npub1psdfxfpxz2cwmmnsk60y3nqpn2tqh9n24h4hstvfkwvr6eaek9js499sr7 (npub1psd…9sr7)
Bayesian classification is fairly effective at detecting spam email, so maybe mastodon can implement a "spamminess" score for posts that you can filter your feed by (say, auto-mute posts with spamminess > 0.8) and a spamminess threshold on the posts of a new server for accepting federation from them?