What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 134 Stack Exchange communities.

This edit to a question was rejected by three reviewers.

The irony is that the edit is entirely correct, and was made by one of the most knowledgeable people in the universe on the topic - a core developer, @avital. While the reviewers were trying to be helpful, they unfortunately lacked specialized knowledge.

How can we fix this systemic problem? Here's the best idea that surfaced (thanks @bart):

Propose reviews on edits only to users who have at least one of the question's tags.

Other ideas are below:

  1. Promote a guideline advising reviewers to only deal in their specific matters of expertise. Knowing JavaScript doesn't mean you're an expert in meteor's package system. More simply put emphasize to reviewers,

    Do make use of that "Skip" button if you're not sure about an edit

  2. Providing a mechanism of notifying reviewers that their rejections were mistaken.

  3. Checking who the editor was. Rejecting this edit is like a LISP programmer telling Douglas Crockford he's wrong about a JavaScript statement.

The current default sentiment when a reviewer doesn't have specific knowledge regarding a question, seems to be to just reject the edit.

It's not an isolated incident, and I see it as a problem from multiple reasons:

  1. it discourages highly knowledgeable (but new to SO) individuals from contributing to SO
  2. it reinforces the sentiment that reviewers have absolute rejection powers, despite lacking specialized knowledge
  3. it perpetuates incorrect answers by rejecting useful changes.

How can we fix this?

share|improve this question
1  
That becomes problematic fairly fast though. Define "knowledgeable". And opening up the system for anybody to ping "wrong decision" would not be a great alternative. I've thought about this before from the angle of wrong acceptances. And my conclusion was to simply correct and move on. –  Bart Apr 13 '13 at 9:08
2  
That would require a user to know who they are. And thereby how knowledgeable they are. If you don't, the only thing you have to go on is rep. And beyond 2k for that knowledgeable user that becomes irrelevant anyway. –  Bart Apr 13 '13 at 9:14
1  
It's very easy to click avital's profile and see he's a core developer to meteor. –  Dan Dascalescu Apr 13 '13 at 9:17
3  
@Bart he may be intelligent guy who do senseless edit maybe stackoverflow.com/review/suggested-edits/923837 :P –  NullPoiиteя Apr 13 '13 at 9:17
3  
@DanDascalescu That edit is rather significant though. I don't think it should have been made in that way. Perhaps as an addition to the already available content. As an update notice perhaps. Or as a comment to the author of the answer. –  Bart Apr 13 '13 at 9:23
1  
@Bart: to clarify, the "senseless" edit mentioned by Null is an updated answer left by a core meteor developer, to an old, out-of-date, obsolete and now misleading answer originally left by another core developer. I am not aware of any other instance where a total outsider can essentially tell a core dev that they're wrong. How can I better explain this? It's as if Wozniak corrects something that Jobs said about a Mac OS API back in '99 because it changed in the meantime, but then a random SO editor comes and says "Nah, Jobs was actually right". Do you not see this as fundamentally wrong? –  Dan Dascalescu Apr 13 '13 at 9:28
4  
They can't tell him he is wrong. What seems to be community consensus however is that such edits are too much. If you're essentially replacing an entire answer, leave your own answer. Or comment to the OP of the answer. Or, at most, add it as an update notice to the existing answer. –  Bart Apr 13 '13 at 9:30
4  
You know he is a core developer. We need to go to his profile. And even then anyone can claim to be anyone. And anyone can claim to be nobody. This becomes a really difficult thing to judge. Have this user participate a little while longer and there will be no such problem any more. If there is really a serious problem here of a significant number of mistaken rejections, I'd love to know about it. I simply don't know. But sh*t happens. In both directions. As I said, my approach by now is to correct it and move on. –  Bart Apr 13 '13 at 9:36
2  
I agree with the OP that suggested edits, specifically for answers aren't working too well. If you only do formatting, it is too minor, and if you actually do something to clarify the answer, it is too radical. –  user213634 Apr 13 '13 at 9:40
4  
@DanDascalescu Depending on how you did that, you might well see that reversed then, marking you as a serial upvoter. Edit: yep, having a look at what you just did, that will just not take....not a smart move. –  Bart Apr 13 '13 at 9:53
3  
"It's very easy to click avital's profile and see" nothing, actually. There's no info about him I can see. Well, all his answers are tagged meteor, so I can infer that's where his interest lies. But whether he's just a run-of-the-mill not-completely-clueless user or a developer? –  Daniel Fischer Apr 13 '13 at 10:11
4  
@DanDascalescu So now the argument becomes, "You can easily see he's a core developer by leaving the review queue, going to his profile, seeing his Meteor participation, searching for his first name on Google, perhaps seeing his LinkedIn profile....". That doesn't leave a whole lot of that argument, now does it? –  Bart Apr 13 '13 at 10:24
1  
@DanDascalescu I am somehow rather sure that there's more than one avital in the whole wide world. So what would googling that bring? Clicking to the profile is already more than should be required to judge a suggested edit. –  Daniel Fischer Apr 13 '13 at 10:25
4  
@DanDascalescu your translation of the edit rejection as "a random SO editor comes and says "Nah, Jobs was actually right". " is wrong. Rejecting the edit says "don't use suggested edit to fix this mistake" NOT "that isn't a mistake". This is an important distinction to grasp. Nobody is saying the editor was wrong in their belief the answer needed to be fixed. A comment or a competing answer are better options when an answer is WRONG. Suggested edits are for formatting and language fixes to essentially CORRECT answers –  Kate Gregory Apr 13 '13 at 10:27
2  
Then you might want to make your request to primarily show suggested edit reviews for questions tagged with tags relevant to the reviewer's participation. But that might be happening already afaik. –  Bart Apr 13 '13 at 10:27

3 Answers 3

@Bart: why are we so stuck with rigid rules, instead of aiming for correct knowledge? Core developers of a project like meteor spend their time doing more productive things than learning SO's obscure guidelines.

The rules aren't particularly rigid. The problem, as you've noted in your question is a lack of domain knowledge in those approving edits. The flip side is, as Bart commented, there is no certain way to ascertain whether the edit suggester has this knowledge or not.

There are three solutions:

  1. Enforce domain knowledge for suggested edits - This doesn't work. By the time Stack Overflow has enough proof (important) of this, the suggester is past the 2k barrier anyway.
  2. Enforce domain knowledge for approving edits - This is ridiculous (I'm sorry but it is); 99%1. of suggested edits do not need domain knowledge in order to ascertain whether they are correct or not and of the remainder, 0.99% are in languages like C# where there are thousands of people with that knowledge on Stack Overflow. Would you massively halt the improvement of posts just because of a few suggested edits getting rejected?
  3. What we have at the moment - Be slightly stricter than is necessary in the 0.01% of cases where something like this happens and to give the editor a few other ways of getting the information out there like commenting and answering.

Point 3 takes me onto the remainder of your comment "Core developers of a project like meteor spend their time doing more productive things than learning SO's obscure guidelines". Maybe so, but if they're going to contribute to a community it's polite to learn something of that community first.

The editor in question has over 50 rep; is perfectly capable of commenting or adding another answer. There is nothing stopping them contributing. They are being stopped from contributing in their chosen manner; and they're obviously not looking to see whether these edits get approved or not and attempting to modify their behaviour.


tl;dr

This "problem" is extremely rare and there are already 2 work arounds: commenting and answering. It's not a problem.

Have you commented and asked the editor to maybe move to comments and answers instead?

1. Numbers are made up.

share|improve this answer

I am a C++ expert and I have written dozens of books on C++. If I go into someone else's C++ answer and suggest an edit that fixes their technical issues, that is NOT a valid suggested edit. My C++ expertise is not the issue and reviewers don't need C++ expertise to evaluate edits to C++ questions. You're not supposed to fix technical errors in code with a suggested edit.

Things like changing "a module" to "an npm module" are too minor. Did anyone reading the answer suddenly wonder if it was a Perl module or a Lunar Landing module? If you have full edit rights you can tweak that sort of stuff, but it's not a good use of the suggested edit facility.

There's really no room in the edit review mechanism for "ah, this edit suggester has tremendous technical knowledge in the subject matter." That's just not relevant. Suggested edits are supposed to fix formatting, grammar, spelling, and the like without changing the meaning, and they are supposed to fix all the errors in the post. If there are only a few tiny errors in a post an edit should not be suggested.

Where does technical expertise come in? Recognizing a keyword and formatting it as code. Spotting a typo in a keyword or code (not an error of thought like saying int when it's really unsigned int but a typo like nit) and fixing it. Knowing how to safely break lines in code without changing meaning. Labelling code fragments ("this is the HTML:") when the OP did not. Fixing links to documentation that has moved. Removing irrelevant lines of code from a question that are obscuring the real problem. And very little else, if anything.

share|improve this answer
2  
Regarding the first bold sentence, I'd say that obvious typos are exempt (in answers). So if I mistype fromIntergal, changing that to fromIntegral would be a valid edit. Likewise tempalte -> template in C++. –  Daniel Fischer Apr 13 '13 at 9:39
    
@DanielFischer agreed, edited last para to reflect that exception –  Kate Gregory Apr 13 '13 at 10:22

What @Avital edited is

  1. package.json => package.js which is radical change
  2. To use a module To To use an npm module sounds good
  3. NPM package to npm module which is radical change

and overall Too minor and radical change

so @Avital should comment to know about this to OP instead doing radical change

share|improve this answer
2  
That is more of an advisable "sure to be seen" approach though, rather than that the edit was necessarily wrong. Though I can understand the reviewers following exactly your points. I might have done the same. –  Bart Apr 13 '13 at 9:07
    
@bart: this is the problem - assuming you can tell what's a radical change and what's not. Sometimes we need to admit where our knowledge ends. Clicking on avital's profile is a very easy way to see he's a meteor developer. Once one does that, they should respectfully back away. Perhaps reviewers should look at who the editor is. –  Dan Dascalescu Apr 13 '13 at 9:09
    
I'm not sure I follow the "and overall Too minor and radical change" part. Can you please clarify that? –  Dan Dascalescu Apr 13 '13 at 9:11
2  
@DanDascalescu That is not really a feasible approach though. And (though I can't back this up with figures) we seem to have far bigger problems with wrong acceptances than mistaken rejections. It's unfortunate, but I'm not sure it's a problem requiring a solution. –  Bart Apr 13 '13 at 9:11
    
@DanDascalescu .json to .js and change meaning completely ... –  NullPoiиteя Apr 13 '13 at 9:13
    
@Bart: I see it as a problem from multiple reasons: 1) it discourages highly knowledgeable individuals from contributing to SO; 2) it reinforces the sentiment that reviewers have absolute rejection powers, despite lacking specialized knowledge; and 3) it perpetuates incorrect answers by rejecting useful changes. –  Dan Dascalescu Apr 13 '13 at 9:13
    
What do you mean by ".json to .js and change meaning completely ..."? I'm sorry, I know we had this problem before, and I'm trying to understand you. Can you form a coherent and complete sentence? –  Dan Dascalescu Apr 13 '13 at 9:18
    
Whether or not the edit should have been accepted or rejected, I'm having trouble understanding why it was rejected as too minor. Usually that applies when someone is making a trivial edit that would bump an old post or ignores other issues in the post. In this case, there aren't really any other issues in the post. Changing the package name so the code works would also not be a trivial edit (but may arguably be a "radical change"). I'm also not sure I see why you think #3 is radical since it's just making the terminology consistent with the rest of the post, only #1 looks controversial. –  Troyen Apr 13 '13 at 21:45

You must log in to answer this question.

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