twitter / the-algorithm

Source code for Twitter's Recommendation Algorithm
https://blog.twitter.com/engineering/en_us/topics/open-source/2023/twitter-recommendation-algorithm
GNU Affero General Public License v3.0
62.32k stars 12.14k forks source link

Discourage filing bogus abuse reports #1791

Open TolstoyDotCom opened 1 year ago

TolstoyDotCom commented 1 year ago

Is your feature request related to a problem? Please describe. Many Twitter users file bogus abuse reports instead of alternatives such as engaging opponents in debate. As discussed in https://github.com/twitter/the-algorithm/issues/1786, Twitter lets people use bogus reporting as a way to silence opponents.

Describe the solution you'd like Reports could be handled using a simple system similar to MLB manager challenges (mlb.com/glossary/rules/manager-challenge)

A Twitter user would start with the ability to file up to three reports. If they file a bogus report, the number of reports they can file decrements. If they file a valid report, they keep the number they had. After six months, the number of reports they can file resets. Those who consistently file bogus reports would lose the ability to file reports or have the number they get reduced. Those who consistently file real reports could be given more reports. The number of reports someone has available would be shown to them (but not others) on their dashboard.

Users would be encouraged to only report real violations of the rules. For cases that don't meet that standard, they'd be encouraged to engage their opponent in debate, or to simply unfollow or ignore the opponent.

There might be cases where organized campaigns create new accounts simply to use the reports against certain targets, but that's probably happening now. There's a small chance Twitter already has a system in place to detect that, but if not that could be developed.

Obviously, if the current team at Twitter implemented this feature request they'd come up with a make-work scheme like using AI instead of a simple count. And, they'd bias the count towards favored people: those with lots of followers, Big Biz, NYT reporters, etc etc.

So, the above assumes Musk has cleaned house and replaced the problematic current employees.

aloiscochard commented 1 year ago

Totally agree with @TolstoyDotCom here, it would be also very much appreciated if Twitter do publish some open anonymized statistics about how user/accounts use reports.

It would especially interesting if such statistics about "karma" score could be published and analyzed by relevant people to further help diagnostic mental illness.

TheMemeSniper commented 1 year ago

🤓