1 of 23

Ethereum: Designing the infinite economic machine

1

Barnabé Monnot

Robust Incentives Group (RIG), Ethereum Foundation

2023 NYU Stern Fintech Conference

2 of 23

Ethereum is our best shot at:

  1. 📈 Maximal satisfaction of user preferences�
  2. 🌐 Over a public, cost-efficient network�
  3. 📉 With minimum rent for operators

2

3 of 23

📈 Maximal satisfaction of user preferences

3

4 of 23

What can you do on Ethereum?

Everything! Smart contracts provide general computability

🪙 They power tokens, NFTs and all kinds of digital assets

🏦 They power DeFi services such as payments, decentralised exchanges, lending and others

🏛️ They power other services and institutions such as domain names, identity, decentralized science, DAOs, games

4

5 of 23

How do you do it?

Users send transactions ⇔ “Recipes” for what they want to do �Transactions trigger code, updating the state of Ethereum

Examples:�Bob pays 10 ETH to Alice�Alice swaps 10 ETH for 10,000 USDC�Carol registers carol.eth�David votes on a DAO proposal

These are valuable to users!

5

6 of 23

Maximum expressivity of preferences

⇒ Maximum (potential) user welfare

6

7 of 23

🌐 Over a public, cost-efficient network

7

8 of 23

Who runs the network?

Ethereum requires consensus over state of the chain�This is done with Proof-of-Stake-based mechanism

Validators are first-class protocol operators

Responsible for maintaining a single view of the ledger� Produce blocks, are accountable for safety faults

8

9 of 23

How to become a validator?

Validators put money at stake, in the native token, min. 32 ETH

Protocol pays them for good performance (~block rewards)

Failure to maintain the network properly ⇒ Penalties

Permissionless entry and exit from the validator set�Anyone can become a validator*� *or pool money to provide capital for the network

9

beaconcha.in, retrieved 19/04/2023

🧮 Data point: ~35 billion USD staked

10 of 23

Proof-of-Stake costs

The Merge: Ethereum runs on less energy�Main cost: Capital (stake) lock-up cost

Tensions�Lowering cost of capital (e.g., liquid staking, re-staking)�⇒ 📈 Increase demand for staking�⇒ 📉 Dilute Ethereum security

Still, potentially easier to target optimal social cost,� as a function of Ethereum security

10

11 of 23

Scalability opportunities and limits

Higher welfare is realised with scalabilityhigh tx throughput

Constraint: verification throughput > tx throughput� Ultimate validator failure is backstopped by community� But honest users must be able to verify operator work!

Ethereum roadmap: Scale the system with rollupsDependent chains secured by the Ethereum validator set

11

12 of 23

Public, cost-efficient network

⇒ Maximum scale and operator accountability

12

13 of 23

📉 Minimum rent for operators

13

14 of 23

Value flows

While users transact on Ethereum, three types of value settled:

User surplus ⇒ Maximise for user

Externalities ⇒ Internalise for network

Operator rent ⇒ Minimise for operator

14

15 of 23

Rent 1: Congestion pricing

Operators pack blocks, but block space is scarce�⇒ Users express inclusion preferences via fees

Monopoly without a monopolist (Huberman, Leshno, Moallemi, 2021)�Operators cannot enforce monopoly pricing (Bitcoin-type TFMs)

Ethereum with EIP-1559 fee market (Roughgarden, 2021)�Fees / Congestion costs are internalised by the protocol🧮 Data point: ~6 billion USD captured and removed since EIP-1559 (Aug. ‘21)

15

16 of 23

Rent 2: Operator privilege

Operators include user transactions in the blocks they make�Last look ⇒ Operators capture value from externalities

⚖️ ArbitrageUser makes a swap order for token A against token B on a market 1� ⇒ Creates price imbalance with another market 2�Operator buys B low on 1 ⇒ Operator sells B high on 2� ⇒ Price imbalance is resolved, Operator pockets the difference

16

17 of 23

Rent 2: Operator privilege

Operators include user transactions in the blocks they make�Last look ⇒ Operators extract value from users

🥪 “Sandwich” attackUser makes a swap order for token A against token B�Operator places: 1) Order for A/B before user swap� + 2) Order for B/A after user swap�Operator buys low ⇒ User buys high ⇒ Operator sells high

Permissionless [operators + programmability] ⇒ No “outlawing”

17

18 of 23

Rent 2: Operator privilege

Maximising extractable value for operator is hard�Requires sophistication and/or access to exclusive order flow

Division of labor: Validators source their blocks from builders�~ Procurement auction, builders extract value, bid it away

Future: Protocol is the auctioneer, permissionless auction�Bid values are captured and internalised ⇒ Minimum rent

18

Ethereum protocol

Network of validators

Builders

Users

19 of 23

Recovering max user welfare

Protocol captures operator rent, but user is still hurt 🥪😖🥪

Question: How to protect user, without hurting coordination?

Tensions�Permissionless programmability ⇒ Max coordination value�Defensive “protections” add constraints ⇒ May destroy value

Are we lost?

19

20 of 23

Recovering max user welfare

Operator may have last look, but user has commitment power!

Examples

  • Order Flow Auctions (OFAs): User sells order to bidders
  • Contextual execution
  • ??? ⇒ Permissionless innovation in cryptography and� mechanism design

This is the most exciting place to do research in! mevconomics.wtf ⇒ 7 hours of great content :)

20

21 of 23

Externalities

⇒ Reduce/rebate what you can�Internalise otherwise

21

22 of 23

Thank you!

Go further:

Get in touch! barnabe@ethereum.org

22

23 of 23

Scaling Ethereum

Ethereum roadmap: Scale the system with rollupsDependent chains secured by the Ethereum validator set

23

🧮 Data point: Today, rollups process ~4x as many transactions as Ethereum

EIP-4844 (ETA 2023, “Dencun”)�Provide more scale for rollups

Danksharding�Subcent costs for rollup txs