Angle Explains: ETH 2.0, changes in block production and impact on MEV
On the night between the 14th to 15th of September, the long awaited Ethereum merge will happen. With the consensus change to Proof of Stake (PoS), new Miner Extractable Value (MEV) risks could appear. Let’s go over the major changes brought to the chain in block production/validation, and see how this impacts validators, MEV and DeFi protocols!
ETH 2.0: Changes in consensus and block production
First, with the Proof of Stake consensus mechanism, 2/3 of validators votes are required to publish a block, and the stake of the ones not voting are slashed. This is very different than in Proof of Work (PoW) where no one else had to express their approval of a block for it to be valid.
With PoS, the chain is split in “slots”, and groups of 32 slots are called “epochs”. Each slot is a chance for a block to be added to the chain. Slots (potentially blocks) happen precisely every 12s instead of ~13 previously. The 2 ETH mining rewards per block is decreasing by ~90% to a 0.22 ETH staking reward per block.
This is great for ETH supply, but has an important consequence on validators behaviors, as the share of transaction fees (tips) in their revenue will be much higher. This will make maximizing MEV even more important to increase revenue and gain market shares against competing validators.
Another key difference between PoW and PoS is that the election of the set of validators for the next epoch (32 slots) in PoS is deterministic. In other words, while the miner of a block in PoW is known only after the block is mined, the validators of the 32 slots of an epoch are known in advance.
This knowledge introduces completely new behaviors from validators as they will now be able to choose which transactions to include in their block in advance. In fact, most validating entities control multiple validators, and could even be validating multiple blocks in a row.
Multiple blocks validation
@alrevuelta wrote a very interesting research post where he estimates this probability depending on the share of validators owned by a single entity. His findings reveal that an entity controlling 10% of the validators has a 25% chance of validating two blocks in a row per epoch!
This probability decreases to 7.15% with 5% of the validators, and to 0.3% with 1% of them.
This new situation in block validation under PoS opens up new MEV opportunities, as validators knowingly validate one or multiple blocks in a row.
For example, this could allow an attacker validating two blocks in a row to start an oracle manipulation or arbitrage at block N, and close it only at block N+1.
As detailed in this paper from @euler_mab, these multiple-blocks attacks to manipulate the TWAP (Time Weighted Average Price) of a Uniswap pool would imply very different requirements for the attacker with no guarantee of success, but are still a possibility.
We could imagine a new version of mev-boost, the Flashbots client used by miners to receive blocks from MEV searchers, sending not only one block to the validator, but two in a row, organizing transactions between these two blocks.
But how realistic is it in practice?
Who controls the validators
To see if we can expect such increases in MEV opportunities in practice, we first have to look at the whole validators set, and see who controls them. There are already a lot of great resources on that, like rated.network or beaconcha.in.
Coinbase, Kraken, and Binance are respectively running 14%, 8.5% and 6.5% of ETH PoS validators: it’s very likely that they validate multiple blocks in a row frequently.
Lido represents ~30% of validators, but currently has 28 node operators. They are all public entities that applied with Lido, and none of them controls more than 6% of the pool (source). Then, all the other entities with more than 1% of validators are controlled by public entities and highly scrutinized.
Though this can look scary, in practice I highly doubt that 1) big exchanges will foregone their advantages of constructing their own blocks (and offering the block space to their clients for example) to do MEV, and 2), public entities would openly perform MEV activities.
This leaves us with the other entities controlling ~1% of validators or less. Will they all run mev-boost to profit from MEV searchers and gain market shares?
To remediate to the censoring of transactions that Flashbots implemented in their main MEV-boost client, they recently open-sourced its code. Will we see new “less aggressive” clients of mev-boost pop-up?
Conclusion
My take is that with time, we should see a whole set of different validators behaviors ranging from “maximal extraction — minimal ethics” to “no extraction — public good validator”. Lido and other decentralized pools might have a very important role to play here in the future.
In any case, some DeFi protocols have already started protecting themselves and their users against this potentiality. Euler’s team wrote a governance post to switch their oracles to Chainlink.
On Angle’s side, all oracles rely on Chainlink or the best price between Chainlink and a Uniswap TWAP, making it resistant to any sort of TWAP manipulation.
References
- https://pwnh4.com/posts/the-merge
- https://eth2book.info/altair/part2/building_blocks/randomness#where-does-the-entropy-come-from
- https://eth2.incessant.ink/book/06__building-blocks/02__randomness.html#wait-what-is-randomness
- https://alrevuelta.github.io/posts/ethereum-mev-multiblock
- https://integral.link/mev-dex-frontrunning-merge/
