It’s almost halfway through 2023. Lots of exciting things are happening at Metis, the most affordable Layer-2 infrastructure and community ecosystem available today. From Hummus to Optimism, Aave, and Forward Protocol, many leading Web3 innovators are collaborating on this project, with more to join.

Metis is on a mission to build the infrastructure for a holistic, functional, and accessible Web3 Economy. And Yuan Su, the project’s Co-Founder and now Decentralized Coordinator of the MetisDAO Foundation, is leading the technical flank on this journey. It’s a great pleasure to engage him in a conversation.

Yuan is a visionary and transformational leader, with hybrid skills and experience in both technology and business. Plus he’s a charming personality, super candid and fun, carrying a multiverse of insights.

So let’s poke him to know more about how Metis evolved, what problems they’re solving, how things look for the future, and so on. Brace up—it’ll be a deep dive.

Hello, Yuan. Thanks for your time. I’m super thrilled to speak with you about Metis, Web3, and so much more. Can you please begin by telling us a little bit about yourself?

Hi. Thanks for having me. I’m very excited too. Let’s do this.

I discovered my passion for technology pretty early in life, thankfully. And I’ve always been somewhat of a nerd—learning new things and experimenting with new ideas is my calling, especially if they’re challenging. So I got a Bachelor’s Degree in Computer Science from Acadia University.

It introduced me to things like Machine Learning, which comes very handy now. I did a thesis on life-long learning and knowledge transfer, besides getting involved with the Acadia Center for Small Business and Entrepreneurship.

Then I did an MBA in Corporate Strategy and Leadership from the University of Toronto. But I was already working at IBM alongside college. It began as an internship in 2003-04 and continued for almost fourteen years, till 2018. I was IBM Canada’s Technical Lead for a year in 2015-16 when I co-founded Metis with Elena and Kevin.

I mostly think about Metis since 2016, to be honest. But I’ve been working with CaseWare International as well. It’s been four years, and I’m currently their Head of Development for Hybrid Cloud. And last but not least, I co-founded Nuvo Technology in 2022, one of the EcoNode in Metis ecosystem. That’s more or less who I am.

Awesome! So first, to give us a wide-angle view, please share why Layer-2 solutions are necessary and how it relates to Metis’s journey so far. 

Sure. Layer-2 wasn’t our first mission at Metis. We initially promoted on-chain collaboration of DAOs and focused a lot on stuff like voting, treasury management, etc. But we soon realized it was equally crucial to focus on business activities—operations, internal transactions, etc.—which existing DAO frameworks did not support.

We saw the opportunity to build solutions that’ll enable on-chain business activities, replacing off-chain entities that serve centralized power. But we hit a roadblock. The infrastructure for fast, inexpensive, and reliable on-chain operations wasn’t ready yet. That’s when Metis got into Layer-2. And this also points to why L2s are critical for Web3 in general.

Layer-1 platforms, like Ethereum, aren’t scalable enough to enable full-scale business activities on-chain. Trying to achieve the desired level of scalability on the Layer-1 isn’t advisable either. One, it’ll have severe security trade-offs; two, it’ll be very very complicated. And you’ll anyway face restrictive gas fees and overall transaction costs on the base layer.

L2s are thus mission-critical for building the Web3 Economy. However, the fees on most existing L2s is still too high to bring Web3 to the mainstream. Plus they often have centralized sequencers, making them prone to manipulation or censorship. We thus developed Metis as a SmartL2, innovating a roll-up model to minimize fees and enhance security.

Speaking of fees, Metis is the cheapest Layer-2 solution out there. It’s boosting the migration from Web2 to Web3 by making the latter more accessible and cost-efficient. How did you achieve this?

Great question. Memolabs deserves a big shoutout for this. Their decentralized storage layer was a great help. But I’m proud that we innovated Optimistic Data Availability which, despite the criticism at the time, is critical for the industry-low costs on Metis. It’s roughly 15 cents per swap and 5 cents per ETH transfer.

We optimistically assume that sequencers will make the data available. If everyone behaves honestly, this ensures the lowest availability cost because you don’t need to upload the entire transaction data roll up to the L1 all the time, paying gas and other fees every time.

Verifiers can request data availability if it’s not available. That’s like an on-demand rollup which submits a fraction of the entire transaction data as and when required. To make this work, we innovated a unique sequencer rotation mechanism where the sequencers in a pool—each with a finite number of allotted transactions—take turns to serve the role. We’ll fully implement this soon, using threshold signing to reach a mini, pre-rollup data availability consensus among sequencers as the first line of defence against attack vectors like Denial of Service (DoS), etc. Metis also has a mechanism to pause sequencers through tiered community voting, which makes the cost of tackling rogue sequencers both finite and feasible.

People start looking for security trade-offs at this point. I’m sorry to disappoint them. Metis’s security is backed by fraud proofing and block consensus on L1, i.e., the Ethereum Mainnet. Plus we introduced a Hybrid Rollup framework combining the agility of Optimistic Rollups with the security and fast finality of Zero-Knowledge (ZK) Rollups.

We adopted a zkMIPS approach to this end, altogether offering a secure but developer-friendly SmartL2. I’m also happy to share that our Hybrid Rollups infrastructure, including all internal discussions leading to its creation, is available open-source on our GitHub channel for anyone devoted to the future of Web3.

That’s truly groundbreaking. You mentioned governance which, I gather, is another key domain for innovation at Metis. Please shed some light on that.

Yes, Metis has a contribution and reputation-based governance model that enabled the transition from DAOs to DACs or Decentralized Autonomous Companies.

We developed this idea to address the various issues with popular token-based governance models: whale voting, early adopter influence, vote sales, and abuse of power by delegates. These models are also time-consuming, while voters often lack the experience or knowledge for informed decision-making.

DACs are very efficient in autonomously handling business activities like administrative tasks, payrolls, communications, etc. Their governance model is based on the participants’ actual contributions and their reputation within the community. We use Soulbound NFTs for this purpose, inspired by Vitalik’s insightful research on the subject.

Thanks to Metis’s low-cost, high-performance framework, DACs ease decentralized business launches. And Web3 community members can thus enjoy complete ownership of their reputation—it’s a win-win situation.

You’re onboarding the Web3 Economy, indeed. But I’m sure working at this level involves many challenges. Can you tell us about them? It’ll benefit everyone building in this space.

Of course. Innovation and challenges go hand-in-hand. You can’t have one without the other. And since Metis—like Web3, in general—is constantly evolving for the better, we face new challenges almost every day. So one thing I’ve learned over time is that you must always be ready for the heavy lifting to succeed.

Applying MIPS was our first major challenge. L2s usually have a computation C that the L1 must verify before finalizing blocks. We had to transform this C into an equivalent polynomial for our zkSTARK-based approach. But this posed questions regarding processor specifications, number of registers, applicable instruction set, etc. Plus combining Optimistic and ZK Rollups is complicated as their validity-proving methods are apparently contradictory. Implementing Optimism’s Cannon framework helped us overcome these issues, besides providing a MIPS-based alternative to the complex process of generating zk-proofs using EVM opcodes.

The next big challenge is high data storage costs—both on-chain on L1s and off-chain on L3s. This isn’t exclusive to Metis. Every protocol running smart contracts on L2 and bridging contracts on L1 faces restrictive storage costs at some level. But for us, it was also about ensuring high TPS despite the Hybrid Rollup framework. So we adopted a recursive proof method called Fractal Scaling—zkSTARK proofs on L3 help verify whether the transaction batch on L2 correctly computed the L1 state roots. This reduces costs because a single proof represents thousands of transactions per batch.

Besides these, there are several associated challenges: ensuring resumability for proofs, deciding a consistent polynomial commitment scheme that generates a feasible proof size, accommodating various transaction types between Optimistic and ZK Proofs, and bolstering security through formal verification methods. Though we’ve solved some of these issues, we invite industry experts and researchers to join us in the effort to develop alternative proof systems.

Thanks for such a superbly enriching conversation. Would you like to leave us with some parting words?

Absolutely. At Metis, we’re still committed to our original goal of fostering on-chain collaboration for a stronger, better, and more relevant Web3. So we always welcome new ideas and constructive criticism from community members, innovators, and all kinds of experts.

Web3 is much more than short-sighted business interests. It’s a visionary endeavor, promising a better future for us all. We must go together on this journey, for the road ahead is long and winding. But with all the tools we have now, the time is perfect for collaborative innovation.