Why do existing Layer2 Rollups like Arbitrum, Optimism, zkSync, and Starknet not open-source and decentralize their important sequencers? Instead, they all release their own network construction frameworks. Cryptocurrency researcher Haotian points out that the reason is not a technical issue, but rather an issue of interests.
Development strategy of centralized sequencers in Layer2
Background knowledge: Sequencers in Rollups
What problems does centralized sequencers cause?
Decentralizing the sequencer is technically simple but difficult to relinquish power
Compromise solution: Sharing trust assumptions through multi-layer2 ecosystem strategy
Project strategy from the perspective of decentralizing the sequencer
The mainstream Layer2 solutions in the Ethereum ecosystem typically outsource the computational demands of smart contracts to external servers. Due to the hardware and quantity of these external servers, they can usually achieve better performance. This is why Layer2 networks can complete contract computations at a higher speed and then settle the results (fraud proofs or zero-knowledge proofs) on Ethereum.
Among these servers, there is one called the sequencer, which is responsible for accepting, ordering, calculating, and packaging transaction contents, and then uploading the required information to Ethereum. Its role is somewhat similar to that of an Ethereum node.
Currently, the mainstream Layer2 solutions, including Arbitrum, Optimism, zkSync, and Starknet, are running networks with centralized sequencers due to the need for rapid iteration and security considerations. The plan is to open them up and decentralize them only when the technology matures.
On Ethereum, when users submit transactions to the Mempool, they need to wait for miners (builders) to select and package the blocks based on gas fees or their own rules. During this process, users’ transactions may be influenced by Miner Extractable Value (MEV).
In theory, after receiving transaction information, the sequencer in Layer2 will fairly order and package transactions based on Nonce, receiving time, gas fee price, and other algorithms. This will make it more secure and cheaper than Ethereum.
However, due to the centralization of existing sequencers, they have greater power. There is a lot of uncertainty about whether the sequencer will engage in malicious behavior, insert transactions, or manipulate MEV, which actually creates a greater possibility of the Dark Forest than Ethereum.
There are two main reasons why Layer2 projects are currently less willing to decentralize the sequencer:
Security issues: The sequencer is a core component of Layer2. If a decentralized sequencer is directly adopted at the beginning, there will be a greater potential risk of failure, which will affect the user experience and be disadvantageous to the early development of the project.
Interest issues: When the project grows to a certain extent, the transaction volume and accompanying benefits will also increase. By controlling the sequencer, the team not only controls the pricing power of gas fees but also has the ability to charge other hidden fees. Therefore, the team is less willing to promote decentralization and give up their power.
The technical logic of decentralizing the sequencer is not difficult, but the real challenge lies in relinquishing power.
In fact, many Layer2 projects have already attempted to operate with decentralized sequencers. For example, Metis has implemented a distributed sequencer operating network and ensured the normal operation of malicious nodes through POS token economics, achieving “hard decentralization” of the sequencer.
However, Haotian believes that although the mainstream solutions in the Rollups market currently use centralized sequencers, they may still be worth trusting.
Arbitrum, Optimism, zkSync, and Starknet all choose to open their own network construction frameworks, aiming to create a multi-network ecosystem and share critical technological components.
Arbitrum introduces Arbitrum Orbit
Optimism introduces the OP Stack
zkSync introduces the ZK Stack
Starknet introduces the Starknet Stack
By sharing the core sequencer and creating a multi-chain combination framework, and then jointly governing the development of the sequencer through multi-signature and governance voting, a social decentralized consensus is achieved as a transitional phase before the true decentralization of the sequencer.
By attracting other projects to join the ecosystem, for example, in addition to the OP Mainnet, the OP Stack ecosystem includes well-known projects such as Base, opBNB, Zora, Mantle, Celo, and Debank, making the market trust the sequencer more, although it is not truly decentralized.
These soft-decentralized solutions require a certain level of trust in the project from the market, so they are only used by mainstream Layer2 projects.
Both soft decentralization and hard decentralization are strategies to gain market trust through decentralization. The former is suitable for large projects that have gained more market trust (and are not easily willing to give up their advantages), while the latter is suitable for small projects that have not received much market attention and are catching up with mainstream technology.
By understanding the differences in these technological approaches, it becomes easy to understand why Layer2 project teams are slow in promoting the decentralization of the sequencer and to see the calculations behind the network construction framework promoted by the project teams. This helps make more accurate judgments and even predict trends in the cryptocurrency market and projects.
Layer2
MEV
Rollups
Sequencer
Decentralization