1. Merge — Improving on Proof-of-Stake
“Merge” went live in September’22 when Ethereum changed its consensus mechanism from Proof-of-Work (PoW) to Proof-of-Stake (PoS). This ambitious transition went smoothly. Not only has Ethereum been running stably, but the much-feared centralization has also been avoided.
Can Ethereum’s Proof-of-Stake consensus be made better? Absolutely. The following are some of the ideas being developed:
Single Slot Finality
Currently, a transaction is not considered final even after it makes it to the blockchain. It is considered final only after several blocks have been built over the block in which that transaction sits. While this offers security, it slows performance.
Ethereum is working on Single-Slot Finality, which means transactions will be considered final immediately after they’re confirmed.
Single Slot Finality could have been achieved quite easily with a small validator set, but Ethereum is not sacrificing decentralization by reducing its validator set size. Given the constraint, Ethereum is deciding between the following options:
Brute Force — Figure out how to process millions of signatures much faster than now.Orbit Committees — A clever solution where a committee decides on chain finality in a manner that doesn’t have the limitations of a committee. When committees, instead of the full validator set, decide on chain finality, there is a security risk. The cost of voting maliciously goes down substantially with committees because only the stake associated with the malicious vote can be slashed. If an Attacker gains control of 51% of the committee, they can undertake an attack at minimal cost. With Orbit Committees, you get around this issue by taking advantage of the difference in the stake sizes of validators. A committee would have some small validators and some large ones. The large validators will stand to lose large stakes if they act maliciously. So, while the security guarantee will be smaller than the status quo, it will still be substantial.Two-Tier Staking — There will be 2 classes of stakers, those with a high or low stake. Only high-stake validators will directly contribute to transaction finality. The rights and responsibilities of the lower deposit validators will have to be worked out, including the right to delegate stake and create inclusion lists.
Faster Confirmation
A block currently confirms every 12 seconds. Considering that Tradfi confirms seconds in micro-seconds, this needs to come down. To have faster confirmation, Ethereum can
Either reduce slot times to, say, 4 seconds, orHave validators publish pre-confirmations during a slot instead of waiting the full 12 seconds to publish the entire block.
Both solutions have pros and cons. Shorter slot times could lead to centralization, as it’ll be viable to have validators only in a few geographies with low latency. With pre-confirmations, we’d have to think about creating incentives for proposers and attesters to publish pre-confirmations, given that they might want to wait the full 12 seconds to maximize revenue per slot.
Maximize the number of Validators
The higher the number of validators, the more decentralized the system.
While Ethereum is clearly more decentralized than most other blockchains, staking pools risk decentralization. Today, Lido controls 27% of all ETH staked, and Coinbase and Binance together control more than 50% of the stake.
Ethereum wants to attract more stakers in general and more solo stakers in particular. It can do so by:
Reducing the amount of ETH required to be staked: Currently, 32 ETH are required to be staked per staker. Several polls have indicated that the biggest challenge discouraging people from solo staking is that the minimum stake is too high. Ideally, we’d want to reduce this to 1 ETH.Reducing the technical difficulty associated with staking: Other blockchains tend to have fewer validators than Ethereum as they impose far greater technical burdens on validators, making the pool of potential validators small as only resource-rich entities can discharge validator duties. However, staking on Ethereum is also becoming more complex. Proposer-Builder Separation (PBS) divides validation into 2 parts — (i) Proposing and (ii) Building. Building blocks is a fairly difficult task requiring expensive hardware and sophisticated software and may become centralized over time, but Proposing will be technically manageable and within the capabilities of Solo Stakers.
Even with the above, we’re unlikely to get to 51% Solo Stakers, which means that a few large validators may come to control the network. Today, a block is considered final after 67% of validators support it. What if this was increased to 80%? Then, if Solo Stakers form 21% of all Stakers, they would constitute a quorum-breaking minority.
Reduce technical overhead
For finality, every transaction must be signed by every validator twice. The higher the number of validators, the longer it takes to finalize signatures.
This conflicts with the previous goal of maximizing validators. Currently, 32 ETH is both the minimum and maximum amount of ETH a validator must stake. MAXEB, in the next Ethereum upgrade called Pectra, suggests increasing the maximum ETH allowed to be staked by a single validator to 1024 ETH. This should help reduce network overhead.
Single Secret Leader Selection
The leader selected to build the block is known in advance. This makes it possible for an attacker to identify the leader’s IP address and undertake a DOS attack right before their turn to propose a block.
A solution being considered to prevent such an attack is to keep the leader’s identity secret until the block is actually produced. This can be done using certain cryptography techniques, where validators have blinded IDs, and the relevant validator must provide proof that they own the blinded ID before they’re allowed to propose a block.
Having said that, the leader selection in secret may not be worth the effort, given the complexity it would bring to the protocol.
Automated Actions
Clients should be able to automatically refuse to accept blocks that are clearly malicious. For example, if a valid transaction has been sitting in the mempool for a while and the validator is not including it, the proposed block should be automatically rejected.
Quantum resistance
Quantum Computers might break cryptography as early as 2030. This means that Ethereum needs to get serious about replacing all elliptic curve cryptography with hash-based or other quantum-resistant cryptography.
Conflicting goals
Many of Merge’s goals conflict with each other.
Increasing the number of Solo Stakers will increase decentralization, which is positive, but also increase network overheads, which is negative.Quantum resistance, automating actions, and Single Secret Leader Selection will increase network complexity for a protocol that aims to be simple.Faster confirmations require higher-performance hardware and software that could lead to centralization.
The trick will be to find the right balance.
1. Merge — Improving on Proof-of-Stake was originally published in Coinmonks on Medium, where people are continuing the conversation by highlighting and responding to this story.