12

I've always had a hard time getting reviewers to approve my edits pending peer review. It's very often either "too trivial" or "too much change", with "should be a comment" thrown about here and there. How do I hit the silver lining?

Just today, in this post, I thought I would improve the accepted answer by adding some useful information that would help the next person. In the process, I also decided to improve the formatting a little bit:

Android code won't generate R.Java after cleaning

I suggested this edit first with the following comment:

improved formatting, add footnote about errors that can be ignored

Within seconds it was rejected 3:1, so I thought I'll try again, this time with this comment:

(why reject? I'm trying to add some useful information here so that the answer is more clear to someone who would encounter the same problem again. how is this edit "minor" or "trivial" or "not substantial"?) improved formatting, add footnote about errors that can be ignored

Again it was rejected 3:1, with one of the rejects explicitly saying "Should be a comment instead.". I would argue against this saying that making the extra information a comment would defeat the purpose of a website like StackOverflow, where seekers of answers should be given correct and complete answers.

In this specific case, the phrase "Make sure you don't have any errors" is misleading, because in the Eclipse IDE, even after other errors have been fixed, the error which I was talking about will still remain, which can only be fixed by doing the step that follows it ("Fix Project Properties") or the instructions in the second paragraph (modify XML file and save). This is what led me to suggest the edit to make the answer more complete and comprehensive, so the next person does not have to sift through the comments for such a valuable piece of information.

I think very often, peer reviewers at SO are rejecting suggested edits too harshly, too quickly and without fully understanding the original problem and the context in which the edits have been suggested. They seem quite often to follow a gut feeling when making decisions (how can you reject an edit so quick within mere seconds of posting it?) rather than analyzing the question/answer properly, understanding the context and thinking about whether the suggested edit is actually something worth keeping.

Suggestion:

I hope the SO/SU community can improve on the peer reviewing process so that actual useful edits are not lost in your battle against useless ones. Perhaps reviewers should be forced to spend some time reading the original question and the original answer, and deciding whether the edit adds valuable piece of information that could be helpful to the next person who has the same question. (This would mean edits can't be approved or rejected within seconds of viewing them, but the reviewer could be delayed by a minute or two* - I believe this will reduce the chance of impulse reviewing as the wait will encourage the reviewer to use the time productively.

* The time to wait could be tied to the caliber of the reviewer: a new reviewer could be given 3 minutes of wait; a learning reviewer 2.5 minutes; a graduated reviewer 2 minutes; a seasoned reviewer 1.5 minutes and a veteran reviewer 1 minute (the minimum wait for any reviewer).

Alternative Suggestion (courtesy CodyGray and Mr.Wizard):

The edit queue could be filtered based on tags of questions on which the edits were made, meshed with the top five? tags the user had been most active on (their past questions and answers combined, with answers bearing more weight and high-voted/accepted answers bearing even more weight). Those edits which had lapsed a couple of hours or a day without review could be placed in the unfiltered queue for anyone with enough reputation points to review them even if their top tags do not match the tags in the question. Both the filtered and unfiltered queues should be made available to the users with enough rep to see the review queues, without any specific restriction towards one or the other.

23
  • 8
    "I thought I would improve the accepted answer by adding some useful information that would help the next person." Never change the answer to something it was not. Don't add extra content.
    – Bart
    Commented Jul 27, 2013 at 7:08
  • 7
    You commented too quickly. See the problem? You just saw some keywords in my long question and thought I'm doing something I'm not doing. I did not change the answer to something it was not. Please, be kind enough to read and spend some time with my question before making your judgments about it.
    – ADTC
    Commented Jul 27, 2013 at 7:10
  • 8
    Besides, what's the purpose of "improve the answer" feature if no one is allowing you to actually improve an answer?
    – ADTC
    Commented Jul 27, 2013 at 7:12
  • 9
    I did not comment too quickly. You added extra content to the answer, modifying it to say something the OP didn't. That is the kind of stuff that should go into a comment. Then the OP can pick it up if he wishes to do so. If that is not enough for you, add your own answer. Edits are not meant for anything else than formatting, language corrections, minor typos and the like.
    – Bart
    Commented Jul 27, 2013 at 7:13
  • 2
    This is why we should look at adding a "What not to edit" section in Help Center, as proposed by Bart: meta.stackexchange.com/q/187493/225020 Commented Jul 27, 2013 at 7:21
  • 5
    @Bart Not everyone agrees with you, e.g. (15839), (19477)
    – Mr.Wizard
    Commented Jul 27, 2013 at 7:22
  • 1
    @Bart I don't know about Stack Overflow, but in case of minor additions like these, which definitely improve the answer, we'd accept that on SU.
    – slhck
    Commented Jul 27, 2013 at 7:25
  • 1
    "Edits are not meant for anything else than formatting, language corrections, minor typos and the like." I don't understand this philosophy. Why is it unethical to improve an answer by doing "anything else" when it could be very useful for someone else in the future?
    – ADTC
    Commented Jul 27, 2013 at 7:26
  • 2
    I don't see how claiming ownership to an answer is helpful in a community-based Q&A website - it seems to go against the idea of a community where everyone works together to improve the quality of content available. It looks like I can't improve the quality because "it's not what the OP intended to say". And the argument that an OP can pick up the comment and edit his own answer is weak considering the original answer was made in 2011, and the OP may no longer be active in SO (look at the comments, my edit has already been in a comment for so long - the OP didn't notice).
    – ADTC
    Commented Jul 27, 2013 at 7:26
  • 6
    Keep in mind @Mr.Wizard that those posts are fairly old now. In my experience, the community seems to have moved away from that to some extent. While I appreciate the idea of editable-for-all content, and the community creating "the perfect answer", like it or not, there is a sense of "ownership" for posts. And subsequently edits should not change a post beyond corrections I've stated.
    – Bart
    Commented Jul 27, 2013 at 7:27
  • 2
    @slhck Maybe this is borderline. Maybe, if I would have known it to be correct, I would have even accepted it. (Not likely, but maybe...and outside chance?) But in my experience on SO that's not going to happen. Add to that some of the reviewing problems we already have, and I see allowing such edits of substance lead to various problems.
    – Bart
    Commented Jul 27, 2013 at 7:30
  • 1
    @Mr.Wizard Keep in mind that on that site the traffic (with all due respect) is less. And I assume that the range of topics is significantly more limited than on Stack Overflow. So you're more likely to have both knowledgeable editors and reviewers for a single post. Unfortunately for a site as large as SO, that is becoming more problematic. So while this might all be peachy and fine on smaller sites, I don't think it's a good idea for SO.
    – Bart
    Commented Jul 27, 2013 at 7:36
  • 4
    @Mr.Wizard Yes, it works as long as Mathematica experts tended to review suggested edits on Mathematica-related tags. That would, in fact, work out for all tags and "sub-communities" on SO, if only people would learn to use the "Skip" button for things that are either (A) not trivial formatting edits or (B) outside of their area of expertise. That's what I try to do, for whatever it's worth. Commented Jul 27, 2013 at 7:47
  • 3
    I think your updates (especially the idea under the alternative suggestion heading) would work better as a new question with a feature-request tag on its own. As it stands, the request is somewhat lost in this discussion. Commented Jul 27, 2013 at 9:21
  • 2
    The 'duplicate of' question is only indirectly relevant here, and not a duplicate. I did not edit any code.
    – ADTC
    Commented Jul 27, 2013 at 19:30

2 Answers 2

10

The problem here is that you're asking the people reviewing the suggested edits queue to judge the technical accuracy of your additions to the answer. Since the site's domain is so extensive (there are a lot of programming languages and frameworks)*, there is little to no guarantee that the people who review your suggested edit will understand what you're trying to add. They have no way to tell whether it is obvious and trivial as you assert here, or whether it's out and out wrong.

So instead of trusting you and clicking "Accept", or recognizing their inability to accurately judge the merits of the edit and clicking "Skip", the majority of folks are clicking "Reject" and arguing that you should do little more than fixing typos and improving formatting.

The good thing about that is that it means anyone is qualified to review a suggested edit—you don't need any particular knowledge about the post's topic. The bad thing about this is, as you point out, that it makes it impossible for users without full editing privileges to contribute anything substantive to an existing post—the only improvements they can make are cosmetic.

Joel's words from 2008 are as true today as they were then: the purpose of this site is to accumulate superior answers to a diverse array of programming questions. There is no reason why you should have to post a second answer just to add some additional background information or a helpful hint.

But it just doesn't always work out in practice as elegantly as it sounds in theory. What you have to understand is how many genuinely awful edits get suggested and end up in the suggested edits queue, fatiguing and jading the reviewers. If it looks like you've suggested one of these bad edits, your edit gets rejected. In this particular case, your edit might have been a gem, but it was hidden among so many ugly rocks that people just assumed it was ugly, too. Honestly there are still so many bad edits getting approved that I'm hesitant to suggest that we should be more lenient in the edits that we accept. It's a much easier guideline to tell people that they should just not accept anything that changes the meaning of or adds additional content to the answer. The reviewers can understand that, and it'll lead to the right decision 80–90% of the time.

So what's the solution? Focus more on contributing your own content in the form of answers. That will earn you reputation, and it won't take long to get to the 2,000 reputation that you need to make edits that go beyond cosmetic. Invalid edits can always be rolled back if a mistake is made, but the idea is that we trust people with full edit privileges to make the right call on their own edits, using both the information they know about the post's topic and the site's guidelines.


* This explains Mr. Wizard's observation that edits of this nature are better handled on smaller sites (read: not Stack Overflow), where the size of the suggested edits queue is more manageable and more users have broadly-applicable domain knowledge. That's great for those sites, and that's why we focus so much on community building with the Stack Exchange 2.0 platform. But Stack Overflow redefines the notion of "critical mass", and I don't see this problem being solvable there.

16
  • 1
    Okay, so I won't have to add my own answer. Thanks. :)
    – Bart
    Commented Jul 27, 2013 at 7:43
  • A fine answer. Before I go looking do you know if the idea of restricting review edits to those with Reputation in that Tag has been discussed?
    – Mr.Wizard
    Commented Jul 27, 2013 at 7:47
  • @Mr.Wizard, I think it has been proposed before on Meta. But the answers argued that less-popular tags won't get enough reviewers (or something along that line). Commented Jul 27, 2013 at 7:48
  • 1
    @Old (1) If you run across the discussion you're thinking of let me know. (2) It would always be possible to move old, unreviewed edits to the main queue after a certain period of time.
    – Mr.Wizard
    Commented Jul 27, 2013 at 7:50
  • 1
    @Mr.Wizard We sort of already got that with the "Filter" link/button. But people aren't forced to use it because the majority of edits can be accurately judged by people with no domain knowledge. I don't know squat about Java or R or Mathematica, but I can tell if an edit positively improves the formatting of the answer, or if it just adds inline code formatting to random words. If it looks like an edge case, that's what the "Skip" button is made for. Commented Jul 27, 2013 at 7:50
  • @Mr.Wizard, here's the Meta post, it didn't get that much attention though. Commented Jul 27, 2013 at 7:52
  • 1
    I understand what you mean. But your statements saying that (1) there's no reason why I should add a second answer just to add more info and (2) focus more on contributing your own content in the form of answers are kinda against each other. In this case itself, I only have two choices - to add a comment to the original answer, and to add a new answer. But adding comments to improve an answer is often a bad idea because comments are usually seen as discussions, not part of answer - plus an OP may never improve the answer based on comments even if they are good suggestions.
    – ADTC
    Commented Jul 27, 2013 at 7:58
  • Now I want to contribute my own content on top of original answer as a new answer as per (2), but this would go against (1). It's easy to just ask the reviewers to only accept minor changes (but what about the 'too minor' reject reason?) but this leads to a site with multiple low quality answers lost in lack of vote counts, rather than a single complete comprehensive high quality answer highlighted by high vote counts and acceptance. I think it's better to ask reviewers to always reject edits of only minor changes and accept quality edit suggestions that improve the content.
    – ADTC
    Commented Jul 27, 2013 at 7:58
  • @CodyGray, the first two of your dump are about allowing reviewers themselves to choose which tags they want to review; the last one (the one I linked) is about the system restricting (reviewers have no say in this) what a reviewer is eligible to review based on his tag score. Commented Jul 27, 2013 at 8:00
  • @ADTC I realize that. My answer is basically conceding that the system is imperfect. Ideally, you should not have to do (1). But practically, you do—hence (2). Or, if you're more of a purist, just answer other questions that haven't already been answered. Then once you get the required rep, you can go back and improve the existing answers. And yes, just add a comment. Again, it's not ideal, but it works. If people have a problem understanding the answer or making it work, they'll look at the comments. Commented Jul 27, 2013 at 8:14
  • 1
    Furthermore, the problem is 'what constitutes minor'? We've tried to hash that out repeatedly here on Meta, and don't really get anywhere. It's subjective, there's no clear defining point for something that is 'too minor' and something that is 'substantial'. Moreover, even if we force 'substantial' edits, that's not necessarily the same as substantive edits. The difference is making significant improvements to style, formatting, and presentation, versus making significant improvements to the content. That's what I talk about in my answer here. Commented Jul 27, 2013 at 8:16
  • @Old Yes, I get that. My comment before that talks about the "Filter" button. I didn't realize you were going to come in and link-snipe me before I could get it posted! :-) Commented Jul 27, 2013 at 8:16
  • That's a great explanation, but I don't really agree with the proposed solution to avoid such suggested edits. I'm as fatigued and jaded a reviewer as it gets, and I do skip suggested edits that look like they're correcting a mistake if I'm not able to verify that they're correct. These are a very small fraction of the suggested edits, a vast majority do not require any subject knowledge. Commented Jul 31, 2013 at 12:51
  • 1
    @CodyGray False positives can be rolled back, and have a good chance of being noticed since the original author is notified and the thread is bumped. A user making many such mistakes will eventually be blocked and discouraged. False negatives get lost, and a user getting many such rejections will eventually discouraged and we'll lose a knowledgeable contributor. When it comes to edits that make corrections, bad rejections do more harm than bad accepts. The trade-off is different for formatting and copy editing edits. Commented Jul 31, 2013 at 13:00
2

@CodyGrat's answer is spot on, but I wanted to reinforce something that I don't think he highlighted enough.

The idea of editing vs commenting vs answering has been debated since the site was new and is still finding its niche (and thanks to Mr Wizard for finding these links):

Both suggest that editing to add content in encouraged. However, this seems to run against the current behavior (or does it?)

If you haven't noticed there is something that is not discussed or even mentioned at all in either post, suggested edits (although Joel Spolsky's answer does lay the foundation). That's because suggested edits didn't exist at all until January 2011. Before then, the only edits were full edits, not subject to peer review. At that time, editing to add relevant content was actively encouraged. This is also part of the idea behind Community Wiki posts, it lowered the editing bar to let anyone edit it/contribute to it and the post became a community owned post rather than one that "belongs" to the original poster.

Fundamentally, suggested edits have different unwritten rules, especially on Stack Overflow due to the volume of incoming posts, that have been refined over the years and what it comes down to is that "how to edit" is different than "how to suggest edits". Clear rules on reviewing edits hits on some of the key points from the reviewer side for suggested edits. Ultimately, your added content fails on several of these items, so they were right to be rejected based on the current rules.

The idea behind suggested edits is was to help moderators and the users with full edit privileges keep the site clean by cleaning up grammar, spelling, etc. Before suggested edits existed you only had three options, (1) flag, (2) comment, or (3) post on Meta Stack Overflow. This made it very difficult to fix posts that had grammar or spelling mistakes, in fact users were not encouraged to use (1) and (3) unless it was a serious problem. Suggested edits though were not intended to lower the bar for any kind of editing because edits need to be reviewed my multiple people to ensure they are correct.

But once you hit 2000 rep, then the "rules" relax as the community "trusts" you enough to know what you are doing and to make good edits. The kind of edits you want to make should wait until you hit the 2000 rep barrier and then they are more acceptable. Note that I said more acceptable as changing the meaning of posts through editing is still not encouraged, but adding some additional information or adding some relevant code is perfectly acceptable.

9
  • 1
    Why is the purpose of 'suggested edits' so different from that of 'full edit privilege'? I would have seen 'suggested edits' as a way for users below the 2000 rep bar to help improve the site by doing exactly what those with full edit privilege can do, but with peer moderation checking on their work. But now it seems this 'suggested edit' feature was merely added as an afterthought to ease the workload of correcting grammatical, formatting and spelling errors by offloading it to (and thereby taking unfair advantage of) new users who currently fail the 2k rep bar.
    – ADTC
    Commented Jul 27, 2013 at 9:10
  • @ADTC I'm not really following the taking unfair advantage of line of thinking, but you nailed the reasons. Suggested edits would probably not exist (or would be extremely limited) if it wasn't for the need to help fix grammar, spelling, formatting, etc. Commented Jul 27, 2013 at 9:15
  • @ADTC FYI - The reason I am encouraging you to post your updates in a new post is I think there is some merit to your alternative suggestion (your initial suggestion will get downvoted heavily). I'm not advocating the rules here, I'm just stating the facts that have been decided by the community. If a good way to permit any user to edit in this manner and still keep the site clean can be found, I would support it. Commented Jul 27, 2013 at 9:19
  • 2
    I nailed the reasons, but my question is why. I would say it's unfair because the reviewers are using a whole mass of people under the 2k bar by deluding them into thinking they can contribute good edits the same way those above 2k rep can, but in reality all they can do is make trivial changes - which surprisingly would get rejected too if they are too trivial. I will post the alternative as a new question some other time, but it would probably be flagged as a duplicate. :(
    – ADTC
    Commented Jul 27, 2013 at 9:27
  • 2
    It's a delusion because SO presents itself as a community site, but in fact it behaves as an ownership site. The thinking that a community site should encourage is here's a good question, let us work together to create a good answer to it, but the thinking I see encouraged here is take my answer. no, my answer is better. forget their answers, look at mine. can't you see my answer is a lot better than those other crap? I'm sorry, but I'm honestly disappointed that this is how SO functions - the fact that I have to comment or post a new answer just to add info is just plain sad.
    – ADTC
    Commented Jul 27, 2013 at 9:28
  • 2
    @ADTC Think about it this way: most users without edit privileges (i.e, under 2k reputation) are not able to positively improve existing answers. We allow them to suggest relatively innocuous formatting edits (although that feature is actually relatively new—before they couldn't do anything but comment) because those are easy to filter out. Granted, by doing it this way, we're robbing ourselves of a handful of contributions from the users who are the rare exception. But it's not long before those users earn the required 2k rep to edit. Then the site does work the way you think it should. Commented Jul 28, 2013 at 10:17
  • I get your point. But is the bar of 2k reasonable? I may not be as active as I should be to earn that much but I would like to think myself as being able to make valuable edit contributions to the site because it has proven very useful when I had many programming problems and I want to make it better for the next person. On the SO site, I have more than 75% of 2k to gain before I can get full edit privileges. It seems like I have to make it a full-time job to gain rep in order to get the privileges fast, and I find that quite over-bearing. Anyway thanks for your comments :)
    – ADTC
    Commented Jul 28, 2013 at 15:20
  • @ADTC there have been mentions of raising the bar in a few posts, but nothing substantial has ever been proposed. The level is a trade off between how much of a suggested edits review backlog can be tolerated and how low of a rep threshold is enough to know the rules of the site. If you want to propose something, feel free, but be prepared for downvotes. Commented Jul 28, 2013 at 15:35
  • 1
    There is no reason to hold suggested edits to a higher standard than edits from users with the edit privilege, except when it comes to minor edits (because they additionally waste the reviewers' time). It is perfectly fine for a suggested edit to go beyond spelling and grammar and correct minor mistakes that do not radically change the answer, just like direct edits. Commented Jul 31, 2013 at 12:48

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