A Visual Guide to Sequencing MechanismsBased SequencingCentralized SequencingApplication-Specific SequencingConclusion

From placeholder by Gurnoor Narula

Sequencing is the mechanism by which incoming transactions are ordered into atomic bundles or blocks to be processed and finalized on a blockchain. All sequencing mechanisms are rooted in the concept of the write-lock, which is a fundamental concurrency control mechanism used to manage access to a shared blockchain state. Most economic value associated with sequencing flows to the actors that control the components of the write-lock, namely the execution, inclusion, and ordering guarantees of transaction data.

Traditional sequencing methods delegate ownership of the write-lock to the off-chain operator(s) of the underlying network. However, new mechanisms enable verifiable sequencer rules that are flexible, programmable, and can exist at any infrastructure layer. These mechanisms make protocols sequencing-aware: value that flowed out-of-protocol can instead be internalized by the protocol, thereby enabling alternative ways of distributing value at the individual application and user levels.

There are three main types of sequencing:

Interestingly, there are also solutions that are fluid across these mechanisms. Generic programmable sequencing layers like Astria can simultaneously serve as based sequencers for a certain settlement layer (Ethereum or otherwise) and shared or application-specific sequencers. They do so by implementing static, well-known, and repeatable sequencing rules embedded within consensus at the level of individual rollups or applications, thereby allowing for a spectrum of configurations, each with its own implications for decentralization, performance, and value distribution. 

Outlined below are visualizations of different sequencing and their value accrual mechanisms. 

Based Sequencing

The write-lock is owned by the off-chain actors that constitute the L1 transaction supply chain.

Equation*: Profit = Application Fees - DA/Settlement CostsFrom the application layer perspective, there also exists the cost of extracted value accruing entirely to the operator(s) of the underlying network

Based Sequencing on Astria

Based Sequencing on Ethereum (Taiko’s Implementation)

Centralized Sequencing

The write-lock is owned by the centralized rollup operator.

Equation*: Profit = Application Fees - Congestion Costs From the application layer perspective, there also exists the cost of extracted value accruing entirely to the operator(s) of the underlying network.

Application-Specific Sequencing

The write-lock is owned by the application layer.

Equation*: Profit = Application Fees + Internalized MEV + Network Fees - Settlement/DA/Execution FeesFrom the application layer perspective, additional value is captured through MEV. From the network layer perspective, there is a lost opportunity cost of synchronous composability between applications.

Application-Specific Sequencing through off-chain actors and auctions

Application-Specific Sequencing on Astria

Application-Specific Sequencing can be implemented on Astria by utilizing their traditional Based Sequencing model, but placing additional rules on the state transition function on the application-level, as well as on the rollup- and protocol-level. This becomes more nuanced as the line between applications and rollups becomes more blurred (appchains, etc.). An important distinction to make is that Application-Specific Sequencing rules are an additional restriction, or subset, on the rollup- or underlying network-level sequencing rules.

Conclusion

As things stand, general-purpose L1s/L2s aren’t sharing sequencer value with their applications. On the one hand, this can be viewed as a necessary tax for the ecosystem, composability, and execution benefits provided by these base layer networks. On the other hand, in this setup, applications can only profit by charging additional application fees. In a highly competitive environment with low switching costs, this is a margin-minimizing business model with questionable long-term sustainability.

To address this issue, application builders can deploy sovereign app-chains (including through RaaS), integrate with external sequencer networks (and by extension, superbuilders), or implement their own app-specific sequencing mechanisms. Each of these options has varying degrees of application-level value capture, as well as different performance tradeoffs. Assuming that general purpose L1s/L2s don’t change their value-sharing model, applications will likely gravitate toward app-specific alternatives, thereby incentivizing more research and engineering efforts to address the composability and atomic execution issues created by moving away from shared, general purpose chains.

It seems likely that, as competing architectures begin siphoning off the most valuable applications, general-purpose L1/L2s will be compelled to implement value-sharing mechanisms with their application ecosystems. This would combine the performance benefits of a unified monolithic architecture with the value redistribution that applications are starting to demand. Either way, the end result will be a more competitive market for verifiable sequencing options that applications can choose from rather than being at the mercy of default sequencing rules set by base layer networks. While this competition will further highlight the economic tension between infrastructure and applications, it will also result in a more diverse ecosystem with improved optionality for both developers and consumers.

You can model and visualize how value accrues to different actors under different sequencing mechanisms with the Sequencer Value Capture Simulator and export the supporting calculations and data from the Sequencer Value Capture Spreadsheet .

*: Equation is adapted from Terry’s presentation, You only need one rollup , at Modular Summit 3.0.