14

This question has been asked more than ten years ago already, but was de-facto declined by a staff member.

Many questions in the field of HTML, CSS, and JavaScript can be presented as a runnable snippet. This allows for easy reproduction of the problem, and also to copy the given code to the answer with the simple click of a button.

Often times I find questions that contain potentially runnable code that is often badly formatted and not integrated in the form of a runnable snippet (mostly from low reputation SO users (< 5k)).

With these, many times I've run into the following situation:

I click on "Edit (1)" to edit the question, only to find someone who lacks enough reputation to edit questions without approval has already made the exact changes necessary to improve the question in the way described.

I have three options now:

  1. Click Approve. This results in the question remaining in the unedited form for a long time, until another reviewer approves the edit too. You have already reviewed this item. It needs more reviews from other users to be completed.
  2. Click Improve Edit, simply add a space character somewhere irrelevant and boom, the edit is live, but this seems like a bad workaround, because instead of what needs to be done - approving, I'm forced to pick improving (which doesn't happen since I have nothing to improve), only to achieve approving through the backdoor.
  3. Click Reject and Edit, resulting in the user getting their "Rejected edits" count increased by one for all the wrong reasons, and redo what the current editor already achieved.

Given the situation that the suggested edit already does exactly what I would have done, had I edited the question myself, the correct choice appears to be to accept the edit, because factually there is nothing to improve about the edit, and also no reason to reject it.

Doing so, though, results in a question that still has no Copy snippet to answer button yet, more often than not discouraging me from answering, which originally I intended to do.

Wouldn't it make more sense, if you trust a user to edit a question with no approval necessary, to also allow that user to singledhandedly accept edits on those very same questions?

If the argument from that old answer is still considered valid, what else could be done to improve the described situation where neither 1., 2. nor 3. seems ideal?

Edit: Based on the comments, I'd like to suggest for discussion that reviewing edits should be treated differently based on whether the reviewer came from the review queue (which does not give the reviewer any control over which suggested edits they are being presented), or the reviewer came from the question itself (which should be safe to assume they wouldn't open if they were not interested in the question).

8
  • 2
    I don’t think it should matter where the edit is accessed from... If we trust a user to unilaterally approve it in one place, we should trust them to do that action everywhere. If I’m in the review queue and want to push a review through, all I have to do (if your FR is implemented) is navigate to the post and approve it from there. That’s not a good design. – ColleenV Apr 28 at 18:33
  • Maybe suggested edits stay in the queue until two people have looked at them, and it’s more like the “first post” queue where actions are more free-form. Anyone can apply a suggested edit, improve the edit, rollback and edit, etc. but at least two people review it even if one just says “everything is fine with it”. – ColleenV Apr 28 at 18:40
  • 1
    @ColleenV since the requirement of two approved reviews is intended to counter robo-reviewing it seems reasonable to allow this. Once a robo-reviewer takes the effort of navigating to a post they are not a robo-reviewer anymore, or at least considerably less so (and more likely to be somebody that actually wants the edit approved). – Marijn Apr 28 at 18:42
  • I don’t think it should matter where the edit is accessed from... I disagree. Ignoring where a user came from discards the most relevant piece of information - which is "is the user/reviewer interested in the question itself", which, as I said, should be safe to assume if they opened the question first, then clicked "Edit (1)", but much unsafer to assume if the user didn't actively choose to review this edit because the system picked to present it to them when they were working in the review queue. – connexo Apr 28 at 18:43
  • @Marijn It would be much cleaner to allow one person to apply the edit but not remove it from the review queue. If too many reviews are disputed, it’s the same effect. Or, you just put a pause on whether the button is enabled to slow people down if that’s what’s really preventing robo reviews. – ColleenV Apr 28 at 18:44
  • 1
    @connexo I reject the entire concept that denying me the ability to use a privilege I’ve earned simply because I’m in a review queue instead of on the post page is the best way to ensure people are exercising appropriate care when approving suggested edits. In my opinion, if that sort of hoop-jumping is required to make the suggested edit queue work properly, it’s broken and should be redesigned. – ColleenV Apr 28 at 18:53
  • That redesign would imo have to involve choice from the reviewer, such as listing suggested edits just like questions to have reviewers pick from the list rather than the system doing it for them. Active choice is the signal revealing interest. The review queue, as it is designed currently, imo actively promotes getting done with it by setting a daily goal (for me it's 20). – connexo Apr 28 at 19:00
  • I think most of the suggested edit mistaken approvals could be prevented with a better display of the changes, and some “training” that has people review some sample edits and gives them feedback on ones they incorrectly approved. I would like to see the ability to step through each change, similar to the way my merge tool works, and be able to see both the markdown changes and the rendered changes simultaneously. I understand how much work that would be, but you will never know what is possible if you don’t dream big ;) – ColleenV Apr 28 at 19:14
8

You're coming at this question from the vantage of seeing the edit suggestion on the post and wanting to improve it. So you're already somewhat invested in the post and that's good. What you're maybe forgetting is that most suggested edits are reviewed in the review queues and there are (unfortunately) some people who will go through those reviews extremely quickly without actually reading them - we refer to that as "robo-reviewing".

Most reviewers do read and review accurately but, because it happens often enough that people don't, by requiring two people to approve edits, it means we're more likely to have at least one of the two people actually reviewing the edit to make certain that it's complete and valuable. This is, essentially, what Grace Note says in their answer to the older question. We had data when we first released this feature that many terrible edits were just slipping through because people weren't reviewing carefully - so we require two reviewers.

So, in your case, you're investing time in working on editing the post - reviewing it carefully to ensure that nothing was missed in title, body and tags... so feel free to improve the edit because it's not uncommon for one or two small things to be missed and then save another reviewer the effort of reviewing the post, too.

But, all we ask is that you be thoughtful and thorough - that's what we want and why we require two reviewers in the first place.

13
  • 4
    So does that mean that it makes no difference if I'm clicking "Edit (1)" in the question itself, compared to being presented the suggested edit in the review queue? Imo that should make a difference. As it stands now, I would often like to give an answer based on that runnable snippet, but as I'm forced to wait on the edit to be approved a second time, the process is taking too long, resulting in me not answering at all. This results in me giving less answers than I intended to. – connexo Apr 27 at 14:18
  • 1
    Also, I understand that users that lack the required badges to edit questions without approval need an additional reviewer. But to allow a user to edit singlehandedly, and not trusting the very same user to accept edits the same way doesn't make sense to me. – connexo Apr 27 at 14:20
  • 2
    Our system doesn't treat them any differently, no. I think, in fact, when you click on the "Edit (1)" you're opening a review window - so you're actually performing the review on the question rather than in the queue. – Catija Apr 27 at 14:21
  • 1
    I don't understand why an edit not being live prevents you from answering the question, though. If you would edit yourself and then answer, then just answer assuming the good edit will be approved. Can you explain why you're only interested in answering if the edit is live? – Catija Apr 27 at 14:22
  • Not treating that differently ignores the addtional information a click on "Edit (1)" below the question grants, which is that I actively opened the question, showing my interest in it, whereas in the review queue I have zero control over what question's suggested edits are presented to me. – connexo Apr 27 at 14:23
  • 2
    You're asking for how it works and I'm answering how it works. If you want to suggest that that should change, write it up as a FR saying "If I'm looking at a suggested edit on a post that I opened directly, outside of review, allow my approval to go through immediately rather than waiting for a second reviewer". That's not a discussion, though - it's a FR. – Catija Apr 27 at 14:25
  • As long as the edit is not live, there is no "Copy snippet to answer" functionality. – connexo Apr 27 at 14:26
  • 1
    @connexo Except it is impossible to validate how a user opened the review task server-side. At some point all requests are going to end up at the same place, and the best you could so is attach a variable for "opened directly from the question." Except client-side data is inherently unreliable and there are no ways to validate the variable on the server. There is no way to reliably determine how a user viewed the task that isn't open to abuse (e.g. a simple user script that would make every task in review think it was opened from the question page). – animuson Apr 27 at 14:28
  • 4
    @connexo That seems like a valid reason to just make a tiny change and improve the edit rather than approving it. I understand your concerns but I'm trying to get you unstuck now and that means getting you to understand that it's OK to slightly circumvent the system by improving instead of approving. – Catija Apr 27 at 14:28
  • 2
    But that is clearly nothing more than a workaround due to technical limitations currently restricting the system's smartness. It may be true as of now (and probably has been for those 10 years +), but I'm sure there's ways to solve this technically. – connexo Apr 27 at 14:38
  • 2
    @connexo If the workaround is "click this button instead of that button" and the fix is a lot of redesign and development work that could be spent fixing or improving other things, is it worth changing? – ColleenV Apr 27 at 14:39
  • 4
    The workaround, while it procedurally might be as simple as you outlined, still requires me to make a wrong choice. I am forced to click "Improve edit" even though there is nothing to improve in that case. The correct choice for a perfect edit semantically is "Accept", not "Improve". – connexo Apr 27 at 14:43
  • @connexo Life is messy. It is often a waste of energy and resources to make every little thing perfect. Humans are great at adapting and sometimes workarounds are the best solution. – ColleenV Apr 27 at 14:53

You must log in to answer this question.

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