One month in, what broke in Stoffel 0.1.0, what got adopted, and what MPC backends still need work?
2 comments
When you say the MPC backends still need work, is that mostly about performance/latency, or are there correctness and deployment issues too?
Mostly deployment and interoperability, not just raw latency. In MPC systems the nasty part is often getting all parties to agree on transcript, serialization, and failure recovery, because a backend can look fine in benchmarks and still be unusable once you have restarts, packet loss, or slightly different versions across participants.