“The true method of knowledge is experiment.” — William Blake
Background
As Metis delved into its development, there were a lot of boundaries to test, like those of scalability. In the past, the lack of exploration and pushing the boundaries of innovation in the rollup space was highlighted by some of the biggest personalities in the industry, including Ethereum co-founder Vitalik Buterin.
“Are there techniques that put some state-update-relevant data on chain but not all of it, and is there anything useful that could come out of that?”
Responding “YES” to the first part of Buterin’s question, Metis rolled out its Smart L2 architecture to achieve lower and more stable transaction fees. The result? It immediately proved that something extremely useful could come out of it: Metis became the most cost-efficient scalability solution on Ethereum.
For more than a year, Metis has enabled users to transact with fees at least 100x lower than on Ethereum Mainnet, while leveraging Ethereum’s underlying decentralization and security. Metis also offered lower transaction fees than any other Layer 2 network, establishing itself as the most cost-efficient L2 on Ethereum.
As we grow as a scalability solution, we have recognized that in order to achieve our full potential, we must close an even more essential gap that remains unsolved by industry participants: Rollup Decentralization. Rollups can become highly scalable, but unless they decentralize their sequencer, they will never be as secure as Ethereum: a key milestone in order to achieve mass and institutional adoption.
Therefore, we’ll get it done.
Metis will soon release the first-ever sequencer pool, cementing itself as the first decentralized Layer 2 on Ethereum. Nevertheless, there’s something that could be even better, enabling the highest security standards and Data Availability standards: the first decentralized rollup on Ethereum.
Moving Data Availability (DA) back to Ethereum Mainnet
Three weeks ago, the Metis team submitted a proposal to Candidac.org (Metis Governance Forum) to seek community approval for changing Data Availability from its current framework, storing data in Memolabs, to storing all data on Ethereum Mainnet. Memolabs is an efficient, secure, and resourceful alternative for Data Availability needs. For Metis’s specific case, though, and ideal implementation of a sequencer pool, Ethereum Mainnet would be the only alternative. Some of the main benefits that would come from moving Data Availability (DA) back to Ethereum are:
- Highest security standards for Optimistic Rollups
- Rollup Decentralization, by facilitating the implementation of a sequencer pool
- Potential for liquid staking fully secured by Ethereum
- A wider range of dApp integrations
Moving data availability (DA) back to Ethereum would mean Metis going back to being a pure Optimistic Rollup. The only tradeoff stemming from this move would be transaction fees, which would increase to be similar to those of other Optimistic Rollups, such as Optimism and Arbitrum.
We consider this to be a reasonable tradeoff to enable the implementation of a sequencer pool, and therefore, to become the first decentralized rollup on Ethereum.
PoS Sequencer Pool
The main goal of rollups is to serve as execution layers for Ethereum. To leverage its security, its decentralization, and utilize it as the global settlement layer. There’s only one problem. Ethereum’s underlying security can only be guaranteed if a rollup counts with two key elements:
1. A fully functional fraud-proof system.
2. A decentralized block producer (sequencer).
A few very talented teams in the space have already delivered (almost) fully functional fraud-proof systems. However, so far, no rollup has decentralized its sequencer. Doing so is a very difficult task, but vital for any ecosystem’s security.
If there’s only one (centralized) sequencer, users are obliged to trust that entity, creating a potential single point of failure. The whole system would be compromised if that one entity is.
A fully functional fraud-proof system could alleviate this, but not solve it.
Even by having a fully functional fraud-proof system, how would a Layer 2 network continue to work efficiently in the event that the one sequencer turns malicious, has its keys stolen, or is legally forced to cease and desist?
If there is only one sequencer, there is no possibility for rotation (i.e. no alternatives for that one and only sequencer). Let’s do a quick exercise to put this on display:
Scenario 1
The single sequencer turns malicious, or has its keys stolen by a malicious actor.
Outcome
All TXs would require fraud proofs executed, therefore posting ALL data at ALL times on ETH (astronomical costs).
Scenario 2
The entity is legally obliged to “cease and desist”.
Outcome
The block producer (sequencer) would not be submitting these blocks to L1, and all txs/data would have to be submitted to Ethereum via forced inclusion (astronomical costs).
In both of these events, the only reason why the network and users would be impacted is because of the centralized sequencer. Costs of this sequencer being off-line or being hacked would be so elevated, that users might as well just use Ethereum directly, and avoid the L2.
We’ve seen these examples of potential risks come true repeatedly and recently on web3, which is why it’s so important to contemplate them and eradicate them.
That’s precisely where the “Second Key Element” for rollups comes in:
A sequencer pool.
With a sequencer pool, neither of the two previously mentioned scenarios would escalate to a negative outcome. A single sequencer being compromised, or being legally obliged to desist operations, would have no effect on the censorship resistance, security, and liveness of the system.
Seamless business continuity.
