Ep.48 [EN]: Joanna Zeng: Bringing the Best of SVM to scale Ethereum SOON
36:58 Share

You are now listening to Hotlong. I'm the host, Mabel Zhang. This show attempts to shed light on the broader Web3 community. The guests will include Web3 operators, investors, or ecosystem participants of other forms. The show will also have a special focus on the Asia-based Web3 world. Connect East and West. Enjoy the warmest Web3 conversation. Hello, everyone. Welcome back to the latest episode of Hotlong.

Today we're extremely excited to welcome Joanna from Soon. I'll let her do a bit of intro about herself and then maybe say hi to our audience as well. Thank you, Mabel. It's a pleasure to be here. Yeah, so my name is Joanna. I'm the CEO/Co-Founder of Soon. We are a SVM L2 roll-up stack

that is focusing on bringing the Sonata-level performance to every L1 ecosystem, starting with Ethereum, where it needed the high performance the most.

So yeah, I'm happy to give a little bit more introduction about the project. But first off, a quick intro about myself. I have been in the crypto industry since 2017. I actually had a pretty long career before getting into crypto. I was a currency trader for almost a decade. As I was starting my currency trading career,

It was back in the days, I think it might be like a year right after the white paper of Bitcoin came out.

And one of my economists on the trading floor introduced me to this concept that maybe one day that the fiat currency is going to be replaced by crypto. And that got me interested in dabbling into trading, buying some Bitcoin shares at the time. And sadly, I was using the one cryptocurrency.

exchange that quickly gone down to zero with Mongov. So yeah, so that turned to nothing. Fast forward until 2017 when the smart contract capability came out of Ethereum. That's when I started to look into it more seriously again.

this time more as an investment, not so much as an investment vehicle, but more as a technical vehicle. So I actually took a coding lesson on the

on the Solidity side and that started my journey as a crypto developer, community builder within New York City where I live in. We started this whole coworking space. This was pre-Banana Time. So we had this very nerdy like white paper reading group every Wednesday we get together called crypto. I think it's just called White Paper Wednesdays.

Yeah, and then that got me, sent me out to actually joining the crypto industry first as a listing institutional listing manager on the Coinbase team. And then later I moved over to the NFT marketplace launch team overseeing the Coinbase NFT marketplace.

marketplace launch and later on moved to a real web3 company as a coinbase person would say so worked at op labs and later on late alio at zkl1 blockchain as the bd head so that was my journey towards finally founding my own crypto startup that's funny what were you trading back then

It was currency like emerging market currencies. So a lot of the Mexican pesos, Southeast Asian and Che currencies as well. Yeah. So that makes sense to look at Bitcoin. Yeah, exactly. I was curious, though, you were trading, but then you decided to start coding first.

within the industry. What was the reason? Yeah, I think that was just a very quick realization that I did not speak the same language.

you know, my nerdy friends do, right? When I first get introduced to the industry, you know, reading these white paper was must. And I just realized I didn't have the background, even though I went to a technical school, Carnegie Mellon for my undergrad, but I never got a computer science degree there.

I did take some coding lessons just because it was a mandatory course. But yeah, nothing to the extent that I can fully comprehend the lingos. So yeah, so I was introduced to this. I think they no longer have this anymore. It's like a coding camp.

organized by some local developers in New York City. And they basically gave me this fast course in three months.

Yeah, a lot of that kind of based on like Java and, you know, very basic programming language. Yeah, so that was very helpful to start with. And then I started to meet more people through that course. And then a lot of them became my co-founders of this crypto developer community. Yeah, the rest of history.

I guess before you deep dive into what Soon is about, I wanted to genuinely ask at the current point of time, there are quite so many solutions available, scalability solutions. So what's the thinking behind launching Soon at this time and what's the mission and vision behind? Absolutely. I think...

Maybe just go back to the current stage of things, right? So despite there is a multitude of L2 solutions, the most commonly used roll-up stack still are based on this single-threadedness runtime environment that processes transaction one at a time. That's the EVM tech stack, right? So, you

You know, we see a lot of EVM L2s that are quite undifferentiated during these periods when we see high transaction volumes peaked. You know, really good examples of this, right? Like these Perpdexes, DYDX, GMX, Hyperliquid, right? They all try to solve that chain limitations by just resorting to building their own app chains at the end, as opposed to focusing on building their own apps.

So I think that is the current bottleneck that as an EVM L2s just have not figured out yet. And then secondly is the quality of the dApps in the EVM space have just unfortunately paled in comparison to those in the Solana ecosystem. So Jupyter is a great example of this. As a dApps aggregator, it's really...

just user experience wise is so different from some of its competitors right in the in the evm land and you see this with the with the fdv with you know mau all these uh user stats right why it's just

by far the number one DEX aggregator out there. I think the users are voting with their own transactions that the product experience subsequently the community engagement far surpassing the other similar products on EPM side. So I think the main thing when it comes back to why we did architect soon the way that it did, why we chose to bet our entire livelihood on

SVM, right? Why SVM is the end game? It really goes back to, you know, not only just the high performance piece, right? It's blocking the innovations. And I think more importantly is this

overall product quality and the developer experience and subsequently more importantly, right, is the user experience. So we wanted to bring that developer quality to the Ethereum world where frankly is facing all these lack of user adoption and you know just the overall cultural you know downside right like decline. So that's the main reason.

But then choosing Ethereum as a settlement layer means that you will still bet on Ethereum in a sense. Right, exactly. That's a good point. So how does that all ties back to, you know, still settle down to Ethereum? It's because we still have to face the matter of fact that even though Solana is attracting all these new users and kind of creating these new narratives, the Ethereum world is not going to zero, right?

that value will never just entirely disappear, right? I mean, there's still so much old money there. There's so much mindshare there, right? There's so much developer existing communities there. The reason for what we're doing is we want to really find that common ground to not compromise the

everything we just talked about that's great about Solana ecosystem, but still marry that best of the two worlds, right? What is great about Ethereum, right? Which is the user liquidity, the user base, et cetera.

Right, so I want to challenge you a little bit here, if you don't mind. Sure. So first off, I think a lot of the user experience on Solana was not just on the speed, but rather also on the interface, meaning the wallet and the DEX itself. But it's not really related to Solana in a sense.

I think the developer there, they were just able to create that sort of sleek experience. And second is that a lot of the native devs on Solana, they would rather...

find a solution that can help Solana scale even further if there need be rather than like using SVM scaling something else. So I guess that's like the two main challenge that I would give you. Yeah, so I think that wouldn't be an issue in this scenario because we are utilizing the same tooling and the same developer experience that a Solana developer will have.

So, yeah, so essentially, like, you can still use the same, you know, wallet to the same user interface, as you mentioned, right, to access soon the network, just like what you would have done with the Solana network. So it's a simple kind of, you know, redirecting the...

the application to a new RPC endpoint, but everything else is like a simple redeployment process. So that's actually one of the biggest issues when we were comparing the three metrics that we were looking at, like all the next-gen VMs, right? Because there's also the Move VM, the Fuel VM, right? There's a couple other even newer ones. Parallel EVM, I guess, also included.

we really wanted to go with SVM because Solano-Versus-Hu Xin not only had the more mature ecosystem, but also the more mature developer tooling experience. So that's the first question. I think the second question is more about going back to the long debate on Twitter. We've seen this

I think like for a month now, the whole, you know, R2 for Solana, is it parasitic or is it going to be network extension, right? So you might get different perspective from different people, but my ultimate answer on this is

is the L2 actually adding value to the overall ecosystem? And there's some definitions around that that we can add here. So one is from a technical perspective, we are coming up with a new design that

we're going to open source this and we're going to generalize it. So we're working with Anza, the Sonana developer shop, to basically contribute back the innovations that we made on top of Agave Ripple.

to the overall community as far as how to add Merkleization to the Sonana virtual machine, how do we add basically decoupling the user threads from the voting threads. So as an L2, we don't have to care so much about the consensus layer, so that strip out all that about 30 to 50% of the overall Sonana mainnet transaction. So we don't just fork the Sonana client side.

So yeah, so all those modifications is net positive, right? To the Solana ecosystem. And you can use it for not only just the rollup use case, but maybe for like permission chain or other type of use cases. And then the second thing is,

not only just from a technical perspective, I think that the overall, like how do we grow the pie for Solana is also coming back to like the ecosystem building side, right? So how do we get net new people to come into the SVM space? I think in this case, we just expand the definition of

the ecosystem as svm ecosystem not just a solana ecosystem right so along with other you know svm projects we are attracting net new developers into the space because now they can for instance in our case because we're not solana l2 or ethereum l2 we're actually getting a lot of ethereum builders who are so not curious to come over to the svm space

Or even we are in emerging markets, right? That's our expertise. So we are getting all these emerging developers in emerging markets to try out this new tech. Excited to hear about how you exactly design the Soon infrastructure, I mean architecture. But before that, what exactly you guys did recently? Because I definitely saw the mindshare around Soon was rising quite a bit lately.

What was the focus? Yeah, good question. We obviously launched our mainnet. That's our biggest milestone to date. So we were really shipping like no other. We did this entire mainnet launch within six months time.

with the first three months on the DevNet and the next two to three months on the TestNet. And just last Friday, we made that announcement of the AlphaMainNet. So we announced our tokenomics recently, and specifically the big part of that is the 55% of the tokens will go to community via Coming Soon NFT Mint.

There are two ways to get discounts. One is via Kaido scores based on your social following on Twitter. We're really happy to be the first one to integrate Kaido's API to directly benefit Yappers and through its referral bonus.

So all the token design is based on this principle that tokens should go towards those who actually contribute to the ecosystem, not just the whales. So even though we got some major VC commitments early on, we are all about the long-term community growth and we chose to give both VCs and users a fair shot at the token allocation.

The mechanism here is through full transparency and fairness. From pricing of the four NFT tiers to the release schedule, everything is now in the open. Some folks may have concerns about the fact that we are not using an existing playbook. This token issuance approach by no means is a copy of somebody else's.

but very much a first of this kind. First round or the last round, the only round that we raised was from co-builders, from people that we have integrated services with or have co-built a narrative with. Yeah, so we really kind of

not only just like talk the talk but actually walk the walk and launch the mainnet we wanted to further that community first narrative even more with our token launch model so the key question to ask here is how is our approach different can normal folks participate in the community rounds that really are reserved for KOLs no but you can participate in our NFT mint

And that's why you should be bullish. Before we go into the differentiation of your architecture, I wanted to touch upon what you think about the differentiation of student other than the scalability issues compared to other solutions. My whole team is a few people like myself.

are all strategically located outside the US, outside of the Western market. We really did this because we thought that we're really bullish in areas like Southeast Asia, like India, like China, like UAE, Turkey, right? All of these regions where we see there's a huge demand coming from the developer side that are underserved.

And we want to be there to provide that like dev rel support, right? The underground support, as well as this incentives layer to get these folks into the crypto world. We have the expertise in these regions,

Just by virtue of like, I can give you some examples. My head of ecosystem, he was the previous founder of this accelerator that's really well known within APAC region. So, you know, he has all these connections with really high quality teams, hackers and teams that come from this Chinese speaking community.

We have another APAC lead, BD lead, who's been AKL herself in the meme coin space. And she also has been working at, before joining us, she worked at DPNVC. So she knows a lot of these specific consumer DPN projects.

that we also found a lot of attractions with our tech stack. And then another individual, our DevRel, was part of Super Team India. The other DevRel was part of Super Team UAE, and they were all part of this previous hackathons within the Solana ecosystem, so they all know the local community very, very well. So I think all of this is just to show you that we have a

really strong tie with the local markets and the expertise. So when it comes to what we're doing right now, for instance, you know, I'm at this soon village, right? We call the hacker house in Malaysia to, to co-work with our ecosystem partners to make sure that, you know, we unblock anything that they experienced when they move over from test net to main net and

and we're doing this in a very coal building way. I'm also like really surprised by non-Western market, like there's not so much of the

the East Maxi versus Salama Maxi, right? So people kind of care less about being in one of these tribes, but more about like, okay, so the best developer in the country is building on soon. Let's go build on soon, right? So that's another realization. So less pragmatic. I learned a new word from you. You said strategically located. I think I'll steal that word.

It's funny that usually people just say like, my team is globally located or this and that and strategically located. That's a good wording. So now let's talk about the innovation of your design and then how your team came to that design. Yeah, for sure. So the innovation of our design, I think first comes to the fact that we are building a

You can think of us, the mental model could be the OP stack for SVM, right? So we're not just a single L2. I think that's probably like another big differentiator when it comes to like our overall vision, right? It's not just focusing on building like one general purpose chain. And in this case, we are building, we're first and foremost, are building the most efficient roll-off stack using SVM, decouple SVM.

But we are coming out with this first POC chain that is a general purpose chain, SatoSan Ethereum. But the vision is to further expand that SVM to other ecosystem as well. It could be other EVM chains, for instance, or it could be like Bitcoin and Cosmos, other ecosystem. So really, we call it the cosmic ecosystem.

SVM vision, similar to what Cosmos was initially designed for. And definitely has interoperability supporting the overall interconnectedness of the chains.

So that's one big differentiator on the tech side. And then the other thing is we really are purposely choosy about the teams that we work with. So we want to be very pragmatic about, you know, the fragmentation of liquidity issue. Right. So when it comes to who do we work with, who do we work?

sort of give this gift to, right, to spin up their own chains, we want to make sure that the chain is also not going to be competing with the gender purpose chain that we are building on top, right? So

So basically now the launch partners that we have lined up, some of them were already announced, like AI chains mostly, or app-specific chains, right? So if you think about the sort of the vision that we have, we got the Ethereum side. We have like one big L2, SVML2, that's going to take over that Ethereum eco. And then we have another one in Bitcoin, another one in Cosmos. And then we have a bunch of other...

app chain surrounding the single general purpose L2 that we have launched. So that's kind of the way to avoid this existing EVM landscape problem that we've seen, the liquidity fragmentation. You earlier mentioned that you did not directly fork the EVM. And I wanted to ask why you went for a decoupled SVM.

versus a forked SVM? Sorry, I said EVM, but I meant SVM. And what's the trade-off? So we made a pretty conscious decision early on in our development journey to focus our engineering hours on Decouple SVM because we realized with forked SVM, the performance can only be on par with Sonata.

And that was just simply like a side chain approach. So we did a decoupled approach, which actually helped with three things. One is to reduce the DA cost so we can reduce the overall fee that gets passed on to the user.

We also figured out a way to basically make the tech stack more lightweight so we can add fraud proof to make the security of the network improved by a long shot, as opposed to like a multi-sig that can be easily wrongable. Lastly, we introduced this concept of horizontal scaling. So think of like in a web tool,

space you have this message bus model so we can further modularize the transaction processing unit or TPU through like putting it onto the workload onto multiple machines so we can increase TPS performance without sacrificing the reliability of the network. I do want you to explain a little bit on what exactly decouple means.

So the decoupled SVM essentially is decoupling all the user transaction threads from the voting threads, right? So as an L2, you know, we don't need any of those consensus layer voting threads, right?

So that is reducing the empty block space that we have to post to the DA or Ethereum block space. So that itself, on the high level, really reduces the network congestion issue as well as just the cost to validate those derived blocks. Just that work, we spend a lot of time to figure it out because engineering-wise, it's not easy to do.

Everything that I just described is Ethereum concept, right? You have to first introduce this Merkle Tree concept or NPT.

Merkle Patricia Tree concept into the SVM mini engine. And then we have to really further optimize the performance because usually when you introduce merkleization, it impacts the performance. So we figured out a way to reduce that or to eliminate that by introducing this concept of not having to synchronize the global state at every slot.

on every Merkle tree slot, but actually syncing quite infrequently, like every 15 minutes. It coincides with the batcher upload time. So that's just doing absolute necessary time frequency. But in that way, we can continue to maintain the kind of performance that we want to see. So we basically make a lot of these modifications to make sure that we can build a native system

SVM roll-up that can still basically is using the same concept as like Ethereum roll-up. So we call it like the first native EVM, sorry, the first native SVM roll-up on Ethereum. You guys making the design for fraud proof securities, there were some trade-offs. What's the thinking behind?

Because we use OP stack as the minimum code diff that we did to link L1 and L2, the derivation pipeline. So OP uses something called the MIMPS, which doesn't support Rust. So we ended up having to trust another type of like a DK-proofing mechanism. So we're talking to teams like Succinct,

to figure out how to add ZK Fraud Proof to our current infrastructure. So one big improvement on that is with ZK Fraud Proof, we can reduce the wait-throw window from seven days, we will usually see with Optimist roll-up to a lot less, to days or even hours.

And then the other thing is basically now with the decoupled SVM, we can support the fraud proof mechanism. Whereas like with a fork SVM, you can achieve the same level of fraud tolerance, something called the forced inclusion with the current fork SVM. Yeah, so we basically

wanted to ultimately provide this transparent mechanism for validating transactions in the L1. And then you also took a modular approach for the architecture? That's probably the very first thing, right? When people think about, you know, soon it's a modular L2, right? Using the Solana as an execution environment, but using, you know, Ethereum or other L1 as the settlement layer to be able to get those two

together, put them in a native way, as mentioned.

takes a lot of thinking, right? It takes a lot of design space thinking to make sure that we're basically doing it the right way. But we also try to not put too much engineering efforts to re-engineer something that has been previously battle tested. So we really spend most of our time on the DECAPO SVM, very little time on the linkage between the L1 and L2, which is what I mentioned, the OP stack.

because OPStack has been audited by many, many people and has been better tested. So we didn't want to, you know, we want to be very pragmatic about

were to save our time on. Right now it's still OP stack until you adopt the ZK fraud proof. Exactly, yeah. So even with ZK fraud proof, we can do something like an OP succinct, right? So that's what that succinct team has done with OP team to add ZK onto the OP stack. So not necessarily turn the rollup into a ZK rollup,

but just simply OP roll up with a zk fraud proof. I was curious to hear, so after all of these mindful and thoughtful design, what's the developer programs or incentive that you guys are proposing? So this goes back to the whole token distribution model that we announced by just this past or this week, actually this Monday.

So as you can probably see when you go see that pie chart, right? So we have 25% of the overall tokens reserved for the ecosystem building.

So that's through, not in one go obviously, but through many seasons, we're going to continue to incentivize the early builders, folks that will come in to continue to provide these new narratives to us. So specifically, I think what will be really relevant for the audience is

We're going to also do an airdrop period after the NFT mint. So that period is when we introduce some top applications that we have done the quality control. We know are really going to be killer apps within the ecosystems. And we want to incentivize users on those either DeFi apps or we

We have a prediction market coming out. We have a consumer DPN coming out. So yeah, so those are the applications that you will be able to explore right away with our Mainnet launch campaign, basically. And I wanted to actually spend a quick moment on like what I mean by consumer DPN, 'cause this is something that really started to dawn on me, right? So with our really high TPS and low block time,

What this test that really uniquely enables are these really high frequency but low notional, what I call micro transactions. A lot of these consumer apps that are not so heavily leaning on the hardware that is really heavy compute driven, like a GPU type of hardware.

but something like a everyday device. What we've seen, like a decentralized Wi-Fi router, or we have a project that uses like an electric toothbrush, right, to record your everyday, you know, healthcare data and upload that on chain and then really gamify it, you know, just kind of

borrowing the inspiration from Stephen and then really kind of incentivize these healthy behaviors to compete with your friends and family. So I think that's a really interesting

use case that we've seen a lot of attractions in, like what we call like consumer deeping. And then earlier you mentioned your partnership with Anza. Anything else that you would like to share? Outside of Anza, yeah, we have a few other partnerships, especially from an ecosystem standpoint would be very beneficial for the ecosystem partners. Obviously with the DA layer, so we're partnering with Eigen.

So, I can also provide a really generous usage grant to the ecosystem participants. So, we'll be able to basically use the IGON token as additional incentive token to help drive and bootstrap those user liquidity of some of the top apps that we are partnering with.

That's one. And then secondly, we are partnering with Hyperlane. That's our interoperability partners. So we use them to help with the soon chains, everything within the soon universe, as well as some of the out ones with the major user projects.

Yeah, so the other partners that we're excited about is Hyperlane. So they've been the underneath tech for our inter-probability solution. So not only we're using them for the cross-chain messaging within the Soon universe, all the chains that exist within Soon, but also with other L1s that has high user adoption like Solana and like Tom.

So Hyperlane also has provided an ecosystem grant that we will be able to utilize to help further that user incentive. We've come towards the end of this podcast. Any last words you want to share with the community? Yeah, I think the main thing is we have been building for the last six months and we're

we've been taking a lot of user feedback and we're building these tech stack that is really essentially tailored towards the use cases that we've seen that have found attractions such as consumer deep in like MIMFI, like AI, right? So we will continue to hear the feedback from the community.

but also focus on the next few months building the user side of things, right? So I think on the developer side, we are...

Not set, that's not the right word, but we're very fortunate that we got a pipeline of great quality teams that has done building. But we wanted to turn our focus towards the user side and through these incentive campaigns and certainly with the community round, we're hoping to get more and more people to come in and explore the Sunmei Land.

Cool. Thank you, Joanna, for coming. It's definitely exciting roadmap upcoming four or five months and looking forward to see the other updates on the social media. Yeah. Thank you so much for having me, Mabel.