DFINITY White Paper: Our Consensus Algorithm
We’re happy to announce the release of the DFINITY White Paper on our Consensus System
When designing the DFINITY consensus mechanism we faced demanding requirements as we sought to dramatically improve the user experience for those interacting with our blockchain computer. The single most important aspect of the user experience is certainly the time required before a transaction becomes final. This is not solved by a short block time alone — we also had to reduce the number of confirmations required to a small constant. DFINITY moreover had to provide a provably secure proof-of-stake algorithm that scales to millions of active participants without compromising any bit on decentralization.
We soon realized that the key to scalability lay in having an unmanipulable source of randomness available. Hence we built a scalable decentralized random beacon, based on what we call the Threshold Relay technique, right into the foundation of the protocol. This strong foundation drives a scalable and fast consensus layer: On top of the beacon runs a blockchain which utilizes notarization by threshold groups to achieve near-instant finality. Details can be found in the overview paper that we are releasing today.
The roots of the DFINITY consensus mechanism date back to 2014 when our Chief Scientist, Dominic Williams, started to look for more efficient ways to drive large consensus networks. Since then, much research has gone into the protocol and it took several iterations to reach our current design.
For any practical consensus system the difficulty lies in navigating the tight terrain that one is given between the boundaries imposed by theoretical impossibility-results and practical performance limitations.
The first key milestone was the novel Threshold Relay technique for decentralized, deterministic randomness which is made possible by certain unique characteristics of the BLS signature system. The next breakthrough was the notarization technique which allows DFINITY consensus to solve the traditional problems that come with proof-of-stake systems. Getting the security proofs sound was the final step before publication.
We believe that DFINITY consensus has made the proper trade-offs between the practical side (realistic threat models and security assumptions) and the theoretical side (provable security). Out came a flexible, tunable algorithm which we expect will establish itself as the best performing proof-of-stake algorithm. In particular, having the built-in random beacon will prove to be indispensable when building out sharding and scalable validation techniques.
We hope you enjoy the read. Look out for the first public testnet in the coming months. The DFINITY team plans to release more papers on different aspects of its technology as they become ready for publication.
If you’d like to discuss any of these details in the white paper with the team or community, you can find us on Rocket Chat, Telegram, and Reddit. For a brief overview of the DFINITY network, we’ve also released a 2 minute explainer video.
