Skip to main content

Fault Proofs

Nexis Appchain’s security model relies on fault proofs, a mechanism that allows anyone to challenge invalid state transitions. This document explains the technical details of how the fault proof game works.

Overview

Optimistic Security

State roots are assumed valid unless proven wrong

Bisection Game

Binary search to narrow disputes to a single instruction

73-Step Depth

Maximum game depth allows 2^73 instruction traces

Economic Bonds

Both parties stake ETH, winner takes all

How Fault Proofs Work

The Problem

Traditional blockchains validate every transaction on every node. This is secure but expensive. Rollups optimize by executing transactions off-chain and only posting commitments on-chain. But how do we ensure these commitments are correct? Fault Proofs: Instead of validating everything, we optimistically assume correctness and allow anyone to prove incorrectness.

The Solution

The Fault Proof Game

Game Structure

The fault proof game is a turn-based bisection protocol that narrows a disagreement over an entire execution trace to a dispute over a single instruction. Players:
  • Proposer: Claims state root X is correct at block N
  • Challenger: Claims state root Y is correct at block N
Goal:
  • Identify the exact instruction where they disagree
  • Execute that instruction on-chain to determine who’s right

Game Tree

Bisection Depth

Why 73 levels?
The 73-level tree can accommodate up to 2^73 ≈ 9.44 quintillion instructions - far more than any realistic execution trace.

Implementation

Dispute Game Contract

Fault Dispute Game

Position Encoding

The position in the game tree is encoded as (depth, indexAtDepth):

MIPS VM

At the deepest level of the game, a single instruction must be executed on-chain. Nexis uses a MIPS emulator for this:

Why MIPS?

  • Simple ISA: Easy to implement on-chain
  • Deterministic: No undefined behavior
  • Well-supported: Go, Rust, C all compile to MIPS
  • Small: Minimal on-chain execution cost

On-Chain Execution

Example: Dispute Resolution

Economic Security

The fault proof system is secured by economic incentives:

Bond Mechanics

Attack Cost Analysis

To successfully attack the network with an invalid state: Attacker Costs:
  1. Proposer bond: 1 ETH
  2. Must defeat all challengers in bisection games
  3. Each honest challenger requires defeating: +1 ETH
  4. With N honest challengers: N+1 ETH total
Defender Costs:
  1. Single honest validator: 1 ETH bond
  2. If correct, receives attacker’s bond: +1 ETH profit
Equilibrium:
  • As long as one honest party exists with 1 ETH, attacks fail
  • Attacker loses all bonds
  • Network remains secure

Griefing Resistance

Learn More

Block Validation

How blocks are validated and finalized

Consensus Mechanism

Understand the consensus architecture

Run a Validator

Help secure the network

Infrastructure Overview

Complete architecture documentation

Want to deep dive into MIPS VM? Check out the Optimism Cannon documentation which Nexis builds upon.