Superchain and the Monolithic Experience - A cross-chain guide for Superchain

1

Hey Optimism Collective, I’m Sandman, a researcher associated with Delphi Digital who worked on the cross-chain report for Superchain.

For context, here’s the RFG for the Cross-Chain Interoperability research (RFG-8): Link
Here’s Delphi’s proposal: Link

And here’s the full report we worked on: Link

In a nutshell
To start with, we go into some important concepts in the cross-chain context to help understand the whole report better. We define Token Bridges, Messaging Bridges and the rest of the components in the interoperability stack as we will be discussing which parts of the stack Optimism should focus on and why.

We then dive into various stakeholders like Users, dApp Builders and Chain Builders, and what are their preferences are when it comes to interoperability. For example,

  1. Users care about user experience, costs of interoperability and the security of the tokens they are holding
  2. Whereas dApp Builders care about interop standards, composability and abstraction of permissions, etc.
  3. And Chain Builders care about the sovereignty, MEV extraction (or prevention) and reorg dependencies, etc.
    Optimism needs to satisfy each of these stakeholders before making major changes to the Superchain stack.

Then we look at how other ecosystems similar to Superchain have approached the interop problem and how they are planning to. We also compare how these solutions affect each of their stakeholders (users, dapp builders, and chain builders).

We believe shared sequencing is a powerful solution that can solve many of our problems. The next evolution of shared sequencing, which is based sequencing, is the best solution to the Ethereum interop problem. We believe this decision is an important one that should be made with a lot of discussions.

The Main Part
We later make suggestions on the directions Superchain should take. There are many directions Superchain can choose to go into. Every option we list trades off one factor for another.
1. No Changes to the Base Layer: We explore what it would look like to let the market figure it out for themselves. Interop will be solved one way or the other. Superchain may choose to solve it or let new teams pop up to solve it. Chain builders benefit here the most as they get to retain their sovereignty. We do not recommend this, for the sake of our industry.
2. Asynchronous Composability: In this approach, Superchain chooses to offer trustless interop between chains thus solving interop to a very good extent, but does not offer the complete solution. This may limit what dApp Developers can build on Superchain.
3. Shared Validity Sequencing: This seems to be the approach Optimism is taking right now. Although a great approach, we believe it trades off permissionless interop with all the old and new OP Stack chains. Chains will have to be super vary about who they interop with as there can be frequent rollbacks via fraud proofs.
4. zk-Aggregation Layer: If OP Stack chooses to turn into a zk-rollup, this solutions seems to be the best option available. The main tradeoff is giving up the benefits of Optimistic rollups. On the other side, it is possible to achieve secure atomic execution natively across all of the aggregated chains.
5. Shared Sequencer Networks: A shared sequencing network solution built to offer interoperability as a feature. The chain builders lose part of their sovereignty if they opt in. But gain interop and decentralization.
6. Based Sequencing: Based Sequencing is about interop with the L1. We believe this should be the ultimate goal for rollups. With this approach, Ethereum can work as a monolithic chain while retaining the benefits of the fragmented rollups.

Finally, we roughly gauge the amount of effort required by various stakeholders to offer the monolithic experience to their stakeholders.

This TLDR version just scratches the surface. Please go through the entire report, ask questions and voice your opinions in the thread below. Thank you.

And I thank Mark Murdock, Can Gurel, Ceteris, Facundo and the rest of the Delphi Team for comments and feedback.

8 Likes
2

Really appreciate your report here. Just finished reading the entire thing.

Still need to do a more background research and process my thoughts but you increased my understanding of shared sequencing & the tradeoffs of potential solutions by a lot!

1 Like
3

I’m glad it was helpful to you.
You can hit me up on Twitter if you have any specific questions. Just sent a DM to say hi. :slight_smile:

1 Like
4

hey @sandman very interesting

can you elaborate on the differences between shared validity sequencing and shared sequencer networks?

is one of them the fact that the former is stateful while the latter is stateless? curious to learn more!

any thoughts on the feasibility/complexity of this? i guess this would not be trivial…

can you elaborate which type of dapps would be limited? feel like for most apps synchronous composability isn’t crucial since you can create user interfaces that are asynchronous and yet still appear and works as if they are synchronous - what do you think?

how would based sequencing/composability look like if there are both optimistic and validity rollups opting into Ethereum as a shared sequencer? (or are you assuming a world with essentially validity based rollups?)

1 Like
5

Sure.

1. Execution:

Shared validity sequencing design by Umbra expects a sequencer of a rollup to run full nodes of all other rollups to watch for cross-chain messages and execute them on their rollup. Of course, block buidlers can do this instead. So this design can offer execution assurances to the users.
Shared sequencing networks design by Espresso/Astria does not offer execution assurances, but can offer inclusion assurances (by filtering transactions using STFs, read more here). Execution is done by full nodes. This allows for this design to be decentralized.

2. Verification:

Shared validity sequencing relies on fraud proofs for verification. If there is a malicious cross-chain transaction, the states of both rollups are rolled back.
Shared sequencing networks relies on filtering invalid inclusion of cross-chain transactions submitted through a shared bundle. This means, it’s cryptographically not possible to modify transactions in a bundle or include invalid ones as a bundle is signed together by the user. This does not rely on fraud proofs for verification.

3. Dependencies:

Shared validity sequencing makes it more risky for rollups to choose to interop with any rollup out there as there are frequent rollback risks involved here.
Shared sequencing networks do not have such significant risks as they often do not have to rollback. Thus allowing a larger number of rollups to interop with each other.

4. Decentralization:

It is harder for Shared validity sequencing to be decentralized than Shared sequencing networks.

This discussion has been going on for a while. Even Vitalik has mentioned several times that in the long term most rollups will be zk rollups. I have not looked into the complexity of this switch, but I think it is not out of the picture.

The 2 main kinds of dApps are the ones that require Flashloans and the ones that require CoWs.
Flashloans require you to repay a loan in the same atomic bundle, hence cross-chain flashloans are impossible in an asynchronous environment.
CoWSwap is not possible across chains because it’s not possible to settle bi-directional intents at once.

Based sequencing vision by Justin Drake and the Espresso guys involves opt-in interoperability. I believe it is possible for ORUs and ZKRUs to interoperate with each other as they work the same way until their state transitions are posted on L1. This system should work similar to shared sequencer networks, where an atomic bundle is guaranteed to be included into two rollups in their corresponding blocks. I have not thought too deeply about how AggLayer can work with Based Sequencing. There is new research in this area called Pessimistic Proofs, I still have to look into this.

Thanks for the questions. Feel free to ask follow up questions and point out any issues with the report :).

1 Like