Improvements to a suggested edit cannot be submitted when the suggestion is approved during editing

Question

I was reviewing a suggested edit and decided that while the edit was incorrect, there were things to improve about the post. So I clicked “Improve” and set about my business.

When I tried to submit my edit, I got the following error message, telling me that the post had been edited in the meantime by the suggester:

User already edited the body of this post; your edit must be more substantive to override the current edit.

When I tried later on a tag wiki, I got the altogether unhelpful indication that “an error has occurred”, and again no way to do anything but cancel and lose my work.

But refreshing the page brought me to the suggested edit page, showing the suggested edit as approved, with no trace of my editing. (Fortunately I'd saved my text on the side!) With a later suggested edit on a tag wiki, I was marked as having reviewed the suggestion, with the indication “edit” — but there was otherwise no trace of my alleged edit.

So, if you improve a suggested edit and it gets approved in the meantime, you get into a situation where

• there is no link to the post;
• if you go back, you get to what you were doing before and lose your edit;
• if you reload the page as instructed, you lose your edit.

The bug here is that unless you take extra precautions such as copy-pasting your edit into an external editor, you will lose your work.

Expected behavior

Unlike normal edit conflicts, an edit made to a suggested edit is guaranteed to be a derivative of the suggested edit - so if that edit was approved and then an "improved" version is submitted, that "improvement" edit should be applied to the post (the helpful status indicated by the checkbox shouldn't affect anything though).

When two different reviewers both opt to improve a suggested edit, a conflict is unavoidable - this should be resolved based on the size of the edit, as is the normal behavior.

Answer 1

Improving a suggested edit is now considered more substantive than an approved suggested edit for the period of 10 minutes after the suggested edit has been approved. There can however still be a conflict if 2 reviewers submit an improvement after the suggested edit has been approved - in that case we fall back to the standard edit conflict resolution.

Implemented in build 2014.2.3.1913 on meta and 2014.2.3.1343 on sites.

Editor's Note: Since this request was filed, there is now a period of exclusivity of 3 minutes when users get a suggested edit review task. In other words, for the first three minutes after you load a review task, the same review task will not be shown to others, so this situation should occur much less often. There are still some edge cases (e.g. manual navigations, accessing the suggested edit from the post and not the review queue, from the notification on suggestions to one's own posts, etc.), but those are rare.

Answer 2

Normally when you're editing and someone else makes an edit to the same post, you get a nice orange box saying the post was already edited, and it makes you refresh the post to see the updated version (if your edit isn't "more substantial").

When reviewing, often crappy, suggested edits and opting to improve, you only seem to get a very, very small timeframe to complete your improvements before it gets approved by ignorant reviewers.

At this point in the process, when their edit gets approved, the post thinks it's now been edited and won't let me submit my improvements anymore (I don't lose my edit like you claim). This is just plain stupid, as I already know what improvements they made. Thanks for telling me that ignorant people just approved the edit that I was going to mark as not helpful, now why are you preventing my edit from being submitted?

My specific case I just encountered was on this post where the user only suggested an edit to the title. It wasn't a great edit and definitely deserved to be rejected as too minor, as clearly there was tons of other stuff to fix in that post. So I finish my improvements and attempt to submit, I'm still greeted with the orange box. How is my massive edit that changes almost everything in the body, plus more changes to the title not considered more substantial than this title-only edit that I've already reviewed and determined needed to be improved, hence I clicked the Improve button.

TLDR;

When I'm improving a post, it shouldn't forbid me from continuing to apply my edit anyways just because it got approved. As stated, I've already seen that edit. I know exactly what they changed, and I know I wanted to improve it. Please stop making it more inconvenient to improve crappy edits.

Answer 3

The main problem I see with the current implementation is that it discourages reviewers to improve posts. This is because it is very likely that other reviewers will accept or reject a post while you are spending time improving it, especially on sites like SO with many concurrent reviewers. The «punishment» in form of lost work or increased burden is conditioning reviewers to stop improving and instead directly accept or reject suggested edits. The pressure to not improve is further increased by the gamification, which rewards reviewers with badges based on the number of reviews they make. And finally, reviewer bias towards rejecting instead of improving may discourage the people who suggested the edits from editing in the first place. I don't think this is what we want.

A solution would make it as easy as possible to successfully resolve the «accept/reject while improving» conflict, as not to «punish» reviewers for improving. For most cases, the resolution could/should be automatic, as suggested by others.

Answer 4

Stack Overflow should create an automatic merging feature when edits are made simultaneously. This means that if different lines are edited, it could merge them into a new revision. If the same lines are edited, the system could attempt to identify if different sentences in the same line is edited, and merge in these cases.

I mean, we expect changes to source code to merge cleanly usually, why not this?

In cases where a conflict occurs, a new screen should show what has been committed by a previous editor in a form (with parts which can be merged automatically merged), along with what you have proposed, along with a diff view of any conflict locations (similar to the reviewing suggested edits).

That would mean that most conflicts can be resolved automatically (perhaps with a notification that a merge has been made), and even for conflicts which cannot be resolved, no work is lost.

Answer 5

I've encountered this issue many times (I was thinking of posting it here), and what's especially frustrating is that it seems to be caused by the post being approved rather then just edited, and quite often it probably should not have been approved at all.

Implementing a CheckIn/lock process for reviewing should resolve this issue as well as encourage more thorough reviews as opposed to fastest gun in the west style reviews.

Answer 6

The simple solution is that if a 2k+ editor has done an improve (and I was thinking you had to be 2k to get an improve option), then just have their edit go through, overwriting the approved suggested edit, just as if they had gone to the question and hit the edit button.

That dialog stopping the process should only appear if someone has edited the post after the suggested edit was approved (which might happen if the editor took a ton of time to do his edit).

Answer 7

Stack exchange should create locks on the posts. It should allow only one user to edit the same post at the same time.

Once a user is editing a post, then any other user that tries to edit the same post, should be notified that the post is already locked by another user and you could edit the post, when the current user is done with editing.

Answer 8

Personally, I think you should be allowed to edit your own questions no matter what (so it overrides any other edits). Your edit should take priority. If you feel that your question needs editing, surely it should be your responsibility. In your case, I think the other person should be notified of your editing, then have to wait until you've finished, then check your edit.

Answer 9

Yes, I agree it is really frustrating after you make a lot of edits and you got the alert.

I would suggest whenever a person edits a question and the post got already updated, we can show a Modal-popup with the message.

This post has already edited and now its content is given below. If you still feels that your review is more helpful then submit your edit

In the popup we also show the current edit content and if user after reviewing the post submit own edit, then it should get priority.

We can solve the problem of losing edited content and save the steps of copy-paste by this.

Answer 10

I agree with animuson, so I propose the following:

• Add a grace period of 5-10 minutes. People that have the review already open can vote, others can't.
• Because we could get 4 rejects after the edit already has 3 accept, I suggest changing the rules for accepting/rejecting a post in the following way:
  • The minimum difference of the majority needs to be 2 (or maybe 3 on SO)
  • The review is removed from the review queque once it reaches the that thereshold.
    • Don't display the "edit (1)" button that leads to the review.
    • The grace period starts now.
    • If the outcome was accept, show it?
  • If the thereshold changes in the grace period, the suggested edit review is added back to the queue.
  • A vote from an moderator/the OP is binding. Don't use the grace period in that case.
  • An improvement let the Community user cast a vote (like before). I assume that only people that want to improve the post do that and that improving an edit is too much work for the robo-reviewers.
  • A moderator and the OP can also cast a binding vote on the review AFTER the grace period until an other edit has been done. An improvement count as an other edit.

The effect of this is:

• You have at least 10 min time to review/improve the post
• Controverse edits are reviewed by more people (the suggested edits queue is almost always empty, so it does not sound like a problem)
• It allowes the OP (and ♦) to override the outcome of the suggested edit review.

But when to grant the +2 rep for the suggestor?

• I suggest after reaching the thereshold with the outcome accept.
• Revoke the reputation if the review does not longer hold the required thereshold.