Read the Original Article on Linkedin:
In my last post, we went over that L3 cannot provide significant scaling for data availability, practical challenges of long wait time when tokens are not issued on L1 (optimistic) and high batch transaction cost (optimistic + ZK). In this post, I will explain Vitalik's L3 transaction cost calculation and the proposed alternative “batch mechanism” as the middle layer approach.
The inclusion of an L3 layer can indeed reduces the Batch Transaction cost incurred by the rollups, consequently lowering individual transaction costs. On L2s, there is a clear trade-off between confirmation time and per-transaction cost. The formula “Gas per tx = tx cost + Batch cost / (TPS * batch interval)” illustrates this relationship. In a Two-layer structure (L1 + L2), where fixed batch costs on L2 are high (e.g., optimistic rollup at 21k gas per batch, ZK rollup at 400k gas per batch), rollups must choose between a prolonged batch interval with cheaper costs per transaction or shorter intervals with higher costs. However, in a Three-layer structure (L1 + L2 + L3), where the fixed batch cost on L3 is low (e.g., ZK rollup at 8k gas per batch), rollups can achieve similar gas per transaction even with different time intervals! Hence having L3 is useful in this case!
To achieve the same objective of reducing the cost per transaction (ideally similar costs as L3 enables), Vitalik proposes an alternative approach akin to ERC-4337. This approach involves a three-layer structure consisting of a "Base layer — Batch Mechanism — rollup/validium" without the need for a complex full EVM system in the middle. The key role of the middle layer in this scenario is strictly to aggregate proofs from any compatible ZK-rollup, utilizing recursive proofs for hyper compression. It then verifies the aggregated proof and communicates with respective rollups, affirming the validity of transactions. Notably, the middle layer is designed as a "simplified and highly specialized object" with perhaps enhanced security. Most importantly, this structure eliminates the requirement for a new token, and hence alleviates the need for intricate tokenomics and governance mechanisms to coordinate a complex system with decentralized agents.
Interested in learning more about blockchain’s real-world adoption from both a technical infrastructure and strategic perspective? Follow me on LinkedIn and X (formerly Twitter - @ceciliasubmari1) and comment on the post below. I am always up for a good discussion!