MT Capital Research Report: Analysis of the Decentralized Sorter Track
TLDR
Decentralized sequencers, as an emerging technology, aim to optimize the transaction ordering process of blockchain networks in a decentralized manner to improve transaction efficiency, reduce costs, and address the MEV issue. The development of this technology marks further efforts in the blockchain field towards achieving higher performance and stronger decentralization.
Metis's "self-operated store" model and Espresso's "outsourced module" approach showcase two main paths for building and maintaining decentralized sequencers. The former emphasizes the security and stability of internal management and operations, while the latter offers more flexibility and openness, promoting technological universality and reducing operational burdens.
The development of decentralized sequencers heralds potential advancements in blockchain technology regarding network security, censorship resistance, transaction efficiency and costs, as well as ecosystem diversity and interoperability. Further optimizations and innovations in these technologies, such as batch processing and state channels, will enhance the performance of L2 platforms, lower user costs, and foster the formation of a more open and interconnected decentralized ecosystem.
Although decentralized sequencers face challenges such as technical implementation, network performance optimization, and governance model design, their critical role in building a more efficient, secure, and open decentralized world cannot be underestimated. Future developments may focus on researching more efficient consensus mechanisms, scalable network architectures, and developing user-friendly interfaces and tools to meet the growing market demands and user expectations.
Introduction to Sequencers
Sequencers, as the name suggests, are responsible for ordering the originally unordered transaction data in the blockchain, organizing it into ordered block data for execution. Each L1 blockchain has its own ordering system, but for L2, centralized sequencers have become an increasingly serious issue.
For L2, sequencers are not strictly necessary. L2 can also choose to use L1's sequencer. However, for cost and speed considerations, running its own sequencer can provide users with a cheaper and more convenient experience. L2 running its own sequencer can compress hundreds or thousands of L2 transactions into a single L1 transaction submitted to L1, significantly saving on gas fees. Moreover, users can enjoy the fast soft confirmation experience provided by the L2 sequencer without being constrained by Ethereum's transaction throughput. Therefore, for L2, running its own sequencer is also a necessary choice to enhance user interaction experience.
Current Status of Sequencers
While running its own sequencer can greatly enhance user experience, the centralization of L2 sequencers has become an issue that cannot be ignored. Currently, the locked value of Ethereum L2 has reached 22B, and a massive number of L2s are continuously emerging, but almost all L2 sequencers are centralized, relying on a single sequencer to determine the ordering of all transactions on L2. Centralized sequencers face numerous problems, such as a single sequencer theoretically having the power to exclude users' transactions, being able to extract MEV from transactions without restrictions, facing censorship resistance issues, and the risk of single points of failure.
source:https://l2beat.com/scaling/summary
In addressing the complex challenges of MEV, rollups face the delicate balance of maintaining user protection while generating profits. This challenge involves preventing harmful MEV behaviors such as front-running and sandwich attacks while effectively utilizing block space for revenue. Traditionally, rollups have relied on a single operator model and used a first-in-first-out (FIFO) order to protect users from MEV impacts, but this approach may miss out on revenue opportunities from block space and overlook the importance of economic incentives in promoting rollup stability and growth. Additionally, ensuring adherence to FIFO principles and maintaining transparency in block ordering presents further operational challenges. Furthermore, utilizing underlying block space as a source of income, while potentially lucrative, raises trust issues among users, who must trust that operators will not exploit this space to harm their interests through methods like sandwich attacks, which could undermine transaction integrity and user trust.
Shared sequencers provide an innovative solution to the MEV problem by introducing a more secure and fair transaction ordering mechanism in blockchain networks, particularly benefiting Ethereum's layer two solutions like rollups. It effectively balances the needs and interests of network participants by dividing the rollup's block space into top block space that protects user transactions and bottom block space that allows builders to utilize MEV. Using Practical Verifiable Delay Encryption (PVDE) technology, shared sequencers ensure that user transactions are invisible to malicious actors, thus preventing harmful MEV practices such as front-running and sandwich attacks. Additionally, by allowing beneficial MEV activities in the bottom block space, shared sequencers generate revenue for rollups while maintaining network integrity and user trust. This mechanism not only enhances the security and fairness of transactions but also supports the sustainable development of blockchain networks through innovative revenue generation methods. In short, shared sequencers bring positive transformation to the blockchain ecosystem through their unique handling of MEV, achieving a balance between protecting user interests and promoting healthy network development.
Overall, the issues with centralized sequencers stem from the excessive power and risk exposure of a single node-based sequencer, while decentralized sequencers composed of multiple nodes can effectively address the problems faced by centralized sequencers. Decentralized sequencers can ensure the robustness and effectiveness of L2 ordering while also bringing additional benefits. For example, decentralized sequencers represented by Metis can further empower tokens while achieving revenue sharing, and shared sequencers allow L2s to avoid building their own ordering networks while providing more convenient interoperability for multiple shared sequencer L2s. In the long run, the trend of modularization and L2 will undoubtedly drive the decentralization of sequencers, and the decentralized sequencer market still has vast potential.
source:https://joncharbonneau.substack.com/p/rollups-arent-real
Decentralized Sequencer Projects
Metis
Elena Sinelnikova, co-founder and CEO of Metis, has previously been dedicated to blockchain industry education and advocacy, co-founding the educational non-profit organization CryptoChicks, which is currently the largest female blockchain community in the world, with members in 56 countries. Kevin Liu is the co-founder and product lead of Metis, as well as the co-founder and CEO of ZKM, and is an active researcher in token economics, DAOs, DeFi, and blockchain governance.
Metis is the first Ethereum L2 to propose and test decentralized sequencers.
Metis has transformed the originally single sequencer node into a sequencer pool composed of numerous nodes, achieving decentralization through a mechanism of random rotation.
First, there will be an Admin role in Metis's decentralized sequencer network. The Admin is responsible for managing the decentralized sequencer system, including adding qualifying sequencer nodes to the Sequencer List whitelist, setting individual node staking limits, block reward release speeds, and so on.
Subsequently, Metis introduces a node staking mechanism. Any node that stakes 20,000 METIS tokens can become one of the nodes in the sequencer pool. Nodes in the sequencer pool have the right to see the contents of the transaction pool, and selected sequencer nodes have the right to package transactions.
Next, Metis introduces a PoS node rotation mechanism. Metis randomly selects block producers based on each node's staking amount combined with randomly generated hash values. The selected sequencer nodes can then package block transactions.
Next, the packaged transaction batch requires at least 2/3 of the sequencer signatures to be considered valid and submitted to L1. The signing keys of the sequencer nodes are managed by Metis's PoS consensus layer, which generates and distributes multi-signature keys when sequencer nodes join or leave the network.
Finally, to prevent sequencer malfeasance, Metis will also introduce a validator role to randomly sample blocks and check whether the transaction order within the blocks is correct, among other tasks. Malicious nodes will face penalties through the forfeiture of their staked funds.
source:https://www.metis.io/decentralized-sequencer
Based on the above process, Metis has built a decentralized sequencer architecture based on PoS network consensus. Staking 20,000 METIS allows one to become a sequencer node, making the sequencer nodes more diverse and avoiding single points of failure, single-point control, and malicious MEV extraction. The node rotation mechanism and multi-signature confirmation make the selection of sequencer nodes fairer and can also mitigate the risk of sequencer node malfeasance to some extent. The validator's sampling checks and penalty for forfeiture further reduce the risks associated with malicious behavior by nodes.
To further incentivize more nodes to participate in Metis's decentralized sequencer network, Metis has also introduced additional incentive mechanisms. After successfully producing a block, sequencer nodes can not only earn the original gas revenue from the sequencer but also receive additional METIS token emission rewards. Metis's incentive mechanism has the potential to create a positive growth flywheel. The prosperity of transaction activities in the Metis network will drive an increase in sequencer node revenues. Increased sequencer node revenues will attract more users to stake METIS, becoming sequencer nodes and capturing sequencer income. The reduction of circulating METIS and the increased demand for METIS due to staking will further raise the market price of METIS. The rise in METIS prices will increase the asset value of staked nodes and the value of staking rewards, thus creating greater appeal for nodes and attracting more nodes to stake, forming a closed-loop flywheel.
Metis's PoS decentralized sequencer network is the first attempt for L2 to implement decentralized sequencers. The implementation of Metis's decentralized sequencer is expected to drive other L2s to advance their decentralization plans for sequencers.
Espresso Systems
Espresso's team has an impressive background, with co-founders Charles Lu and Ben Fisch both holding PhDs in computer science from Stanford University. Team members have also worked at leading Web2 and Web3 companies such as Binance Labs, Coinbase, and Google. Previously, Espresso successfully raised $23 million in funding from top venture capital firms including Sequoia Capital, Coinbase Ventures, Polychain, and Robot Ventures.
Espresso positions itself as middleware between L1 and L2, decoupling ordering from execution, aiming to become a decentralized shared sequencer network that provides decentralized sequencer services for different L2s. Similar to the DA outsourcing concept in modularization, the service provided by Espresso is more akin to transaction data ordering outsourcing. Like DA outsourcing, the sorting outsourcing service provided by Espresso is also independent of the chain and virtual machine, allowing any type of L2 to use Espresso's sorting services.
source:https://hackmd.io/@EspressoSystems/EspressoSequencer
Espresso's core idea is to provide a modular sequencer middleware for L2. After users send transaction data through the client, the transaction data, along with the identifier of the L2, is sent to Espresso's sequencer network. Espresso's nodes (nodes of the Espresso Hotshot proof-of-stake system) will sort the transactions and, after sorting, broadcast them to subscribers (L2 nodes). Subsequently, L2 will execute based on the packaged ordered transaction data. At the same time, Espresso will submit a block commitment containing transactions to the L1 sequencer contract. Finally, L2 needs to send the new state to L1, and L1's Rollup contract will use the block commitment from Espresso to verify the state updates submitted by L2 to ensure the correctness of the execution.
source:https://docs.espressosys.com/sequencer/espresso-sequencer-architecture/system-overview
In the future, Espresso also plans to reuse existing Ethereum validator nodes through Eigenlayer to participate in ordering, achieving higher security.
Overall, Espresso's decentralized sequencer solution aligns more closely with the concept of modular blockchains, utilizing its own PoS network to achieve decentralized ordering through the form of sorting outsourcing, forming a decentralized sequencer network middleware between L1 and L2. The universality of Espresso's sorting services also allows it to become a shared sequencer network, where any L2 can use Espresso's sequencer services. Furthermore, L2s that share Espresso as a sequencer service provider can enjoy more seamless interoperability.
Astria
Josh Bowen, the CEO of Astria, is the primary driving force behind the project. Josh Bowen previously worked at Edge & Node, the startup behind The Graph, and Celestia Labs. His past work experience has given him a deeper understanding of concepts like modularization and decentralization. He has shared important insights on the significance of shared sequencers in maintaining blockchain space speed and decentralization. Bowen emphasizes that most application-specific rollups may not need their own sequencer, and nurturing a more decentralized, modular shared sequencer network is beneficial for building a more decentralized and efficient blockchain system. Josh Bowen and Astria's vision have also garnered support from institutions such as Maven 11, 1kx, Delphi Ventures, and Figment Capital, raising $5.5 million in seed funding.
Similar to Espresso, Astria aims to provide a decentralized shared sequencer network. Astria's shared sequencer network is a middleware blockchain that has its own decentralized sequencer set, capable of accepting transaction data from multiple L2s. Similarly, Astria can handle ordering requests from any type of L2. Furthermore, L2s using Astria can also enjoy atomic-level interoperability provided by Astria.
The ordering process of Astria is illustrated in the following diagram.
- After users submit transactions, L2 submits transaction data to Astria through an interface.
- Astria's shared sequencer reaches consensus on transaction ordering through the ComeBFT PoS consensus network and packages them into blocks. Astria's shared sequencer network uses CometBFT as its consensus algorithm. During the network consensus phase, the proposer decides the transactions for the block and creates a commitment to the ordering data for each rollup. Subsequently, other nodes in the network need to verify and reach consensus on it to form a final determination.
- Once the transaction data ordering is complete, Astria's Conductor will parse the data required by different rollups for each ordered block and validate the batch of data, including verifying whether the block has been finalized, whether the extracted rollup data is complete, correct, and properly ordered, etc. After validation, the Conductor will convert the rollup's ordered data into a transaction list and pass it to the rollup's execution engine for execution.
source:https://docs.astria.org/docs/overview/why-decentralized-sequencers/
- L2s pursuing faster user experiences can accept soft commit ordered blocks from Astria through reading interfaces to provide users with quick block confirmations. L2s can also read hard commit ordered blocks written by Astria through the DA layer.
source:https://docs.astria.org/docs/overview/why-decentralized-sequencers/
Astria's decentralized sequencer network is very similar to Espresso's solution, both aiming to provide decoupled decentralized ordering services for any L2. L2s can further simplify their development processes and operational costs by outsourcing sorting services, enjoying atomic-level composability between L2s.
Radius
Radius focuses on developing a trustless shared sorting layer, aiming to address the challenges of harmful MEV extraction and censorship in the blockchain space. Radius has successfully secured $1.7 million in pre-seed funding from investment institutions such as Hashed, Superscrypt, Lambdaclass (Ergodic Fund), and Crypto.com.
Radius also aims to build a trustless, censorship-resistant shared sequencer network, and compared to Espresso and Astria, Radius's biggest feature is its ability to effectively reduce harmful MEV through an encrypted memory pool.
The overall architecture of Radius's shared sequencer network is similar to mainstream shared sequencer networks. Users submit encrypted transaction data and proofs to the sequencer layer through Dapps. The sequencer verifies the transaction data and proofs provided by users and packages them for sorting. Subsequently, the rollup accepts ordered blocks from the sequencer network, executes transactions in order, and submits the executed state and state proofs to the settlement layer.
source:https://docs.theradius.xyz/developer/architecture
Interestingly, Radius introduces an encrypted memory pool to prevent sequencers from extracting harmful MEV. Transactions submitted by users are encrypted and submitted to the sequencer network in encrypted form. When the sequencer sorts transactions, it cannot access the keys and cannot decrypt to view the specific content of each transaction. Therefore, the sequencer cannot extract MEV through malicious ordering or inserting transactions.
Radius divides block space into top space and bottom space. The top space is exclusively for user transactions, effectively avoiding harmful MEV through the encrypted memory pool. The bottom space introduces an open market based on auctions for traders, where cross-rollup beneficial MEV bundling transactions can be created, such as beneficial arbitrage and liquidation. Subsequently, traders submit bundled transactions and bids to the sequencer, which selects the highest bidding bundled transaction to include in the block, maximizing rollup profits and fostering a healthy MEV competitive market.
Compared to Espresso and Astria, Radius has two significant advantages. First, by introducing an encrypted memory pool and dividing block space into top and bottom spaces, Radius can effectively eliminate harmful MEV transactions, fostering a healthy MEV competitive market and maximizing rollup profits. Second, the introduction of the encrypted memory pool means that individual sequencer nodes cannot engage in MEV malfeasance, thus eliminating the need for additional consensus mechanisms to ensure the correctness of ordering, greatly improving the final confirmation speed and scalability of the sequencer network.
SUAVE (Single Unifying Auction for Value Expression)
The SUAVE solution was proposed by the Flashbots team, a pioneering group dedicated to addressing the MEV issue in the Ethereum ecosystem, composed of professionals with deep backgrounds in computer science, mathematics, psychology, and economics. According to LinkedIn, the team currently includes 28 employees with expertise spanning a wide range of fields from Python programming, blockchain technology, and machine learning to C language.
The founding team of Flashbots includes Philip Daian and Stephane Gosselin, the latter of whom left the team in October 2022 due to disagreements over censorship mechanisms. Additionally, Alex Obadia, another co-founder and top strategy researcher, left Flashbots in June 2023 for personal reasons. Core member Andrew Miller is known for his research on breaking Intel SGX code and currently serves as the research director for trusted execution environments and SUAVE. Miller plans to take a temporary leave from his assistant professor position at the University of Illinois, where his academic work focuses on electrical and computer engineering. Another core member, Hasu, serves as the strategic director of Flashbots and has a wide influence in the blockchain field, including serving as a strategic advisor for the liquid staking protocol Lido and a research collaborator at Paradigm investment firm. Hasu is committed to advancing industry development and education through writing, social media, and podcasts.
SUAVE is a unique decentralized builder and sequencer, distinctly different from other shared layers or sorting layers. It aims to provide transaction ordering services for Ethereum and other blockchains but does not directly embed into the protocol of any chain. Users can send transactions to SUAVE's encrypted memory pool, while SUAVE's executor network is responsible for outputting blocks or partial blocks for the chain. These blocks will compete with blocks generated by traditional centralized Ethereum builders, chosen by Ethereum proposers.
Source: https://foresightnews.pro/article/detail/28930
SUAVE does not replace the mechanism for rollups to choose blocks, nor does it change the chain's fork choice rules. It focuses on providing optimal profit ordering for any chain, typically having complete state to simulate the outcomes of different transactions and create the best ordering. This design allows SUAVE to collaborate with shared sequencers or other MEV-aware builders to provide services such as atomic cross-chain arbitrage, ensuring that several transactions are executed atomically or canceled simultaneously.
Source: https://foresightnews.pro/article/detail/28930
In the long run, rollups may be a better choice. Rollups ensure their security, censorship resistance, and liveness through L1, while SUAVE, as a chain focused on transaction ordering, is not suitable for ordinary users. Its goal is to limit users' need to bridge funds to SUAVE and instead focus on providing an operational platform for seekers/builders. SUAVE focuses on providing the most advantageous ordering for transactions rather than completely replacing existing ordering mechanisms. It can handle fully stateful transactions to create the best transaction ordering.
Source: https://foresightnews.pro/article/detail/28930
Regarding MEV handling, there are various mechanisms to reduce potential competition and negative externalities associated with transaction ordering and inclusion. For example, Arbitrum's time-lifting mechanism and Flashbots' proposed FBA-FCFS model both attempt to reduce the motivation for latency competition by allowing users to express their preferences for rapid transaction inclusion through fees.
Arbitrum's Time-Lifting Mechanism
Time-lifting mechanism is a security measure designed to prevent a specific type of attack known as the "time bandit attack." In this attack, an attacker may attempt to reorganize already confirmed blocks to profit from previously unknown information (e.g., using knowledge of a transaction after the fact).
Arbitrum defends against this attack through a unique mechanism that allows anyone to submit a "challenge" when they discover someone attempting a time bandit attack, proving the attacker's behavior. This mechanism is based on an economic incentive that ensures the potential benefits for the attacker are offset, thereby protecting the network's security and fairness.
Flashbots' FBA-FCFS Model
The FBA-FCFS (First Bid Auction - First Come, First Served) model is a transaction ordering mechanism proposed by Flashbots. The purpose of this model is to address traditional transaction selection and ordering issues, especially in the context of MEV extraction.
- The First Bid Auction (FBA) part means that traders can prioritize their transactions by bidding (usually paying additional fees to miners). This is similar to an auction, where the highest bidder gains priority.
- First Come, First Served (FCFS) means that under certain conditions, transactions will be processed in the order they are submitted, ensuring fairness and transparency.
The FBA-FCFS model attempts to balance fairness and efficiency by allowing bidding on transactions to optimize the use of network resources while ensuring that certain users are not completely excluded due to insufficient payment ability.
These mechanisms each have their pros and cons, but their common goal is to improve the efficiency and fairness of transaction processing.
By collaborating with rollups and other MEV-aware builders, SUAVE aims to provide higher economic security and efficiency for cross-chain operations while exploring new economic security models and MEV mitigation mechanisms to improve the decentralization of blockchain transaction ordering and execution.
Summary and Outlook
Projects like Metis, Astria, Espresso, Radius, and SUAVE, while each focusing on different aspects, collectively aim to enhance blockchain scalability and transaction efficiency, address the MEV issue, and strengthen system decentralization and interoperability.
Metis focuses on optimizing Ethereum's transaction processing capabilities through its Layer 2 solution to reduce costs and improve efficiency, aiming to provide developers and enterprises with a more convenient development platform. Astria and Espresso propose the concept of decentralized shared sequencer networks, supporting transaction data processing for multiple Layer 2 solutions, which not only simplifies development and operational processes but also strengthens composability and interoperability between systems. The Radius project aims to create a trustless, censorship-resistant network by introducing encrypted memory pools and block space division, intending to reduce the harmful impacts of MEV while enhancing transaction privacy and security. SUAVE focuses on addressing the impact of MEV on transaction fairness and transparency through a decentralized sequencer network, demonstrating a commitment to improving the fairness of the transaction environment.
In exploring the development directions of decentralized sequencers, Metis and Espresso provide two distinctly different models: the "self-operated store" model and the "outsourced module" approach. These two models reflect the community's different thoughts and strategies on how to build and maintain decentralized sequencers.
Metis's "self-operated store" model emphasizes internal management and operation of its decentralized sequencer network to ensure the security and stability of the network. This approach allows Metis to directly control the nodes within its network, maintaining a healthy network environment through staking and incentive mechanisms. Although this model can enhance the security and reliability of the network, it also requires Metis to bear significant operational responsibilities and resource investments, which may somewhat limit the network's flexibility and scalability.
In contrast, Espresso's "outsourced module" approach offers a more flexible and open solution. By allowing any blockchain project to access its sorting services, Espresso promotes technological universality and diversity while reducing the operational burden on individual projects. The challenge of this model is that it may introduce additional trust issues, as project parties need to rely on Espresso to handle transactions fairly and securely. Moreover, any issues or attacks targeting Espresso's services could potentially impact a wide range of client projects.
Metis's "self-operated store" model and Espresso's "outsourced module" approach showcase two main development paths within the decentralized sequencer field. Each model has its unique advantages and challenges, and the choice between them depends on the specific needs of the project, resource conditions, and the emphasis placed on decentralization and security.
The development prospects of decentralized sequencers indicate the immense potential of blockchain technology in enhancing network security, increasing censorship resistance, improving transaction efficiency, reducing costs, and promoting ecosystem diversity and interoperability. With the continuous advancement of decentralized sequencer technology, we can foresee a more secure and efficient blockchain network, where decentralized ordering mechanisms can effectively defend against single points of failure and malicious attacks, protecting user assets and data security. Additionally, optimizations and innovations in decentralized sequencers, such as batch processing and state channels, will further enhance the transaction processing capabilities of L2 platforms, lower user transaction costs, and achieve high throughput and low-latency transaction confirmations, thereby improving user experience without sacrificing security and decentralization.
At the same time, the widespread adoption of decentralized sequencers is expected to foster the formation of a more diverse and interoperable blockchain ecosystem. Shared sequencer networks, such as Espresso and Astria, will not only provide services for multiple L2 platforms but also facilitate the flow of data and assets between different platforms, creating a more open and interconnected decentralized world. Furthermore, innovations in incentive mechanisms and token economic models will provide reasonable incentives for participants in decentralized sequencer networks while achieving network governance and revenue distribution through token economic models, attracting more participants and stimulating community vitality.
Despite the bright prospects for decentralized sequencers, they still face challenges in technical implementation, network performance optimization, and governance model design. Therefore, future development directions may focus on researching more efficient consensus mechanisms, exploring scalable network architectures, and developing user-friendly interfaces and tools to meet the growing market demands and user expectations. In summary, decentralized sequencers, as a key factor in promoting the development of blockchain technology and applications, will play a crucial role in building a more efficient, secure, and open decentralized world.