What is Snowman (Snow Protocol)?

Snowman is a linear chain consensus protocol from the Snow protocol family developed by Ava Labs, implementing the Avalanche consensus approach for blockchain use cases that require a totally ordered sequence of blocks. While the original Avalanche protocol uses a DAG (Directed Acyclic Graph) structure suitable for simple asset transfers, Snowman adapts the same repeated sub-sampled voting mechanism to produce a single linear chain of blocks, which is necessary for smart contract execution where transaction ordering matters. Like Avalanche consensus, Snowman works by having validators repeatedly sample random subsets of other validators, querying their preferences and building confidence in a particular chain tip. When a validator's confidence in a block exceeds the configured threshold, it accepts the block.

Snowman inherits the key properties of the Avalanche consensus family: probabilistic finality with tunable security parameters, sub-second time-to-finality, and efficient O(k·log(n)) communication complexity per decision. The protocol achieves high throughput because the sub-sampling approach means each validator only communicates with a small random subset of the network, rather than all validators. Snowman provides strong consistency guarantees — once a block is accepted, the probability of it being reverted can be made negligibly small by tuning the protocol parameters. The protocol maintains safety as long as a sufficient fraction of the stake is held by honest validators.

Snowman is used by the Avalanche C-Chain (Contract Chain), which is the EVM-compatible smart contract chain in the Avalanche network. It is also used by the Flare Network, which implements the Snowman protocol for its Layer 1 blockchain focused on providing decentralized data to smart contracts. Within the Avalanche ecosystem, Snowman runs on the C-Chain and P-Chain (Platform Chain), while the X-Chain uses the DAG-based Avalanche protocol. The Snowman protocol demonstrates how the Snow consensus family can be adapted for different data structures while maintaining the same fundamental consensus properties, allowing the Avalanche platform to use different chain structures optimized for different purposes within a single network.