28

Whenever I review the suggested edit queue, there are always a number of trivial edits that only fix one of many issues with the post.

As a reviewer, I know that my appropriate course of action ought to be to click the Improve button and make the additional fixes.

However, improving results a couple negatives:

  • I now have to spend my time combing through the post, fixing all of the issues that the original editor could have fixed while they already had the edit window open. Overall time spent by users re-editing posts increases.
  • The user is rewarded for their edit. By giving them them reputation, we're positively reinforcing the behavior.

I know that trivial edits have been discussed here before. Reactive solutions for this have been proposed and discussed. The ideal reactive solution would perhaps be a combination of:

However, perhaps there is a proactive solution?

How can we encourage editors to make more substantial suggested edits?

|improve this question
  • 1
    Reject and Improve was already implemented. When you click Improve, there's a neat checkbox that says "this edit was helpful". You can uncheck it. – animuson♦ Feb 6 '12 at 21:30
  • @animuson - Yeah, I had read that; forgot to link to it. I think it's only really effective if bullets two and three exist alongside it. Also, the checkbox label is a bit... meh - the edit probably was helpful, it's just that it only helped fix, like, 5% of the question. :) – Rob Hruska Feb 6 '12 at 21:32
  • Make your point, reject the edit for being trivial. – Uphill Luge Feb 6 '12 at 21:37
  • 1
    @UphillLuge - While that's a fine reactive solution, without this it's not as effective, since I wager a lot of users aren't readily trying to educate themselves on what the norms are. Also, since some reviewers approve trivial edits, the suggesters still see reputation increases, thinking they're doing okay. – Rob Hruska Feb 6 '12 at 21:42
  • 19
    I bet over 90% of users don't even go check to see which of their suggested edits get accepted/rejected. They really need to be notified when they get a reject. A simple message in their inbox... – animuson♦ Feb 6 '12 at 21:55
  • @animuson - That would be excellent IMHO. Consider posting as an answer? – razlebe Feb 8 '12 at 12:33
  • 2
    @razlebe Already suggested. – Daniel Beck Feb 8 '12 at 12:39
  • I've been trying to be more heavy-handed lately, rejecting trivial suggestions and leaving a helpful "Other" message for the rejection. However, it's frustrating when that rejection gets overruled and "Approved" by a Community edit or an improvement; the original suggester gets their +2, thinks they're doing great, and continues making half-assed edits that require more peoples' time to improve (when they could have just fixed everything on the first sweep). </rant> – Rob Hruska Mar 12 '12 at 12:02
  • 1
    waffles summed it up nicely here (in point #2): "Processing edit suggestions is not a free process, it cost eyes, distraction and time. We want quality in the queue, so high standards is a good first step." – Rob Hruska Mar 12 '12 at 12:18
6

The process of editing could perhaps be.. ahem.. gamed?

On the one hand, we could try and eliminate some of the work required in the suggested edit queue by automating the approval/rejection of trivial edits:

  • Imagine a case where trivial edits are shown on the live posts themselves (like stickers) and can be approved or rejected by anyone (a cleaning the screen game). This could allow many small edits to be approved/rejected efficiently and en masse by the community, and may even encourage people to edit posts more by making the editing process so prominent.

On the other hand, we could encourage 'substantial' editing behavior by making non-trivial edits supersede trivial ones:

  • Imagine you're editing a post and a little counter beside the post notifies you if you're including any trivial edits already in the suggested edit queue. This would introduce a 'find them (trivial edits) all!' game that encourages wide-ranging edits. In addition, approving a substantial edit could reward the author with varying amounts of reputation depending on how many trivial edits they superseded. The authors of the trivial edits would get no reputation for such 'superseded' edits.

My examples clearly have all sorts of problems and possible improvements but this kind of thing just might be something worth thinking about! I've seen the stackexchange team do amazing things with features which at first glance seemed completely impractical or downright wacky.

|improve this answer
  • An interesting solution; it does kind of encourage the opposite of what I'm hoping for (more smaller edits vs. larger, all-encompassing edits), but I think any solution that makes the review process easier for reviewers ought to be considered. – Rob Hruska Feb 7 '12 at 14:16
  • You are quite right, I was thinking more about automating dealing with smaller edits than encouraging the all-encompassing edits. I've added a second suggestion to my answer that hopefully addresses your (very good) question a little more directly. – Daniel Gill Feb 8 '12 at 12:31
  • I have seen several large edits of questions lately that tampered with the code to such an extent as to possibly invalidate the question. Even very trivial edits to questions can make the question irrelevant. Just adding a stop could be all that was needed for the code to work. Cumulative small edits could add up to a substantial edit. I think it would be even easier to mistakenly accept a small screen edit than an edit you have taken the time to review. And what about spelling wars? – Remou Feb 8 '12 at 13:07
5
+50

  1. The reason so many trivial edits get approved is the reviewers are not trained when to approve a suggested and when to reject it. All we get is thrown a link to the suggested edit queue, and get told;

    Approve edits you know are correct; reject those you know are wrong. Leave ambiguous edits for other users to judge.

    When users first enter the queue, they should be guided through which edits to approve and which edits to reject (e.g. show them what types of edit are too trivial and should be rejected). Make them read this guide and accept it, then let them go nuts on the approving and rejecting. I envisage showing examples of edits and saying why it should be accepted, or why is should be rejected.

  2. The consensus what should be approved and rejected is diverged even on Meta. I've come across many discussions where there are equal votes to both reject and improve the same edit. We need to agree as a community when a edit is too minor. Personally, you'll find me as one of those reviewers who will accept most trivial edits. A minor improvement is still an improvement, right?

  3. Additionally, as well as giving an editor an in-your-face notification that their edit sucked and was rejected, we should also be telling a reviewer when their vote was wrong. We already do it for flags, so why not do it for edit votes? Users who get a number of edits rejected loose the ability to make edits, so why not remove the review privilege for users who give the wrong verdict on edits?

  4. Also, I'd put down the minor improvements down to the shoddy 2 rep you earn from doing so. So a <2000k user comes across an abolishment of a post... why should they spend their time rewriting the post to make it awesome, to get 2 reputation in return (especially when they know they can edit 6 chars and get the same reward)? To make things worse, the improved post is then good enough quality for the user-who-couldn't-be-bothered-to-format-a-post-in-the-first-place to start getting his own reputation for the good-users' edit; at a rate of 10 (or 5) rep a pop! This isn't fair!

    I propose the idea of the reputation reward for a suggested edit changing depending on the size of the edit. Give trivial edits the minimum rep (2 (or 0?)), and give large, substantial edits more (10 rep?) and make the edit worth the users time. Maybe a post can be triggered as community wiki by a substantial (suggested?) edit as well, to stop the OP riding the upvotes gained by another user's time and effort and making the whole process fairer.

|improve this answer
  • 1
    Great points. I think basing the rep gain on the size of the edit has some potential. I fear there might be some gaming, though. It would only work if reviewers were trained (like you suggest) to recognize appropriate and inappropriate edits. This does offer to solve the issue in a proactive way. – Rob Hruska Feb 8 '12 at 17:05
  • I would also say that 10 is probably too much. I'd drop trivial edits to zero, and scale up to, maybe three for a nice, tasty, entire-post-fixing edit? – Rob Hruska Feb 8 '12 at 17:07
  • 1
    @Rob: I think 3 is too little. A substantial edit to a post can reward the original poster with many upvotes (3 on a question = 30 votes). Their lack of effort (hence the need for a substantial edit) means they do not deserve any of that. It is the time taken by the editor that has caused those upvotes, and they should be rewarded as such. – Matt Feb 8 '12 at 17:29
  • I believe the smaller rep gains for edits are because goal #1 is to encourage asking/answering; if edits were worth 10 rep, edits would be equal to an answer upvote; everyone would stop answering questions and just start editing things! Mostly kidding, obviously, but we would see a lot more editing at 10 rep per edit. Maybe that's not bad, I dunno. There are six other numbers between three and 10. Perhaps one of them will suffice as a nice middle ground. :) – Rob Hruska Feb 8 '12 at 17:41
  • Also, if you think about the time spent editing a question vs. the time spent constructing a really good answer, I think the time spent on the answer (I'm guessing anywhere between 5-20 minutes?) on average well exceeds the (perhaps 2-5 minutes?) on most substantial post edits. That ought to be considered too. – Rob Hruska Feb 8 '12 at 17:42
  • I agree with the sentiment, but not with the solution: providing larger rewards just provides an incentive to users to heavily edit posts, including posts that do not need to be edited. I hate seeing edits that only change the style, or that change so much of the post that the original flavor is lost. – Matt Fenwick Feb 8 '12 at 17:47
  • @MattFenwick - That's part of my concern too. Although if we can encourage people to recognize and reject those types of edits, that's a partial solution. Very tough to do, though. – Rob Hruska Feb 8 '12 at 17:53
  • @RobHruska -- I still don't get how the comment notification system works. I meant to respond to the actual post. Did you get notified? If so, sorry for any confusion! – Matt Fenwick Feb 8 '12 at 17:58
  • @MattFenwick - I actually don't remember if I was notified :P – Rob Hruska Feb 8 '12 at 18:06
  • @RobHruska: I agree the number is arbitrary right now, but just to reinforce where I'm coming from; by giving someone 10 rep for a amazing suggested edit, that's the maximum they'd get; they don't get additional points per vote like a post gets. Such an edit may take ~5-10 mins of their time, but a large proportion of posts that attract 1-2 votes (and as such have also gained 10 points!), will also have taken 5-10 minutes as the answer is merely a short paragraph etc. – Matt Feb 9 '12 at 9:33
  • 2
    @MattFenwick: That's where training the reviewers come in, I guess. I believe it's the us (the reviwers) fault for accepting the wrong type of edits, rather than it being the communities fault for suggesting them. It is our job to be the gate for poor edits to try get through, and at the moment, we're wrongly letting them past. – Matt Feb 9 '12 at 9:36
  • @RobHruska: You might find the following link useful for how the comment @ reply system works ;); meta.stackexchange.com/questions/43019/… – Matt Feb 9 '12 at 9:36
3

<vent>

This question made me decide to take a look at the suggested edits. Many of them were downright wrong, and even more were insubstantial!

I think we should look into revising the rewards system for suggested edits. It seems to send the wrong message to some (albeit very few) people.

Here's an example: first, a user made a radical edit and got 2 points (how was that approved??); now the same user is trying to undoing the edit, which could give him another 2 points. And since the user has (at the time of this writing) earned 0 other points, those 4 points are a big deal.

</vent>

The system we have now is okay, even though it can be abused. Trying to make it better may have some unintended consequences and might just make everyone's SO experience a little more complicated.

But I still don't like users getting rewarded for poor/lazy/insubstantial/wrong edits.

|improve this answer
  • I completely agree with your vent. :) – Rob Hruska Feb 8 '12 at 16:23

You must log in to answer this question.

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