Cornell CS Ph.D. candidate Benjamin Chan presented a new paradigm for consensus protocols at the Stanford Blockchain Conference (SBC) on February 20. The family of protocols, called “Streamlet” (in joint work with CS Associate Professor Elaine Shi), is surprisingly simple, performant, and promises to be a unified scheme both for pedagogy and practical implementation. Watch Chan's talk beginning at 05:56:10. Review slides from Chan's presentation and also from Professor Shi's presentation.
Consensus protocols formerly had a reputation for being difficult to understand and implement, both in the classroom and in practice. Decades of research have focused on simplifying classical, well-known protocols like Paxos and PBFT. Recently, various blockchain projects have shifted to a Proof-of-Stake paradigm (rather than Proof-of-Work), and interestingly Proof-of-Stake, unlike Proof-of-Work, is rooted in the classical approach of “permissioned” consensus. Numerous blockchain projects have attempted to design new consensus protocols, after realizing that classical protocols like Paxos, PBFT, and Raft are difficult to understand and to implement.
The Streamlet paper notes that although the recent body of work in consensus is difficult to navigate, in fact a big leap has silently taken place—in terms of making consensus simpler than ever—thanks to the joint efforts of the community. Streamlet illustrates novel ideas with an extremely simple and natural embodiment, and has the potential to unify consensus implementations across the industry.
The SBC conference is a gathering of the the most prominent projects in the blockchain space as well as thought leaders from the research community. The conference is established and organized by The Stanford Center for Blockchain Research. This year there were more than 300 attendees, including engineers and founders from the blockchain industry and cryptographers from academia. Interest in the new protocol was substantial.