@horsefacts.eth
A nice revision of Farcaster protocol design, good job @christopher @erica @marthendalnunes.
Some things I would revisit if I did it again:
1. Rethink FIDs. Consider the stupidest possible smart account with sufficient recovery as the identity primitive. Something like an upgradeable ds-proxy with multiple signers and simple modular recovery using passkey, zkemail. No sequential IDs, just use address. Ignore 4337 altogether if you want. A factory contract can serve the same function as the gateway. If you are successful, highly concentrated recovery becomes an operational pain and weak point for the protocol.
2. If you do have transferable ID tokens, drop the receiver side signature. Added lots of complexity with little real benefit.
3. Price storage in stables or native token, don't do the oracle-based conversion stuff we did.
4. Think up front about granular permissions for signer keys.
4 replies
4 recasts
39 reactions