Establish whether graph-traversal order is proposer-independent

Establish whether leaving intra-round block order to graph traversal in Mysticeti is beyond a proposer’s influence, so that the round-only ordering rule cannot be manipulated through the traversal inputs.

Background

The paper compares several intra-round tiebreak rules for Mysticeti, including author identity, block digest, seeded digest, and round-only ordering with the remaining order determined by graph traversal. The round-only rule produced the fairest measured baseline among the tested alternatives and is not directly grindable through the block digest.

However, the authors did not establish that a proposer cannot influence graph traversal order through other aspects of its block or the DAG. Consequently, they decline to recommend the rule until its resistance to such manipulation is determined.

References

Round only, with intra-round order left to graph traversal. This measured the fairest of the three rules we ran ($52.7\%$, against $53.7\%$ for the digest rule and $56.3\%$ for production), and it is not grindable through the block digest, because the digest plays no part in it. We stop short of recommending it because we did not establish that traversal order is beyond a proposer’s influence, and an ordering rule is only as good as its worst manipulable input. Establishing that is worthwhile future work.

— Competition, Collusion, and Corruption: The Spectrum of MEV Attacks on DAG-Based BFT Consensus Protocols  (2609.20069 - Mirzaei et al., 17 Sep 2026) in Section 6, subsection “A grind-resistant tiebreak,” third bullet point

Defending against silent-except-leader means changing what a leader round is worth, which is a change to the ordering rule (P2) rather than to the tiebreak, and we leave it open.

— Competition, Collusion, and Corruption: The Spectrum of MEV Attacks on DAG-Based BFT Consensus Protocols  (2609.20069 - Mirzaei et al., 17 Sep 2026) in Section 6, subsection “The Limits of The Whole Family”