Consensus

The consensus mechanism for verified compute coordination

Experimental Feature

This feature is experimental and may change significantly in future versions.

Overview

ComputeNet uses a specialized consensus mechanism designed for verified compute coordination. Unlike traditional blockchain consensus that orders transactions, ComputeNet consensus focuses on agreeing on compute results and their proofs.

Consensus Goals

The consensus mechanism is designed to achieve:

  • Finality — Verified results cannot be reversed
  • Liveness — Jobs are processed in bounded time
  • Safety — Invalid results are never finalized
  • Efficiency — Minimal overhead for verification

Consensus Flow

The consensus process for a compute job proceeds as follows:

  1. Proposal — Primary validator proposes a result with proof
  2. Verification — Committee validators verify the proof
  3. Attestation — Verifiers sign attestations
  4. Aggregation — Attestations are aggregated
  5. Finalization — Result is finalized when threshold is reached

Consensus Parameters

Verification threshold

2/3 + 1

Committee size

Variable (testnet)

Epoch duration

TBD

Finality time

TBD

Validator Selection

For each compute job, a committee of validators is selected to participate in consensus. Selection is based on:

  • Stake weight
  • Historical performance
  • Geographic distribution
  • Computational capacity

Slashing Conditions

Validators can be slashed for consensus violations:

  • Attesting to invalid proofs
  • Double-signing conflicting results
  • Extended downtime during assigned duties
  • Collusion with other validators

Experimental Status

The consensus mechanism is under active research. Parameters and mechanisms may change significantly before mainnet.