It seems that ppl are mixing up the goal of a blockchain full node, which is providing a trustworthy copy of the entire blockchain state, and one of the ways to achieve that. Surely downloading and re-executing every single transaction locally allows for that, but it is

I guess I did my part of facial expression management when being asked why L2 nodes need to verify transactions when optimistic fraud proofs and zk-SNARK/STARKs are designed exactly to eliminate that need!

> Optimistic fraud proof does not prove a valid state transition/blockchain state; it's a fork choice rule that determines the canonical chain Fork choice rules usually pertain to consensus algorithms. Not sure why this term is brought up to describe something that is not

> Another strawman that avoids the point that it does not prove valid state transition, weak sauce :/ A fraud proof does not prove valid state transition, as indicated by its name. However, a lack thereof does. Sorry for skipping this point––I thought this is the basics of how

Generally, a MegaETH user wishing to confirm a transaction has three options 1. Running a node that does not validate any transaction, but rather listens for state updates from the sequencer––the kind that Keone mentioned. Security comes from the sequencer's preconfirmation and

I will be happy to repeat the benefits/importance of having optimistic full nodes (that do not re-execute every transaction) as many times as needed. Such nodes will be the kind that most users care about, because they offer a meaningful trade off between confirmation assurance

What should we call an L2 node that stores a full copy of the blockchain state but relies on external proofs (zk or op) for verifying its correctness?

Existing p2p networking protocols in crypto are overwhelmingly focused on transporting individual short messages and ensuring almost every node receives them. This has a lot to do with them being initially designed for synchronous and partially-synchronous consensus algorithms.

DA's role is not cloud storage but reliable broadcast. It ensures that online nodes can receive new blocks for purposes like generating proofs and updating account states. However, it does not care about offline nodes or nodes that come online in the future––they can sync with

Love the sound of "read assurance"! And here the chain is a data stream as in radio/TV not a growing record as in Youtube.

Built an Intel Ice Lake workstation/server a while ago (x.com/yangl1996/stat…) hoping to play with SGX when I get some time. I guess it's going to host my MegaETH node instead now.

Thanks Logan Jastremski for having me on his podcast! Really enjoyed the conversation!

Today's podcast has been one of my favorite engineering chats to date! @Yangl1996 the CTO of MegaETH and I discuss the benefits and limitations of pushing L2s to their logical extremes Our conversation encompasses: - Single sequencer endstate - Removing consensus -

Working with Sreeram Kannan for the past 6 years has been such a privilege and a thrilling intellectual experience. Cannot wait for the many more years to come! EigenLayer ♾️ MegaETH Sreeram and I met in person for the first time at Stanford Blockchain Conference (SBC)

Good question! Here's the high-level logic 1. to maximize performance we want to use a single active block producer (rotation is possible and a top priority) 2. we need to bootstrap trust for the single producer 3. we use the L1 with the best trust as well as DA/restaking infra

With leader rotation it is still only one active sequencer at any given point. Personally I don't think consensus over a small––compared to Ethereum––cohort of nodes provides enough marginal trust to justify the lost efficiency & perf. I'm sure there are ppl who will disagree.

It's been a blast working with Michael Wee and the rest of the EigenDA engineering team––seriously strong people working on a seriously solid product. Looking forward to continued collaborations!