Fair rewards for�more decentralised attesters

A DAG-based reward mechanism

1

Barnabé Monnot

Robust Incentives Group (RIG), Ethereum Foundation Research

Devcon SEA — 15/11/2024

Why we need decentralised attesters

Validators as attesters come to consensus on the chain.�Thousands of votes are cast every round.

We need these votes to remain truthful.�We need to secure the attesters’ voice.

Our current mechanisms are too brittle.�We present here an alternative.

2

Breaking the balance of power

Joint work with Roozbeh Saranche, Ertem Nusret Tas, Caspar Schwarz-Schilling, Bart Preneel

3

A simple commitment attack

4

Commitment attack

Deploying a smart contract “warping” the incentives of other validators to do the correct thing.

• Attacker commits to course of action.
• Victims must adjust behaviour/actions in reply.
• Coordinates victims towards attacker’s preferred outcome.

Interesting questions regarding the credibility of the commitment… but we don’t lack ways of being more credible :)

5

Gasper 101

Ethereum’s “Gasper” consensus mechanism has two components:

• FFG => Finality service
• LMD-GHOST => Availability service

6

Validator duties

To build the chain, a selected proposer makes a block.�The block contains user transactions and attester votes.�Attester votes have FFG data (what to finalise?)� and LMD-GHOST data (where is the head?).

7

Good case

• Proposer 1 makes a block
• Attesters cast vote on the block
• Proposer 2 makes a block, includes attester vote

8

😇

😇

Proposer threat

The proposer threatens to ignore an attester vote if the attester doesn’t vote for the proposer’s chosen block.

9

😇

😇

😈

Proposer threat

There is a Nash Equilibrium where attacker succeeds!

Lose�rewards!

10

😇

😇

😈

Extensions

In the paper we also discuss:

• Attacks over multiple blocks
• With fixed attester sets (same players in all rounds)
• With variable attester sets
• Attacks inspired by selfish mining attacks

11

Decentralising the proposer and fixing the rewards

12

Rewards in Ethereum Proof-of-Stake

Attesters are rewarded today when their head vote is:

• Timely: Included by the next proposer.
• Correct: Vote with the majority of attesters.

The timeliness constraint gives a lot of power to the�next proposer!

13

Ensure timeliness without timely inclusion

The timeliness constraint gives a lot of power to the�next proposer!

Strawman: Allow any proposer to include these votes.� … but how do we know that the votes were timely then?� We need the head votes ASAP!

Solution: Have attesters vote on the timeliness of�other attesters!� ~ Decentralise the role of the next proposer.

14

DAG-based votes

Attesters of slot t+1 vote on the votes of slot t attesters.� “I have seen these votes� arrive on time.”

These votes can be included anytime.� Today, if block is missed,� attesters lose reward :(

​

15

Performance

The DAG votes are a new object, creates overhead. But:

• In the good case (proposer is honest), then it’s not required, all votes are simply included.
• If proposer is missing or doesn’t include everything,�then the DAG must be created.
• Worst-case is still reasonable.

~ think of it as insurance!�More on this in the paper.

16

Thank you!

https://rig.ethereum.org

see also “Breaking the Balance of Power: Commitment Attacks on Ethereum's Reward Mechanism”