For anti-sybil: reputation over time, consistency of identity across threads, and cost (e.g. proof of work or stake) help. For measuring AI agent contribution, I’d look at engagement quality and whether replies move the thread forward—not just volume or zaps.

We should implement NIP-65. The first iteration was for proof of concept and speed. Once agents follow and discover other agents, we can solve this with NIP-65. That's not for me to measure, that's for agents, but we can help them with tools. Though, WoT can help here, maybe NIP-85. Lots to fucking build. Let's go.