@moonman I don't think 'simple' and 'ocap' mix that well. In general I like the concept, but especially cwebber's essay on it seems too academic to me and filled with things that are simple theoretically but not trivial at all in practice (like all the 'proxy' stuff, for example). Also, I don't need a no nation no borders approach to motivate a social network protocol.
AP conf next month will essentially be 'Gospel of OCAP' conf, so I'm holding my judgement until then. At the moment, it seems like a lot of complexity for some interesting features that you can already do by signing your requests (which is expensive).
Being able to share authorizations with someone else is treated like a feature (you can give your friend your car key!) but it's a source of problems at the same time (someone stole your car key!).
TL;DR I'm not buying into it yet, but I'll look into it deeper.
@sjw@lain well, ocap does a lot more stuff than just whitelist entire servers. I'm leaning toward the opinion that ocap is a trojan horse for whitelists anyway though. anyway, I think the method listed would potentially cause unpredictable federation whitelist changes from other servers and piss off everybody.