Scaling Web 3, with Anatoly Yakovenko of Solana


Scaling Web 3, with Anatoly Yakovenko of Solana

November 2020

Posted by

Roland Spencer

Creative Producer

Roland is also responsible for the identification and implementation of new creative strands, with a specific focus on building community, communication and content strategy, and overseeing community output and engagement for the team....read more

We talk with Anatoly founder of Solana which claims to be the fastest unpermissioned blockchain in the world (without using sharding) and his 18 year career working as a senior engineer at Qualcomm, scaling distributed systems used by millions of people.

We explore the ‘blockchain trilemma’, its history, and what he proposes is it’s false trade-off between security, speed and security and why by solving for the hard concept of ‘time’ and shifting focus to unlocking gains from hardware based on Moores Law could allow us to leapfrog the limitations of blockchains today and enable fully automated highly decentralised discovery markets for machines and the convergence of blockchain, mixed reality like AR / VR and AI. We discuss 5G and we ask him ‘can we ever have a fully decentralised censorship resistant Web 3?’ when state actors can control the physical infrastructure of the internet.

Posted by Roland Spencer - November 2020

November 2020

Posted by

Roland Spencer

Creative Producer

Roland is also responsible for the identification and implementation of new creative strands, with a specific focus on building community, communication and content strategy, and overseeing community output and engagement for the team....read more

Listen on iTunes

Jamie Burke 0:13
You’re an early stage web three founder apply to our award winning accelerator programme base camp at Outlier ventures.io slash base camp, we write your first $50,000 check and give you access to 200 mentors, including many of the leading web three founders, and a network of 1000 of the world’s leading investors and exchanges. We’ve helped over 30 startups from 15 countries from all around the world, raise 100 and $30 million in crowdfunding. It can help you fast track product market fit and where relevant the launch of your token economy. Today I’m really happy to welcome on the show co founder Solana and Tilly yakko vanko. I was really worried about getting out wrong. I think I just about flute it. Why don’t you show

Anatoly Yakovenko 1:00
up? Great to be here. Thank you for having me.

Jamie Burke 1:02
So I describe salon as web scale blockchain for fast secure scalable decentralised apps and marketplaces. The picture I’ve heard used a couple of times is layer one security for layer two speeds, ultimately, highly performant without sacrificing security. And the claim and I know you’ve been very humble about this claim, like you kind of openly said, Well, you know, depends. The claim is the fastest permissioned. Without sharding Oh, sorry, permissionless. Okay, at 50,000 TPS transactions per second. So the reason why I wanted you on the show, obviously, scaling web three is a big topic, it hasn’t gone away yet. In fact, increasingly, so especially as people want to make DeFi more accessible. And as we’re kind of looking at things like NF T’s non fungible tokens to allow us to potentially cross over into mainstream, but have a totally different requirement for volume and kind of transaction perspective. So saloner, you’ve taken a particular design approach to solving for this problem at layer one, and we’re going to we’re going to talk about that a little bit later. The project has great backers, you know, good friends of outliers, like multicoin block tower, and he’s making really good traction. I don’t know if these numbers are still still relevant. I know where three moves quickly, but 2 billion transactions since March 2020

Anatoly Yakovenko 2:25
are billion now. Oh, wow.

Jamie Burke 2:27
There you go. Okay. 478 Global validators?

Anatoly Yakovenko 2:32
Yeah, between our tasks that are made now that’s where we’re at about 478 validators. Okay, wow, congratulations, or 80 or something like that? Yep.

Jamie Burke 2:41
Okay. And, you know, but I think so that’s all really interesting. But equally, you personally have close to, but not quite, you’re not that old, two decades of optimising networks, I think it’s like 18 years or something like that. Most notably, over a decade as a senior engineer at Qualcomm working on mobile operating systems. For those that don’t know, Qualcomm is a major multibillion dollar hardware and software networking firm,

Anatoly Yakovenko 3:10
if it feels like if you have a cell phone, there’s some component by Qualcomm built in, they’re typically the most important one the modem. Right, exactly.

Jamie Burke 3:21
But you’ve also worked at several other firms, including likes of Dropbox working on distributed systems prior to saloner. And so and this is something I really want to go into on the podcast is, you know, for me, what’s really exciting about web three is that it continues to attract people like yourself, who, you know, could be quite happy in a large corporation, you know, solving equally very complex problems, but somehow web three is pulled you into its domain. And so I kind of want to leverage, you know, your wider view as an engineer, having built systems used by millions of people that have also had to solve for performance, scalability security. So it’s not just a blockchain trilemma. And I want to go into different approaches, like the history of scaling really in web three over the last decade, seven years or so. And I also know that you’ve been speaking a lot about some of the more demanding use cases at the moment. These cases, while still, you know, most blockchain struggle to kind of serve them. You know, we’re still nowhere near some of the requirements for the systems that you’ve built in the past. And especially as we look forward to, like 5g or ubiquitous AR, you know that the requirement is just going to be an order of magnitude more complex. And so I kind of want to look, look forward at that scaling in that context, too. So I way of contextualising you as a guest, I’m going to do my best to describe your kind of origin story. So you said before we came online, originally born in the mother country, Russia, Arabia, Ukraine, sorry. Okay, there you Get in during the Soviet period, obviously. And but moved to the USA what age?

Anatoly Yakovenko 5:07
I was 11. So yeah, 1992 April sometime around there. Oh, I don’t remember that. I think it was actually March 30, the day before April Fool’s.

Unknown Speaker 5:18
Was that an exciting experience or a terrifying one?

Anatoly Yakovenko 5:21
Uh, it was exciting for me. Yeah. But you know, this was like in a period where the USSR was going through some crazy like transformation, right? Because when we left, it was no longer the USSR, like, so. It was like really collapsing like, in front of everyone’s eyes. It was kind of a transfer transformative, I think. I think it probably the reason why I’m in blockchain this a lot to do with living through that. And you’re not the first founder that said that actually, we’ve had a few who grew up in the Soviet area, era watched it collapse. And that really informed their view on economics and society and systems and stuff.

Jamie Burke 5:59
So you studied at University of Illinois, graduated in 2003. BS in computer science? You then did is it your Was it your first venture the I don’t even know how to pronounce this LS Ra.

Anatoly Yakovenko 6:13
Yes, LS sear this was like a bunch of college kids with some help that tried to build the voiceover IP system, in you know, this is 2099 to 2003. In Illinois, like outside of Silicon Valley, but like in that kind of start, like tail end of the gold rush for for like, web, two companies. And we’re building something that was pretty advanced. Like, if you look at like Google Voice or Grand Central Dispatch, that was basically kind of what we had built in college.

Jamie Burke 6:46
Right. And that was kind of promoting targeting SMEs, as I understand it. And then you went straight from there into into cocoa.

Anatoly Yakovenko 6:54
Yeah, so after the.com crash, that kind of like, scared everyone that was in computer science. And I think like, if I was all smarter, I would have gone to Silicon Valley and just tried to build a company anyways. But Qualcomm was using the exact same technology and kind of hardware on the spot, and flew me out to San Diego. But every engineer there was wearing board shorts, coming straight from the beach. It’s hard to say that.

Jamie Burke 7:20
Yeah, right. Well, and it looks like it was definitely informative to a lot of the stuff you’re doing now, as long as I’m glad you did make that choice. So maybe, maybe to kind of summarise this, as I said, you were there for over a decade and you worked your way up from an engineer to a senior staff engineer manager. I don’t know the hierarchy. But that sounds pretty important. aisle six,

Anatoly Yakovenko 7:39
cell seven and like Silicon Valley speak but yeah, I was fairly senior. Basically, driving huge major design decisions and technical specification. So I’m kind of pushing a lot of stuff through really focusing on performance optimization. Like some of the I was in the advanced technologies group at Qualcomm working on some really cool ahead of the curve, things like we got the first 4k video post processing running like this was five years ago, we had a AR, like implemented, we took like an AR those running in a desktop that was overheating and got it running in a phone form factor in like a span of six months. So like things like that were were literally a tonne of fun.

Jamie Burke 8:30
So that was in the advanced technologies group. So as I understand it, you are a lead architect team of 10 working on high performance software, effectively software stack that would operate on Qualcomm chipsets for next generation applications. So you mentioned virtual, you mentioned AR, but there’s also augmented reality, there’s virtual reality 3d camera processing. Google X is Project Tango. And as part of that, you, you basically also had to kind of build this technology that would work in a while pretty quickly. So you know, thousands 10s of thousands 10s of thousands of third party developers, including a lot of major OEMs. So Samsung, LG, Google, what have you, and so you left in 2016. And then he did a brief stint in relatively short term contracts or whatever, as a software engineer, but really focused on distributed systems that was Dropbox and mesosphere. mesosphere,

Anatoly Yakovenko 9:25

Jamie Burke 9:26
Yep. So focus on distributed systems and compression. And he said the I mean, that’s like you’re optimising for 1% if, if you’re lucky for in, in these systems. So I guess maybe we kind of pause there because as we kind of go into, you know, your career on web three, why would a call comm engineer be bothered or interested in in blockchain? Right, as I said, You’ve you’re solving for you know, problems that have very high requirement. You know, Why be bothered with this? nascent technology that has all these limitations still has all these limitations.

Anatoly Yakovenko 10:04
I think actually like the limitations are part of the challenge. The reason I was even thinking about blockchain was because it was. It’s transformative because it gives people access to cryptography, like in a very kind of simple way you can think of, like me working at Qualcomm for 10 years. Qualcomm booth side code, like we would generate a signature once every six months. And Bitcoin users and by Nance and all these major financial institutions now sign a million signatures per day, like theirs, it just kind of like went from cryptography being this weird thing you did once or twice a year to now it’s part of every daily life. And it’s such a roots, power, like kind of, it gives humans so much power, because it can displace all these centralised authorities like Google like Twitter, like Facebook, no, like, Facebook gives us a guarantee that these are my tweets, right? But cryptography should be doing that I should actually be signing the stuff and then this data can propagate anywhere in the world that nobody has to hosted. There is no single authority and any what is like the source of my opinions, right? my opinions not Twitter’s right. Like really transforms the world. And to me, that was the reason why it’s important for engineers to work in blockchain is because of the cryptography aspect. And I think like the details of consensus and all this other stuff, these are implementation details. These are hard computer science problems, hard software problems, but the human transformative part is giving people the power of cryptography. And to me that that’s, I think, really, you know, change the world, like in very dramatic ways for the for the good. The really empower all the all the citizens all the users of the web.

Jamie Burke 11:58
Yeah. So I mean, I guess, is it fair to say that it was a kind of socio political poll. And then actually, once you’re drawn into it, you realise that there’s some really interesting technical problems to solve for. And that kind of triggers your, the engineering you. So as I said, I want to kind of before we go into saloner, and some of the design choices that you’ve made, perhaps trade offs, it would be good to just start at a very high level around scaling, the problem of scaling web through the history of scaling web three. So maybe we could start with the blockchain trilemma. I’m sure a lot of guests have already heard of it. But I think just in case, we’ve got some new people coming into it, we can start there. And it’d be really interesting to have an engineer like you with your background, who has this pedigree in in distributed systems more generally, to kind of just walk us through the different attempts at scaling, both, you know, in the past, and what we’re seeing happen now at layer one, layer two.

Anatoly Yakovenko 12:58
Yeah, so the trilemma is this kind of like set of trade offs between decentralisation security and performance. And the surface of it that kind of talks about that this is without expose that if you optimise for performance, and you’re kind of giving up either security where you have double spans or more double spans, or users can kind of muck with the state, or you’re sacrificing decentralisation, which is hard to define term, but it kind of means the number of validators or participants in the network. And I think the nuance that he missed, or potentially wasn’t forward thinking enough is that performance is fundamentally tied to hardware, like bandwidth, computer CPUs, and that’s an exponential curve. I got I got to experience that curve. Working on mobile Qualcomm, we started with like, you know, 16 bit single core 200 megahertz arm devices with two megabytes of RAM. And in 10 years, we got to eight core 64 bit four gigs around doing 4k video post processing. Like basically like, you know, the designs for silicon there were three dimensional, right? They were stacking silicon, the top of each other memory CPUs, like a cube of organised sand, totally different kind of, just don’t envision this, right? Like, if you don’t think about, you know, 20 years ago, me saying that we’re gonna have one gigabit internet to the home sounds crazy. Sounded crazy, as crazy as me saying today 20 years from now we’re gonna have one terabit internet without one terabit internet in your hand, right? You’re literally gonna have one terabit internet in your hand 30 years from now. So that that’s basically kind of like the trilemma is this idea that if you want to exceed the capacity of the network, like the bandwidth capacity, then then you have to take these approaches like sharding or something like that. And sharding is trying to compromise security. And decentralisation. And this is where by kind of like 10 years a Qualcomm brain says, Don’t worry about any of that, because by the time you’re done with the software hardware is going to be two, four times faster. So just bet that you’re never going to run into the bottleneck there.

Jamie Burke 15:18
Is it true that you refer to this as well, it has been photo’s Moore’s law. But a lot of people say it’s coming to an end. So do you believe that it’s going to continue to be exponential? Or do you think that something else then takes its place?

Anatoly Yakovenko 15:30
There is a totally nuanced, I mean, people have been predicting the death and Moore’s law for, you know, 20 years now, the way like first ended was, frequency scaling became harder, it became harder to increase, like double frequencies every two years. So now they kind of bump by 25%. But the density of the chips gets larger, the wafers themselves get bigger. So the number of transistors per dollar keeps doubling. And this is because there’s just kind of like, very many dimensions to optimise. And that’s not going to stop until you literally have like cubes of organising, like everybody carries around, it’s probably 100 years of innovation there just just to like optimise that. So I think like that that’s a, there’s kind of just so many directions at this thinking go, that it’s really hard to predict how fast things are going to be when we start needing something like quantum computing, you know, or like, if I say 100 years, the reality is it might be like 25, you know, or, or, or 200, right? We just don’t know.

Jamie Burke 16:39
And I think what a lot of people don’t understand about Moore’s law, it’s an economic law. So as you as you alluded to the idea that with every dollar invested, you kind of get a certain level of return over a period of time. And so I guess the question is, could something replace Moore’s law in the context of quantum computing? Ultimately, that would have to make it a business case, right? Is it you know, for every dollar invested, am I going to get a certain rate of return? And I think that’s where a lot of people struggle to say, well, it’s impossible at this stage to put a price on investment into something like quantum right. I’ve also had a lot of people say, and I’m going well out of my comfort zone here, but you because we’ve been getting these efficiency gains in the hardware, we’ve got a lot of bloat in the software layer. So there’s still a lot of room to optimise within most software,

Anatoly Yakovenko 17:29
software is, is become much, much harder to build like it, because it’s just much more complicated. But much, much harder to build it like people are building, you know, Atari Games in the 80s. Like, it just requires more tools. And then optimizations become harder and harder to analyse. But we do have better tools, better languages, better compilers and kind of it’s again, also a self scaling problem, because the beauty of software is engineers typically don’t solve the same problem twice. You know, that we do like we do two or three times, but then we generalise the problem and then everyone can do it. So but that that’s kind of part of it, that there’s a lot of human scalability around software development.

Jamie Burke 18:16
So I guess, is what you’re saying that the blockchain trilemma is a false trade off, actually, you can solve for it in different ways. So presumably, and we’re going to get to this, I guess, eureka moment that you had that inspired salada. But before we go into that is the is the assumption that a lot of the attempts that are being made by other projects are almost pointless, right? That they’re kind of trying to solve a problem that that doesn’t really exist. So everything we’re seeing at layer one, layer two, or even projects like polka dot, where you have this, you still retain this concept of shared security. Is that is that the premise,

Anatoly Yakovenko 19:03
there’s some interesting implications of having shared security, or like systems like cosmos, where you have an integrated, like protocols that can talk to each other, that are outside of just the performance trilemma, like, so it’s not like I don’t want to discount that these are pointless. But for scaling something to handle more transactions per second, and reducing the cost to the users, charting isn’t strictly necessary. And might actually make the system more expensive, overall, like for the end user. So to give you an idea of why like, you know, you have a slower design system that’s not taking advantage of the same things we are but it’s using sharding. You’re effectively have 1000, smaller pipes, and the price to send, you know, information depends on the capacity of that pipe in that channel. Right. If that channel fills up price shoots up. If you don’t have sharding, you have just one giant pipe, then you have more capacity, and therefore the overall prices are going to be much lower. So does that make sense? You kind of like, you know, you have 100 different markets that are not connected, then there’s a lot of demand and like San Francisco for you know, avocado toast, the prices are gonna go up. And that’s not going to be reflected in like, you know, Chicago, but if we had one Jen market for avocado toast, and you kind of have this like global capacity, that’s that’s fungible. Right. So that that’s kind of like the main, I think, unexpected, or at least, it was, it’s it’s a, it’s a trade off that folks often don’t talk about that prices for users might actually increase in a sharded. system.

Jamie Burke 20:46
Yeah. And again, I think that comes back to the economics, right. So it’s easy to look at technical solutions out of the context of the economic principle. And again, alluding back to Moore’s Law, you need the economics to work for to get that exponential gain. Course, let’s go into salon and so you know, that the story goes, you had, there’s a very specific mix, I want to make sure we get this right. It was was it two coffees, and one beer, two beers, one coffee, it was two coffees, one beer. Okay.

Anatoly Yakovenko 21:19
It was actually a caffeine and silay. In San Francisco, with a friend of mine, we were working on this cheesy startup taking deep taking hardware design for deep learning, and offsetting the capital cost of it for with mining. And this is how we got like, I started really getting into kryptos, we had this idea, because we both of us were former Qualcomm engineers, working at our separate companies now, but we wanted to kind of do more hardware stuff, because that’s kind of our blood, sweat and tears. And I started really getting into prefer work and why I was slow, and like the fundamentals of consensus, and had, you know, two copies on beer, which kept me up the whole night, but kind of put me in this kind of alpha state or to get, you know, idea generating state. And I have this eureka moment that you can take shelter 56, which is the proof of work algorithm used by Bitcoin, or the hash function used by Bitcoin and run it recursively in a single thread, such that when you sample this process, it generates a data structure that represents time passing for somebody somewhere. And why this was like, kind of such a eureka moment, for me fundamentally, outside of blockchain is that, you know, there is no mathematical representation of time, right? There’s no math function that actually defines time, you look at a lot of like physics, problems like that are described in math. You can run them in any direction, there’s no guarantee that they all that they can run always run forward in time. And this is this been like, probably an unsolvable problem, right? This is for for the world. And this thing is a mathematical implementation of their own time running on a digital system. And it’s not like I’ve solved an unsolvable problem. It’s just this was so cool, because it actually allows you to have some way to track map. There’s a bunch of, you know, I wasn’t the first one to do this. There’s actually RSA accumulators that have been done in the 90s. And a bunch of really smart folks out of Stanford, and other universities working on these things called verifiable delay functions, which are far more sophisticated than what salami uses. But our approach using Sha 256 is just very hardware efficient and very easy to use to actually code with without breaking cryptic, crypto cryptic, cryptographic security parameters. So this was like what kind of kept me up for like, a week. And finally, like, convinced my wife who’s an engineer to like, Listen to me, like, hey, this seems like okay, this is cool. And that was really the start of it, that that’s what kind of pushed me over the edge to go starts a lot of the reason why, like, time is so important in in this whole thing is that if you look at like our radio protocols evolved, you know, this was one of the first optimizations people tried radio towers at the same problem as Bitcoin or performed block producers, to block producers produce a block at the same time the chain gets into this unknown state, nobody knows what the head state of the chain is, two radio towers transmit at the same time or the same frequency you get noise, right information can pass. So in radio, they gave everybody a clock that they trusted, and they alternate by time. This is this is the hard part and blockchain is that everything has to be trustless and open and permissionless to achieve the kind of goals that we want to achieve. So once I knew that I had that piece solved I kind of saw the Rest of the system. And in front of me like, Okay, this is like everything that we’ve done at Qualcomm to optimise mobile phones to get 4k video faster, all those memory optimizations, all those horizontal scaling techniques that modern operating systems and, you know, like databases have used now that’s applicable to blockchain. And that was really the kind of how we went for it.

Jamie Burke 25:24
Well, I’m going to go and immediately drink two coffees and a beer. Hopefully, I don’t get it wrong. So So this led up to the point of being able to scale without sharding, avoiding splitting, security, avoiding the problem of having to, you know, move state from one chain to the other. Could you go into a little bit about the proof of history proof history is how you describe the kind of main innovation? I believe there are like several others. But could you unpack that a little bit in a way that’s accessible to non hardcore engineer?

Anatoly Yakovenko 25:59
Yeah, so proof of history is that implementation of time, and the mechanic just talked about why it’s important in distributed systems. And to kind of give you another take on it, you know, imagine you like you’re in an island somewhere and you saw a thumb drive in a bottle, right, you can plug it in, and you have this proof of history based ledger, you can examine that thing locally. And, and determine that, from the point of reference of this ledger, I see that there’s a bunch of validators that have been active in the last, you know, five minutes or 10 minutes, that based on this thing, and this point of view, I can then make assumptions about time from the rest of the network without actually talking to anyone else. So I can eliminate a lot of the messaging overhead of verifying the state for the network. And that’s this fundamental optimization, like we’re removing a tonne of the work that every other design up to up until ours has had to do to keep track of the rest of the system, the rest of the state, and make sure that everything is synchronised, and kind of in the same state machine. So that that’s been the main thing that allowed us to really move forward. And the hard part with like other approaches is that the shiny like kind of take on two or three computer science problems, we got really lucky with the one computer science problem that we just had the right idea at the right time. And everything else after that has been hard engineering problems. But they’ve been solved before. So the cool like something like turbine, if you go read through that should be very familiar to users that are aware of what how BitTorrent works. So it’s a way to propagate data across the internet globally, in a very efficient way that uses actually increases the bandwidth of the network when you add more nodes. So that it’s hard to implement hard computer science problem for streaming financial data, but hard engineering problem, but not like a brand new computer science problem. So that that was like in a one one technology that we built. The other one cloudbreak, again, database, kind of databases have figured out how to scale across multiple hard drives and multiple machines. It’s a hard engineering problem. And this allowed us to effectively kind of build our chain. So it can, you know, you can throw if our account size grows beyond four terabytes, we can just throw more terabytes at, right? Like if if we have, you know, thousand transactions and need to read and write, you know, two different states that that’s easily scalable as well. So those techniques are typically called horizontal scaling. The reason it’s called horizontal is that, you know, if you imagine like a set of machines in a grid, you can add more machines at the same level, and the capacity goes up. And if you think of it, like from a if your radio engineer worked with radio, this is very much like the Shannon Hartley theorem, where to increase capacity, you have to increase bandwidth. So you double the double the bandwidth, double the number of machines, double the number of lanes of data can propagate. And that that’s been kind of this foundational thing that we’ve been applying to blockchain. But the reason we can do all this stuff, is because our nodes don’t have to talk to each other to synchronise on time or abouts, they can kind of trust the data itself. So,

Jamie Burke 29:12
you know, as you well know, the best technology doesn’t always win. And as we were alluding to earlier, you know, most blockchains are more than kind of just a technology stack. They’re an economic social system. So let’s say you know, you’ve solved the problem. Let’s say you do have the best technology.

Unknown Speaker 29:33
Do you?

Jamie Burke 29:35
What’s your go to market, right is your intent to augment, like, fix or replace, and it’d be interesting to think through as a founders and entrepreneur, how you approach that.

Anatoly Yakovenko 29:48
So I like we’ve been pretty focused on the kind of from the start, like if you look at our first pitch decks, you know, the tagline was salona blockchain At NASDAQ speed, and the idea there, if you get rid of proof of work, and you have a smart contract platform, I think it no longer becomes a store of value, like critical use case, but it becomes a price discovery engine. But if you look at like aetherium, and all of these contracts, even something like crypto kitties, that are financial contracts, that may be entertaining, but there’s still financial contracts. And you have this massively complex exchange that everybody’s evaluating and trying to discover the price like, of any of these things, and those things are moving through all these complicated markets, like ATMs and other things. So from our perspective, like what we’re building was this really, you know, revolutionary kind of price discovery engine, it’s a global price point for for anything.

Jamie Burke 30:52
So presumably for compute, right? I mean, that’s its main is price discovery for its

Anatoly Yakovenko 30:57
its price discovery for arbitrary things that people have markets for. So even for something like, and it’s applicable to like an incredibly wide range of things. So because of our background, one of the first things we were focusing on was 5g. Because it’s really expensive to deploy, you need a tonne of hardware, but you can make that problem, a market problem if you have people that bring in self serve hardware, and you have a, like a Verizon or dish or whoever that’s providing spectrum and our users can dynamically bid and get the hardware that they want for the spectrum that their phone has access to. Right. So I can use tokens and these like ATMs and auctions to connect, you know, 1000 different providers for hardware with a million different people in like a small market like San Francisco and do this at like 100 miles an hour as a phone is driving, you know, 65 miles an hour as a phone is driving across the five right and has to switch between these towers, you know, like, you know, basically every second so that is like a that is a really cool high performance market that that you don’t typically think of, but actually does does exist. And obviously like trading and a Mavs are like a very simple version of that. But you end up with, like where blockchain can revolutionise is things like, you know, ads and adware and like all this other stuff that we’ve seen the internet that’s been kind of poisoning us, right? If you have a browser like brave it can cryptographically sell ads in a privacy preserving way against people that want to, you know, provide you information that isn’t like stealing your data. Right. But that’s all not possible in open decentralised platform. So this is kind of we were how we were thinking about this this problem from from day one, is it fair to say that effectively the moat,

Jamie Burke 32:53
this idea that a proof of stake network with this economic social moat, actually, it’s not much of a moat, because in a proof of stake network, the store value concept is is not relevant. And then also the the use cases that you’re targeting our use cases you are making possible for the first time of the blockchain. So in a way, you’re kind of leaping over this incremental innovation of rolling out DeFi, for example, on Ethereum, where you’re constantly breaking things and kind of back testing the boundary of the limit. You know, with every iteration here, you’re just kind of saying you can go straight over the top, and enable use cases that you just couldn’t even consider on a proof of stake network, there’s an existence,

Anatoly Yakovenko 33:40
you can scale, a lot of these use cases to real world, like user basis. So even like something like brave, but you know, to like one to 2 million active users, they can’t really run an ad exchange on an Ethereum. It’s just impossible. Right. But like, running that on salon, I can scale it to you know, 500 million users per day. But it’s it’s something like that, I think, is a moat. It’s a moat, because once those markets exist, they become intertwined, because just simply because of finance and arbitrage, and then this is what markets do. So my vision for cilona was that this thing could be the marketplace for everything. Because even if you take something like NASDAQ, but promises, you know, one millisecond trading, it’s those one millisecond high frequency trades are only trading against the statistical noise and kind of an incoming order queue at NASDAQ. But reality you know, like, you know, Meteor strikes a cornfield in Russia and corn prices need to adjust that news is going to propagate at about, you know, 80 milliseconds around the world because that’s the speed of light through fibre. So salona by design can propagate state transitions at the same speed, that as soon as I’ve made it, hits, you know, cornfield in Russia, somebody can take a trade and adjust the price of corn, right, and this information propagates globally. And because we have this old, you know, like rudimentary pbft system that guarantees full state propagation, everybody in the world has a guarantee that they’re seeing the exact same market, as us reflected by the current set of news in the world. And that is, I think, kind of a moat. If we accomplish this, like, I think there’s no reason for any other markets exist. That’s the goal that could be still local high frequency trading markets. But I think like, being a global price discovery engine for everything is a pretty good goal, right?

Jamie Burke 35:43
Well, I certainly like the ambition. And I guess coming back to this concept of the economic paradigm, the idea being that, presumably, I think you reference our comments on our offer. But this idea of convergence with blockchain AI is something that’s been very dear to central to our thesis and close to my heart. Ever since we we started investing in blockchain. But presumably, the idea is that you could have greater levels of complexity and automation in these markets on chain, as a consequence of lowering the economic barrier that the cost of actually using the system. So let’s go into some of those use cases in a little bit more detail. So you reference the kind of the 5g use case. I know you’ve got wormhole, which is your hackathon, right? We’ve also got an accelerator.

Anatoly Yakovenko 36:30
Yep. We have a for use cases, specifically, I think, like kind of, I think what everyone has heard of a serum, which is, was almost like, I met my counterpart of the this idea that there’s a global market that can be decentralised, which is Sam, and he’s not an engineer, but he’s basically like, understands crypto and finance, do I think the level that I understand, you know, software, an operating system, and we had this, you know, like week long, three hour long calls about, like, how does this stuff work? How should it work? What can the space accomplish, and like, in his words, he was he was like, you know, I don’t know if DeFi can be the entire world’s finance, but it can definitely be 25 to 35%, or 50%. And that is an absurd, large enough opportunity. And they kind of went for it. And really a crazy sprint started knowing nothing about salona. And about four weeks, they built a demo, project serum, and then two weeks later was live. So this was a fun time for both teams, because I think we were, you know, handling a lot of support requests about like, how does this work? Or are these API’s doing and things like that, but they were sprinting, like mad building this thing, and pretty amazing to see him accomplish it.

Jamie Burke 37:52
Yeah. And it’s interesting to hear that process of kind of collaboration with your initial use cases, to kind of bring bring the network to live. And I know that you’ve also been exploring NF T’s I think the second NF T was sold on something called solvable.com. On October 25 of this year. Could you talk us through the NFT use case? I mean, NFC is obviously very broad. But

Anatoly Yakovenko 38:19
yeah, so these are like independent teams that are started building stuff on top of salada. And the cool thing about it is that project zoom is a protocol like uniswap as a protocol, and anybody can take those markets that the hard work that they built was implementing, for the first time ever, a full central limit order book, that is the primary function of a matching engine, like f dx or by Nance, but running entirely in a decentralised platform. No one has done that before. They did it for weeks. But now that now not only is that code, shareable, the actual like, physical implementation that’s running is reusable by anyone else. And this is the beauty of composability in DeFi. So folks that wanted to sell NF T’s they took this very sophisticated and not battle tested marketplace and just plug them in. And that to me, it was like kind of like, they didn’t have to go do this work, right. They didn’t need the expertise, but they can all reuse it like instantly and for free, essentially for free. So it’s

Jamie Burke 39:16
just a new form of collateral basically leveraging your DeFi composability.

Anatoly Yakovenko 39:21
Yeah, this is a I think another transformative piece of blockchain is that it’s almost like we went from like analogue phone lines where we have to have a direct copper line that can send voltage across between any two people in the world to this like digital packet switching approach to I can build software and not only can you borrow my software through GitHub, you can actually take my programme that’s running in this decentralised platform and interface without without talking to me, like we don’t need to have a business deal or like a big pile of contracts and negotiation that you know, you’re going to like Use my server rack at Dropbox it just automatically happened. So that’s kind of like the crazy cool thing about all of this right like, and I think I imagine like this is going to be as transformative as open source was to develop it. We went from like these, like, box stores selling software right at boxes and CDs, do like open source software that anybody can remix and now to open source or like open accessible services that anybody can plug in and like remix, and that that’s kind of insane.

Jamie Burke 40:31
Yeah, I mean, I think that’s why open finances in a way, a better term, right? Because it carries through those principles. And I agree, you know, for me, defy and finance is unstoppable because of this composability. Because effectively, you’ve got these specialised protocols that are just solving a single problem really well. But of course, they’re interchangeable. But within this DNA of this hyper aggressive pursuit of efficiency and yield, and, you know, somehow trying to find an equilibrium between those two things, so it’s a closed off to come as zoom, zoom back out, and to kind of ask some of the bigger questions. So in a software sense, in the context of blockchain or web three, we talk about decentralisation and decentralising the web. But you know, you’re a networking guy you work to the interface of hardware, the physical infrastructure of the internet hardware, software, networking. How decentralised can the web be if the kind of hardware and networking isn’t also decentralised? And as a consequence, how censorship resistant can it be? So I remember, I think the beginning part of the year or midway through COVID, there was some through Huawei, China and Russia are actually lobbying the UN to propose to break the West Coast consensus on how the internet is structured to allow for greater centricity because their argument was, well, we need greater performance if we want economic games innovation of 5g and, and VR and AR. So, you know, you’re probably in a better position than most to answer that question. Is it fallacy we’re ever going to have a decentralised web? That’s censorship resistant?

Anatoly Yakovenko 42:23
I have a hard stop in 10 minutes, right. But I can I can ask you this question. So is a decentralised web possible. So to bring this back to kind of the first point that I made, is cryptography is in crazy powerful tool with like, really crazy implications is that when you look at like validators in our network, they’re not identified by physical hardware for IP addresses. They’re identified by cryptographic key. And that key is like a small set of data that can be like, transferred over, you know, over shortwave radio bouncing off the ionosphere if you have to. But that that is, that thing is what is creating decentralisation that key itself and having as many of those keys that have some stake distributed to the largest number of folks out there, that that’s kind of that is the core part of decentralisation, it’s not the machines are where the actual execution happens. So from that perspective, I think the the old world is operating in this like physical level, and it can definitely, like make it hard to for open access, and for folks to kind of get access to the stuff but there’s enough kind of like global environments where things are free enough to where that access wants, as it exists, it becomes accessible to everybody, you know, like no one in the United States is going to ban cryptography, right? They’ve tried but like, I think that’s the kind of cat out of the bag. And I think that will allow like, the systems to be open and free and you will end up in like, I think, interesting paradigms, where, you know, we can deploy like a all your UI and something like are we ever filecoin but talks to salon is its execution engine, and now you have effectively unstoppable systems, right, like truly unstoppable systems that are indistinguishable from web to like, you can tell that the performance here is any worse. If you look at if you go to like dex projects here calm the prices are updated as fast as by Nance.

Jamie Burke 44:30
Well, I think we have to leave it there I on that optimistic note. What a great way to end. Thanks so much for coming on the show. really grateful for everything you’re doing. I think it’s a really interesting experiment. You’ve definitely got me more excited by having spoken to. So wish you the best of luck and hopefully we get to speak again soon.

Anatoly Yakovenko 44:50
Awesome. Thank you so much.

Jamie Burke 44:54
If you enjoyed today’s podcast, please make sure you subscribe, rate and share your feedback to help us reach as many people as possible with the important mission of web three