Discover more from Web3 Roundup
Building Web3's foundation for scale, an interview with Mysten Labs Co-Founder/CEO Evan Cheng
Web3 is still in the very earliest phase of development. There are so many interesting and exciting applications being built, but many people ignore the fact that the foundational infrastructure of many public blockchain protocols are deficient for any level of scale. Many widely adopted protocols such as Ethereum have significant limitations to scale such as limited transaction throughput, unpredictable transaction (gas) fees and limited storage.
This week I interviewed Evan Cheng, Co-Founder and CEO of Mysten Labs. He is addressing those limitations by building Sui - a Layer 1 blockchain built for the next billion users in Web3. His previous experience at Facebook (Novi/Libra) and Apple give him the perspective to build a foundational layer for Web3 that focuses on scale.
The following has been edited for clarity and length.
What was the journey that led you to start Mysten Labs?
I got exposed to crypto relatively late, in 2016 or 2017. Once I dug in, I saw its potential to change consumer experiences, how we build products, business models — everything. What I didn’t see was infrastructure that could actually fulfill that vision.
The fundamental idea of the decentralized internet is that you can put anything on it and transfer ownership seamlessly. Everything about the product experience is going to change. But you need to be able to keep up as the product scales and more people interact with these products. I realized that what was on the market would never meet that demand and keep up with the ecosystem.
I decided to contribute to the Libra Project [run by Meta, later renamed Diem]. I knew I could draw on the expertise of this big company with a lot of talent. But I also thought that some of the fundamental design mistakes that were very evident in Ethereum back then would be solved by someone else, because surely they would realize the model doesn’t work. Nobody solved it. Blockchains today still follow pretty much the same model.
What was the jump from working on the Libra Project to starting your own company?
It was a great experience working on Libra. Initially we were just learning about blockchains. But ultimately it wasn’t satisfactory in a lot of ways, because what the product needs is a high fidelity, high integrity blockchain that serves as a payment rail and settlement layer between banks. It doesn’t need that much. We focused a lot on the security, safety, and usability aspects by building a smart contract platform.
Ultimately the design was very basic. We tried to do things like a check or a money order, where you issue a money order, put it on-chain, send a ping to me, and I take that ping to a local 7-Eleven and cash it out. That’s something we wanted to do for social goods, for giving people in poor countries without bank accounts access to money in that way. But we realized we couldn’t even do that because of the object model and storage costs associated with it. So you realize you need to push the technology forward.
We started developing ideas about an instant settlement protocol and started figuring out all these different components. We realized there’s something here. That’s when we decided to build a company.
When you started looking at blockchain technology, what were those fundamental issues that you saw?
Blockchain is built for financial transactions. It keeps a history of transactions, and the only states it keeps are fairly basic. It works reasonably well for these simple transactions, which led to early adoption with DeFi. This whole concept is about assets, and back then it was focused completely on fungible assets like currencies and tokens, and how you facilitate the transfer of them.
Then you want to adapt the concept to any kind of asset. NFTs don’t do it for something like fine art. The NFT contains a single URL, but you want to keep a lot more information than “this is the work.” For it to be authentic, you need to keep its history. For a piece of art, that history might include where it’s been exhibited, how many times it’s changed hands, the history of the value over time. You should be able to represent this on-chain, and you can’t do that with blockchain technology today. It’s extremely constraining for product builders.
Why do you think it’s been hard to evolve some of the earlier blockchains that people were working on?
When you look at TPS (transactions per second), everyone recognizes that the bottleneck is consensus, which is super slow. You send your transaction, but you’re waiting for a whole bunch of other transactions to be ordered first. That doesn’t make sense. If I want to modify an asset that I own, why am I waiting for a whole bunch of other transactions? If no one else can touch my asset, my transaction should just go through.
People have tried to improve upon consensus, sometimes with disastrous side effects, or improve upon efficiencies in other parts, like Merkle Tree, which is very hard to parallelize. Nobody has solved the fundamental problem, where you have all these transactions that happen at the same time and you should be able to tell them apart.
The breakthrough is that we model each asset as an independent object, fully encapsulated. All the data belongs to the object. The blockchain is keeping a pool of objects. That’s a fundamental change that allows us to change everything, including scalability.
Tell me about the protocol you’re building at Mysten Labs. How are you solving those infrastructure challenges you mentioned earlier?
When we started Mysten Labs, we thought maybe we’d build the infrastructure closer to the product layer. We quickly realized that the foundations were broken. You can’t stitch together a bunch of poorly performing, non-scalable networks and call it internet, right? It doesn’t keep up with the growth of the ecosystem. We feel like we’re very close to figuring that out. We’ve spent the last few months coming up with a suite that solves the problems of scalability, performance, storage, and developer experience. Those four components are all intertwined.
With our model, we’re essentially tracking the object evolution, the state changes per object, that allow us to move this aggregated state off the critical paths. This allows us to store things on-chain with no limits. And latency is shaved significantly, to keep it under a second for the vast majority of these owned objects. We move things off the critical paths.
Blockchain is a bunch of things that are hard to parallelize, so you have to redesign everything to make that possible. That’s the only way you can solve all these problems. And the developer experience is night and day.
What’s your prediction for what will happen to products built on protocols that are starting to show their cracks?
Most of these crypto-based products are not the same level of sophistication as their Web2 counterparts, to be completely honest. I’m old enough to remember Prodigy & AOL. That’s the best analogy I have for it — it’s very much in that state. I think that’s an infrastructure problem.
There are a lot of interesting ideas happening around legacy blockchain, like play-to-earn. But ultimately gameplay is fairly limited in terms of what you can do on blockchain. We’ll see that transition over. There’s still a lot of experimentation happening with current blockchains.
How about on the Layer 1 front? Do you think the end state will be many protocols for many use cases, or one to rule them all?
I don’t have a prediction, but there should be a few dominant players that capture most of the market, that just have a much better experience and product than the others. We believe we’ll be the first one, and maybe people will come up with better ideas. When you talk about protocol, wireless protocols evolved over years. Early wireless was so slow, so basic. That’s what you’re seeing.
If you were giving advice to someone building a Web3 application, who’s looking at all these Layer 1s to build on top of, what would be the most important thing for them to think about?
The most important thing is not to pick and choose Layer 1s. It’s: Who is your audience? Are you building for the masses or for a niche audience? If you want to reach hundreds of millions of users, you’ll realize that you can’t build this on Ethereum or Solana, or whichever one it is.
If you want decentralization to be pervasive, you have to change the game. My advice for product users is to think long-term. Think big. You’re in it to make waves, not just to make a quick buck.
NYC Web3 Meetups
I’ve been hosting a regular series of Web3 in person get togethers in NYC. They range from small table breakfasts to larger happy hours. If you’re interested in attending or sponsoring one of these events sign up here:
Hi, I’m Andrew Chang - I created the Web3 Roundup to share what I’m learning in this space. I’ve spent my career at the forefront of the technology industry in areas such as crypto/blockchain (Former COO @ Paxos, co-founding partner of Liberty City Ventures), video and adtech. I learn by meeting with founders, investors and other thought leaders and approach Web3 with the same enthusiasm – and skepticism – I had about crypto/blockchain technologies 10 years ago.