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.

I recently posted an answer in which I used the little-known jQuery .one function. The question asker then edited my post to use the (incorrect, but similar) .on function instead, and that edit was unanimously accepted by three edit reviewers who did not have any substantial jQuery experience (though note the question was tagged CSS, and not jQuery).

While I understand the confusion about the answer, I am a bit concerned that this sort of well-intentioned "correction" could happen to other posts too. (I've already reverted the edit in question and added a clarifying note).

Given the typical length of the edit review queue (i.e. very short), I'm wondering if it might make sense to restrict (or preferentially weight) the reviewers. Specifically, would it be reasonable to give reviewers with substantial related tag experience higher weight (or even exclusive access) to suggested edits? It may also help reduce the "serial acceptance" issue currently seen with the review queue.

share|improve this question
1  
I have been watching the suggested edits come into the mod queue for awhile now and, in the balance, I think the review queues are doing far more good than harm. The bad edits seem like a problem because you're hearing about them here on Meta, but the fact is, there really aren't that many, relative to the number of flags passing through the review queues. And you're never going to get the error rate down to zero. –  Robert Harvey Oct 26 '12 at 7:04
14  
Note that people really shouldn't be changing code with suggested edits, unless they're absolutely certain that the change respects the intent of the author. It's better to put a comment below the post, and allow the author to decide for himself whether or not the change is appropriate. –  Robert Harvey Oct 26 '12 at 7:08

5 Answers 5

up vote 19 down vote accepted

If an edit harms your post in any way (which it seems to have done in this case) by all means feel free to roll back the edit. That's what the functionality is there for.

share|improve this answer
6  
Not only there are rollbacks, but we're also notified when our own posts are edited. –  bfavaretto Oct 25 '12 at 20:55

I wish I'd recorded video of the screen, but I've watched the Review overview and seen the same two users churning through the queues, spending about 8 seconds per post. (Alas, SEDE is currently over 4 months out of date, so it's useless for scraping up the "Science" on the new review queues.)

I propose that no review action be allowed (except Skip) until a time interval has passed since the reviewer has seen the question. The interval should be random, to make scripting/gaming harder. Maybe between 23 and 66 seconds.?.

share|improve this answer
5  
Skip should always be available. There are edits I immediately know are outside of my expertise, or beyond the scope I care to deal with. –  Jason Sturges Oct 26 '12 at 1:51
2  
@JasonSturges, Good point. Edited answer. –  Awesome Poodles Oct 26 '12 at 1:52
    
This was suggested under the "current system encourages fake reviews" post, but its OP deleted it. –  Pops Oct 26 '12 at 19:18
    
Oh, I totally forgot the second part of my comment. Plenty of legit reviews take less than 66 seconds to complete. Even if this doesn't hurt as much as it helps, it'll still hurt a decent amount –  Pops Oct 26 '12 at 19:45

That would only work if there were enough items in the queue that it would spit out something in one of "your tags" instead of something else (which, by the way, it already does do when it's given a choice). When there are only a few items in the queue it doesn't (often) have the option of giving you something in one of "your tags".

If you change it so that rather than weighting your tags it actually restricts tags you don't have sufficient activity in then it will be a major problem for suggested edits to posts in low population tags. That, and there's the fact that most suggested edits really shouldn't require knowledge of the relevant tag. Code changes such as this should just be rejected as an "invalid edit" to begin with. I don't need to know anything about jQuery to do that.

share|improve this answer
    
I guess my point is that it's less likely for someone to blindly accept a change if they have specific expertise in that area. –  nneonneo Oct 25 '12 at 20:27
    
@nneonneo Why would anyone need to know anything about CSS or jQuery to know that you shouldn't be editing the substance of the code of someone's answer, correct or not? If it's wrong it should be downvoted or commented on; only the OP should be making a code change such as this (barring exceptional circumstances, such as a comment by the OP indicating it should be changed). –  Servy Oct 25 '12 at 20:28
5  
@nneonneo Nobody should change the code shown in an answer written by another user. If there is a typo, or the code is wrong in any way, users should leave a comment, pointing out the error. It is then the author that should edit her/his answer. –  kiamlaluno Oct 25 '12 at 20:44
    
@kiamlaluno: Seasoned users, of course, should know this. But, novice users (who would be the ones in the suggested edits queue) may not, and those are the edits that are getting through and potentially causing problems. The problem is exacerbated by some of the blind acceptance issues on the edit queue. Reviewers should be rejecting code changes, but they aren't. –  nneonneo Oct 25 '12 at 22:12
    
@nneonneo The problem is accepting suggested edits that should not, not the suggested edits. If something should be done, it would be avoiding such edits are accepted. –  kiamlaluno Oct 25 '12 at 22:36
    
@nneonneo The problem is most certainly that reviewers are accepting edits that they shouldn't be; it's not the fault of the suggestor. The issue is that someone who would incorrectly accept such an edit is not going to do any better at reviewing it if they have answered questions in that tag before. –  Servy Oct 25 '12 at 23:46

A vast majority of suggested edits do not require any deep subject knowledge. Most only modify generally-understandable aspects of the post such as the English, presentation, formatting, title, tags... For those few edits that purport to be a factual correction, but for which it is not immediately clear that the edit is an improvement, a responsible reviewer will click “Skip”. An irresponsible reviewer will click “Approve”, but they would do that even if they happen to have tag knowledge anyway. So restricting reviewers to certain tags would have no impact on serial approvers.

In my experience as a reviewer on Stack Overflow, it is rare that I see an edit that I'm not sure about, even though I am unfamiliar with most of SO's top tags. I don't feel any need for a tag filter. (In contrast, I found the close queue unusable until the tag filter was added.)

Restricting reviewers to tags that they have reputation in would be a show-stopper in smaller tags that don't have users who both do suggested edit reviews and have whatever threshold you set. If it's merely a bias and not a ban, it would either have no impact if the review queue is most often empty, or could potentially delay the suggestions in smaller tags significantly if the review queue was most often full as it once was. Furthermore, users with 2000 reputation are able to edit posts regardless of any tag-specific reputation. All in all your proposal would have very little benefit if any, but a nontrivial cost in implementation and a potential hindrance in review speed.

There is a precedent of sorts: when tag wikis were introduced, in order to edit a tag wiki, you had to have a minimum reputation or be one of the top answerers in that tag (which took care of smaller tags). The system was abandoned in favor of the current system where any registered user can suggest an edit and any high-reputation user can review.

share|improve this answer

I'm thinking the rubber-stamping of bad edits wouldn't be solved by weighting based on tag experience. Keep in mind it is almost entirely being done by people checking the review queue, meaning they already have a not-insignificant amount of rep. If they're not picky about what questions they throw answers at, i don't see setting a threshold high enough to bar them, but low enough to make it worth putting in the queue in the first place.

As for the edits in question, if you think they make the post wrong in some way, revert them.

share|improve this answer

You must log in to answer this question.

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