Speculative MEV
In the previous discussion, we covered ordinary MEV attacks, but when attackers cannot see specific transaction content and cannot target specific transactions, they may adopt another attack method: speculative MEV attacks.
This attack method is specifically manifested as: in MEVless, attackers can pre-position a transaction and submit the transaction hash to the chain. When the execution block begins and everyone submits transaction content, attackers can judge and calculate whether they have attack profits based on the transaction content submitted by other users. If there are profits, they submit their transaction content; if not, they choose to refuse to submit their transaction content, achieving sandwich attacks.
This approach is difficult to succeed in the MEVless protocol because speculation requires costs. When attackers find it unprofitable and choose to abandon submitting their transaction content, the prepayment they paid when submitting the txHash earlier becomes wasted. The more attackers want their transactions to be ranked higher in the ordering, the greater the prepayment cost they need.
Then, some readers might wonder: if block-producing nodes themselves perform MEV attacks, then this prepayment is essentially paying themselves, completely offsetting the attack cost. In this case, how should it be solved?
Consensus Layer Optimization
Indeed, when block-producing nodes themselves perform MEV attacks, this prepayment will be covered by their own block rewards, so there will be no waste of attack costs. Our approach to weakening this speculative method is: making it impossible for miners to predict whether the next block will be produced by them, thus dramatically increasing their speculative costs. For this purpose, We need to make the following design at the consensus layer:
-
Access consensus: Proof of Burn L1-token, people must burn some L1 tokens (ETH/USDT/USDC) on L1 to gain the right to join the miner group.
-
Block production consensus: VDF based on L1 hash, we use the hash of the latest block on ETH L1 as input, generate random output values with VDF, and compare all miners’ output values. The node with the maximum value of (output value * burned L1 token amount) will become the block-producing node for this round.
-
Final consensus: Upload the blocks produced in the above steps to ETH L1 every 3 ETH slots as a cycle, and select the fork with the maximum value of (output value * burned L1 token amount) as the finalized branch. Once finalized on ETH, it cannot be rolled back.
Moreover, this consensus design has better MEV resistance as decentralization increases, because the more miners there are, the harder it is for each miner node to predict whether they will be the next block-producing node, and the higher their risk of speculative MEV.