Epicenter - Learn about Crypto, Blockchain, Ethereum, Bitcoin and Distributed Technologies - Anna Rose & Kobi Gurkan: Zero Knowledge Proofs – State of the ZK Ecosystem - Part 1

Episode Date: December 30, 2022

Blockchains are immutable record keepers, storing information about every transaction ever recorded. However, this attribute comes at a cost: as more and more transactions occur, so does the size of the record expands. In turn, the hardware requirements for running a node increase which ultimately affects decentralization. Another aspect frequently criticized is how transactions are being processed by legacy blockchains in terms of speed. For both of these aspects there is a cryptographic solution that addresses scalability: zero knowledge proofs (ZKPs). These allow one entity (prover) to prove to another entity (verifier) that a statement is true, without revealing any additional information about the process or data involved. When applied to blockchains, ZKPs improve scalability and privacy by certifying that state transitions occurred, without the need to relay the underlying information.In Part 1 of this 2-part episode, we were joined by Anna Rose, host of Zero Knowledge Podcast and founder of ZK Summit, ZK Validator and ZK Hack, and Kobi Gurkan, Head of Research at Geometry, to discuss the state of the ZK ecosystem and how this promising technology is evolving.Topics covered in this episode:Anna’s background & her involvement in the ZK ecosystemKobi’s background in cryptographyTrusted setups, a brief history: Plumo & AztecZK Hack(athon)The State of the ZK ecosystemHow the ZK landscape evolved: from SNARKs to STARKsTrade-offs: compatibility, security, performanceAztec & Noir DSLThe need for performance benchmarksZK L1s: Zcash, Mina, AleoEpisode links: Zero Knowledge Podcast on TwitterZero Knowledge Podcast on YouTubeThe paper Kobi mentioned (by Ariel Gabizon)Anna Rose on TwitterKobi Gurkan on TwitterSponsors: Omni: Access all of Web3 in one easy-to-use wallet! Earn and manage assets at once with Omni's built-in staking, yield vaults, bridges, swaps and NFT support.https://omni.app/ -This episode is hosted by Friederike Ernst. Show notes and listening options: epicenter.tv/476

This is Epicenter, Episode 476 with guests Anna Rose and Kobe Gurkin. Anna Rose is the founder of Everything ZK, so ZK Pod, ZK Summit, ZK Validator and ZK Hack. And Kobe is the head of research at Geometry and also works with Anna on ZK Validator and ZK Hack and there's also Cryptography Advisor at C Lab. So lots of hats here. Before we speak with Anna and Kobe, about the state of the ZK ecosystem. Let me tell you about our sponsors today.

Our sponsor is Omni. Omni is your new favorite multi-chain mobile wallet. It supports more than 25 protocols, so you can manage all of your assets in one place across all major EVMs, their twos, so ZK Sync and StarkNet are coming soon, and non-EVMs. But what's really special about Omni is that you can do

are the most important things in Web3 directly within the wallet itself. Want to get yield. Omni allows you to get the best yield with zero fees in three taps, be it's taking, liquid, staking, lending via Ava, or volts via yen. Need to exchange USCC on ETH to Atamon Cosmos. Omni aggregates all major bridges and dexes, so you can bridge and swap across all supported networks in one transaction directly from within your wallet. Love NFTs. Omni also offers NFT support, so you can collect and manage your favorite

NFTs across all chains in one place. And it's a super easy way to use Web3. And most importantly, Omni is fully self-custodial, meaning you never have to trust anyone with your assets other than yourself. And if you want, you can even use Omni's ledger integration, so all of your funds stay on your hardware wallet. Join tens of thousands of users in this next generation wallet by downloading it today on iOS and Android at Omni.app.

Thank you for coming on shortly before the year. holiday season. Good to be here. Yeah. Fun to be on the other side. It is crazy, isn't it? I was, I was recently on Sebastian's podcast to talk about Nosis Chain, and it's really funny because it's so different on the other side. Oh, yeah. On my show, actually. Yeah, I've been on your show before, like ages ago. So, cool. Fantastic. So we're here to talk about the state of the ZKEE, because obviously it's been, there's been a lot of changes over the last year or so. And it's been extremely difficult to kind of keep up. For me, as someone who does not pay attention to this full time, but maybe also for you guys.

So maybe kind of. Definitely. We need your help. But before we kind of dive into the ins and outs of the ZK ecosystem, let's get some background on both of you. And I'll start with Anna. over 10 years ago, I think or almost 10 years ago, I was the founder of a web two startup. And I was, you know, learning. I was working a lot with engineers at the time, but it was very web two.

It was video. It was good, good experience, good education. And then come 2017, I sort of started to enter this like blockchain world. And it was really like going back to start. Like everything I had learned in the previous. tech phase, I sort of had to unlearn. And at the same time, like, I wanted, I remember like 2017, I wanted to actually do a new project. I wanted like a second technical startup that I would start. But like right then, at least the people that I was coming into contact, they were working

on like deep client software. Like it was like the deepest, most technical stuff. And it was like, yes, there were some people pitching these ideas for like, you know, Denta coin or like, yeah, we'll use blockchain plus XYZ, but it was so kind of like phony. To really get involved, you kind of needed to be like an engineer. You needed to be like deep in it. And so I was like, okay, I can't really be an engineer. I can't really even do product at this point. So what can I do?

And somebody, so I had joined parody sort of for like a special project just to get a taste for what was going. And Frederick, who became my co-host, pitched this idea like, why don't we do a podcast? We're both new in the industry. We have a bit of a technical background. Why don't we do it? And so that's where the ZK podcast starts. And that's where, yeah, I think if anyone's listening and if they know anything about what I do.

So I think that's where my work probably starts for them. The ZK kind of family kind of proliferated or at least, I mean, you started with the ZK podcast, but now you have the ZK Summit and ZK hack and ZK validated. and they're all like independent but interrelated kind of projects, right? Yeah, you also just, you introduced me just earlier as like the, you know, in charge of all things, ZK, that is not true. There are so many projects with TAR with DK that I have nothing to do with. And actually, I mean, I always love telling the story of like, why did we call it the zero

knowledge podcast? Like a lot of people ask me like, oh, did you work in cryptography before? I was like, absolutely not. I was around engineers a fair bit. But we, for like, and if anyone listens to the early episodes, like, I think until episode 11, we just refer to the podcast as like this podcast because we had no name. And we like floated some really like crappy, crappy names before that. And then someone kind of came around at some point and said, hey, why don't you call this?

Like we know about this weird form of zero like of cryptography over there. Like this, it's called zero knowledge. you have zero knowledge. Why don't you call your show zero knowledge? It was kind of a joke. We laughed a little. It's very flattering. It was a little bit of a day. You clearly know nothing about this. Might as well, you know, be upfront about it. The thing is we thought, why don't we do it? I mean, literally at the time, the only people who were listening were like the people in the room with us. So it really didn't matter what we called it. There was very low stakes. But yeah, so we picked that name. And then around that time, I wanted to do an event in

Berlin actually about some, like with some of the people I was meeting in the ecosystem. So I just wanted to throw a small blockchain event. But because the name of our show was Zero Knowledge, I called this event the Zero Knowledge Summit. And then I shared that event with a bunch of people. It was like people could apply to talk and I shared it as far as and wide as I could. And it reached the ear of or the desk of Howard Wu over in California. California. And he got that like, oh, there's a zero knowledge summit. This is awesome. And then he applied with like eight of his friends. And I at the same time had contacted Zucco and I did get

Strad from the ECC now to come over. And so even though it really just started as like this podcast summit, but called zero knowledge summit, it became by the time we actually had that event, the first zero knowledge summit. And I actually do get to claim. first because there was an event called ZK Proofs just a month later, but we were, I think, I'm pretty sure the first. And then from then on, like, I was doing this twice a year. So, like, the ZK summit also just introduced me to new ZK topics, new ZK people, and that community grew. And at the same time, the podcast, like, it really, I mean, and I really don't recommend anyone listens to the early ones. They really suck. But, like, they really aren't very good. But, like,

around episode 21, that's where we actually first start mentioning Zero Knowledge Tech. And we do like an intro to ZK. But from that point forward, as we were learning more about it, so now we're talking like 2018, right? It's still really early in ZK. There's only, I don't know, maybe a hundred people in the world who know what this is, really. So it's like, it just was kind of lucky. Yeah. I'm really glad we picked that name. ZK. validator started in 2019. I met my co-founder Will and I had pitched him. I mean, actually it was on my show. I think it was the Bison Trails guys sort of pitched me on being a validator. They were like, you could be a validator too. And I'm like, really? Anyway, I pitched

it to him and then we started. And there we started validated on Cosmos and then Pocodot, near, extended. And then yeah, I mean, it was, this was super cool because it gave me, like, I still think, like if you go back to that earlier part of the the story like I have always wanted to be building something in this space but it's been really hard or was really hard in 2017 to figure out what that was and this validator role was like a really good way for us to start engaging in actually doing something like participating adding value to what we're what we're talking about and it was a great way to also be introduced into cosmos and poca dot and understand deeper I mean I kind of already need polka dot but like these other ecosystems

What was happening? Who are the players? It definitely influenced my thinking about the entire space. It probably changed a little bit like who I would have on his guess, what I was curious about. Well, and then ZKV has also developed. So we expanded past just sort of Cosmos and Pocod and we actually started working with Kobe in 2020, or 2022, officially. But I know we started talking about it in 2021. Yeah, like the near the end. Yeah.

Yeah. What we hadn't been able to do before, because we had been really partnering with existing infrastructure providers before that, right? Like we worked with Bison Trails in Chorus One. And, but when we joined force with Kobe, it was like we could actually start to be early on the networks and actually run the infrastructure ourselves and learn even more what was happening. And it's been really, really exciting. So that's ZKV. And this is also a super nice way to actually hand over to Kobe. Kobe. Tell us about, you know, you're part of, you know, the story, what happened so far before you met Anna and, you know, started working on DK validators with her. Yeah, so I, like, about 10 years ago, I think, like, I kind of had this opportunity in my master's degree to give a lecture about any topic that I wanted.

And, you know, I heard about Bitcoin fears before that. Like, I heard about it in a podcast, actually, like the Security Now podcast. and how about how he was trying it out and how cool it is and how he was mining 50 bitcoins on his personal computer and that was cool. But I never got into it. And like when I got into it, like back then

and got to understand how it works, that kind of what captured me. But since then, I left my job back then and started working on this space. Like I've done some products in Bitcoin, like multi-sig bullets and things of that sort. But quickly after that, I doubled down on cryptography because that was my favorite topic. And like going forward a few years, I also discovered the topic of Snarks, which I really like. Was like looking very deeply into Zcash and how things work there because basically they were the only ones that deployed Snark. back then like nobody else know how to do that and yeah I was I led a development

in a startup that was doing snarks for about two years and afterwards I joined CLABSs leading the cryptography there we were doing very interesting things around light clients with Snarks which are starting to become more common now actually. Also, we worked with Anna Bid there, like we can share about it later. That's where I kind of got to to even do more advanced cryptography and like create products from scratch and deploy them to users. So around this topic of scenarios, but also threshold cryptography, which is a topic that I also really like and I was working with Ethereum Foundation a bit around this. And I Jumping forward again, last year, about a year and a few months, we started the geometry with Tom.

And that's the main thing that I do today, but also do ZKHack and ZKV with Anna. And others, actually, there's four of us on ZKV. We should mention our co-founders on that, which is Philip and Will. So there's four of us who do it. Yeah, maybe Kobe, you just sort of hinted at this thing. So the first project we worked on together was the Plummo trusted setup. Yeah. And it was, I had been doing sort of a series on the show about trusted setups,

like finding out how everyone was doing these things and like everyone had heard sort of the radio lab version. So that's the earliest one that Zcash did. But there was like advancements by the time we did it and Plumo, this ceremony, like Kobe, was it the first time that we did? that we did like that that parallelization or like the batching was happening i think so there were other forms of it that were happening so as tech was doing something that was also optimizing by pipelining so that you start that subsequent contributions start before you verify the fully the previous ones but in plomo we did do this out-of-order execution where people could contribute in

parallel, but if any of them fails, then this whole batch is like fails. And but if they don't fail, then you get like 10x times better performance. So that was, that was cool. I'm not sure anyone else is doing that today. And especially because Plumol was a huge setup. Like it was more than like a few hours, like even more than 10 hours in some computers for every participation. Yeah.

I took part in the Aztec one that you just referred to. Basically, you were given a slot and you basically had to be online at that point. And because you had to have like your Docker image running. And basically it was, I think mine was on Christmas. And so clearly a few years back, I don't, I don't remember when exactly it was. But yeah, I remember kind of having to be there, you know, at the right time and kind of getting everything started and took a couple of hours. And yeah, it was. I think Ignite, that was the Aztec one.

I think that was 2019, 2020. I got a T-shirt. I still wear it to bed every now and then. Oh, cool. So my husband's always particularly excited when he gets to go to bed with, you know, like ceremony participant, which is says on the T-shirt. So it's, yeah. Number, which number participant you are?

No, no. Unfortunately, there's a private key on the back, but basically it's not yours. I mean, there's a key, but it's all the same one. But, yeah. Ah, okay. That would be a pretty intensive swag print, actually, if every single piece. That would have been really nice. Fantastic.

So, yeah, do you want to talk about geometry a little bit because you said that you founded it? Go for it. Yeah, we started geometry last year. It's kind of a combination of a research house where we do many cool cryptography things that we try to show how to use. new kind of constructions or protocols in a way that people may not, may have not seen before, and maybe inspire building new products around those. But also it's, we invest in companies that have this kind of deep research DNA within them. So there is a bunch of cryptography-related companies, some bridge-related companies,

some of them around scaling, some of them touch other aspects of cryptography or others. But yeah, basically we would like to talk with all these founders that have something unique in their approach to how to build more products that utilize interesting research in it. So ZKHack is like the project that, for the last year at least, I've been super excited about. ZK podcast is sort of general. ZK Summit is very much new research. ZK validator is running validators on various networks and then using some of those funds to do events and bring more ZK into the networks.

ZK hack, and this is something that came from, I guess, a brainstorm with you, Kobe. I had wanted to do a ZK hackathon forever, and it is happening. It's still coming, but I've been wanting to do it forever. But it was like late 2021, and it was still too early to really get people together on such large scale, a lot of teams weren't sure. They weren't ready even, like the tech wasn't

quite ready. And so we were talking about this. It's like, I really wanted to do this hackathon. What could we do instead? And Kobe had this idea of doing a like CTF like competition where you break ZK protocols. You'd find a bug and you'd exploit it. And it would and it was, it was like very, very challenging. And then around this, we designed this event called ZK Hack, which is like, and we've now done it a couple times, but it's, It's usually like a weekly or bi-weekly workshop. So you have like every week a workshop or twice a week of workshop. And then during the sort of break between these workshops, there's a puzzle competition running.

And we did the first one in 2021, kind of through ZK Validator actually. And it was seven weeks and it was super intense and it was super good. Because what we found out is like you find out who are the experts in these different teams. Like people you might not really know, they're not necessarily at the events. They're not necessarily presenting the papers. Some of them are. But then they were hacking and you got to know their work. And it was really exciting.

It introduced, I know both of us to a whole new group of people that we hadn't met before. Yeah, definitely. And how often do you have the ZK hacks? Oh, well, so far we've done three. I think we're going to try to do them twice a year. So we just wrapped one up. We just wrapped up ZK Hack 3. which we decided to do four weeks this time,

which was probably a better, more sane amount of time to be running these events. And we had three puzzles, and I think we had almost 30 hackers. So the thing is the puzzles are really hard. So they're not like, everyone come. It's like only the most hardcore can really do them.

At least the most stubborn. Yeah. But the workshops, I think we had over like 2000, at least 2,000 unique sign-ups. this time around for the workshops. So we like, these are, the workshops are much more focused for people who are either beginner or beginner intermediate who want to use the tools. They want to learn how to like use noir or, you know, what exactly some of these architectures look like under

the hood. So, yeah, it was really fun. So seeing that they're like several weeks long, I assume they're all online. It's like people. Yes, there's a big playlist. It's nice. Yeah, I mean, there's the whiteboard sessions, right? Oh, yeah, yeah. We also started to create content around this. So we partnered with Polygon to do the ZK whiteboard sessions, which are like, actually, and then we got Dan Bonnet to do the first three.

So it's like a mini ZK course and then a whiteboard session sort of inspired by the near whiteboard sessions. I always want to give them the shout out because like, yeah, we were really, we were like, that's awesome. Why don't we try doing it for ZK? Super cool. How do people access this? This is all on the Zero Knowledge Podcast YouTube channel. Fantastic. So they can check that out. We'll link to this in the show notes.

Are you guys ready to talk about the state of the ecosystem? Let's do it. Fantastic. So we had our first Zero Knowledge episode seven years ago with Eli. And obviously back then, I mean, and the next couple of years, it wasn't all that difficult to kind of keep up with things that were happening because there were and all that many, right? So basically, and even maybe like a year or a year and a half ago, basically there were like, there was Starknet, there was Matter Labs with ZK Sync.

There was MS that's now part of a polygon with the ZK EVM. And then there was maybe Aztec and tornado. But in the realm of kind of like ZK stuff, that was more or less it, right? There was. So in like fall of 2019, there was this like flurry of papers. So around that time, there was like all these new ideas. But I think you're right that like the implementations, I mean, there weren't that many of them. There were maybe just like academic implementations.

And those companies really came out during the bear market, the last bear market. And I remember like some of the hackathons, like the Heath Global hackathons, that's where some of those teams actually formed for the first time. Like they would have like done these little projects that sort of like tried to take some of this research and put it into practice. And those formed the companies that you mentioned. But yeah, there was still only a handful. That's wild to think about that. And I think one thing that, you know, made the change or like caused the change was the fact that the tooling developed a lot. back then, if you wanted to build something that is best on Zika,

you had to take, let's say, Zika implementation and try to write your own smart contract, it would verify it, and you would have to write your own, like, JavaScript implementation and a complete web assembly or whatever. It was super complex. But around that time, like, and that actually came both from the Ethereum Foundation with Socrates, but also from the IDN3 team, which is now the Hermes team, right? And from Jordi, they created this circum environment.

And those two made a big difference in terms of tooling, because now you could build something that was end-to-end, and you could have client implementations sprout from it, and then verify on Ethereum. And it unlocked this ability for a lot of people to create ZK solutions. and the thing that made a big change. Anna, you're nodding. You agree? What I think a lot of people, when they thought of the idea of ZK stuff,

they kind of, they did leap ahead to the use cases. And then as they started to like try to do those, they were like, oh, wait, there's like none of the stuff you need to do that at all. And so it was like catching up with the dream over the last few years. And those toolkits are so key. And there's new ones, too, that have come out since then. Yeah, there's many. Yeah.

There's a ton. I mean, if you look at ZKDS, even, there's just like, we've done entire talks on that because there's a map of the DSLs. Yeah, actually, just gave a talk about it in like ZK Proof. And I had a list. I couldn't cover it all even. Like, there's so many.

Wow. Yeah, it's crazy. I mean, it's also, it's super exciting. Maybe let's talk about the OTs that we just mentioned where they are at now. So give us, which one? do you want us to start on? Maybe like any one of the layer twos in whichever order you want. So, uh, CK Sync, Starkex, CKEVM, you know, ad tech to a certain. I mean, I think tornado is the one that we have covered, you know, to a large extent on this podcast before, but we, we both had

someone from tornado on before the recent events and afterwards we had some lawyers on. So I think tornadoes probably were good on that. But yeah, let's talk about Stocknet Matterlapse, ZKEVM, AdStech. Well, I think we can say that, like, if you look at Matter, I mean, they, a few years ago, they were, like, they were building a DSL. There was, like, a lot of the core research. So I actually had them on the show a ton then to talk about, like, circuits and exactly, like, that part.

But then since then, I think, like, as far as I've seen, they've kind of gone heads down to implement and ship. So there is a shift often. Like it's, it is much more about delivery and it's a little bit less about experimentation. I mean, this is projecting, but this is what I sort of what I got from them. And I don't know, Kobe, you can probably talk more about the Starknet emergence. Yeah, I think that Starknet was a, what was a nice development, let's say from Starkware ecosystem. Like before we had Starknet, which I think started like, end of last year, like August, something like this. If I remember correct, I'm not sure.

Before that, if you wanted to use the offering from Starkware, you basically had to partner with them and they would build with you this custom solution based on StarkX. And Starknet was giving this ability to people to participate in a general purpose roll-up. So that was a bit of a business role-up. So that was a big change.

That was a really major thing. I think that back when Starknet started, I didn't understand what it was. I didn't understand the documentation. But when I tried it, it clicked, then I got it. And the ecosystem there grew very rapidly. There are a lot of people that are learning Cairo, which is the language that is running inside Starknet. It gives you a lot of flexibility and power.

I still think there's a long way to go. Like it's still not decentralized, like most of the L2s today. Like there are some failure modes that are not very pleasant, like things can get stuck, that's fine. Like still early days in some sense. But it's a very welcome and necessary development to give this power to users that they didn't have before to run things at some sense.

scale an application that they couldn't run on L1. So that that was really cool. They were really, I mean, I think they were incredibly influential on the ZKVM team. I mean, the ZKVM teams. There's all these other ZKVM teams now. And I think even like in the conceptualizing of this like super fast environment that's still connected to a blockchain, but you can like do a lot more, it inspired people to like, oh my God, I could do this kind of app, but this kind of app.

And it goes so much further than what you often could do on blockchain. But yeah, I know that there's a number of different teams like approaching that problem, usually still using Starks actually or features of Starks. Like they're using some step that's Stark-like. But yeah, you have over at Polygoni of Maiden and you have Risk Zero, which I think kind of falls in that category as well. What about Hermes and the ZKIVM? Is that kind of like, where's that at?

Well, I mean, so ZK. Evm, we just did a wrap-up episode actually over on my show, and we talked about like the summer of ZK. Evm announcements. There was this funny week in July where like all the teams were like, we're doing ZK. Evm. But I think, I mean, they all have different approaches. I think they do work together fairly well. I think there's like a friendliness in it.

Yeah, I think they're trying to find a way to really scale Ethereum. I think that's a very Ethereum focused solution. It's like you can, it's solidity. You don't have to change the workflow, the style. You can really redeploy things. And I think that we don't know yet if there's a winner or if they're just like a, these cool separate environments that all have their own unique benefits to using. Yeah, I think it's, it's actually Kobe, we were talking about this.

Like, do we know what phase they're really at? Like, I think you can sort of do something with them, but they're not necessarily like full on. Yeah, I think like from, like if we go even higher level, we have like two major approaches to ZKVM today. Like there was the one that Zika Sync was taking and the one that StarCquare was taking, which was called Warp there that they say,

you know, you don't need full compatibility. You just need to take solidity programs, which is what people write and compile them to some ZKVM. So some people also, you know, thought about what the term ZK. Evm means. Like, is it compiling solidity or is it running EVM code? Like, there was this, I think there still is this battle around this term. But other kind of teams like Scroll and Hermes and Polygon Zero, they're taking another approach, which is actually running the EVM bytecode.

directly, and which ensures greater compatibility. So it's not like full compatibility still, because some things are more expensive when you run it in ZK, like hashes. So you have to sometimes modify gas costs. But maybe you can carry over some security analysis. Maybe you can carry over a lot of the more special functionality that you've developed. Maybe you've even written some EVM assembly code so you can use that. But yeah, I think that in the last few months we've seen a bunch of them launching test nets, which is cool. So a lot of them are functional.

Like you can use crawl public test net. You can use Hermes test net. All of them are at an increasing level of compatibility, I think, each week. So you can run things. but sometimes not everything is proven, like not all the op codes are supported. So you get the feeling of how is it to use it. So you have the appearance of a proof being created,

but sometimes not everything is being actually proven. So it's more like sometimes a simulation of the experience. But it's like an honest simulation of the experience. Like this is what they aim, that basically you can use every tool you have today. and it will work as is. But yeah, I think that that's the state where they are today. Like they're functional, they're usable, not maybe fully compatible,

and nearly not at all audited. Like their audits are going to be very complex. Yeah, I want to talk about audits definitely later, just because it seems like such a fraud thing that, yeah, that could go wrong. Anna. Kobe, didn't they just, I think Jordi just shipped it off to audit, right? Yeah. So like, I think that they shifted off to audits and like I think Jordy has kind of been,

or like Georgian and the team, right? They've been, I think, kind of training auditors in some sense. Like they've been teaching them about what it means to audit that code. Wild. So yeah, I think things are starting to be audited. But that's Polygon Hermes. That's the project you had asked about, Frederica. So that's, I guess, the stage that that one's at.

So if you kind of zoom out and kind of look at, you know, the landscape of all these different L2s, I think kind of the high school question, you know, your essay question would be like contrast and compare how they. I mean, so basically, if I'm going to deploy on a ZK. layer two. How do I decide which one to deploy to? How are they different from one another? Wait, question for you, Kobe, but if you're writing solidity code, would it look very different to you to deploy on any of the ZK EVMs? Depend, like if you're talking about like, let's say, the Hermeson scroll variants, it would feel very similar. Like it would feel, I think, also similar to how you date on Ethereum. You just need to change the RPC basically and then you're done.

Also, basically, those just kind of just transliterate like upcode by opcode, right? So basically it's kind of like you just translate like word by word and then basically using the same grammar that you used before. You just put them one after the other. But that's different from, say, Starnet and ZK. Sync, no? Yeah, exactly. There they just transpile solidity into their own VM. So that becomes very, very different.

The security analysis is very different. And there you have to audit the compiler from like this transpiler. That's what you have to focus on. And then also their VM. But yeah, it's very different. So but let's maybe let's let's not talk about the audits now. So I think we are all happy to agree that, you know, audits are important and difficult. And we will talk about them later.

So say everything is audited and in principle, we know that it, works as designed. How do these different layer twos differ from each other? Kobe, you explained it, didn't you? It's this idea of are you transpiling or not, right? Yeah, I mean, like, do you mean L2s in general? Like, or just ZK EVMs? No, just the ZK one.

So I mean, I totally get that there are some that kind of transpose them upcode by upcode and some that don't. But other than that kind of they arrive at the bytecode differently, do they have advantages and disadvantages? Yeah, so I think that if you compare the ZK EVM once, I think that it remains to be seen. So we don't know yet. Basically, maybe some teams will go further in the decentralization efforts and they will do that earlier. Maybe they'll create some more interesting environments for validators or provers to participate in.

and maybe some of them will have better performance, although I think all of these teams are extremely strong and will reach adequate performance. And maybe some of them will be more receptive to feedback or will do upgrades faster or have better bug bounties so I'll be more confident and things are more secure there. But I think choosing a Ziki EVM is really hard today. So it's more about the team, yeah.

And maybe correct me if I'm wrong here, Kobe, but I think that like on the Starkware side, it's all Stark's under the hood, like full stark's the way that it was kind of originally presented, whereas on, I know some of the other ZKEVMs, they're using like Fry, right? Like they're using like, there's like one step that's sort of like starkifying a snark,

but it's quite a different thing under the hood. Yeah, I think that basically every team is taking another. a different approach. So like you say, StarCwer is doing starks, like pure Starks with some additions. The Hermes team, they're taking this kind of layer approach that you mentioned. So basically, they're doing Plunk, but with Frye. So like they use this transparent polynomial commitment that came from Starks, but they use it with Plunk. So that's interesting. And then they do snarks on top of that to optimize the verification.

And the Scroll and Ethereum Foundation teams, what they are doing, they're doing plonk with this KZG commitment, which is needed to trust a setup, but it's a universal one so that you can reuse it. And they're doing a very complex architecture of many circuits that interact with each other. And they do kind of recursive layer composition

until they reach the EVM itself or the layer one itself where they're verified. But just as a side note, there is a very good write-up from Ariel Gabison that describes how you go from air to wraps, which wraps is kind of plonk. And it shows how similar starks and plunk with fry are. which kind of bridges this gap which which we chose how like that basically everything is very very similar you just latch on another thing and then it becomes one with fry instead of stark so it's really cool okay if you if you got none of this we will link to the to the paper that we'll be referred to in the show notes I'm sorry so basically maybe can I can I kind of summarize this

this with in principle, all of these altos do the same thing and which one is best remains to be seen when they're live and operational and some of them have failed or not failed and, you know, gone up in flames. For ZikiVM, I think that's true. Yeah. Yeah. It's best by which by which metric is probably also worth because there is that aspect of it being decentralized or not. Like you might actually suffer.

Exactly. You might suffer some performance, but you'll get like a more secure. system in a way. Yeah, exactly. But aren't all of them currently, I mean, none of them actually have a decentralized sequencer, right? No, not yet. So all of them have like a centralized choke point. So it doesn't really, I mean, as long as you have like one centralized choke point, it doesn't really matter how many more centralized choke points you add, right? Depends. It really depends because if you have a centralized sequencer, then you can do so much damage. Like, you can do all the damage. If the L2 is designed correctly, you can still force the sequencer to take transactions in

so they can't forcibly censor you. And they can hurt liveness, but they cannot hurt security because the security is guaranteed by the proving process itself. So there's so much damage that this central choke point can do, which is a really nice feature. That's what's really compelling about this kind of Zika roll-ups, because you can limit the damage that the centralized party can do. On the scale of decentralization, where do all of these sit? Very early, all of them. Okay. But I think all of them at least have the intentions, the right intentions.

And I think they have this honest intentions to do the right things. But it's just they need to ship what they have first. and move towards decentralization later when things are functional. So, yeah. We all know that this is incredibly complex technology. So we talked a lot about L2s, but mostly ZKEVMs or ZKVMs. But there is Aztec, which we haven't talked about, which we might want to say a few words on.

So like Aztec, first of all, just a team that I adore. I think they're such great people. I love hanging out with them when I'm in London. And they launched ZK Money, I think in the last two years. They recently have launched sort of a, I don't know if you call it competitor DSL to CERCOM. What do you think, Kobe? I think it's the same level.

It's not a competitor. I mean, if you treat all the ZKDS, like all the languages as competitors, then sure. But I think it brings something really new to the table. So I really like it. And they do talk about private computation, but I wouldn't put them in the category of ZKEVM. No, I think they're very different kind of structure. They're not a ZKEVM. They're not even a ZKEVM.

But and they did not have this recursive power yet, like that would allow private smart contract structures. But that's where this new language that Noir that they released is aimed to enable that. So basically, it would write the circuits with Noir, which looks like a normal programming language. And then you would be able to deploy it inside their ZK roll-up. And that's why they call it a ZK-ZK roll-up.

So that you can verify within their ZK roll-up more ZK. So that's what they do. So that's what Noir would give them. But basically, before we move on, just to make sure that we gave the right picture, there are other L2s, right? Like there's Starknet and there's, like you say, ZK Sync and all of them are not even EVM, like they're just VMs. And if you want to choose them, it might be for other reasons, right? Like you would choose them for the reasons that you're willing to learn a new language or a new environment. And you want to enjoy this greater performance that is good.

but you suffer from maybe other reasons because maybe you need to do new security analysis. Maybe you need to train more developers. You're taking a bet on a new VM. Maybe it won't be as expressive as you want because in the EVM you know what you can do and what you cannot. Maybe here you will find that you can do more, but maybe you'll find that you can do less. So there you just have to do some homework and maybe experiment and see. see whether it even fits your use case. So how is the performance different if I go to?

I think isn't it still? Kobe, can you answer that? Is there clear metrics on these? No. Yeah. I think it's a community. This is one of the things that we need to develop. Like I actually had a chat about this with Tom today.

Like we need to develop some framework for benchmarks for all of these L2s. So it's interesting, like to do apples to Apple's comparisons here. It's not clear. Yeah, absolutely. So basically, if, if you know I'm going to build an app that I want to run on one of those L2s, and basically it's going to need a shitload of compute or it will need lots of data availability and so on. And basically kind of having like this click through experience where basically tells you

you should definitely deploy to Starknet or something. So I think that would be super interesting. Yeah, or even some common use cases and like create this common use case on all of the different L2s and just evaluate how performance it is. That would also already be big progress, I think. So we've talked about L2s for a bit now. let's talk about L1s. It used to be that Zcatch was the ZKL1, and everything else was sort of like building ZK

on top of things. So there's a lot of teams that are like also there may be in other ecosystems like Penumbra or Manta. Let's talk about kind of where the L1s are at and what their usage is like and whether they're life at all or whether they're kind of in proof of concept stage because basically the legacy privacy presence. having L1s. To me, that would be Zcash and Monero. And it's gotten a bit quiet around those. But obviously, the L1s that you just cited, they've gotten a lot of press recently.

So where are they at? Maybe just one thing about the ZCats. So Monaro, I was never, I've never really explored very much. But I will say something about the Zcash crew is like, in terms of research, they released Halo 2. And it's being adopted by almost like, like just all over the ecosystem. So in terms of like output on research, they're still really top in a way. But it's true that like I think a lot of the other projects are bringing, because they're often at least exploring programmability or being able to deploy things on top of them and not just being money transfer. A lot of them are also proof of stake or they've like evolved the consensus to to fit a little bit more of the narrative of today. I think so the one that went live,

probably the earliest of that batch is Mina. They went live in 2021, I believe. And yeah, side note, I'm an advisor to them. So just disclosure there. So they used to present this idea of snaps. And this was like ZK apps on like on Mina. But they've rebranded that. So it's ZK apps now. And as far as I know, like there are, there's a cohort of people building things on TestNet now. So like it's if you want to start really playing with these things, it's there to play with. It's maybe not yet live on the main net. And I don't know what it's going to take to get there yet. But yeah, that's kind of cool.

That's part of the reason why I think the ZK hackathon idea can come to fruition this year, because there is finally environments where people can really build stuff without like having studied this stuff for months and months and months in advance. And going on, I mean, okay, so ALEO is. I know they're in like TestNet 3. Kobe, I don't know. Have you tried it out? I don't know actually how much you can do yet.

Well, I haven't tried the TestNet 3 itself, but I've like double with the tool chain that they have. Like they've like Elio tried to develop a whole stack of things. So they developed a new language. They develop this new blockchain that has programmability in ZK. And the language that they have is also very interesting because it combines, combines both the blockchain part and the circuit part in the same syntax, which is really fun to see and really fun to work with.

And so I've tried that and that's really cool. But yeah, it's like, they also have like this concept of proof of succinct work, which is an interesting approach to mining because like they're kind of in a hybrid, I think, right now with proof of stake and proof of work. And like they use this part because the proof of work itself uses snarks. So they use this part to basically incentivize creating hardware accelerated snark provers in some sense. So it's kind of a whole circle.

So it's a nice ecosystem. Yeah, we might want to mention actually. So the ALEO team also like they put together the Z prize this year, which was an attempt to like push and fund hardware acceleration, which makes sense with their product. And I think the results just came in. We did a little, Alex did a presentation at ZKHack last week. So yeah, there's like, I think, and from what I understood, like, there were actual, there was progress made in hardware like ZK proving.

It's pretty cool that that worked, that it happened. It was like industry wide. ZKV partly sponsored it too. So like, yeah, we all got involved to try to. to push that. Yeah, I think like hardware acceleration, like the results maybe were surprising for some people and you got better than expected. And you actually got some interesting breakthroughs, like in even in things that we haven't

seen progress in like decades. And people now that they see that the market is serious about paying them for hardware acceleration, they're not looking at this and actually making interesting breakthroughs. breakthroughs. So that's really cool to see. There's also a few teams that have emerged like projects that just work on like hardware too. Like Omer's project whose name I always messed up. Ingoniama. Ingoniama. There we go.

Exactly. Yeah. I should mention ZK validators as an investor in a few of these projects and I think geometry might be as well. Yeah, geometry is an investor in Gnogamma. Yeah. There's other projects like other L1s but they live within the other ecosystem. So we have PNumbra that's like it's going to be a privacy decks within the cosmos ecosystem like using IBC. You have a NOMA, which I feel like sort of is in Cosmos, sort of is in Pocodot. It's like also this like bridged to other ecosystems project. And I know they're doing a

trusted setup right now that's going really well from what I could see. It's like a lot of people wanted to do it. We both participated. Yeah, we both participated and ZKV participated. We also have on the Pocodot side, you have Manta that is like their ZK environment. And actually I will say a little plug like ZK validator, we're actually doing a workshop series with them to use XCM with Moonbeam. So it's actually called privacy and Moonbeam. And it's cool because like I think by putting the project out in the world, we sort of pushed the developers on both sides to make sure that we could like actually make exactly this combination. So actually there will be applications building kind of over that through our workshop, at least like experimental applications.

Yeah. And like beyond the L ones that are, let's say, ZK focused, you also have more and more support in other L ones that are not generally ZK focused, but they add ZK support. So Ethereum started with that like that. That was a big leap. But SEL did that and I think Sui, So he's doing that. And Falcon, to some extent, are using, like, I think they're also, like, maybe the biggest snark consumers at this point, maybe by, like, proving power. So you have this kind of deep ZK integration in a lot of these other L ones as well.

Thank you for joining us on this week's episode. We release new episodes every week. You can find and subscribe to the third. a show on iTunes, Spotify, YouTube, SoundCloud, or wherever you listen to podcasts. And if you have a Google Home or Alexa device, you can tell it to listen to the latest episode of the Epicenter podcast. Go to epicenter.tv slash subscribe for a full list of places where you can watch and listen. And while you're there, be sure to sign up for the newsletter, so you get new episodes in your inbox as they're released. If you want to interact with

us, the guest or other podcast listeners, you can follow us on Twitter. And please leave us a review on iTunes. It helps people find the show, and we're always happy to read them. So thanks so much, and we look forward to being back next week.