Briefing
The core research problem is the systemic vulnerability of blockchain consensus to selfish mining , a threat significantly amplified by Miner Extractable Value (MEV), network rushing, and bribed miners, which historically drops the profitability threshold to zero in practical DAG protocols. The foundational breakthrough is the Mutual-Assured-Destruction Directed-Acyclic-Graph (MAD-DAG) , a novel protocol mechanism that strategically leverages the DAG structure to align participant incentives by ensuring that a selfish action results in a self-inflicted loss of block value. This new theory’s most important implication is the establishment of a practical, non-zero security threshold (up to 31%) against rational, MEV-aware adversaries, fundamentally securing the integrity of high-throughput decentralized systems.
Context
Before MAD-DAG, the challenge of securing high-throughput consensus protocols, particularly those based on Directed Acyclic Graphs (DAGs), was the inability to maintain security against rational, profit-maximizing miners under realistic network conditions. Established protocols like Bitcoin and previous DAG-based designs, when analyzed under a model that accounts for MEV and network advantages (rushing), were proven to have a 0% security threshold , meaning any miner, regardless of size, could profit from selfish mining. This theoretical limitation represented a foundational barrier to creating truly secure, high-latency, high-throughput decentralized systems.
Analysis
MAD-DAG introduces a new incentive layer to the DAG structure, fundamentally shifting the game-theoretic landscape of block production. The mechanism is rooted in the principle of Mutual Assured Destruction , which is formalized by a conservative reward rule within the protocol’s block validation logic. This rule is designed such that a miner attempting to withhold a block to execute a selfish mining strategy (e.g. to capture MEV or exploit rushing) will find the value of their hidden block significantly diminished or destroyed upon eventual publication. This disincentive is rigorously modeled using a Markov Decision Process (MDP) to analyze the rational miner’s optimal strategy, demonstrating that the protocol makes the selfish strategy sub-optimal unless the miner controls a significant fraction of the network’s computational power.
Parameters
- Security Threshold (Adverse Conditions) → 11% to 31% – The minimum fraction of computational power a miner requires to profit from selfish mining, a critical metric that is 0% in prior protocols.
- Modeling Framework → Markov Decision Process (MDP) – The formal game-theoretic tool used to model and analyze the rational adversary’s optimal selfish mining strategy.
Outlook
The immediate next step is the formal extension of the MAD-DAG principles to Proof-of-Stake environments, as the current model focuses on Proof-of-Work. In the long term, this research provides a new blueprint for designing incentive-compatible consensus algorithms where MEV is not merely mitigated but is structurally weaponized against the selfish actor, paving the way for a new generation of high-throughput, low-latency, and provably secure decentralized finance (DeFi) architectures within the next three to five years.
Verdict
MAD-DAG provides a rigorous, mechanism-design solution that elevates the security threshold of DAG-based consensus, marking a significant theoretical advance in protocol resilience against rational adversaries.
