6

A few minutes ago I was reviewing low quality posts.

A question appeared containing some link looking like a spam. I hovered over the link, but it didn't show the URL.

I pressed "Edit" to see the link and immediately got a message that I am banned for doing too many wrong reviews.

That's funny.

P.S. This is the review. It clearly states that my action was "Edit". It is a bad post. So the audit is blatantly wrong. How does one know it is spam?

P.P.S. I looked at the review stats (now I can do it thanks to @ByteCommander) and see that I am #2 in Close Votes all time and #4 in Low Quality.

When a user with this stats is banned, that obviously mean that something goes wrong. The system is not perfect, but the main task of moderators is to fix this type of issues.

It's a shame that a moderator instead of reviewing the audits and lifting the ban starts to write an answer, explaining why that happened with some wrong statements, and only after some hype and suggestions from other users payed attention and finally started looking for the button.

| |
  • Is that the only one you failed recently? I see you in review most days, you must have some good review weight. – Mark Kirby Jun 1 at 22:06
  • I don't remember any other fails recently. Maybe one a day or two ago. I wanted to close one as a dupe and that was a fail ;-) This system works very poorly. – Pilot6 Jun 1 at 22:12
  • 1
    Seems very harsh then, I failed two on the same day the other day and didn't even get a warning, I assume because I don't fail often, so you should be safer than me. Also, choosing edit should never be a fail, even the best posts on the network might need an edit. – Mark Kirby Jun 1 at 22:17
  • I don't fail often either. I agree about Edit. I planned to flag as spam, but wanted to see the URL first. – Pilot6 Jun 1 at 22:18
  • I can add that I passed a few ones today. After I failed maybe yesterday once. Can you see the history? I am not allowed anymore. – Pilot6 Jun 1 at 22:20
  • As far as I know only a mod can see your audit history. I can see all your reviews but not the audits. – Mark Kirby Jun 1 at 22:21
  • 1
    The review system needs improvement. I was banned recently, the last audit I failed the answer was the correct solution. It was a duplicate answer though, and the other answers are not presented to you in the audit. – rtaft Jun 2 at 14:24
  • 4
    "How does one know it is spam?" IMHO, a post titled askubuntu.com/questions/1242792/where-to-watch-bigg-boss-14 should not require much thinking to realize it is totally irrelevant to our site. Whether it contains a link or not doesn't seem like a strong requirement to press the delete button. But that's just me. – jokerdino Jun 3 at 9:37
  • It could be a game or something. I don't think it is something wrong to look first before voting to delete. @jokerdino – Pilot6 Jun 3 at 9:39
5

According to https://meta.stackexchange.com/a/158678/280883

Here are the full details on how the ban works. Most of the steps take into account the length of any previous review bans you've faced, including manual bans imposed by a moderator:

  • If you fail three audits within a (rolling) 30-day period, you will be banned for 2 days if you have no history of getting banned, or for half the length of your previous ban if you do.

  • Once a review ban ends, you'll be placed on "review probation" for a period of 30 days. If you fail a single audit while on review probation, you'll be banned for double the length of your previous ban. Review probation is also imposed at the end of a manual moderator's ban

There is no indication that the total number of reviews or passed audits is taken into account for the thresholds to block someone, only the failure count, as I see it.

The system says that you seem to have actually failed even more than 3 audits within those last 30 days.

While I am not sure why you weren't already blocked earlier with these numbers, or whether the post I linked is not up to date, your current block is going to last three days and will be automatically revoked again then.

Regarding how edits should be treated in audits, that is a network-wide implementation and nothing we can directly influence here. Complaints or suggestions about this would need to be posted on Meta Stack Exchange instead. https://meta.stackoverflow.com/a/368118/4464570 clarifies the behaviour (unless that behavior got changed again by now):

if you try to edit spam, you fail; if you try to edit anything else, you pass.

As a tip for the future: Instead of using the built-in "Edit" button of the review queue UI, open the small "link" on the right side of a post in a new tab to visit the original post on the main site. This will allow you to interact with it normally without accidentally triggering any review action. Keep in mind you still need to go back to the review to complete it.

| |
2

Review audits are frequently intentionally tricky and weird. It's a common practice on the Stack Exchange network designed to teach the reviewer to stop, look and listen. It's not unusual at Stack Overflow to find audit questions in the close vote review that have an accepted self-answer posted by the author of the question. My first instinct would be to automatically vote to leave all such questions open in order to not close a question that has accepted answer, but the only way to pass the review audit is sometimes to close vote the question anyway because its accepted answer is not good enough. For example, the accepted answer is just a code snippet with no explanation. Even though it doesn't seem fair to close a question just because someone posted a bad answer to it, in this case it is fair because the question was self-answered with a bad answer that no one can benefit from and the accepted self-answer locked out anyone who posted a better answer from getting an accept vote for it.

There are endless variations of honeypot review audits. The only sure way to not fail review audits is to stop, look and listen.

| |
  • There is no way to never fail, unless you open the real question every time and check if it s an audit. And there is no way if you are dong >1000 reviews a month to never fail 3 times. I have a good record not to be banned for 5 years, but this is pure luck. In 2015 I just started really using the site. – Pilot6 Jun 3 at 7:44
  • The length of the reviewer ban increases exponentially with each successive ban results in some users being permanently banned from all review activities, but the length of the last reviewer ban wears off back to the length of the previous reviewer ban if the user goes 30 days without getting another reviewer ban. Your protection is to take it easy if you get one reviewer ban, and wait for 30 days until it wears off before resuming your reviewer activities. – karel Jun 3 at 7:55
  • re your last sentence: Another way would be to stop reviewing altogether - those audits are sometimes really annoying and break my concentration :-(( – guntbert Jun 6 at 21:00
  • 1
    @guntbert That's basically what I've done. When SE do all they can to alienate reviewers with an opaque and random process, I don't really feel like giving back. – vidarlo Jun 7 at 19:28
  • The process is not really opaque. It is random in the sense that posts for audits are selected randomly from a pool of eligible posts meeting some criteria, because it works good enough in most cases and curating them manually would be even more human work again. I really don't get why people are so upset about this all. So what if you fail a review once? So what if you can't review for a couple days? It does not mean your voluntary work is not appreciated, in contrary. The audits are there to make sure it isn't done mindlessly and automated ("robo reviewing"), to ensure quality. – Byte Commander Jun 7 at 19:39
  • @ByteCommander Opaque, as in no well defined complaint venue and process for remediation if you fail audits. In addition, many of the audit examples are plain bad; they may work better on other sites tho. – vidarlo Jun 8 at 4:43

You must log in to answer this question.

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