has had that many edits already, then maybe it should be CW.)

All of the below can be summarized very simply:

Do no evil. Edit responsibly.

Taking your primary concerns, in order:


Correctness is not only highly desirable, but should be at the heart of any edit. If you are not editing something to make it correct (whether that be grammatic, typographical, syntactic or other), then why are you editing? Yes, always edit for correctness.


Credit already comes with the edit trail of any given post. As far as I'm concerned, editing an answer should never change the original intent of the answerer, and thus the answer is still the original author's. My edits are done with my own guidelines at heart -- don't let someone make himself look foolish, fix anything glaringly obvious, do not change intent, and make sure it is still (at its core) its author's post.


Ideally, you should remove cruft. Cruft includes greetings, salutations and signatures. (Sorry folks, but that stuff is very much cruft. We are not an e-mail or a social networking site.) However, I wouldn't edit an answer (particularly close to its CW boundary) just to remove cruft. That's just me; others may disagree and I respect their opinion on that.


Ideally, answers don't repeat each other too much anyway. However, we are in anything but an ideal world, and answers will repeat. I wouldn't remove repetition if it harms the answer's intent, though -- how do I know for fact that the user wasn't acting in good faith while writing the answer? In cases of blatant copying, instead of editing, I'll downvote the answer and usually call out the user for this.

Copying a 1,000 word article to fix one typo is ludicrous; I'd just fix the typo and leave the answer as the author intended it.


Disputes, unfortunately, cannot be avoided. It'd be nice if they could, but that's not feasible. Some people will always object. However, I feel that responsible editing will, by its very nature, avoid this issue.

  • Are answers automatically made into CWs after so many edits? (I hadn't realized this.) – Gnome Nov 18 '09 at 0:41
  • Yes, I believe the threshhold is 10 edits by the original user or 5 by outside users. – John Rudy Nov 18 '09 at 14:19
  • @Fixed-Width: According to the following it is 8 times for the original user and/or 6 times by 4 or more editors. From meta.stackexchange.com/questions/11740/…, near "How does a post become a Community Wiki post?: "The body of the post has been edited six times by at least four different people." and "The post has been edited eight times by the original owner." – Peter Mortensen Jan 2 '10 at 14:18
  • @Peter: Thanks for the corrected numbers on the CW threshold. – John Rudy Jan 2 '10 at 14:58
  • Regarding the correctness part: I've seen many answers that clearly state the author is not versed in the requested language but provide useful code in other (somewhat similar) languages. What do you think about changing that code to the requested language? (maybe I should make this a question unto itself...) – R. Martinho Fernandes Jan 2 '10 at 16:59
  • @Martinho: Good question, and I think it deserves its own meta question. My personal opinion is mixed on it. On the one hand, it is changing the author's intent. On the other hand, if it's what the author would have written if he/she were an expert in the target language, maybe. Usually, I'd probably just proffer it as an answer myself. – John Rudy Jan 2 '10 at 23:32

If you have the ability to edit the answer and correct the error, do so. Regardless of whether you do or not, leave an explanatory comment.

This is precisely the primary usecase for having editing abilities: to make the content on the site more correct and consistent.

  • 1
    Agreed on almost everything, but leave a comment regardless? Aren't these type of edits self-explanatory? – Gnome Nov 20 '09 at 0:31
  • 2
    For trivial changes the comment is just in the change history, but for larger edits sometimes it's useful to indicate to the reader what was wrong and got changed. – Ether Nov 20 '09 at 0:50

I'll usually leave a comment to indicate the 'incorrectness' regardless of how trivial it is. If I happened to see it not corrected say after more than 8 hours, I will edit and leave a comment.

Going slightly out of scope, I've seen many answers (almost correct like the cases mentioned above) got down-voted. I will try to give the benefit of the doubt to the answer-er and up-vote it to neutralize the vote count.

  • 2
    If the answerer cannot be bothered updating their answer after a helpful comment, then a downvote should stick and not be neutralised. Neutralise it after the edit. – random Nov 18 '09 at 0:21
  • 1
    @random: Agree +1. Usually from the answer content, it's not hard to tell if the answerer puts in sincere effort or not. But that can be subjective. – o.k.w Nov 18 '09 at 0:35

You must log in to answer this question.