Layer 2 (L2) solutions are, at their core, applications deployed on Layer 1 (L1) blockchains, defined by their settlement contracts. They rely on default “Application” Specific Sequencing (ASS) provided by their sequencers.
In contrast, for applications explicitly leveraging ASS, this functionality depends on a bundler operating the user-operation (user-op) mempool. Effectively, this bundler serves the same purpose as a sequencer: it bundles and sequences transactions, submitting them to the application for execution. While similar in functionality, there are key distinctions between ASS and L2s that are worth exploring.
The primary differences between ASS and L2s are as follows:
1.Permissionlessness: Becoming a bundler for ASS is permissionless, unlike being a sequencer for L2s, which often involves a centralized entity. Bundlers only need to conform to the specifications defined in the Entry Point contract to participate, enabling anyone to run one.
2.Execution and Security Models: L2s execute transactions off-chain and settle results on-chain, relying on economic security mechanisms. In contrast, ASS requires execution to occur on the settlement layer itself, providing cryptographic security rather than economic security.
3.Composability: ASS preserves L1 composability, whereas L2s inherently lose this due to their off-chain execution model.
When comparing the overarching characteristics of these systems, ASS outperforms L2s in several critical dimensions: permissionless sequencing is superior to permissioned sequencing, cryptographic security outmatches economic security, and composability triumphs over isolation.
This makes ASS more censorship-resistant, secure, and composable than L2s.
The sole advantage of L2s is scalability, but this trade-off comes at the expense of significant downsides, including reduced censorship resistance, diminished security, and a loss of composability.
If the base layer (L1) is scalable, the rationale for adopting L2s weakens significantly. Why accept worse censorship resistance, security, and composability if it isn’t absolutely necessary?
Examining the fee structures highlights another critical point.
With ASS, execution fees are entirely directed to L1 validators, while MEV flows to the application itself. In contrast, with L2s, both execution fees and MEV are claimed by the sequencer, while settlement fees are shared with the L1. If the L2 employs an alternative data availability (DA) layer, even some settlement fees may bypass the L1.
This means ASS unambiguously directs more fees to L1 validators compared to L2s.
This discrepancy in fee allocation raises serious concerns about the impact of L2s on the L1 ecosystem.
Consider this: if an application on the L1 could perform all the same functions as the L1 but returned only a fraction of its fees to the L1 while pocketing the rest, would this be beneficial for the L1?
The answer is a clear no.
Such an application effectively diverts the majority of the L1’s potential revenue for its own benefit, undermining the health and sustainability of the L1. In essence, this is what L2s do.