5

One of the special things that makes Stack Exchange different from sites like Reddit or countless discussion forums is the ability of anyone (with a minimum of rep) to suggest improvements to any post. And many do. But occasionally, whether simply due to ignorance or to deliberately game the system, some users suggest edits that don't really improve the post being edited.

Fortunately, we do have a review system that usually — though, alas, not always — catches such mistakes. But when it does... nothing happens. The edit just disappears, vanishing into the dusty bowels of the system.

This has two harmful consequences:

  1. Users who make poor edits by mistake or from ignorance get no feedback, and so never learn to make better edits.

    At best, those users may notice that the post they tried to edit hasn't changed, or that they didn't receive the expected +2 rep (if they know about that). But since there's no message of any kind telling them that their edit was rejected, they may assume that it was simply a random software glitch and keep trying to make the edit again until a) they give up in frustration, b) the edit gets approved by mistake, or c) a frustrated reviewer contacts them personally to tell them to stop it. I've been that reviewer several times.

  2. For users who try to game the system, there's no disincentive to making a huge number of crap edits (like, say, adding random backticks to posts) in the hope that some of them will stick. This wastes reviewer time, and may degrade the quality of the site if some of the edits get incorrectly approved.

To fix both of these problems, I'd like to propose the following feature:

Make rejected edits give -1 rep to the user who suggested them.

In particular, this rep loss should be reported in the same way as rep loss from downvotes is currently reported, and the report entry should link to the rejected edit suggestion (/review/suggested-edits/number), or (better yet) to a new page that would list the reasons given for the rejection, and would also provide additional advice on how to make better edits and what to do if one disagrees with the rejection.

A -1 rep penalty on rejected edits would not actually have much direct effect on anyone's rep score; a "robo-editor" could still break even, as long as at least 34% of their edits were good enough to be accepted. But to editors who are not completely cynical, it would create a powerful psychological feedback effect, encouraging them to find out why their edit was rejected and to adjust their editing habits to minimize rejections.

Perhaps more importantly, however, for many new editors, the mere fact that the rejection was reported at all in the achievements drop-down would be a great help. By tying the notification to a rep change, it would fit neatly into the existing Stack Exchange notification system, while linking it to the review page would, for the first time, actually let editors see the feedback provided by reviewers.


Ps. In case you think this suggestions would be too harsh and likely to discourage even useful edits, I wouldn't object to simultaneously increasing the rep rewards for accepted edits to, say, +3 or even +5 rep. The latter option would make the rep changes from suggested edits equivalent to those from question up/downvotes, and ought to encourage more good edits.

Also, obviously, I'm not suggesting that a change like this be made simply by jumping into the deep end and hoping for the best. Rather, it should be e.g. first rolled out on a subset of the SE sites for, say, a one-month trial period to see what the effect would be. If, as I hope, this change would decrease the number of poor edits and increase the number of good ones, it could then be more widely deployed; if not, it could always be rolled back.

Pps. Yes, I'm aware that both parts of this feature request (rep penalty and notification) have been suggested before. I don't think the combination has been, though, even if, in hindsight at least, it's kind of obvious.

  • As a practical example I disagree with several points you've made; but I haven't downvoted because you've made your case in a clear fashion and I can understand much of the reasoning even if I ultimately disagree; the same resistance to reject could occure if this was implemented – Richard Tingle Dec 29 '13 at 1:28
  • I have used the Custom reject reason to communicate to the editor. No way to know they read it. I also do go to the posts and comment. This does seem to work at times. – Bill Woodger Dec 29 '13 at 1:32
  • @Pompous: I've done both too, but IME, the only people who seem to even know that there is such a thing as custom reject reasons are a) those with enough rep to have participated in edit review themselves, and b) those who have had a reviewer contact them by other means and explain why their edits are "getting lost". – Ilmari Karonen Dec 29 '13 at 1:50
  • 1
    I'd use the "Custom" reject reason more often if I didn't know that someone else is going to approve the thing while I'm still composing a thoughtful message – Pëkka Dec 29 '13 at 3:33
26

Rejected edits do have a consequence; sufficient bad edits will land you with an edit ban.

However feedback is always a good thing and I would support increased feedback. Expecially as failed reviewers cannot be @replied to to give manual feedback directly on the edited post itself.

However, a rep penalty may have the opposite effect to what you hope. If a rejected edit means -1 an edit reviewer may be more kind to mediocre posts and skip rather than reject (leaving the robo reviewers to approve)

  • Re: unintended consequences, that's possible, I suppose. I don't see it as a particularly likely outcome, but one can't really say anything definitive without at least trying it on a small scale first. Besides, I'm not particularly worried about mediocre edits (like, say, "too minor") being approved, especially if reviewer workload is decreased anyway. It's the edits that actually make posts worse that are the real problem. – Ilmari Karonen Dec 29 '13 at 1:10
  • I don't recall rep changes ever undertaken on a "trial" basis. – Robert Harvey Dec 29 '13 at 1:14
  • @TheGrinch: Do you know if there's an architectural reason for that (as in, would it mess up rep calculations somehow), or is it "just not done here"? – Ilmari Karonen Dec 29 '13 at 1:22
  • @Ilmari would seem a little unfair if some people were under one set of rep rules and other people were under different rules – Richard Tingle Dec 29 '13 at 1:26
  • @Richard: I was suggesting that the trial be limited either by time and/or by site, so everyone on a given site would still be playing by the same rules. Between sites, there are already huge differences in practical rep gains due to things like different voting practices and sheer visitor volume (typically, older and bigger sites are easier to gain rep on, although there are exceptions like Academia), even if the nominal scores are the same, so a brief trial like I suggested would not really increase the inequality. – Ilmari Karonen Dec 29 '13 at 1:33
10

I like the idea (I actually made a post on the very same subject myself a few days ago), but penalizing them with reputation subtraction is not a good idea.

I can see several issues with that:

  • Incorrect judgement by the community subtracts reputation when it shouldn't (with the amount of poor reviewers, this is a real issue).
  • As Richard remarks: friendly reviewers will go out of their way to prevent a reject, leaving robo-reviewers free play.
  • Inflated reputation: if you give +3 or +5 rep for accepted edits the overall worth of reputation is going down since it's a lot easier to receive reputation.

What instead should happen is some sort of warning system. When users are getting their edits rejected too much they should (gradually) receive warnings AND positive notifications when they're heading back the right direction.

I have seen the argument about displaying negative notifications, but at that point we should ask ourselves the question: do we really want to do nothing when their decisions are affecting the community negatively? It might not be severe, but a friendly warning message isn't that intruding either.

This will allow users to revise their editing behaviour earlier on instead of just getting an edit ban out of nowhere without any indication why. Likewise adding the positive notifications will reinforce their behaviour and in its turn provide feedback that they're doing better.

What's key here is feedback: there should be a system that allows you an overview of your edits (both for reviewers and proposers) on a regular basis where you can review your decisions. They can serve as a tool for people to learn from, possibly with a few links to relevant meta posts for each rejection reason as further explanation.

2

In the position of Stack Exchange, you don't want the user to have negative feedback. In the words of Jeff:

I understand the desire to educate, but there's something deeply wrong about the proposed design.

The net effect is the user seeing this broadcast in their face in the most obtrusive way we can:

You've done something wrong. Click here to learn why you suck.

You should only send the user obtrusive 'in-your-face' messages about how awesome they are.

Messages about non-awesomeness should always be delivered quietly, via a backchannel, in a way that minimizes their impact. Otherwise, we're slapping them in the face.

And I agree with him. Sure, something needs to be done to tell people if they're messing up. But for one edit rejection I think it's unnecessary. Maybe we could have a notification displayed to the user if x number of edits are rejected, somehow, but unobtrusively. Otherwise, it's too much in a person's face that they messed up.

In regards to your second point on rep-loss, that, I think, would have a negative effect on a user. It also would make people second-guess Rejecting an edit, since it is removing rep from a user. It makes people less likely to edit posts, given the fact that they have to second-guess the edit they're going to make.

I would suggest making a system in which users who've had a percentage of edits rejected (say 70% of 10, or something like that) be warned about this. Maybe there could be an extra pop-up shown when they are editing, or, possibly even editing suspensions. However, I don't think that rep-loss and the in-your-face notifications are good to tell a user about their bad edits.

  • 3
    We do notify users when their posts are downvoted, although we try to keep the notification as gentle and "not in your face" as possible. I'm suggesting to make use of that same notification method for edit rejections. – Ilmari Karonen Dec 29 '13 at 1:19
  • @IlmariKaronen What do you mean, in terms of rep? – hichris123 Dec 29 '13 at 1:23
  • Sorry, I don't understand your question. What do you mean by "What do you mean, in terms of rep?" (And please don't ask me what I mean by that. ;-) – Ilmari Karonen Dec 29 '13 at 1:26
  • 1
    @IlmariKaronen What do you mean by We do notify users when their posts are downvoted? The only way I can think of a user getting notified that they got downvoted is them loosing rep. – hichris123 Dec 29 '13 at 1:28
  • ...which gets shown both in the top-bar achievements menu (although there's no in-your-face alert for negative rep gains, they're still listed) and in the user profile. – Ilmari Karonen Dec 29 '13 at 1:36

You must log in to answer this question.

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