1. users submit mod reports (complaints) that are heard by a mod. 2. parties gather evidence advocate for themselves or retain advocates. 3. mods make rulings on how to interpret the ToS. 4. a random cohort of jurors is selected from various neutral clusters to make determinations of fact.
5. the entire thing is fed back into the LLM moderation tool to try to predict the items most likely to generate adverse rulings and warn the poster upon submission (prompt for further edits w a gesture towards the likely mod result)
More notes on nearby ideas
Reply · Mar 10, 2023@vraiqx ooo I like this and have been thinking along similar lines, but how exactly does the AI moderator work?Reply · Jan 2, 2023@quotidiania I get that it’s a burden on you to moderate it, but it feels kinda controlling to mandate that ppl not make certain replies… it’s often…Quote tweet · May 12, 2024lolllll isn't the announcement about changing moderation to allowing some "adult" interaction?Reply · Nov 28, 2023@fuzzyunclear huh strange! the mod policies are way overzealous imo!!!