You make a very good point about edits getting

not only rampantly suggested, but also rampantly approved

but I think that your idea of

1. Mark[ing] the edit as rejected [...]

2. Revok[ing] the +2 rep bonus [...]

doesn't address

the real problem

which is not the low reputation users suggesting bad edits, but the high reputation users approving them.

This is why I came up with the following

COUNTER PROPOSAL

which consists in

• sending a notification to a user if a suggested edit that has been approved by them (or a certain number of edits) get rolled back.

• temporarily suspending the approving privileges of a user if too many of their approved edits have been rolled back.

The content of this post has been sanitized in response to anonymous feedback that the formatting was making it completely unreadable by assistive technology (AT). I strongly value making content accessible over proving a point against users who willfully contribute to making content inaccessible. Read this post as it was originally intended to be read, at your own peril.

Some caveats and other concerns:

• As people blindly approve these edits, people may just as blindly roll them back. In a more serious case, one could also use one's edit privileges to serially roll back a user's suggested edits, thereby making it basically a different flavor of revenge downvoting — and one much harder to reverse. I don't like the sound of that...

• If we limit who can perform such rollbacks, who exactly do we limit it to? Users of a certain reputation level? Moderators? Someone else?

• What about the users who are approving these edits? How do we address them?

In fact, now that I think about it, I should agree that the problem probably lies in these users who are blindly approving these edits in the first place, more so than the edits themselves which could have been rejected just as easily.

I disagree. The fact that 2 reviewers have voted to approve the edit should (note, should) indicate that the edit is acceptable.

It would be wrong for one user (note, I accept it's different for you as a moderator) to be able to override the decision of those two users, and to undermine them by rolling back the edit.

What I propose instead is;

1. Rather than simply needing two approval votes to be approved, an edit is only accepted if the number of approval votes is at-least 2 greater than the number of rejection votes.

   e.g, if one user rejects the edit, the edit then needs 3 approval votes.

   This gets more pairs of eyes on the controversial edits to make sure the right decision is made.

2. Better information to the editors about what should be accepted. This problem is not restricted to backticks, nor is it restricted to approvals-that-shouldn't-have-been. I see rejected edits that should have been approved and vice-versa all the time.

Instead of concentrating on rollbacks reversing the rep gains from edits that have already been made (which I agree pretty much needs to be done by diamond moderators; it would be silly to require 3 reject votes in the suggested edits queue by ordinary users to block an edit, but then allow a single non-mod to overrule the approve votes), the system should concentrate on weeding out the reviewers who approve these in the first place.

How? Audits. We've all seen audits designed to catch people who aren't paying attention at all and who won't notice blatant vandalism. We also need audits that are clearly under the "too minor" category. Throw purported edits where some backticks are placed around a couple of words into the audits. Include some fixes to a single misspelled word. These problems will disappear quite quickly.

Instead of introducing a new feature, I think it is better to educate both the editor and the reviewer - for the first time they gain privilege to edit/review. I doubt people check meta frequently, and I doubt people know what to check for when they edit/review for the first time.

We can give specific example of how to transform a badly formatted post to a clean post. There are enough suggested edits on SO to pick out one example. Or we can just create an example on our own.

The guide for editor/reviewer should mention:

• Salutation
• Tagging
• Various stuffs about formatting code (indentation, syntax highlighting, <pre>/4-space/back-tick)
• Quoting
• Do not overuse formatting
• Substantial edit
• (Esp. tag wiki) Do not copy content.
• Visit meta site for changes in policy.

The guide should be paginated - something look like cartoon, so that reader will not be overwhelmed. Each of the page can contain a link to the relevant part in the FAQs or relevant meta question.

Despite the fact that this is a reviewer problem, as all the other answers seem to point out against this feature, I would love to see your request implemented, given the sheer number of bad suggested edit reviewers, the inefficacy of audits (compared to the other review queues), and the effect rejection rates have on suggested edit bans (though being able to add those manually would be fantastic too). Abuse isn’t much of a problem, either; rolling back the rollback can be taken into account, for example.

To sum up: a rollback is usually well-thought-out; it being prioritized over queue reviews is fine.