Metis: How do we achieve decentralization goals under the "EVM" equivalent framework?

Author: MetisLab Foundation

Compiled by: Perry Wang, Chain News

Metis was originally a hard fork of Optimism. Our founders appreciated the effectiveness and simplicity of Rollup technology, and Optimistic Rollups can bundle a large number of transactions to create a faster, cheaper, and more scalable user experience. Additionally, we greatly appreciate that Optimistic Rollups can be built as Ethereum Layer 2 (L2) scaling solutions, meaning they can benefit from all the security and decentralization of the Ethereum blockchain.

The upcoming release of OVM 2.0 and related literature published by the Optimism team confirm significant adjustments to their plans: Optimism will no longer provide the originally envisioned level of decentralization.

Metis plans to follow most of the overall architecture of Optimism. However, we refuse to compromise on decentralization. Therefore, our future plans will reflect our commitment to providing maximum scalability without skimping on decentralization.

The Shift to EVM Equivalence

In a recent article, Optimism introduced the concept of "EVM Equivalence" to replace its original vision of "EVM Compatibility." In the article, Optimism describes how EVM compatibility has become a significant barrier for current Ethereum DApps migrating to L2, as "merely settling for compatibility means you are forced to modify, or even completely redeploy, lower-level code that Ethereum-supported infrastructure also relies on."

We can confirm that this is true. The Metis team has also received painful feedback from developers regarding such compatibility issues. Therefore, we understand the vision of switching to EVM equivalence, as it allows some Ethereum projects to experience an easier process when migrating their DApps to L2 networks.

However, all of this comes at a cost. As we delved into the design of OVM 2.0, we found that switching to EVM equivalence (without any other adjustments) would compromise the level of decentralization originally proposed by OVM. In short, OVM 2.0 will fundamentally change the implementation of its Optimistic Rollup, creating an environment where, as Optimism's article explains, "fraud proofs are dead." In the new design, the mechanism for challenging on-chain transactions will be disabled, meaning we can no longer rely on L1's strong consensus to challenge transactions.

This new change from Optimism does not align with Metis's vision, mission, or DNA. By shifting to EVM equivalence without making other adjustments, the Optimism team is building a system with more centralized characteristics that requires network users to simply trust: Optimism will not make mistakes. Everything that Metis is building and everything we represent is in direct contradiction to these changes.

From the very beginning, Metis's goal has been to make blockchain accessible to everyone.

• An L2 Ethereum Rollup scaling solution that supports fast and cheap transactions
• No-code middleware that allows anyone (even complete blockchain newcomers) to migrate to the Metis L2 solution in just a few clicks and minutes
• Decentralized Autonomous Company (DAC) infrastructure that goes beyond simple voting and governance functions of DAOs, allowing anyone to build decentralized enterprises on-chain with real-world functionalities of "real-world" companies while enjoying all the advantages of being built on blockchain.

At every step, we seek only the highest levels of decentralization and security. We will not and do not wish to compromise.

Thus, our team spent weeks discussing and debating whether we should completely deviate from OVM. This was a very difficult decision because we knew that EVM equivalence is what DApps and the Ethereum community need, both now and with the upcoming Ethereum 2.0 release. Ultimately, we distilled all the discussions into one big question: Can we achieve our decentralization goals within the EVM equivalence framework?

The answer is… yes.

How Metis Achieves Decentralized "EVM Equivalence"

Our outstanding development team has found a way to achieve the dual goals of decentralization and "EVM equivalence" after weeks of tireless research. This path aligns perfectly with the design of the Metis Ranger system detailed in our white paper.

It is important to note that once we shift to EVM equivalence, complete on-chain proof becomes impossible. Therefore, here is how Metis adheres to its original DNA while fulfilling its commitment to decentralization:

First, our sequencer pool model increases the level of redundancy and decentralization that a single sequencer Rollup architecture cannot achieve. In our upcoming release, we will also introduce a peer-to-peer network, and once the sequencer pool is open, validators participating in the peer-to-peer network will automatically join the sequencer pool.

Next, consider the evolution of the Metis Ranger system. In our initial design, our Ranger system required significant consensus from L1. In our new design, Rangers will achieve consensus more efficiently. Through the consensus reached by these Rangers, validators will be able to challenge transactions and initiate fraud proofs through the peer nodes we are building. These peer nodes will collectively act as validator entities that can verify and challenge the sequencer. When a challenge is needed, each validator entity will reach consensus on L1 before the window time expires. When a challenge is initiated and confirmed by other validator entities (through consensus reached on L1), the problematic sequencer will be cut off, penalized, and taken offline. Another sequencer will be randomly selected from the sequencer pool to open a correct state. The chain will also be reorganized, and the problematic blocks will be discarded. Due to the distributed nature of the storage in the Metis peer network, this process will occur very quickly.

It is worth noting that another benefit of the Ranger system is the reduction of transaction verification time, such as the time users wait for withdrawals (from L2 to L1). Each validator entity will track and report the height of verification based on Ranger consensus to ensure the quality of verification. This acts like a pulse checking the overall health of the validator network, further enhancing asset security.

Our design will also help achieve the goal of network self-correction, as we firmly believe that relying solely on fraud proofs is insufficient; the network infrastructure should have the ability to self-adjust and correct to achieve the security and scalability needed for long-term sustainable operation.

Choosing to commit to both EVM equivalence and decentralization is undoubtedly a more challenging path. Taking this route means we must adjust and review all the code we previously drafted to verify if there are any conflicts or potential risks. For our development team, this means many sleepless nights.

Nevertheless, we are still pushing forward. From day one of the project, we have been committed to building a network that meets the highest standards of scalability, decentralization, and security. This is the network we plan to provide for excellent partners and user communities, as well as the entire blockchain world. It will begin with the delivery of our Andromeda L2 network, which is the first phase of our mainnet launch.

What happens next?

We will be shutting down the Metis Dragonfire testnet (chain ID: 488) this week and deploying a new testnet with EVM equivalence called "Stardust" (chain ID: 588). To ensure everything goes smoothly, we will conduct a new round of testing for DAC staking and mining products. After testing is complete, we will launch Andromeda based on the new framework, achieving a combination of EVM equivalence and our firm commitment to decentralization.

Therefore, Andromeda will launch in the week of November 15, 2021. The creation of DAC and METIS token staking will be launched within no more than a week afterward.

What to Expect from Andromeda?

It is important to note that our upcoming products are far more than just the release of a single application.

Andromeda is a complete architecture with a long list of innovative features.

The architecture includes:

Andromeda L2 Network

• Multiple Ethereum Virtual Machines (VMs): Decentralized virtual machines supporting the scalability and security of Metis L2 infrastructure
• Peer Nodes: Conducting fraud proofs and reaching consensus during load balancing
• EVM Equivalence: No longer needing to customize or modify EVM smart contracts
• Metis Explorer: Checking all on-chain transactions and verifying smart contract calls (limited to advanced users)
• Metis Bridge: A bridge for transferring tokens between the Ethereum mainnet and Metis L2. In the current version, you can only bridge $METIS tokens; in our upcoming L2 network versions, more tokens and NFTs will be available for bridging.
• Polis: A place for users to create accounts (wallets) or connect MetaMask, as well as a place for developers to manage their own domains, smart contracts, and APIs.
• The first DAC template and numerous DApps will run on the Metis L2 network Andromeda.
• DAC staking and mining: Create your own DAC on L2 using the power of the community, stake tokens, and earn rewards.

(Important notice: Regarding the release of Andromeda, we have adjusted and extended the deposit time from L1 to L2 to make assets safer)

What Follow-Up Actions Will There Be After the Andromeda Release?

• A high-speed "bridge," as we integrate with multiple third-party cross-chain bridges (we will announce more details after selecting suitable "bridges") to shorten withdrawal times. In our current L2 network version, withdrawals from L2 to L1 still require a waiting period of 7 days. In future network versions, withdrawals will be completed within minutes. Metis supports a native withdrawal "bridge" for fast withdrawals, and our Ranger system will be launched next year.
• After launching peer nodes on Andromeda, our next network version (next month) will support load balancing.
• Transaction costs on Andromeda will be significantly lower than those on Ethereum L1, but still relatively high compared to our future plans (currently around $1-2). Metis will launch its IPFS technology early next year, enabling cheap on-chain data storage and transactions on our L2 network for just a few cents.
• Future versions of our Polis middleware will make no-code integration into the Metis L2 network possible. This means that mature blockchain projects will no longer need to spend valuable developer time on integration into Metis, blockchain startups will no longer need to hire expensive and hard-to-find developers for integration, and companies unrelated to any blockchain will only need a few clicks to integrate into Metis.