Competition, Collusion, and Corruption: The Spectrum of MEV Attacks on DAG-Based BFT Consensus Protocols

Competition, Collusion, and Corruption: The Spectrum of MEV Attacks on DAG-Based BFT Consensus Protocols (arXiv:2609.20069) introduces a unified attack space for Maximal Extractable Value (MEV) in Directed Acyclic Graph (DAG)-based Byzantine Fault-Tolerant (BFT) consensus protocols.

• Attack Taxonomy: The paper categorizes MEV attacks into four families: the adversary (actor, coordination, cost), the protocol (DAG construction and linearization), the target (victim and success metrics), and the deployment (committee size, stake, latency).
• Key Findings: Experimental evaluation across six production protocols reveals that protocol structure, rather than attacker effort or resources, dictates which attacks succeed. For instance, a single bribed validator can outperform multiple colluding Byzantine nodes, and leader-anchored DAGs reward lone silent attackers while certified DAGs are vulnerable to exclusion attacks.
• Mitigation Limits: The study finds that standard mitigations like reordering after the fact or intra-round tie-breaking rules are insufficient, as they often leave residual advantages or simply shift the attack vector rather than eliminating it.

Conclusion: The research emphasizes that defending DAG-BFT protocols requires identifying and closing specific structural gates within the protocol design, as these gates determine the feasibility of MEV extraction independent of the adversary's budget or size.

Competition, Collusion, and Corruption: The Spectrum of MEV Attacks on DAG-Based BFT Consensus Protocols examines how Byzantine Fault-Tolerant (BFT) systems can remain safe and live while still exposing a significant economic attack surface: proposer-controlled transaction ordering. The paper starts from the observation that BFT guarantees are primarily about agreement and fault tolerance, not fairness or neutrality in ordering. In DAG-based BFT protocols, where blocks or proposals may be organized in a directed acyclic graph rather than a single linear chain, proposers can still choose which client-submitted transactions to include, in what order, and when. That discretion enables maximal extractable value (MEV) behavior: front-running, sandwiching, priority extraction, or other forms of profit that arise when the committed order differs from the order in which users submitted transactions.

The central contribution is a taxonomy of MEV attacks along a “spectrum” from market-level competition to explicit collusion and outright corruption. In the competitive regime, proposers or validators may independently optimize their own inclusion decisions, leading to race-like or auction-like dynamics over profitable transaction order. In the collusive regime, actors coordinate—proposers with users, proposers with other validators, or multiple proposers with each other—to extract value that would be harder to obtain in a purely competitive setting. In the corrupt regime, the paper considers stronger failures such as compromised proposer power, bribed ordering, or protocol-level manipulation that turns ordering discretion into a direct economic weapon. This framing is useful because it separates MEV as an emergent incentive problem from MEV as a coordinated or malicious abuse of protocol roles.

The work matters because many BFT and DAG-based consensus designs are evaluated mainly on throughput, latency, and fault tolerance, while under-specifying the economic security of transaction ordering. For systems used in finance, institutional ledgers, or high-throughput decentralized applications, ordering power can be a first-class source of value and harm. By mapping how MEV can arise across different levels of coordination and corruption, the paper helps motivate stronger ordering mechanisms—such as proposer-independent ordering, private transaction submission, commit-reveal schemes, fairness constraints, or incentive-compatible ordering auctions—and clarifies that safety and liveness alone do not guarantee a neutral or economically secure transaction layer.