To facilitate future developments of parallel data load, we need to encourage as many transactions as possible that contain as complete and valid access lists. Current access list pricing does not sufficiently incentivize their inclusion, and this EIP aims to address this gap.
This would massively increase the theoretical max number of state accesses in a block, to the point of making statelessness and even running regular nodes in some contexts unviable.
Today, the theoretial max accesses per block is 30000000 / 1900 = 15789. This EIP would increase that to 30000000 / 320 = 93750.
Given that we’re already struggling to make Verkle tree updates, and in the future STARKs over Binary tree branches, fast enough, I don’t think we can handle this increase.
I feel like if we want to take this path, we should actually do the other direction, and make outside-the-list accesses more expensive, though start with a gentler ratio (eg. 1.5) at first.
1 Like
If increasing the non-access list price, what about significantly dropping the warm SLOAD, TLOAD price at same time?
2 Likes