DMQ Framework: On-Chain Penalty Execution & MEV Attack Vectors

To move beyond theoretical modeling and address potential execution vulnerabilities, I have deployed the DMQ framework on the Sepolia testnet and compiled the empirical execution data. 1. Proof of Execution: On-Chain Penalty Enforcement The core thesis of DMQ is that panic states must programmatically enforce subordination without relying on secondary market liquidity. The following transactions demonstrate this exact mechanical execution: State Transition (Triggering Panic Level 3000): Tx Hash: 0xe0691600dfd119b9dc8d4878f1f57f2b105d2ff1e85d474618a5e1e59ec81420 Penalty Execution (The Haircut): Tx Hash: 0x22563198f8946fc7d3125bdf66eca131d5205a4e74ba6eec678a88232704c959 As verified in the second transaction, an explicit withdrawal request for 100 units under the maximum panic state bypassed T+0 liquidity constraints. The contract strictly enforced a 40% penalty (burning 40 units) and returned exactly 60 units. 2. Interactive PoC Architecture You can test the front-to-back logic and trigger the panic state directly via this simulator: https://codesandbox.io/p/sandbox/jovial-yalow-9xfx4q?file=%2Fsrc%2FApp.js 3. Open Questions for MEV Searchers & Quants While the PoC functions technically, I am analyzing two critical attack vectors where I’d appreciate the community’s critique: Continuous Curves vs. Step-Functions: The current PoC utilizes a basic if-else step-function for the penalty tiers. This clearly introduces an MEV attack surface (e.g., sandwiching withdrawals exactly at the boundary of a 1499 to 1500 velocity shift). Would transitioning to a continuous penalty curve—such as a Sigmoid function $P(v) = \frac{P_{max}}{1 + e^{-k(v - v_0)}}$—mathematically eliminate this extraction vector, or simply shift the MEV to oracle latency? Decentralizing the Velocity Oracle: The mockDepletionVelocity needs to be replaced with a trustless on-chain metric. If we rely on a TWAP (Time-Weighted Average Price) of reserve depletion, how vulnerable is this to flash-loan manipulation during a true macro panic? I invite researchers to break this model. Where does the math fail? 1 post - 1 participant Read full topic