Verification Protocol

How compute results are verified across the network

Experimental Feature

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

Overview

The verification protocol defines how compute results are validated across the ComputeNet network. It establishes the rules for proof generation, verification, and dispute resolution.

Verification Layers

ComputeNet employs a multi-layered verification approach:

Layer 1: Cryptographic Proofs

Zero-knowledge proofs or SNARKs that mathematically verify computation correctness without revealing inputs.

Layer 2: Validator Attestations

Multiple validators independently verify results and sign attestations that are aggregated for consensus.

Layer 3: Economic Security

Validators stake collateral that can be slashed for incorrect attestations, providing economic incentives for honest behavior.

Verification Flow

The standard verification flow proceeds as follows:

  1. Client submits a compute job with verification requirements
  2. Primary validator executes the job and generates a proof
  3. Secondary validators verify the proof and add attestations
  4. Once threshold is reached, the result is finalized
  5. The compute receipt is anchored in the consensus layer

Dispute Resolution

If validators disagree on a result, the protocol initiates a dispute resolution process:

  • Disputing party submits a challenge with evidence
  • Additional validators are selected to re-verify
  • Majority decision determines the correct result
  • Incorrect attesters are penalized

Experimental Status

The dispute resolution mechanism is still being refined. Current testnet implementations use a simplified model.

Verification Modes

Clients can choose different verification modes based on their requirements:

  • Full verification — Maximum security with cryptographic proofs
  • Optimistic verification — Faster finality with fraud proofs
  • Threshold verification — N-of-M validator attestations