There needs to be consequences for getting lots of downvotes.
One of those could be something like this, where everyone can see that you get downvoted a lot and decide whether and how to engage with you in future.
@rimu @Andres Not related to bad behavior, but Phanpy displays a ratio of interaction types on the profile page. See “Last 20 posts in the past X days” section. You have to click it to see the ratio tho. Might spark some ideas.
Also, this is sounding a bit like Spoutible’s Bot Sentinel rating. An algorithm decides if you’re good/bad and assigns a letter grade. Users can decide to hide or limit others with a low letter grade.
https://help.spoutible.com/support/solutions/articles/150000050110-ratings-explained
@box464 @Andres Oof, I'm really not keen on AI-assigned scores. And they let the D-rated accounts continue to post??? Now that's weird.
There could be value in using simple "AI" to flag accounts for manual mod review. I've been thinking a lot about fascist detection. It should be doable, for the most obvious ones (pepe profile pic, 88 in the user name, posting 4chan screenshots, that kind of thing). Some image recognition machine learning and keyword filters would get me 90% there.
@JoeGermuska @box464 @Andres Yes, possibly... But any data you make available over ActivityPub is technically public so it's not great to put sensitive information in there.