This question got me thinking: what edits, suggested or "full," are generally welcome in our community?

The reason for this question is that I see lots of edits that change ("correct") an author's position/solution.

Or this edit by yetanothercoder (I don't have suggestion number for it)

In some of these cases editors might be arguably correct, but I always had the impression that you're not supposed to change an author's statement (CW excluded). You can leave a comment, and if author persists in being wrong, it's his problem. You don't fix every answer on SO to be the right one.

Yet, such suggestions are not only made, they are routinely approved by long-term users.
So, am I getting it wrong? Are we drifting to default CW mode?

PS. According to the edit privilege page (which is linked for users suggesting edits too), you should edit any time you feel like it.

3 Answers 3


Edits that "correct" code on a question make me a bit uneasy, unless it's completely obvious from the question that the error being corrected was an inadvertent typo or transcription error and has no connection to the asker's real problem.

However, corrections to answers are far more appropriate, especially if the original author clearly had the right idea but messed up or omitted something that would cause confusion for the asker or other readers.

You don't fix every answer on SO to be the right one.

There's usually a difference between fixing mistakes in an almost-right answer and re-writing a completely wrong answer to be correct. In the examples you gave, the first three answers were mostly correct to begin with - the edits either fixed issues likely introduced by rushed writing, or addressed issues that had been previously raised in the comments. The closest I see to a "point of view" change would be the first, but even then it didn't really change - the edit merely clarified the context in which the answer was correct, and removed the lack of confidence in the original author's words (which was now inappropriate considering the clarifications that had been made in comments).

SO depends on this to an extent - letting errors go uncorrected save for an increasingly lengthy list of errata in the comments is a slip down the slope toward forum noise. Whenever the choice exists between confusing or misleading future readers and [editing|down-voting|deleting], option #B is the lesser evil...

That said, the last edit you cite (by yetanothercoder) appears to be utterly wrong, and should not have been approved.

See also:

  • The problem is, there's usually no way to be sure about correctness. E.g., the second example required very specific knowledge, which none of approvers had. In the first example, it's the same: the guys don't even deal with web-development, let alone Internet Explorer compatibility issues. It looks, nobody's checking 'correctness' now. Tossing a coin would be just as effective. Commented Feb 17, 2011 at 17:03
  • In fact, I'm curious, why do you call those edits 'correct'? Do you have any knowledge of OpenGL or WPF? Commented Feb 17, 2011 at 17:11
  • @Nikita: approvers approving things that shouldn't be approved can be a problem. What you found objectionable in your second example?
    – Shog9
    Commented Feb 17, 2011 at 19:24
  • @Shog9 Both approves have zero posts in [opengl]. Do you trust such approval? It's even more useless than me answering questions about linux kernel. Commented Feb 17, 2011 at 19:35
  • @Nikita: If I were thinking about approving either of those edits, I would first have done just what I did when I saw them linked here: searched for the relevant documentation and verified that the edits were technically sound. If I was unable to do that, then I would not have approved them. Granted, someone with significant experience in the subject matter would have an easier time verifying such things, but it's hardly impossible for someone else to do so... Note that the WPF correction had been suggested several months ago in a comment by someone with more involvement in WPF.
    – Shog9
    Commented Feb 17, 2011 at 21:21
  • @Shog9 How exactly did you verify that that OpenGL edit is "technically sound"? How long did it take you? Commented Feb 17, 2011 at 21:58
  • @Nikita: I checked the SGI docs, and verified that the glVertex functions are suffixed with a two-character parameter count and type indicator (since it's a C API, this is necessary for overloading). Granted, he's passing integers, but he also does this later in the code (and there aren't any real problems with converting 0 and 1 to floats). Donno how long it took me; I spent maybe 20 minutes on the whole answer, with frequent interruptions.
    – Shog9
    Commented Feb 17, 2011 at 22:20
  • @Shog9 You spent ~20 minutes, but you didn't really verify whether original code is broken or whether new code is better in any way? That's why I think people, for public good, shouldn't answer questions (approve edits) in areas they know nothing (very little) about. Not to mention, the idea that average approver spends even 5 minutes verifying some edit is a pure fiction. Commented Feb 17, 2011 at 22:55
  • @Nikita: I said I don't know how long I spent on it, but can pretty much guarantee it was well under 20 minutes. What does it matter?! Do you have a real problem with the opengl edit, or are you just offended that folks who aren't obviously opengl gurus dared to edit/approve/review it? If it's the former, then do not hesitate to roll back the edit! If it's the latter, then I humbly suggest that you stop making unwarranted assumptions and come up with better examples.
    – Shog9
    Commented Feb 17, 2011 at 23:03
  • @Shog9 I'm also getting tired of it, but I made no assumptions in my last comment. (except the last sentence, which is not important for the point) And my concern is the quality, the same as you claim to have. PS To make a rollback, would mean to become like those people: to take a side in the argument, where you have no idea who's right. Commented Feb 18, 2011 at 1:19
  • @Nikita: sorry, felt like this was getting bogged down. Of course you're right - if you or I don't know that an edit is a clear improvement, then approving it is careless - when reviewing such an edit, one should either reject it or simply leave it for someone with experience in the subject. However, I maintain that such edits are not, in and of themselves, inappropriate, nor is it beyond the realm of possibility for readers without significant on-site experience in the topic to independently verify the validity of the changes. Whether this actually happened in these cases, I cannot say.
    – Shog9
    Commented Feb 18, 2011 at 1:52
  • @Shog9 I suppose, it means we need some clear guidelines for edits and approvals. And in any case, I reserve my grim view on the nature of such approvals in general :) Commented Feb 18, 2011 at 2:20
  • @Shog9 Actually, it might make sense to pass 'proving' edit onto editor. Not like 'I fixed it' comments we have now. Commented Feb 18, 2011 at 2:37

If you are able to judge the edit is correct and absolutely correct, then you can usually approve the edit, else its always best to leave the decision to some other person.

As you have already mentioned in your question, its not good to change author's exact words (unless correcting spelling mistakes) and good to add comments for authors to notice. If they notice and act accordingly, well and good, else its their problem. Subsequently other users who visit the question might see your comment and act accordingly ignoring the author's mistakes.

So finally, if you see such things, leave a comment and carry on with other questions.

  • 2
    Your first paragraph seem to contradict the second one. First you say 'do it', then 'do not do it'. Commented Feb 17, 2011 at 16:33
  • 1
    @Nikita - My first para says if you are abolutely sure do it, else rest everything says, do not do it. Commented Feb 17, 2011 at 16:38

I would've rejected all of those. In my opinion (and I'm still not sure how widely accepted this is), you shouldn't change a post to say something different than what the author said. You can rephrase what they said, you can fix spelling, you can fix formatting, but don't fix code errors, and don't actually change the meaning. This means if you see:

#define DAYS_PER_WEEK 8

you shouldn't change it to

#define DAYS_PER_WEEK 7

I don't care how certain you are that there are 7 days in a week, leave it alone. Leave a comment saying "uh...are you sure about that #define?", and let the OP fix it

Unlike most people, I'm not convinced suggested edits is a net positive; I see a lot of bad edits in the queue that have an approve vote; this gem from yesterday had one, and it's not even disputably good like the code correction edits

  • 1
    I like your example with DAYS_OF_WEEK, conveys the point effectively. Commented Feb 17, 2011 at 16:31
  • 2
    Why, Michael, should we resist promoting correct information in deference to the author? Is our goal to answer questions, or avoid stepping on toes? (And if you're gonna leave blatantly incorrect information in place, then why even bother fixing the formatting - down-vote it or delete it). As for the poor edits... If reviewers aren't doing their job, then perhaps we need to raise the bar on that.
    – Shog9
    Commented Feb 17, 2011 at 16:53
  • @Shog9 Same point as in my comment to your answer: it's impossible for a random 20k user to verify correctness of a random statement. And raising the bar won't help: 40k users don't have specific knowledge either. Commented Feb 17, 2011 at 17:10
  • 1
    @Shog9 Because I've seen people "correct mistakes" that were not mistakes at all; now the answer is wrong, people rightly downvote it, and the answerer gets upset. I understand the argument that the goal is the best answer regardless of how we get there, but it's too easy to mess up. If an answer is minorly wrong, post a comment; people reading the answer will see your upvoted comment at the same time and factor in it. If it's majorly wrong, post your own answer Commented Feb 17, 2011 at 18:48
  • @Michael: frankly, I'm more concerned that folks are accepting bad edits. Stuff like your example, that's not surprising - folks are driven to post crap like that, and use whatever facilities they can find - arguably, an easily-dismissed edit is better than a non-answer answer. An experienced user accepting such a thing though... That's negligent. I'm wondering if slapping the names of the approvers onto the edit would help...
    – Shog9
    Commented Feb 17, 2011 at 21:28
  • 1
    @Shog9 There's an existing feature-request to add it to the revision history Commented Feb 17, 2011 at 21:37
  • 1
    @Michael: very good, voted.
    – Shog9
    Commented Feb 17, 2011 at 22:22

You must log in to answer this question.

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