Snowflake to Avalanche: A Novel Metastable Consensus Protocol Family for Cryptocurrencies

This whitepaper introduces the Avalanche consensus family, a paradigm shift in decentralized consensus that moves away from traditional leader-based or chain-based protocols. The Avalanche protocol utilizes a novel 'metastable' approach where nodes perform repeated, small-scale random sampling of their neighbors to reach consensus. This process, termed 'subsampled voting,' allows the network to reach a consensus state exponentially quickly, even under high adversarial conditions. Unlike classical BFT protocols that struggle with throughput as the number of nodes increases, Avalanche’s performance is relatively independent of network size, making it highly scalable. The protocol family progresses through several refinements—from Snowflake to Snowball—each improving the network's resilience against different attack vectors. By leveraging metastable states, Avalanche enables a high-throughput, low-latency transaction processing environment suitable for a global, decentralized internet of finance. The paper provides both the theoretical underpinning of this probabilistic consensus model and a justification for how it solves the scalability trilemma by favoring rapid convergence over total serial ordering.