Layer2 Sequencer: The Centralization Dilemma Behind High Profits and the Exploration of Decentralization

The main source of revenue for Layer2 at present comes from the Gas fees paid by users when trading on Rollup. After deducting the Gas fees required for Layer2 to submit data to Layer1, the remaining portion is essentially pure profit. Rough estimates indicate that OP Mainnet’s profit from June to December 2023 is about 5.23 million USD, Arbitrum’s annual profit is 16.5 million USD, and zkSync Era’s profit from March to December 2023 is 22.24 million USD.

Such high profits are closely related to the sequencer they operate. The sequencer ( Sequencer ) plays a key role in Layer2, with its main function being to receive and execute transactions from Layer2 users, and then submit the sorted and compressed batch ( Batch ) to Layer1.

The sorter can be likened to a bus driver. In the past, users had to drive into the city themselves ( to trade on Ethereum ), but now they only need to pay a service fee to take the bus (Layer2). The driver will try to fill the bus with passengers before departing and arrange the passengers reasonably to maximize the use of space inside the vehicle.

Currently, the mainstream sorter operation scheme is uniquely operated by Layer2 teams or designated organizations, which is highly efficient and cost-effective due to its centralized approach. Another option is a completely permissionless sorter, where anyone can sort and submit to Layer1, but this may lead to resource wastage.

Sorting is usually done in one of two ways: first-come, first-served or sorted by Gas fees. Layer2 does not have strict rules for sorting methods; theoretically, sorters can sort freely and may even act maliciously. To prevent malicious behavior, different Layer2s have implemented constraint mechanisms such as fraud proofs or validity proofs.

Currently, mainstream Layer 2 solutions such as OP Mainnet and Arbitrum One are using centralized sequencers operated by official or affiliated organizations. This approach facilitates management, improves efficiency, and allows for revenue generation, but it also raises some concerns:

1. Weak censorship resistance: A single entity operating may be required to eliminate certain transactions due to regulatory requirements.
2. Low Activity: Single Point of Failure risk, such as the brief outage during the Arbitrum airdrop.
3. Possible improper MEV gains: Sorters have the power to manipulate the order of transactions.

To address these issues, mainstream Layer 2 solutions have proposed decentralized sequencer schemes, such as geographic decentralization, sequencer auctions, leader elections, and so on. There is also a scheme called Based Rollup, where Ethereum validators lead the transaction sequencing on Layer 2.

Another idea is a shared sequencer, where multiple Layer 2s share a third-party sequencer network. This is beneficial for achieving atomic composability across Layer 2s and preventing MEV extraction, among other things. Currently, projects such as Astria, Radius, and Espresso are working on establishing a shared sequencer network.

Whether decentralized sorters or shared sorters can perfectly solve the problems of centralized sorters remains to be seen. For example, the MEV problem is quite severe in decentralized Ethereum. If the sorter of Rollup is completely open, will a similar market structure also emerge? Compared to the single point of failure that could be caused by trusting the Rollup team, the chaos caused by disorderly market competition and another form of centralization is also concerning.

Although the shared sequencer can improve interoperability, if widely used, it may evolve into a powerful network controlling multiple Rollups, raising centralization issues again. These require further consideration.

The decentralization of blockchain is a long-term process. The attention on sorters is due to their important role in Rollup. I believe that through continuous exploration, these issues can eventually be appropriately resolved.