PaperlayerView exact revision on arXiv

Source-linked AI summary

Transaction Fee Mechanism Design for the Ethereum Blockchain: An Economic Analysis of EIP-1559

Tim Roughgarden

arXiv:2012.00854v1cs.GTcs.DCcs.DSecon.TH

TL;DR

Whether EIP-1559 improves Ethereum’s fee mechanism is hard to assess because its “better UX” goal is vague and must be mathematically defined. The report analyzes the mechanism and alternative designs game-theoretically, identifying potential benefits including easier fee estimation, lower transaction-fee variance, robustness to deviations and off-chain agreements, and reduced inflation from fee burning.

  • Problem

    Whether EIP-1559 improves Ethereum’s fee mechanism is hard to assess because its “better UX” goal is vague and must be mathematically defined.

    Source: §4.2 · p. 13

  • Method

    The report analyzes EIP-1559 and alternative designs through game-theoretic analysis of miner and user incentives.

    Source: §6.4 · p. 28, §8.5.1 · p. 44

  • Results

    EIP-1559’s potential benefits include easier fee estimation, lower transaction-fee variance, game-theoretic robustness to protocol deviations and off-chain agreements, and reduced inflation from fee burning.

    Source: §10 · p. 54

  • Takeaways & Limitations

    The case for EIP-1559 is strong for those who favor continued consensus-layer innovation, while others may prefer the status quo.

    Source: §10 · p. 54

    The base-fee adjustment is underspecified because confirmed transactions cannot reveal whether full blocks reflect a base fee that is too low or exactly right.

    Source: §2.3.1 · p. 6

Abstract

from arXiv · show

EIP-1559 is a 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. This report assesses the game-theoretic strengths and weaknesses of the proposal and explores some alternative designs.

1 TL;DR

EIP-1559 replaces Ethereum’s first-price auction with a demand-adjusted, burned base fee and optional miner tips. The report finds it is unlikely to substantially reduce average fees, while identifying potential benefits, risks, and alternative designs.

1.1 A Brief Description of EIP-1559

  • EIP-1559 retains a first-price auction as Ethereum’s starting point but adds a burned base fee adjusted to block demand, variable-size blocks, and optional miner tips.The base fee rises after larger-than-target blocks and falls after smaller-than-target blocks; blocks can reach twice the target size.Source: §1.1 · pp. 2–3

1.2 Ten Key Takeaways

  • No fee mechanism, including EIP-1559, is likely to substantially reduce average transaction fees because persistently high fees reflect a scalability problem.Source: §1.2 · p. 3
  • Variable-size blocks should reduce transaction-fee variance and some users’ delays, while an “obvious optimal bid” should ease fee estimation outside rapidly increasing demand.Source: §1.2 · p. 3
  • EIP-1559’s short-term miner incentives and game-theoretic impediments to double-spend, censorship, denial-of-service, and long-term revenue-maximizing attacks appear as strong as under first-price auctions.Source: §1.2 · p. 3
  • Burning transaction fees should modestly decrease ETH inflation, but off-chain agreements link fee burning to easy fee estimation.Source: §1.2 · p. 3
  • Alternatives include forwarding base-fee revenue to future miners or using a fixed tip; the update rule may need adjustment, and maximum-size blocks enable an expensive network-overwhelm attack.Source: §1.2 · p. 3

1.3 Organization of Report

  • The report moves from mechanism and market context through user experience, game-theoretic guarantees, and miner collusion to alternatives, additional benefits, and conclusions.Source: §1.3 · p. 4

2 Transaction Fee Mechanisms in Ethereum: Present and Future

This section explains Ethereum transactions and its first-price fee mechanism, then describes EIP-1559’s history-based base fee, variable-size blocks, and tips. It presents an informal rationale for coupling these changes to make fee estimation easier.

2.1 Transactions in Ethereum

  • Transactions specify a gas limit and gas price, while miners select and order transactions within blocks capped by total gas.A transaction’s gas limit measures its computational and storage cost, and its gas price specifies the amount offered per unit of gas.Source: §2.1 · pp. 4–5

2.2 First-Price Auctions

  • Ethereum’s status quo is a first-price auction: confirmed users pay their bid to the including miner, who commonly fills blocks by highest gas prices.Users pay the specified gas limit times gas price if confirmed, and miners receive that payment for included transactions.Source: §2.2 · p. 5, First-Price Auctions · p. 5

2.3.1 Burning a History-Dependent Base Fee

  • A protocol-set reserve base fee depends only on preceding blocks and is burned, but full blocks alone cannot reveal whether demand exceeds its level.Because the chain records confirmed transactions but not those priced out, a sequence of full blocks does not indicate whether the base fee is too low or exactly right.Source: §2.3.1 · p. 6

2.3.2 Variable-Size Blocks

  • EIP-1559 doubles the maximum around a target size and adjusts the base fee up or down with block size, using past sizes as a demand proxy and potentially easing fee and delay variance.With a 12.5M-gas target, the proposal makes the maximum 25M gas; larger-than-target blocks raise the base fee and smaller ones lower it.Source: §2.3.2 · p. 6

2.3.3 Tips

  • Transactions specify a fee cap and miner-paid tip; tips reward inclusion and priority while the base fee is burned.The creator pays per gas min(base fee + tip, fee cap); small tips can suffice under stable demand, while larger tips can seek rapid or preferred inclusion.Source: §2.3.3 · p. 7

2.4 An Informal Argument for EIP-1559

  • The informal case is that bid-dependent first-price fees complicate user reasoning, while inclusion-dependent second-price designs invite miner manipulation, motivating a transaction-independent base fee whose proceeds miners cannot reclaim.The argument says base-fee revenue must be burned or otherwise withheld from miners to prevent them colluding with users off-chain to simulate a first-price auction.Source: §2.4 · p. 8
  • The narrative further argues demand needs an on-chain proxy and user-set tips deter empty blocks while accommodating high-value transactions; burning tips could push their market off-chain.The proposal uses variable block sizes as a demand signal and transfers tips to miners.Source: §2.4 · p. 8

3 The Market for Ethereum Transactions

The market-clearing outcome allocates scarce block capacity to the highest-value transactions at a price balancing gas demand and supply. EIP-1559 may reduce overpayment during stable demand and smooth fees during volatile demand, but it cannot by itself solve high market-clearing prices caused by excess demand.

3.1 Market-Clearing Prices and Outcomes

  • The market-clearing price equates gas demand with 12.5M-gas block supply, yielding an outcome that fully uses capacity for the highest-value transactions.The report treats this as the ideal because it maximizes block value subject to the supply constraint; revenue is a side effect, and positive miner marginal costs shift the relevant threshold.Source: §3.1 · pp. 9–10

3.2.1 The Problem of High Market-Clearing Prices

  • Fees cannot remain significantly below the market-clearing price during stable demand, and reducing an already high clearing price requires more supply or less demand—not mechanism design alone.At below-clearing fees, demand exceeds supply and users with high willingness to pay have incentives to raise fees or make side agreements; scaling can increase throughput or move computation off-chain.Source: §3.2.1 · pp. 10–11

3.2.2 Two Potential Benefits of EIP-1559

  • During stable demand, users can treat the known-in-advance base fee as a proxy for the market-clearing price, reducing first-price-auction guesswork and consequent overpayment.Both the status quo and EIP-1559 strive for market clearing: first-price bidders estimate the price, while EIP-1559 adjusts the base fee toward it.Source: §3.1 · p. 10, §3.2.2 · pp. 11–12
  • 61.69 gwei maximum base fee was about 53% of the 1162.3 gwei maximum market-clearing price under fixed 12.5M-gas blocks in the demand-doubling example.With negligible tips, variable-size blocks absorbed excess demand as the base fee rose, then became smaller than target as demand returned to its original level and the base fee fell.Source: Table 1 · p. 12, §3.2.2 · p. 12
  • During volatile demand, variable block sizes can reduce fee variance, maximum fees, and some users’ delays by borrowing capacity from the near future.No mechanism can guarantee low fees when demand for EVM computation significantly exceeds supply.Source: §3 · p. 9, §3.2.2 · p. 12

4 The Purpose of EIP-1559: Easy Fee Estimation

The section frames EIP-1559’s primary purpose as making fee estimation easier and transaction fees more predictable, while treating other benefits as byproducts. It evaluates better user experience by contrasting posted prices with strategic first-price auctions.

4.1 The Problem of Fee Estimation

  • Users appear to overpay relative to market-clearing prices, and while wallet algorithms may improve, the report identifies first-price auctions as a fundamental cause requiring mechanism change.Source: §4.1 · p. 13
  • EIP-1559’s assumed primary purpose is improving Ethereum users’ experience by making optimal gas-price estimation as easy as possible; other benefits are treated as byproducts.Source: §4.1 · p. 13

4.2 Auctions vs. Posted-Price Mechanisms

  • The Amazon analogy presents posted-price shopping as economically efficient because buyers have higher willingness to pay than nonbuyers.Source: §4.2 · pp. 13–14
  • In Ethereum’s first-price auctions, users must anticipate rivals’ gas prices and may regret over- or underbidding, while aggressive low-value bids can displace higher-value transactions.Source: §4.2 · p. 14
  • A posted-price mechanism offers a take-it-or-leave-it gas price for next-block inclusion, and EIP-1559 acts like one except after large, sudden demand increases.Source: §4.2 · p. 14

5 Incentive-Compatible Transaction Fee Mechanisms … 5.3 Incentive Compatibility (Myopic Miners)

Section 5 formalizes game-theoretic guarantees for transaction fee mechanisms, focusing on single-block incentives for miners and users; longer-term incentive issues are treated in Section 7. It defines miner incentive compatibility and the stronger requirement that fake transactions impose a substantial cost.

5 Incentive-Compatible Transaction Fee Mechanisms

  • The section frames miner incentives alongside easy fee estimation and resistance to miner-user collusion, with both single-block attacks and longer-term issues in scope.The single-block focus includes fake-transaction insertion and off-chain agreements; Section 7 treats issues that manifest over longer time scales.Source: §5 · p. 14

5.1 The Basic Model

  • The basic model assumes a block gas limit G, common-knowledge marginal gas cost µ shared by miners, and mempool transactions with gas limits, values, and creator-controlled bids.A transaction’s value is its creator’s maximum willingness to pay, whereas its bid is the offered gas price and may differ from that value.Source: §5.1 · pp. 14–15

5.2 Allocation, Payment, and Burning Rules

  • A transaction fee mechanism combines feasible allocation, payment, and burning rules; miners control inclusion, while the protocol specifies on-chain payments and burns.Payment and burning rules cannot depend on transactions excluded from the current block, and the analysis restricts attention to individually rational mechanisms.Source: §5.2 · p. 15, §5.2.1 · p. 16, §5.2.2 · p. 17

5.2.1 Allocation Rules

  • An allocation rule selects included transactions from history and the mempool subject to the block-size limit, but miners ultimately control the block’s allocation.The intended first-price allocation maximizes gas-weighted bids net of miners’ gas costs.Source: §5.2.1 · pp. 15–16

5.3.1 Myopic Miner Utility Function

  • A myopic miner maximizes current-block net revenue and may costlessly add fake transactions with arbitrary gas limits and bids.The analysis ignores the block reward because it is independent of the miner’s actions in the single-block game.Source: §5.3.1 · p. 17

5.3.2 Incentive-Compatibility for Myopic Miners

  • MMIC requires miners to maximize utility by creating no fake transactions and following the allocation rule; it is closely related to credible mechanisms.Akbarpour and Li find that first-price auctions with an exogenously restricted bid space are the only static credible mechanisms under their assumptions.Source: §5.3.2 · pp. 18–19
  • Status quo first-price auctions are MMIC, while Vickrey auctions fail because miners can raise revenue by inserting fake transactions.In the example, honest Vickrey execution yields 9 in revenue, but a fake bid of 8 raises net revenue to 16.Source: §5.3.2 · pp. 18–19

5.3.3 γ-Costly Transaction Fee Mechanisms

  • First-price auctions are µ-costly but not γ-costly for γ > µ; EIP-1559 is generally γ-costly for larger γ, more aggressively punishing fake transactions.First-price auctions impose stronger fake-transaction costs when block capacity binds or marginal gas cost µ is large; otherwise, miners can fill spare block space with fakes at little cost.Source: §5.3.3 · pp. 19–20

5.4 Incentive Compatibility (Users) · 5.5 Off-Chain Agreements

The section defines user incentive compatibility through obvious value-based bids that remain optimal when other users follow the same strategy. It also defines off-chain agreements and OCA-proofness, showing how fee burning affects the scope for miner-user collusion.

5.4 Incentive Compatibility (Users)

  • UIC formalizes easy fee estimation as an obvious value-based bid that remains optimal when other users follow the same strategy.The strategy must be a symmetric ex post Nash equilibrium for every on-chain history; users need not reason about competitors’ bids beyond trusting that they follow the recommendation.Source: §5.4 · p. 20, §5.4.2 · p. 21

5.4.1 User Utility Function

  • A transaction creator’s utility is its value minus the miner payment and fee burn per gas unit if included, and zero if excluded.The creator is assumed to choose its bid to maximize this utility.Source: §5.4.1 · p. 20

5.4 Incentive Compatibility (Users)

  • First-price auctions are not UIC because the optimal bid depends on competing transactions’ values; EIP-1559 is UIC except during rapidly increasing demand.Source: §5.4.2 · p. 21

5.5 Off-Chain Agreements

  • An off-chain agreement specifies users’ on-chain bids and per-gas transfers to a miner, who agrees to mine their transactions; coalition utility counts transaction value less fee burns and miner costs.Transfers between coalition members cancel when calculating joint utility, so the coalition seeks to maximize value not lost to the protocol or mining costs.Source: §5.5 · pp. 22–23
  • In a first-price auction that burns half of revenue, users and miners can collude to earn miners 50% more revenue and give users a 25% discount, at the network’s expense.Users bid zero on-chain and communicate bids off-chain; miners keep 75% of those bids and refund 25% to users.Source: §5.5 · p. 22
  • A TFM is OCA-proof when, for every history and outstanding transaction set, some on-chain bids yield joint utility at least as high as any feasible colluding agreement.Without this property, miners and users can sometimes collude to achieve greater joint utility than any on-chain outcome.Source: §5.5 · p. 23
  • First-price auctions are OCA-proof because on-chain bids equal to transaction values achieve the same joint utility as any off-chain agreement.Their fee-burn function is zero, so the allocation objective matches coalition joint utility.Source: §5.5 · p. 23
  • Fee burning is central to EIP-1559’s response to off-chain agreements: adding a burn makes first-price auctions no longer OCA-proof, while a history-dependent base fee has no teeth unless its revenue is burned or withheld from miners.Source: §5.5 · p. 23

6 Formal Analysis of the 1559 Mechanism with Myopic Miners

The analysis finds that the 1559 mechanism is always incentive compatible for myopic miners and proof against off-chain agreements, while user incentive compatibility fails only when the base fee is excessively low.

6 Formal Analysis of the 1559 Mechanism with Myopic Miners

  • The 1559 mechanism is always MMIC and OCA-proof, while user incentive compatibility holds except during periods of rapidly increasing demand.Section 6 translates the mechanism into a formal model and proves these three guarantees.Source: §6 · p. 24

6.1 The 1559 Mechanism

  • The formalized allocation rule maximizes gas-weighted bids net of base fees and gas costs, pays miners the bid-minus-base-fee difference, and burns the base fee.The rule chooses a feasible subset of outstanding transactions subject to the block-size constraint.Source: §6.1 · pp. 24–25

6.2 The 1559 Mechanism Is Incentive Compatible for Myopic Miners

  • Myopic miners maximize utility by following the allocation rule and are disincentivized from fake transactions; the mechanism is (r + µ)-costly, with the burn strengthening that deterrent.If the base fee went to miners instead of being burned, the mechanism would be only µ-costly and fake transactions would be only mildly disincentivized.Source: §6.2 · p. 25

6.3.1 Excessively Low Base Fees

  • A base fee is excessively low when demand at price r + µ exceeds the maximum block size; a sudden demand spike can leave it low for several periods as it adjusts.In the example in Table 2, periods 2–6 have excessively low base fees.Source: §6.3.1 · p. 26, Table 2 · p. 26

6.3.2 The 1559 Mechanism Is UIC Except with Excessively Low Base Fees

  • When the base fee is not excessively low, users have a symmetric EPNE bid: set the fee cap to value and the tip to marginal gas cost; otherwise, scarce block space makes bidding a first-price auction.The obvious bid makes the mechanism a posted-price mechanism at r + µ, but the equilibrium is not a dominant strategy when others overstate fee caps.Source: §6.3.2 · pp. 26–28
  • At the symmetric EPNE, the mechanism includes transactions valued at least r + µ and excludes lower-value ones, yielding a market-clearing outcome for supply G*.The relevant supply is defined as demand at price r + µ.Source: §6.3.2 · p. 28

6.4 The 1559 Mechanism Is OCA-Proof

  • The mechanism is OCA-proof because its fixed base-fee burn makes the allocation rule’s objective identical to miners’ and users’ joint utility at truthful bids.No off-chain agreement can improve on that joint utility.Source: §6.4 · p. 28

7 Miner Collusion at Longer Time Scales

This section examines long-term miner collusion under EIP-1559 through extreme and observed coordination, attacks, and revenue-maximizing strategies, focusing on whether such behavior could benefit miners and withstand defection. It considers the strategic and practical obstacles to collusion without predicting whether miners will coordinate.

7.1 Extreme Collusion: The 100% Miner Thought Experiment

  • The thought experiment assumes one miner controls all hashrate, maximizes fee revenue over many blocks, and knows an unchanged demand curve, isolating fee-mechanism incentives.With extreme decentralization, individual miners have little reason to use non-myopic strategies such as base-fee manipulation; the model instead studies a large miner or perfectly coordinated cartel.Source: §7.1 · p. 29

7.2 First-Price Auctions with a 100% Miner

  • Under first-price auctions, a 100% miner maximizes revenue by setting the monopoly price or quantity, and restricts computation only when monopoly quantity is below the maximum block size.In that case, restricting EVM computation forces high-value transaction creators to submit higher bids; if monopoly quantity equals the maximum block size, the miner does not deviate from the intended allocation rule.Source: §7.2 · p. 31

7.3 EIP-1559 with a 100% Miner

  • When monopoly quantity is at most 12.5M gas, a 100% miner can drive the base fee to its minimum and then use the monopoly quantity strategy, producing an outcome essentially like the status quo.When monopoly quantity exceeds the target, the miner faces a demand-dependent tradeoff and can mix underfull and overfull blocks to raise net revenue; a 51% miner can also control all blocks on the longest chain.Source: §7.3 · p. 32

7.4.1 Types of Miner Coordination

  • The authors do not predict whether miners will collude under EIP-1559, but use observed behavior under the status quo as an analogy.They distinguish strategies by whether they benefit the network or miners, are detectable, and are robust against cartel members seceding; miners have coordinated on actions plausibly intended to improve network health.Source: §7.4 · p. 32, §7.4.1 · p. 33, §7.4.2 · p. 33

7.4.3 The Risk of Undetectable Coordination

  • Miners should not be expected to avoid undetectable strategies that serve their interests, so protocols should avoid them or make them game-theoretically fragile.The cited example is fake-transaction manipulation of Vickrey auctions, which could be difficult to detect and strongly incentivized for miners.Source: §7.4.3 · p. 33

7.4.4 The Lack of Detectable Coordinated Strategies in the Wild

  • Double-spend, censorship, and denial-of-service attacks have been rare to nonexistent in Ethereum; possible explanations include miners’ network interests, ETH holdings, and fear of hard-fork punishment.The authors present these as speculation, not established causes.Source: §7.4.4 · p. 34

7.4.5 Game-Theoretic Fragility

  • Although disguised monopoly price- or quantity-setting could be difficult to detect, there is little anecdotal evidence that Ethereum miners have coordinated to use it.A possible obstacle is rapidly changing, hard-to-predict demand that makes sustained coordination on a price or quantity difficult.Source: §7.4.5 · p. 34
  • Small miners can profit by leaving a monopoly-pricing cartel and filling blocks with the highest-bid transactions, while punishment may deter secession but appears difficult or detectable.The authors hypothesize that harmful strategies requiring short-term sacrifice and costly or difficult punishment may be rare in a well-secured blockchain such as Ethereum.Source: §7.4.5 · pp. 35–36

7.4.6 The Upshot

  • Double-spend, censorship, and denial-of-service attacks fail the network-interest test, while monopoly price- or quantity-setting fails the cartel-robustness test.The section identifies the most concerning coordination as miner-serving rather than network-serving and game-theoretically robust, with members incentivized to remain.Source: §7.4.6 · p. 36

7.5 EIP-1559: Will Miners Collude?

  • Under EIP-1559, detectable attacks remain detectable, and a cartel pursuing the optimal 100% miner strategy is fragile because a myopic miner can earn more by filling blocks up to 25M gas.Such a miner free rides on cartel members who leave revenue uncollected to keep future base fees low; the authors say the game-theoretic impediments remain as strong as under the status quo.Source: §7.5 · pp. 36–37
  • Because EIP-1559 burns fees, fake transactions cannot costlessly disguise a strategy simulating optimal 100% mining, making that strategy arguably less likely than under the status quo.The strategy is consequently either costly or detectable.Source: §7.5 · p. 36

7.6 Caveats

  • Six of nine miners responding to a questionnaire said they would not implement EIP-1559 under any circumstances, suggesting its profit shift could galvanize greater collusion and making deployment and acceptance explicit issues.The passage raises possible approaches including securing support from major Ethereum projects or seeking and incentivizing miner support directly.Source: §7.6 · p. 37
  • Fee-burn-driven miner revenue losses could lower hashrate and attack costs, while coordination to avoid the base fee might be more acceptable if seen as harmless despite its fragility.The discussion says less profitable miners may exit, leaving Ethereum security reduced but still supported by the unchanged block reward; miners may retain significant tips from transactions seeking special treatment if their willingness to pay exceeds the target-size market-clearing price. The proposed 12.5M-gas coordination would leave immediate revenue on the table and could be less damaging than censorship or throttling, yet might gain support if perceived as mostly harmless.Source: §7.6 · p. 37

8 Alternative Designs

This section asks whether EIP-1559’s goals can be achieved better or more simply and surveys alternative fee designs and base-fee rules.

Sections 8.1 and 8.2

  • Sections 8.1 and 8.2 argue that easy fee estimation and fee burning are linked through the threat of off-chain agreements.Source: §8 · p. 38

Section 8.3

  • Section 8.3 investigates forwarding transaction-fee revenue to future-block miners as an alternative to burning, with similar game-theoretic properties.Source: §8 · p. 38

Section 8.4

  • Section 8.4 recaps a recent transaction-fee mechanism proposal by Basu et al. [11].Source: §8 · p. 38

Section 8.5

  • Section 8.5 discusses a design that favors UIC over OCA-proofness relative to EIP-1559.Source: §8 · p. 38

Section 8.6

  • Section 8.6 explores alternative base-fee update rules.Source: §8 · p. 38

8.1 Paying the Base Fee to the Miner

Paying base-fee revenue to miners makes 1559-R equivalent to first-price bidding under off-chain agreements, so a base fee improves user experience only when burned or otherwise withheld. Burning only a fraction is equivalent to a smaller fully burned base fee.

8.1 Paying the Base Fee to the Miner

  • A simple agreement has users bid r on-chain and settle the difference from their standard first-price bids off-chain; the example uses 25M-gas blocks.The footnote says the 1559-R base fee would eventually rise to its maximum level.Source: §8.1 · p. 38
  • 1559-R passes base-fee revenue to miners, but its outcomes correspond one-to-one with first-price-auction outcomes under an off-chain agreement.The mechanism uses the same allocation rule as 1559 and treats bids below the block’s base fee as invalid.Source: §8.1 · p. 38
  • Burning an α fraction of the base fee is economically equivalent to a fully burned base fee α times as large.If the fully burned mechanism stabilizes at r*, burning half the fee would imply stabilization at 2r*, with r* burned.Source: §8.1 · p. 39
  • When demand is stable, monopoly quantity is below the 12.5M-gas target, and miner gas costs are negligible, miners can robustly sustain the monopoly price.They can raise the base fee to the monopoly price using fake transactions, then maintain it by filling blocks with real and fake transactions; the strategy favors miners at the network’s expense and is undetectable.Source: §8.1 · p. 39

8.2 Fee-Burning First-Price Auctions

Burning fees in a first-price auction does not prevent off-chain agreements: payments can move off-chain, leaving no fee burn. The same conclusion holds when any fixed positive fraction of fees is burned.

8.2 Fee-Burning First-Price Auctions

  • The alternative retains first-price auctions while defining the TFM with x = xf, p = qf, and q = pf.It is proposed for a mechanism with fee burn where easy fee estimation is not required.Source: §8.2 · p. 39
  • Fee-burning first-price auctions are not OCA-proof because miners and users have incentives to move payments off-chain.The proposition establishes this for the fee-burning first-price auction; the argument also applies when any fixed positive fraction of fees is burned.Source: §8.2 · pp. 39–40
  • With negligible miner gas cost, an off-chain agreement can yield joint utility V, while positive bids cause fee burn and zero bids may select the wrong transactions.The agreement uses zero on-chain bids and can achieve the value-maximizing transaction set, whereas an on-chain outcome reaches that joint utility only with that set and zero burn.Source: §8.2 · p. 40
  • In the obvious OCA, on-chain bids are zero and there is no fee burn, so non-trivial fee burning requires a base fee.Burning fees in a first-price auction moves all payments off-chain and leads to zero fee burning.Source: §8.2 · p. 40

8.3 Paying the Base Fee Forward

Forwarding base-fee revenue to future miners offers an alternative to burning it. The ℓ-smoothed mechanism retains EIP-1559’s user and myopic-miner guarantees while shifting tradeoffs to miner rewards, issuance, and security.

8.3.1 The ℓ-Smoothed Mechanism

  • Instead of burning base-fee revenue as EIP-1559 does, the ℓ-smoothed variant divides each block’s revenue equally among the next ℓ miners (ℓ=0 recovers EIP-1559), while miners retain current tips.Paid-forward revenue acts as a fixed bonus independent of a miner’s current actions.Source: §8.3 · p. 40, §8.3.1 · pp. 40–41

8.3.2 Properties of the ℓ-Smoothed Mechanism

  • All user and myopic-miner game-theoretic guarantees carry over; the mechanism remains (r + µ)-costly and UIC if the current base fee is not excessively low for demand.These guarantees hold for every ℓ and regardless of how the base fee and pay-forward rewards are defined.Source: §8.3.2 · p. 41
  • For demand with monopoly quantity above 12.5M gas, a 100% miner benefits when ℓ≥1; costly fake transactions deter upward base-fee manipulation, and ETH-holder caveats become moot.Lost tip revenue still discourages cartel members from lowering the base fee.Source: §8.3.2 · p. 41

8.3.3 Pros and Cons of the ℓ-Smoothed Mechanism

  • Burning favors ETH holders and makes issuance variable while keeping miner rewards and security relatively stable; paying forward favors miners and predictable inflation but makes rewards and security variable.The security comparison assumes negligible tips; the ℓ-smoothed mechanism’s security is never lower than under EIP-1559.Source: §8.3.3 · p. 42
  • With small ℓ, variable rewards create reorg incentives: at ℓ=1, a miner may extend an earlier high-fee block to orphan the next block and capture its reward.The paper suggests a large ℓ, such as 1000, so consecutive blocks have nearly identical associated total rewards.Source: §8.3.3 · p. 42
  • A blend can burn λ of base-fee revenue and forward 1−λ, balancing miners’ and ETH holders’ interests and issuance/security variability; Theorem 8.6 and the miner-collusion discussion still apply.Source: §8.3.3 · p. 42

8.4 The BEOS Mechanism

BEOS, a simplified pay-it-forward proposal by Basu et al. [11], uses fixed-size blocks and no base fee. Its pay-forward rule discourages fake transactions, but its game-theoretic guarantees are weaker than EIP-1559’s.

8.4 The BEOS Mechanism

  • BEOS is a simplified variant of the pay-it-forward design proposed by Basu et al. [11] for a transaction fee mechanism not directly related to EIP-1559.The paper calls its slightly simplified version the BEOS mechanism.Source: §8.4 · p. 42
  • With fixed-size blocks and no base fee, BEOS charges every included transaction the lowest included bid, which can make miners exclude transactions to raise revenue.For bids 10, 8, and 3 in a block with room for three, the miner includes the first two and earns 2 × 8 = 16.Source: §8.4 · pp. 42–43
  • Charging only a minimum fee in underfull blocks is toothless because miners can costlessly fill them with fake transactions bidding the existing minimum.The rule alone is equivalent to a mechanism without that minimum-fee provision.Source: §8.4 · p. 43
  • BEOS splits each block’s transaction-fee revenue evenly between its miner and the miners of the next ℓ−1 blocks, so fake transactions cost their miner a net 1−1/ℓ of their fees.The miner pays the full fees but recoups only a 1/ℓ share.Source: §8.4 · p. 43
  • BEOS is not MMIC, is only approximately UIC as the user count grows, and is not OCA-proof for ℓ≥2, making it appear uncompetitive with EIP-1559.Its simpler design has no base fee to track.Source: §8.4 · p. 43

8.5 The Tipless Mechanism: Trading Off UIC and OCA-Proofness

The tipless mechanism replaces user-specified tips with a hard-coded tip, making it always UIC while sacrificing OCA-proofness when the base fee is excessively low. It therefore reverses the 1559 mechanism’s trade-off between UIC and OCA-proofness.

8.5.1 The Tipless Mechanism

  • Users specify only a fee cap; the mechanism packs blocks with transactions bidding at least r + δ, charging r + δ per gas, burning r, and transferring δ to the miner.The hard-coded tip δ can be set to the miner’s marginal gas cost µ.Source: §8.5.1 · pp. 43–44
  • The mechanism is the “monopolistic price” design of Lavi et al., which is MMIC, approximately UIC as the user count grows, and not OCA-proof.The cited work was motivated by maximizing transaction-fee security when Bitcoin block rewards become negligible, at the expense of economic efficiency.Source: §8.5.1 · p. 43

8.5.2 Properties of the Tipless Mechanism

  • The tipless mechanism is MMIC and UIC, and with δ = µ is OCA-proof when r is not excessively low, but may fail OCA-proofness otherwise.When r is excessively low, maximizing block size need not maximize joint utility, so an off-chain agreement can improve the miner’s and users’ joint utility.Source: §8.5.2 · pp. 44–45

8.5.3 Pros and Cons of the Tipless Mechanism

  • The design simplifies bidding to one parameter, makes ct = vt optimal regardless of other bids, and has miners maximize block size using only bids of at least r + µ.Unlike in the 1559 mechanism, the tipless mechanism gives miners no lever to pressure users to increase tips.Source: §8.5.3 · pp. 45–46
  • The hard-coded tip δ is an arbitrary parameter that may require network upgrades, and low-base-fee periods may prompt off-chain deals for inclusion or block position.Such agreements can arise during rapidly increasing demand when OCA-proofness breaks down, or even with a base fee that is not excessively low.Source: §8.5.3 · p. 46

8.6 Alternative Base Fee Update Rules

The section restricts attention to history-dependent base fees whose revenue is burned or otherwise withheld, then examines how update rules can be designed and what assumptions the EIP-1559 choices make. It proposes criteria and alternatives for experimentation.

8.6.1 Assessing Update Rules

  • The ideal update rule tracks the market-clearing price without creating manipulation incentives or attack vulnerabilities; the report assesses rules against five desiderata.These are reasonably quick upward and downward adjustment, restraint against overreacting to brief changes, resistance to robust cartel manipulation, and costly exploitation; thresholds are left to experimentation and community discussion.Source: §8.6.1 · pp. 46–47

8.6.2 Decomposable Update Rules

  • Decomposable rules use the latest block’s contents and prior base fee, but sliding-window rules are nondecomposable and some decomposable rules can oscillate under stable demand.Such oscillations may be unlikely in deployment; clients could inject randomness to avoid pathological cases.Source: §8.6.2 · pp. 47–48

8.6.3 What Should the Adjustment Function Depend On?

  • Using transaction bids in the adjustment function can let users and miners collude off-chain to keep an excessively low base fee, so EIP-1559 uses block gas consumed instead.Such an agreement can encourage a game-theoretically robust miner strategy that favors miners at the network’s expense, although it does not violate OCA-proofness.Source: §8.6.3 · p. 49

8.6.4 The Functional Form of the Adjustment Function

  • The adjustment function’s form and parameters are open to experimentation: exponential updates depend on cumulative gas consumed, unlike EIP-1559’s path-dependent rule.A degree-2 polynomial approximation to the exponential is also proposed as a compromise between exponential and linear forms.Source: §8.6.4 · pp. 49–50

8.6.5 Choosing the Rate of Change

  • The EIP-1559 adjustment rate lets the base fee respond to sustained demand shocks while limiting responses to short-lived shocks, but its rate parameters should be revisited with deployment data.A run of maximum-size blocks would double the base fee in under 1.5 minutes and increase it by an order of magnitude in under 5 minutes; variable-size blocks mitigate sudden increases.Source: §8.6.5 · pp. 50–51

8.6.6 Choosing the Block Elasticity

  • The maximum-to-target block-size ratio balances absorbing sudden demand spikes against node processing costs and centralization; the report recommends making it as large as network resources allow.A ratio of 2 also means a 51% miner cartel is needed to significantly manipulate the base fee or long-run average block size; other ratios create different manipulation tradeoffs. The best choice may change over time.Source: §8.6.6 · p. 52

9 Additional Remarks

EIP-1559 offers side benefits beyond easier fee estimation, including lower fee variance, effects from burning fees, a gas-price proxy, and reduced miner incentives for undercutting attacks. The section also describes EIP-2593’s bidding approach, tradeoffs, possible combination with EIP-1559, and narrower scope.

9.1 Side Benefits of EIP-1559

  • Easy fee estimation and variable block sizes should decrease transaction-fee variance during periods of changing demand.Source: §9.1 · p. 52
  • Burning base fees could reduce Ethereum’s roughly 4% inflation rate or make it negative when fees remain high, while tying ETH value to network usage and giving ETH on-chain functionality.The inflation rate would become less predictable.Source: §9.1 · p. 52
  • EIP-1559’s base fee can serve as a difficult-to-manipulate proxy for the market-clearing gas price, enabling smart contracts such as gas futures markets.Source: §9.1 · p. 52
  • By directing transaction fees away from miners, EIP-1559 makes fee-driven undercutting attacks less attractive, though attacks may persist when a few transactions pay far above the base fee.Source: §9.1 · p. 53, §9.2 · p. 53

9.2 The Escalator: EIP-2593

  • EIP-2593 adds validity-window endpoints with bids, linearly interpolates intermediate bids, and automates escalating bids that users could otherwise submit by rebroadcasting transactions.Users can choose a short interval and higher initial bid for impatience, or a longer interval and lower initial bid when prioritizing price over immediate inclusion.Source: §9.2 · p. 53
  • EIP-2593 adds bidding parameters that may complicate strategy choice and locks users into linear increases, even when a concave or convex strategy might serve them better.Source: §9.2 · p. 53
  • EIP-2593 is less radical and risky than EIP-1559 but lacks its side benefits, and its first-price auction cannot offer EIP-1559’s typically-UIC guarantee.The proposals can be combined by using EIP-2593’s linear bids for EIP-1559 tips while the base fee evolves independently; the added functionality may matter mainly during occasional periods of rapidly increasing demand.Source: §9.2 · pp. 53–54

10 Conclusions

EIP-1559 offers potential gains in fee estimation, fee variance, game-theoretic robustness, and reduced inflation, but also introduces protocol and implementation risks. Whether those benefits justify adoption depends on differing views of the value of continued consensus-layer innovation.

10 Conclusions

  • EIP-1559 offers an “obvious optimal bid” outside periods of rapidly increasing demand and lower transaction-fee variance through increased block-size flexibility.The bid benefit is qualified by rapidly increasing demand; the variance benefit comes from greater flexibility in block size.Source: §10 · p. 54
  • It also aims for game-theoretic robustness to protocol deviations and off-chain agreements within and across blocks, and reduced inflation through fee burning.The robustness claims cover both single-block and multi-block settings.Source: §10 · p. 54
  • Risks include implementation errors, forks, consensus complexity, upgrade parameters, unforeseen consequences, possible miner hostility and coordination, and a costly attack vector enabled by variable-size blocks.Miner hostility may arise from lost revenue due to burned transaction fees; the variable-size-block attack vector is described as expensive.Source: §10 · p. 54
  • People may disagree on whether the benefits justify adoption: some may prefer the status quo, while those favoring continued consensus-layer innovation may find the arguments for EIP-1559 strong.Source: §10 · p. 54

Related papers

Loading 2012.00854v1…