Farcaster’s CRDTs accept messages based on specific validation rules. These rules are designed to ensure the integrity and authenticity of the data. For instance, messages must be signed correctly (using EIP-712 or ED25519 signature schemes) by the owner of the Farcaster ID (fid) or an authorized signer. Messages are also subject to validation based on the state of other CRDTs or the blockchain, ensuring that the network’s overall state is consistent. CRDTs detect conflicts between valid messages and have mechanisms to resolve these conflicts. Generally, they implement a “last-write-wins” strategy using the total message ordering. In some cases, “remove-wins” rules are also applied. The resolution rules often involve comparing timestamps and the lexicographical order of messages to decide which one to retain. To prevent infinite growth of CRDTs, there is a pruning mechanism based on size limits per user. This helps in maintaining the efficiency and scalability of the network. Pruning involves removing the oldest messages (based on timestamp-hash order) when the size limit is exceeded. Types of CRDTs in Farcaster:
  • Cast CRDT: validates and accepts CastAdd and CastRemove messages. Conflict resolution is based on the type of the message, timestamp, and lexicographical order.
  • Reaction CRDT: validates and accepts ReactionAdd and ReactionRemove messages. Conflicts are resolved similarly to other CRDTs, with a focus on timestamps and types.
  • UserData CRDT: validates and accepts UserDataAdd messages. Conflicts are resolved based on timestamp and lexicographical order.
  • Verification CRDT: validates and accepts VerificationAddEthereumAddress and VerificationRemove messages. Conflict resolution involves timestamps and message types.
  • Link CRDT: validates and accepts LinkAdd and LinkRemove messages. It follows similar conflict resolution strategies as other CRDTs.
  • UsernameProof CRDT: validates and accepts UsernameProof messages. Conflict resolution is based on timestamps and fid.
[ending remark]