gm, thinking of ways a client can bootstrap a list of relays and test that they're minimally trustworthy in some way. maybe we end up defaulting to nip-05 relay recommendations - if that happens, would providers injecting their own relays into a user's records be helpful or viewed as an attack vector?

maybe a canary bot that tries to send restricted samizdata - 3D PDW files, politcally sensitive speech, financial transactions to blacklisted addresses - through a list of relays and testing end-to-end delivery would be useful?