Interesting, will take a look. Regarding your questions:

- Historically, reputation and web-of-trust models have been tried with mixed results (see PGP/GPG history)

- Proof of work for human validation can probably be gamed, useful as a potential workaround for rate limiting/DDOS mitigation though (check how Tor uses it)

- I'd be very skeptical providing my full KYC details to a new service, perhaps a host verification a-la Let's Encrypt could be useful as a Layer 1 "KYC" tier?