Recently, I've come across a few questions which had suggested edits. I looked over the edits, and either accepted or rejected them.

Now, for some reason edits accepted or rejected by me require additional people, whereas edits made by me do not. I always found that system a bit odd, but that's not the point of this question. Just pointing it out. :)

The point is that my edits are in the queue. No problem there. Unfortunately, sometimes I will look back at the edits and want to improve them, or realize I missed something and want to reject it. I can't do any of these things. My initial decision is locked in and I can't roll it back until others weigh in.

An extreme and recent example is this question. It looks fine - if a bit simplistic. A low-rep user recently suggested an edit. To me, it looked like he was removing a space after the question mark. Picky, but nothing wrong with it. I accepted the edits.

Then I clicked the edit button again, and looked at the markup. This is what met my horrified eyes:

He is one of two supporting characters that die in my screenplay.

My wife says I shouldn’t make him black if he’s going to die.

Would it be politically incorrect?


The additional lines were comments - I removed the !'s to make them visible here.

I frantically tried to rollback my edits, but of course there is no way to do that. My accept vote was locked in, and sure enough, this morning, I found the edits to be approved (probably because someone else didn't look at the markup either). I have since deleted them as they are purely inflammatory commentary.

Is it possible to add in a button for us to rollback our edit decisions from the queue if we make a mistake?

2 Answers 2


For the "I thought of something to improve" cases, where no actual harm is being done in the meantime, I think the solution is just to wait. Modifying edits that are already in the review queue, that somebody could be looking at right this moment, seems like it would have some complex interactions.

However, the specific case that you described in this question should have been detectable much earlier. If you look at the formatted text rather than the raw markdown, you'll never see HTML comments. (There are other things you also might not notice; I've seen spammers drop a link on a punctuation mark, which is a harder problem.)

When showing edits in the queue the system will occasionally add a warning like "hey watch out, this could be spam". I don't know what rules govern that, but it happens. I suggest that when there is commented-out text, and maybe if there is link text shorter than N characters (3-5?), the system also display a warning like "This post might contain hidden text in comments or links; please check the markdown". If you'd seen such a warning you never would have approved that edit in the first place. Let's try to fix the problem at the earliest point we can.

Update: I just requested this feature on Meta.SE.


I understand your feeling, but I don't see this as a problem.

Edit is completely reversible.

If you've approved an edit that should've not been approved in the first place and it's pending another approval/reject, leave a comment below the post stating the edit might should be rolled back ASAP. The next person with edit privilege who see this can rollback the edit.

Additionally, flag for moderator attention for immediate rollback, especially if the edit is rude/abusive.

  • It seems like those are patches for a problem, not an actual fix though. Wouldn't it be far simpler to just let us be able to rollback our edits? I mean to me, I don't see any negative repercussions, and it simplifies the process you outlined above. Commented Dec 8, 2017 at 5:32
  • @ThomasMyron I don't see it as a problem, because I think hidden malicious edit is rarely happen, and most of the time reviewer is able to catch that.
    – Vylix
    Commented Dec 8, 2017 at 5:46
  • I suppose that's entirely true. Commented Dec 8, 2017 at 15:59

You must log in to answer this question.

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