Bisection Game

A bisection game (also called an interactive fault proof or dispute game) is the mechanism optimistic rollups use to resolve disagreements about L2 state transitions on-chain. When a challenger disputes a proposed state root, the bisection game protocol narrows the disagreement from an entire block of transactions down to a single instruction step, which is then executed on L1 to determine the correct outcome.

How bisection works

1. Challenge initiation — A challenger posts a bond and asserts that a proposed output root is incorrect.
2. Binary search — The proposer and challenger take turns bisecting the execution trace. Each round, the disputed range is split in half, and each party commits to intermediate state hashes.
3. Single-step resolution — After sufficient rounds of bisection, the dispute narrows to a single instruction. This instruction is executed on-chain using a fault proof virtual machine (e.g., Cannon's MIPS-based FPVM in the OP Stack).
4. Bond resolution — The losing party forfeits their bond to the winner.

Chess clock mechanism

Both parties have a limited time budget (a "chess clock") to respond at each step. If a party's clock runs out, they lose by default. This prevents indefinite stalling. However, timer logic itself has been a source of vulnerabilities — in March 2024, Offchain Labs disclosed that timer inheritance through grandparent claim relationships could be exploited to force acceptance of fraudulent roots.

Security implications

• The bisection game is the sole enforcement mechanism preventing invalid state transitions from being finalized in optimistic rollups.
• Its correctness is critical to bridge security — a compromised bisection game allows draining all bridged assets.
• Timing, bonding, and game resolution logic are all active attack surfaces requiring thorough auditing.