118

Having just failed my first review audit from any of the queues, it was reviewing close votes; however, this audit was flawed in that I didn't even cast a close vote.

Often I press the Close button to preview greater detail and sentiment of the closure, spawning the Why should this question be closed dialog box:

why-should-this-be-closed

Especially for duplicates, where functionality is solely dependent on spawning this dialog.

From Closing > Duplicate, I can search or view selected duplicates:

duplicates

Although, I realize general closure reasons are summarized on the page:

summary

This question had 6-upvotes, but seemed very localized referencing subjective query is "taking too long" and even had a sqlfiddle link. Maybe overly broad regarding performance tuning, the OP was posting like a forum with two answer posts to his own question here and here with one of those answers asking follow-up questions.

Wanting to gauge community response, current closure totals struck my curiosity.

So, I pressed the Close button and boom - oh, the humanity! I failed, and didn't even cast a close vote.

Votes are not cast until a reason for closure is selected and Vote to Close is pressed.

Taking any review very seriously on this site, it only seems fair that my review quality is tested through actionable steps instead of clicking for more detailed information, leveraging the functional design of the close vote user interface.

Sure, it's complex to seed fake vote totals, and then there's controversy of mindless voting in the majority or people who simply cherry pick 4-vote questions.

Still though, it seems unfair I should hesitate to review further information of vote totals and duplicates before skipping due to the implementation of this audit.

16
  • 24
    If you aren't supposed to use this information to help you decide whether to vote to close it shouldn't be displayed at all.
    – hayd
    Commented May 10, 2013 at 7:35
  • 12
    I think there should be some sort of Or do you genuinely believe this should be closed? dialog to redeem yourself when failing close audits. I'm pretty sure the questions which are pulled are done at random, and there are always going to be a few which have slipped through the net which should be closed but aren't. Commented May 10, 2013 at 7:54
  • 1
    Erm, wait, you pressed the Close button and didn't expect the system to think that you wanted to close the question? If you don't know the close reasons well enough to need the selection page then how could you possibly review accurately? Commented May 10, 2013 at 8:33
  • 9
    @JamesDonnelly: There is no need of "Do you genuinely believe this should be closed?" dialog. There is already a button called Vote to Close.. Review Audit should consider if we pressed this button for Closure..
    – Aditya
    Commented May 10, 2013 at 12:51
  • 23
    @UphillLuge No, I want vote totals / duplicates functionality of the Why should this question be closed. Casting a vote occurs on Vote to close, not pressing the Close button. After 1,477 close vote reviews, I'm pretty familiar with closure options. Commented May 10, 2013 at 14:08
  • 40
    @UphillLuge I've voted to close 835 questions and even now I'll open up the close menu on a question I'm either probably not going to close, or am only debating closing, just to re-read the specific wordings. Honestly that page has the most helpful information to consider as to whether or not a question should be closed. Do I need to check it every time, no of course, not, but I'll look at every now and again to refresh my memory, pull an exact quote from the description to post in a comment, or just to think over a borderline case.
    – Servy
    Commented May 10, 2013 at 15:26
  • Related: meta.stackexchange.com/questions/168824/… Commented May 10, 2013 at 18:27
  • @EsotericScreenName How is that related? Commented May 10, 2013 at 18:29
  • @JasonSturges It's discussing the same functionality: audit decisions being made as soon as you click flag or close. Commented May 10, 2013 at 18:33
  • 1
    @EsotericScreenName Fair enough, minor overlap loosely coupled per Wouldn't it be better to force the person to confirm the flag or deletion recommendation before displaying the audit results. Commented May 10, 2013 at 18:42
  • Yes, the UI is pretty inconsistent here - you beat me to it by half a day!
    – halfer
    Commented May 11, 2013 at 11:06
  • 9
    I've been running into this lately. I decided I'm just gonna fail every audit in this style, because I'm too used to clicking "Close" to see what people have done.
    – user176088
    Commented Sep 19, 2013 at 9:25
  • 6
    When a legitimate question is voted to be closed, my first reaction is to click the "Close" button to see what the heck. I guess I'm doomed to be banned. :(
    – Athari
    Commented Dec 24, 2013 at 1:26
  • 1
    I clicked on the close button to see how people were justifying closing the question, because I couldn't understand why it was closed. And then I saw, it wasn't... Commented Mar 2, 2014 at 20:27
  • 2
    the close and flag dialog are now fully functional in audits, see meta.stackexchange.com/a/231658/156631
    – m0sa
    Commented May 15, 2014 at 9:01

3 Answers 3

36
+500

However, it kinda falls apart when you want to close a known-good question as a duplicate.

Fixing this would require dragging in code that's currently being extensively re-written; doing so is not feasible at this time. It may or may not become possible in the future - even without the current re-write, the logic for that UI isn't trivial to fake.

Perhaps I'm oversimplifying a lot, but wouldn't it be kind of practical to just move the pass fail check to when the actual close vote is cast?

I'm thinking it could and likely should work in much the same way that suggested edit review audits currently work, with a caveat for duplicates.

  1. Present the user with an audit
  2. If the user clicks the close button, show the dialog as usual
  3. If the user closes as a duplicate, behave as if it were a normal close vote
  4. If the user chooses another close reason and clicks "vote to close" then pass/fail the audit appropriately

Like I said this is likely oversimplifying things by leaps and bounds, but I think this is the expected behavior.

If the concern is generating "fake" close votes so that the audit seems to show that other users voted to close the post for some reason or other... I'm not sure its entirely necessary.
I routinely see close vote reviews that don't have any indication of how many close votes the post has. I assume these are posts that have flags rather close votes, I would guess that most users are accustomed to seeing these reviews and wouldn't really think twice about not seeing the blue numbers.

5
  • 3
    I just got caught by this, do you know of any updates on plans to change this behavior ? Commented Mar 3, 2014 at 7:56
  • @WillemD'haeseleer From the look of shog's answer I don't think they've got any thing in the works yet, but it looks like momentum may be growing on this issue. It'll probably be 6-8 weeks...
    – apaul
    Commented Mar 3, 2014 at 8:01
  • @apaul34208 Any update yet? Commented Mar 17, 2014 at 0:28
  • 1
    @ZachSaucier meta.stackexchange.com/a/19514/217863
    – apaul
    Commented Mar 17, 2014 at 12:38
  • Looks like when they fixed it, they didn't implement point 3.
    – Didier L
    Commented Aug 18, 2018 at 17:50
43

My usual flow when reviewing is:

  1. Read the title
  2. Read the question
  3. Click the "Close" button, skim trough the reasons if I am not sure and verify that what I just read is okay with all the points.
  4. Either click "X" to close the closing window and Leave Open or Skip, or choose the close reason and close.

To clarify; I am not interested in what other closevoters chose. I want to reread reason descriptions and use the searching for duplicates function (which also makes sense if I don't know if the question is a duplicate, I guess).

The obvious problem with that is that the test system slaps me automatically when I click "Close" to show the reasons, when I haven't even made my decision! Frankly, it's a bit annoying, because I often open the close reasons just to close them and leave the question open (so my decision isn't really done hastily), but I can't do that if the system punishes me for my method of reviewing.

In my opinion the test should be concluded only if one of the definitive outcomes is choses, that is one of close reasons, "leave open" or the user goes in edit mode. It should be changed if only for consistency with actual closevote casting process.

-25

I consider this behavior appropriate for most of the audits. If you're clicking "close", you probably do want to close the question.

However, it kinda falls apart when you want to close a known-good question as a duplicate.

Update: This should now be fixed, insofar as you'll have to actually select a close reason and click Vote before passing / failing the audit.

31
  • 25
    This audit needs reevaluated - testing press of Close barely improves reviewers. 4-votes of a duplicate that's totally off, completely wrong closure reason, or even just blank with nothing selected would verify reviewer quality. Logic for all audits isn't trivial, I'm sure. Good time for fresh ideas if it's being reworked. In general, I really like review audits - great solution to many problems. Agreed at a high level, but close review audits could be improved. Commented May 11, 2013 at 2:40
  • 3
    I just got bitten by intention-to-close-as-duplicate of stackoverflow.com/review/close/2321254
    – Neil
    Commented Jun 15, 2013 at 15:27
  • 6
    in current form, this answer looks like rather poor justification for audit test design. Particularly, an assumption that clicking Close is a reliable indication of a desire to close, is wrong, as explained in a duplicate question. If you believe that desire to see CV split or desire to do this from review queue justified audit failure, consider editing the answer to reflect that
    – gnat
    Commented Aug 8, 2013 at 10:07
  • 5
    @Shog9 the more I think about it, the more it looks like it is worth investing effort into coding tests that fail at the moment of unambiguously proven failure. It is a honorable goal to teach people subtle intricacies of moderation, but so far it looks like getting audits there does more harm than good. The very dupe-closures wouldn't be an issue if audit would fail only at voting for the guaranteed wrong reason (that way, you'd have usability complaints anyway but these would be of much lesser severity)
    – gnat
    Commented Aug 8, 2013 at 14:17
  • 21
    I completely agree that the audit system should not presume what you want to do before you do it. No sane audit system should do this. A vote to close is not cast until the "vote to close" button is clicked. An audit is designed to see if you "voted to close" a question. If you have not "voted to close" then the audit should not fail (presuming it is a question that should not be closed). I don't see how a counter-argument to that is even possible. An audit should audit what you do, not what it guesses you might want to do.
    – Jason C
    Commented Aug 16, 2013 at 22:11
  • 19
    If it's not possible to build a functional auditing system from the existing UI logic then it shouldn't even be there to being with. There's no point in an auditing system that doesn't audit what you actually do.
    – Jason C
    Commented Aug 16, 2013 at 22:14
  • 12
    We're still waiting.
    – bjb568
    Commented Feb 8, 2014 at 8:05
  • 5
    This answer looks like an attempt to explain to the boss why a bad coding decision was made. Commented Feb 23, 2014 at 15:12
  • 7
    Failed a 'duplicate' review today because I wanted to check if something is a dupe of anything interesting. This is really annoying. Commented Feb 28, 2014 at 9:04
  • 4
    still being rewritten, I guess? Commented Feb 28, 2014 at 14:57
  • 7
    I often click the close button to reread reasons to validate. And I will definitely click it when I feel a question SHOULDN'T be closed to see what the hell people are talking about. I feel like this is important context. Commented Mar 3, 2014 at 8:08
  • 3
    "If you're clicking 'close', you probably do want to close the question." I think I might not be the only one who uses the close dialog as a checklist. I open it especially when I think the question might need to be left open to double check that the question doesn't fall into a category where it should be closed. Also, sometimes there's text on "off-topic: other" which is important when considering whether or not to close a question.
    – zzzzBov
    Commented Mar 3, 2014 at 17:57
  • 3
    @Shog9 Clicking "(more)" does not provide me with the same context information then when clicking close. I think the OP's post gives enough information on why that is. If you are genuinely afraid that people are just gonna cherry pick the most voted item then that's where audits might actually be useful. But 'clicking close' does not equal 'closing', surely you can see that right ? Commented Mar 3, 2014 at 18:22
  • 3
    @Shog9, "If you're sitting there staring at a perfectly good question trying to think up a reason for it to be shut down, you're doing it wrong" Questions don't typically end up in the queue accidentally, but before I can say that it shouldn't be closed I need to verify that it doesn't match one of the available close vote reasons (including the entire off-topic subgroup). This means that there are 11 different reasons that the question could be closed, and it must not fit into any of them before I can confidently click "leave open".
    – zzzzBov
    Commented Mar 3, 2014 at 19:13
  • 8
    @Shog9, I open the close window because it acts as a checklist. Do you expect everyone to remember all the close reasons all the time? The simple solution for me is to just stop trying to help on the review queue. It sucks, but it beats being punished for trying to do the right thing. It's not my job to change my habits in reviewing questions.
    – zzzzBov
    Commented Mar 3, 2014 at 19:16

You must log in to answer this question.

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