Earlier quoted context omitted.
How about you initially "trust" all of your friends. If one friend blacklists a bad actor, all their friends automatically blacklist them as well. If it turns out the friend blacklisted an innocent person out of spite or something, word should eventually get around and people can manually revoke trust from that friend.
I myself block a ton of stuff, which isn't illegal not offensive, not ... but just boring for me. That list shouldn't spread to others. Also I don't want to have to think about personal vs. global block.
To be a little clearer: you have your own private filters and a separate, distributed block list shared amongst your friends, which is for bad content. You use the big red button when you encounter spam or harmful content, and this bad content/user/node/etc is added to the shared block list.
You initially trust all of your friends to use this block list properly, but if you find one of your friends is misusing it you can revoke trust from them and ignore their contributions.