Evm alt-da + conditional l1 proxy

[askucher]

Idea

Optimism is working on a way to make transactions cheaper by using a new kind of data storage solution (DAL) that isn’t based on the Ethereum Virtual Machine (EVM). However, this new feature is still being developed, and in our tests, we found that using it can cause the blockchain to split into different versions (called “forks”) more often than when we don’t use it. When we don’t use this new data solution, everything runs smoothly and stably.

To address this issue, we came up with an idea: create a L1 proxy that decides where to send each transaction based on what it does. If a transaction involves moving assets between different blockchains (cross-chain transfer) or updating the main state of the network (state root publishing), we’ll use the main Ethereum network (Layer 1). But if a transaction is about publishing data batches (blobs) collected by a “batcher” (a system that groups transactions together), we’ll use our own private Ethereum network that uses a special QBFT consensus.

We also noticed that the “batcher,” which publishes these grouped transactions, doesn’t directly interact with the “proposer,” who is responsible for suggesting new blocks to add to the blockchain. The batcher publishes transactions, and then these can be checked to figure out the correct version of the blockchain (the “canonical fork”).

Our idea aims to improve the stability and efficiency of the system by smartly choosing the best network for each type of transaction, ensuring everything runs smoothly without unnecessary forks.

[jsvisa]

The root issue seems to be the DAL’s forking tendency, so I think maybe we need to investigate why it is more likely to fork after using DAL, could you please provide more data or usecase of

To better analyze why the new data availability layer (DAL) causes more forks in Optimism's blockchain, could you provide more details on how or why these forks occur? For example:

1. What specific conditions or tx types trigger the forks when using the DAL?
2. Have you observed any errors in the batcher or proposer behavior during these forks?

[RISHABHAGRAWALZRA]

Hi @askucher , I also want to understand this more in detail, Can you share more about it, why this happens?