8

Link to the strange behavior: https://stackoverflow.com/review/low-quality-posts/18484797

I identified it's a spam and clicked "Recommend Deletion", then hit "Delete", and there was no doubt I was presented with a "Congratulations" screen. But it looks, um... a bit strange?

I immediately captured the screen for this bug report. Then out of curiosity, I refreshed the browser and was surprised to see the new message:

*Shrug* So he did the audit faster than me??? Then why am I presented with that stub?

Should be a design bug. AFAIK a review is "locked" to the reviewer for 5 minutes after it's presented to them (others won't be reviewing that very item during "lock period"), so why aren't audits "locked" too???

For what it's worth, I swear I was presented with that audit. I wouldn't have been able to capture the first screenshot had I not seen it. Besides, I am completely innocent in Alexa Skills or whatever that question involves, nor is that audit discussed on Meta, so it's plainly impossible for me to find it manually.


Experiment on this

Thanks to Donald Duck for his help. We performed an experiment on the "race condition" of review audits. The full transcript can be seen here.

I proposed the experiment. In the process, I found 10 audits and sent to Donald Duck. D completed the audits in the desired way (i.e. passed them) while I keep the browser page open. Then I deliberately fail the audit from the page that I had kept open. After 10 failures, I was still not review-banned.

Here are the audits: 1 2 3 4 5 6 7 8 9 10

Here are the screenshots taken after I failed the audits: 1 2 3 4 5 6 7 8 9 10

To prove the "race condition" exists, Donald Duck showed me his visual of zero item in the LA queue, while I had one item open. Then we both completed the same review, and found that we both are on the list of completed reviewers of that review item.

So, since an audit can only be completed once, should the "conflict protection" be implemented on them?

  • Are you sure that you were presented with that audit? On the list of audits page, I don't see your name associated with that. There's only their name present on the list. – Bhargav Rao Jan 11 '18 at 8:50
  • Ok, I didn't understand your final comment. FWIW, from the mod interface on Stack Overflow, I can clearly see as to which audit was presented to which user, and what their action on the post is. That particular post was chosen as an audit 4 times and the last one was shown to that user. Perhaps only a dev can clearly tell if there's some other screw up. – Bhargav Rao Jan 11 '18 at 9:00
  • Things could have been more interesting if I deliberately failed the audit and got banned for "failing an audit actually done by another" :) – iBug Jan 11 '18 at 9:00
  • This is definitely a bug. Had a similar thing happened with a real review in a queue where a single user's review dequeues a task (i.e. First Posts or Late Answers), it would have shown both your name and the other reviewer's name. – Sonic the Anonymous Hedgehog Jan 11 '18 at 15:40
  • 1
    Cross-site duplicate – Donald Duck Jan 11 '18 at 21:38
  • 1
    For the confused old-timers here like me, it might be worth explicitly pointing out that since a couple of years ago even sub-10k users should be able to see deleted posts on the review audit completion page. So the "answer not found" is indeed a regression triggered by a separate bug. – Ilmari Karonen Jan 26 '18 at 15:45
  • @SonictheAnonymousHedgehog Thanks for your suggestion but this time I rejected it (clearly conflicts with author's intent). The bug report focuses on the existence of audit results, not the result themselves. So say you passed an audit but found that didn't belong to you, it's confusing. – iBug Jan 26 '18 at 16:12
  • I just noticed a bug with the race condition. This review was presented to me and another user at the same time. Since both of us reviewed it as "Edit and reopen", it caused some confusion with the edits. – Donald Duck Jan 29 '18 at 17:54

You must log in to answer this question.

Browse other questions tagged .