3

As a continuation of my comments here, which weren't understood too well and are more suitable as a separate question/discussion, I'm wondering why does anyone think review audits are any useful on this network?

They are in my opinion implemented very poorly. Some examples from different queues:

  • the worst out of the ones I have access to are Suggested edits, by far. All they do is they show some non-comprehensible mix of words added from combination of other question(s) to the audited question. I've seen extremely weird phrases, like "Windows install horse" added to the title or question body. You have to be basically autoclicking Looks ok or completely drunk to ever fail them. They could have as well not been coded, IMO.

Here are some examples of "audits based on actual edit proposals":

  1. "enter image description of citation style with here": Audit The spelling really got improved there.
  2. "in abu dhabi call me Excel": in abu dhabi call me Excel "removed noise" is quite ironic edit reason, which is the only thing copied from actual edits. "abu dhabi" is 100% taken from some random spam posts (not edits) - perfect place to take audit mumbling from.
  3. "How to excel stores but to install wsa OpenWRT on github device Hyper-V?": How to excel stores but to install wsa OpenWRT on github device Hyper-V? That comma got surely replaced with semicolon there.

Also, some references to old similar meta questions:

  1. Are the suggested edit audits too easy? - one solution proposed, other comments suggest it's to filter out basically autoclick scripts and people who don't even bother looking at the edit at all.
  2. Stop using real accounts in suggested edit audits - this speaks about a problem I haven't touched - real accounts being used for the mumbling. If anyone paid attention to usernames in the 1st place - IMO last thing to observe, then you can also notice the same people are shown as "editors" over and over again.

Not much or anything has improved since they've been asked.

  • Low quality posts haven't been much better, at least since I got rights to use that. 90+% of audits were obvious spam, deleted from spam flags. What is that exactly checking? IMO questions deleted as spam should be skipped there as that's not the purpose of the queue and you can't even flag there.
  • the choice for other queues isn't great either. I had cases of older questions appearing in e.g. First questions or Close votes queues which didn't really meet current website standards. E.g. very obvious upvoted software recommendation request that I tried to flag and failed the audit. Questions like that not being moved causes problems in general (attracting more questions of the same type mostly), but at least it could not mess with audits. I'm wondering if the choice is automatic - looks like it is. It should be manual if there should be older questions as well.

On top of that, there are some absurd implementations discouraging you from using certain features:

  • comments can be written on audits of deleted questions that are supposed to fail. So if you're new to the queues, you can write a lengthy comment just to get an "it's an audit" sort of error when submitting. The way around it is ofc to just type a short comment and submit, then edit in case it's real (or to open the question in new tab), but I think it should state it's an audit as soon as you start, not finish, typing.
9
  • 1
    I am fairly sure, that those audits, are based on actual edit proposals. I have for years opened the question, in another tab, while I perform a review. By doing that for every rule, meet the spirit of the intent behind an audit, despite their obvious flaws at times.
    – Ramhound
    Aug 8, 2023 at 14:45
  • If you are fairly sure then I have no questions. I provided 3 examples of very useful audits now. Are you still sure about that? And are you sure then about your AI detections being on point, as you said somewhere else? Even the most obvious CTRL + C CTRL +V ChatGPTs require a bit more knowledge to notice that they're created by a machine than this nonsense mumbling.
    – Destroy666
    Aug 21, 2023 at 11:19
  • Yes; I am fairly confident in my ability to tell the difference between the quality of contributions that a human generates and the absolute trash that ChatGPT generates. I am not sure what your point is to be honest.
    – Ramhound
    Aug 21, 2023 at 11:58
  • @Ramhound please explain why you wrote these edit audits are based on actual proposals then as I'm still not following where that conclusion came from.
    – Destroy666
    Aug 21, 2023 at 12:03
  • They are based on actual edits to actual contributions just not the question it was applied to, their entire purpose, is to verify you are paying attention
    – Ramhound
    Aug 21, 2023 at 12:07
  • But based in what way? E.g. the shown "abu dhabi" part of the title - where do you think that came from?
    – Destroy666
    Aug 21, 2023 at 12:09
  • As for paying attention, you can replace the whole edit with long string of WWWWWWWWs as well and it'd be nearly as effective IMO, but still checking for any sort of attention and eliminating e.g. scripts. You could also use a smarter algorithm constructing sentences that make some sense or some real scenarios actually based on past edit reviews (although this is less reliable with only 2 passes required). The question is what's the effectiveness of audits supposed to be. IMO it's very low in that queue right now, stopping only bot users.
    – Destroy666
    Aug 21, 2023 at 12:16
  • 1
    Suggested Edit audits are created by the bot. They're supposed to be easy.
    – Tetsujin
    Aug 21, 2023 at 12:17
  • 2
    @Destroy666 - The original text likely came from spam. These audit edits are suppose to be easy to spot. Here is the truth. The entire review system needs a ton of work. The correct action is often not even an option depending on which queue you are talking about. I agree that Audits should be based on actual audits, against contributions, which were already declined. However, not everyone has a problem not approving an edit propsal that is incomplete, so even that has problems.
    – Ramhound
    Aug 21, 2023 at 13:44

1 Answer 1

2

The downvotes so far would seem to indicate people don't like to be told they have to try harder. If you can't get an audit right, you weren't paying enough attention.

Audits are a PITA, that's for sure, but their purpose is to make sure people are paying attention.

If you get an easy one, fine, you passed.
On to the next task.

If you didn't already know - Suggested Edits audits are supposed to be easy - Are the suggested edit audits too easy?

If you get one that's not so obvious, then you need to be more careful.
Open the original question. Sometimes that's a dead giveaway - it's already been deleted, or downvoted to hell & back.
Cool, you pass because you were paying attention.

enter image description here

If at any time you fail an audit, be prepared to take responsibility for not investigating sufficiently, but also accept that no-one is going to get every single one right, every single time.
The system won't start to restrict you until you rack up a score of bad reviews, though I have no idea what the numbers are.

4
  • This doesn't touch on anything specified in the question, which I'd recommend reading before answering.
    – Destroy666
    Aug 21, 2023 at 11:31
  • I read the original question, before you filled it with noise. It's covered in my second paragraph.
    – Tetsujin
    Aug 21, 2023 at 11:33
  • If you think examples for one of the points that were brought up are "noise" then it further confirms what I wrote above. The examples shouldn't have been needed in ideal case scenario of people answering in meta actually doing reviews in mentioned queues. Even then they provide value to newcomers not knowing what I'm talking about, though. I'm downvoting this now as it's very generic reply about reviews that attempts to answer "Why are reviews too hard?" rather than "Why are reviews so bad?", which is partially opposite to few of my included bullet points.
    – Destroy666
    Aug 21, 2023 at 11:43
  • 2
    It's basically three examples of the same thing - what the bot does to generate obvious audits in Suggested Edits. Still covered by paragraph two. And now made more blindingly obvious in paragraph three.
    – Tetsujin
    Aug 21, 2023 at 12:16

You must log in to answer this question.

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