Research Summary: An Economic Analysis of EIP1559

TLDR:

EIP1559 is proposal to make several tightly coupled additions to Ethereum’s transaction fee mechanism, including variable-size blocks and a burned base fee that rises and falls with demand. The proposal has attained considerable interest given its potential to reduce ETH’s free float supply in the long run, improve UI in times of fee congestion, and disincentive miners from accepting payments in any assets other than ETH. The paper assesses the game-theoretic strengths and weaknesses of the proposal and explores some alternative designs.

Citation

  • Roughgarden, T. (2020). Transaction Fee Mechanism Design for the Ethereum Blockchain: An Economic Analysis of EIP-1559.

Link

Core Research Question

  • Can EIP-1559 meaningfully affect the existing game-theoretic impediments in Ethereum to various forms of network attacks (e.g. double-spend attacks, censorship attacks, denial-of-service attacks)?

Background

  • Bitcoin pioneered the use of first-price auctions to determine which transactions miners are incentivized to prioritize in their block templates. In such a model, users bid up fees on the basis of demand for quick transaction settlement.
  • First-price auctions are challenging for users to reason about because a user’s optimal gas price depends on the gas prices offered by other users at the same time.
  • In times of high fee volatility, first-price auctions have led to inefficiencies as users that demand quick settlement, such as those engaging in arbitrage, tend to overpay to have their transactions included in the next block.
  • Wallets and smart contract applications are also impacted by fee volatility since it becomes impossible to determine appropriate fee thresholds.
  • EIP-1559 intends to address this issue by introducing a posted-price auction that regulates fee prices and smoothens the impact of demand spikes on transaction fees.
  • This mechanism enables the network to better adjust for congestion via dynamic block sizes (doubling the current block gas limit), as well as two new fee parameters; the base fee and the tip.
  • The base fee is a protocol-computed reserve price (per unit of gas). Paying the base fee is a prerequisite for inclusion in a block. Much like a difficulty adjustment algorithm, the base fee is adjusted to reflect network demand on the basis of block utilization. The protocol actively targets an average block size of 12.5M gas. If the previous block used less than 12.5M gas, the base fee goes down. If the previous block used more than 12.5M gas, the base fee goes up (following an exponential function).
  • Since miners could cartelize, spoof transactions, or collude with users off-chain to maintain fees artificially high, the base fees paid in every block are destroyed, or burned. This is one of the biggest points of contention of EIP-1559, given the potential impact such a policy would have on miner income. In spite of the negative impact on miner revenue, EIP-1559 carries strong community support because this design would lead Ethereum’s inflation rate to decrease.
  • In order to incentivize miners beyond the block subsidy, EIP-1559 also introduces a tip parameter that, in times of high demand for transaction settlement, can revert fee price discovery back to a first-price auction. Unlike the base fee, miners receive tips as part of their revenue.

Summary

  • No transaction fee mechanism, EIP-1559 or otherwise, is likely to substantially decrease average transaction fees; persistently high transaction fees is a scalability problem, not a mechanism design problem.
  • EIP-1559 should decrease the variance in transaction fees and the delays experienced by some users through the flexibility of variable-size blocks.
  • EIP-1559 should improve the user experience through easy fee estimation, in the form of an “obvious optimal bid,” outside of periods of rapidly increasing demand.
  • EIP-1559 should decrease the risk of economic abstraction, whereby miners accept payments with assets other than ETH. As such, the proposal makes ETH the de facto instrument to pay for transaction fees at the block level.
  • The game-theoretic impediments to double-spend attacks, censorship attacks, denial-of-service attacks, and long-term revenue-maximizing strategies such as base fee manipulation appear as strong under EIP-1559 as with first-price auctions.
  • EIP-1559 should at least modestly decrease the rate of ETH inflation in the long run through the burning of transaction fees.
  • The seemingly orthogonal goals of easy fee estimation and fee burning are inextricably linked through the threat of off-chain agreements.
  • Alternative designs include paying base fee revenues forward to miners of future blocks rather than burning them; and replacing variable user-specified tips by a fixed hard-coded tip.
  • EIP-1559’s base fee update rule is somewhat arbitrary and should be adjusted over time.
  • Variable-size blocks enable a new (but expensive) attack vector: overwhelm the network with a sequence of maximum-size blocks.

Method

  • The author begins by providing a formal game-theoretic framing of Ethereum’s current fee price discovery mechanism, which follows a first-price auction model.
  • The current first-price auction mechanism is then compared to EIP-1559, and the author lays out the key concepts that underpin the proposal.
  • Then, the author formalizes the concepts of a “good user experience” and “easy fee estimation” under EIP-1559’s posted-price mechanism.
  • The desirable game-theoretic guarantees of EIP-1559 are then evaluated, and the extent to which the current proposal satisfies such guarantees is examined.
  • Lastly, the author investigates the possibility of collusion by miners over long time scales and describes worthy directions for further design experimentation.

Results

  • The biggest potential benefits of the proposed changes with EIP-1559 are as advertised: easy fee estimation, in the form of an “obvious optimal bid” outside of periods of rapidly increasing demand; lower variance in transaction fees due to increased flexibility in block size; game-theoretic robustness to protocol deviations and off-chain agreements, both at the scale of a single block; and reduced inflation due to fee burning.
  • Most of the major risks in implementing EIP-1559 are the same as those for any major change to Ethereum’s L1: implementation errors; a fork caused by some parties rejecting the changes; extra complexity at the consensus layer; additional parameters to be tweaked with every network upgrade; and the spectre of unforeseeable downstream consequences.
  • Additional risks specific to EIP-1559 include the possibility of a hostile reception by miners (due to lost revenue from burned transaction fees) and a new (albeit expensive) attack vector enabled by variable-size blocks.

Implications & Follow-Ups

  • Cryptoeconomic monetary policy is a widely debated topic, as it is often portrayed as one of the unique value propositions of crypto assets.
  • Mechanism design, in this context, is intended to optimize a network’s security budget while minimizing risk.
  • Given the infancy of this research field, the intrinsic relationship between transactional capacity, monetary inflation and network security has been poorly understood from a game-theoretic perspective.
  • “An Economic Analysis of EIP1559” is one of the first attempts at fully mapping all economic actors at play and the factors that drive their decisions. Its thoroughness will likely inspire similar analysis of other protocols.
  • A considerable implication of this work is that it shows that it is possible to link on-chain activity with network value through market-impacting operations such as token “burns”, as their impact is similar to a stock buyback, or a fiat currency being taken out of circulation.
  • Such market-impacting operations may prove possible the implementation of a unified mechanism that provides deflationary pressure in the long-run (via token burns) while still maintaining a block-by-block subsidy.

Discussion Topics

  • Do the benefits of EIP-1559 justify the risks of adopting it in light of potential miner/staker collusion?
  • Should ETH1 adopt EIP-1559 ahead of ETH 2.0?
  • Will tips be sufficient to finance network security (by incentivizing miner competition) in a sustainable fashion?
  • Can off-chain agreements such as out-of-band payments undermine EIP-1559’s posted-price auction?
3 Likes

An interesting paper was recently shared on Research Pulse that showcases the predominance of out-of-band payments. Since transaction ordering within an Ethereum block can greatly impact the profitability of arbitrageurs, there has been anecdotal evidence of traders paying miners directly via out-of-band payments (outside of block rewards) to have their transactions included first.

The paper validates the notion that out-of-band payments do occur. It found that the 21 most prominent mining pools on Ethereum, which combined control over 50% of the hashrate, do include transactions in their blocks that were never sent to the mempool. A caveat is that the data also seems to include payouts from mining pools to individual miners.

Even in light of miner payouts, this implies that there is an already-existing direct relationship between 3rd parites and Ethereum miners. In the context of EIP1559, this relationship may strengthen since miner payouts would be affected by fee burning. As a response, miners could further collude with such 3rd parties to require out-of-band payments in addition to a base fee. Tips would only affect transaction inclusion if they are higher than such out-of-band payments.

This is perhaps a reason to evaluate the implementation of EIP1559 exclusively on ETH2, as opposed to ETH1.

2 Likes

Is there any modeling you’re aware of that estimates the changes to miner revenue?

This is perhaps a reason to evaluate the implementation of EIP1559 exclusively on ETH2, as opposed to ETH1.

I’m a little fuzzy on where EIP1559 fits into ETH2, but I think that’s because I’m a little fuzzy on how ETH2 transactions work in general. What resources do you recommend for me to check out to clarify that picture?

1 Like

To my knowledge, no formal work has been done on the economics of EIP1559 under eth2. Structurally, the overall schema would likely be identical. However, PoS would certainly change the weight of some of the parameters.

If eth2 features 1024 shards, overall gas consumption would increase considerably. At the same time, increased throughput would likely make network fees much lower. Combined, those two factors could potentially balance things out and ultimately lead to a proportionate amount of gas being burned. In theory, PoS could also solve the aforementioned issue of off-chain agreements, but only if validator cartelization does not occur.

3 Likes