Love to see web-of-trust implementations popping up! Can you share some more thoughts and/or details? Like, what issues we might run into with this approach?
Discussion
Imposters could still swim into your social network if people start following them. Iris has "publicly flag user" feature, but maybe there should be a distinction for the reason of flagging: unwanted content, impersonation, expired key etc.
Then bots will start flagging everyone. If people follow bots then trust leaks to them. And if a bot is silent then it won't annoy it's followers and they won't unfollow. Not sure what to do with that yet