Priority Gas Auctions on Ethereum and Flashbots
One of the most interesting implementations of auctions in the real world is within the Ethereum blockchain. Mind that these observations hold only to the proof of work implementation, and the proof of stake implementation offers a slight change in mechanics which I have yet to familiarize myself with. Very similarly to the reading given in class, Ethereum miners select which transactions to include based on the gas fee bid by transaction senders. This becomes much more complex when we account for the idea of searchers on the blockchain. A searcher can be defined as a user who is scanning the blockchain and mempool for opportunities where a transaction can be made to extract some sort of profit. This could be an arbitrage, liquidation, or some other asymmetry exposed in a smart contract. These searchers have a vested interest in having their transaction included, as this transaction would result in some profit P for the searcher. Because of this limitation, a searcher would only be interested in bidding an amount B such that B < P. this is because if they were to bid more than P for this transaction, they would no longer profit from the opportunity they are submitting.
Enter Flashbots. Flashbots is an innovation to the ethereum blockchain which modifies the miner’s client to remove the need for auctions to have transactions included. Its goal was to “democratize” the process of having transactions included, by replacing the auction system with a simple bribe. As opposed to searchers bidding to have their transactions included, they would provide a bribe amount to miners which incentivizes its inclusion. This drastically changes the dynamics of searcher methodology.
In the Flashbots system, as noted by this user, bribes will always converge to 99.9+% of the profit P for a searcher. The reason for this is simple. In a bribe, a user has no feedback on the competitiveness of their bribe. Because of the lack of feedback, it is optimal to bribe as much as possible to ensure the maximum likelihood of inclusion. In the traditional auction system, the strategy is significantly more complex. This is due to the concept of “gas” on ethereum. Gas is a function as the amount of operations that will be conducted on the blockchain when a transaction is processed. This can be highly optimized by building efficient software. Because searchers can further optimize their code to reduce gas spent, it is very possible that a strategy that yields a very high profit requires very little amount bid. This opens up the opportunity for searchers to use nuanced strategies to outbid their opponents. This article details some of these strategies in more depth.
The implications on our coursework are of course the idea of auctions, and the game theoretical implications of attempting to extract profit based on an input amount bid. One other interesting implication is in the idea of arbitrage itself. An arbitrage is typically a transaction through several pools, which can be thought of as individual sellers. An arbitrage is a path taken through sellers that can yield a profit. That is, a person could swap an amount X of one asset for amount Y of another asset, and then find another seller to exchange their Y for X+1. This is typically found by way of finding the optimal weight for their path. In efficient markets, the prices typically settle into equilibrium, and in this case there are no arbitrage opportunities present. There are many implications of game theory, equilibrium and auctions when trading on the blockchain.