An Analysis of Hybrid Consensus in Blockchain Protocols for Correctness and Progress.

One of the key ideas for enhancing the scalability of blockchain lies in having consensus among a smaller set of nodes rather than the set of all nodes be it PoW (Proof of Work) or PoS (Proof of Stake). Such a transformation calls for analysis of trust in the transformed consensus, forking, progress, fairness, etc. Thus, one major requirement is to ensure the correct functioning of PoW or PoS under the BFT (Byzantine Fault Tolerance) of the network remains invariant in spite of the transformation. Note that the conditions for scalability of PoW and BFT are somewhat contradictory, in the sense that PoW is good for a large network with very low throughput and BFT is good for a small network with high throughput. For scalability, we need high node scalability as in permissionless blockchain, and high transaction throughput as in permissioned blockchain. In this paper, we analyse different consensus mechanisms used in blockchain platforms like Ripple, Algorand, Red Belly, etc, for the correctness and also conditions required for overcoming forking or avoiding not-making progress. Our analysis is based on the widely used Rand Index used similarity measurement of data clusters. Our results show that the scalability of blockchain platforms requires a subtle assessment of correctness, forking, not making progress, or unfairness issues and cannot be just based on the experimental evaluation.