Well, I keep these posts here then (moderator deleted my thread...).
The biggest problem on 4chan is the spammers. They don't get banned even though they spam same image with same hash 24/7 (5-10 times per day, desuarchive shows that they spam same image over 5000 times per year). Most of the times they even use same exact message. 4chan mods could just click once and deny the hash, message, blacklist words etc. If that is too "dangerous", they could just give IP address a 1 year ban. Weirdest thing is that we don't need any AI to make guesses. This is not confusing or difficult. They literally spam same image with same hash with same text 10 times a day.
I sometimes report them, but immediately after I get banned for 1-4 weeks for "false report" or "trolling". Of course their old spam still exists, they don't get any bans. Only common thing is that all this spam is always pro-Apple, pro-proprietary, pro-left, pro-Microsoft. Every single spam is anti-FOSS, anti-Linux, anti-freespeech, anti-right, anti-<insert-any-of-the-right-wing-youtubers> etc. Who is behind this and why they do this? I know FBI has 8 moderators there 24/7 and after 4chan hack it was proven they spam these, but for who? Why?
It would be fun project to create the best moderation system ever. Only allow AI model to moderate that is trained on rules. You could add black listed words/hashes/etc. first, but randomize the time the threads and posts are removed so anons don't know if it was blacklisted thing or human banning them.
So weird how simple rules can be and then nobody gives a shit about breaking rules and also moderators banning things that were not against the rules. If humans are moderating, then you just do three strike system where they are removed after three times they banned somebody that didn't break the rules. Rules could be changed and updated with version control and when there is potential ban, AI and/or humans could just check if the post was breaking the version of rules at that time. AI could also create a "trending spam prediction" that would see the larger picture and history, checking if some overall spam is being done that humans could not even see as propaganda. There could also be automated system that scans everything and reports once a month to the police as a general "potential danger" reports.
Bonus comment from other anon: "Because the spammers are mods and they find it funny."