Building EigenLayer with Sreeram Kannan

Transcript

[Music] Welcome to CSX week 3. Uh this is infrastructure week and we uh have a lot of great things planned. We'll start uh the morning with a with a talk by uh our general partner Ali Yaha and uh Shriram Kanan the founder of Igen layer. Please join me in welcoming Ali and Shir. Shirram, thank you so much for joining us today. Thank you so much, Ali. My pleasure. Great to to talk about what you're up to and uh and also about your journey and and how you've got gotten to this point in kind of the arc of your uh career as a founder. But let's start

maybe just with a big picture about IEN layer. Uh what is your latest framing for the problem that I layer solves? IGN layer. What is IGEN layer? IGEN in German for your own your own layer. Anybody can build whatever they want. The core thesis of IG layer is open innovation. So what is open innovation? So if you think about decentralized networks like Bitcoin, you can think of these networks as made of many different layers. So there's a layer of security for example in Bitcoin pro work layer of security. Then there's a layer of consensus. is we all agree on something that what's the next block and then there's a layer of execution for example in Bitcoin it's bitcoin script and on top of it is the application in bitcoin

it's like send and receive bitcoins so all these four layers are bundled together in the bitcoin blockchain so I think of you know nowadays people talk about app chains right bitcoin is the first app chain right it's integrated all the way up and this is typically how any industry starts you you have like a fully integrated system because specialization is not possible when you're inventing something off the bat. And what Ethereum did is expand uh or modularize the system so that anybody can build applications. So the application layer becomes modularized.

So anybody can build applications and deploy them. And what does this do? You know, it's a really interesting phenomenon I call the separation of trust and innovation. So the blockchain is supplying trust and anybody can now supply innovation like I can write a piece of code and then deploy it on top of the Ethereum blockchain and you don't need to know who I am, where I'm from, you know, or anything. You can just trust that that system works as expected because Ethereum's underwriting trust into it. So this is a huge deal and uh you know one of the stories I like to say is you know one of our uh teammates now Agajesh he was a 13-year-old kid in India from a random like place he wrote this DeFi program for you know trading

and you know in without blockchains there's no chance anybody would trust this kid and put like a penny into it but there's $7 million that was locked into his contract and people trading because they don't need to know who Gajesh is or where he came from or what his anticidence are. The trust is underwritten by the blockchain. This is awesome. So opens up so much permissionless innovation. That's why we have the whole pseudonmous economy like the idea that I I can be a nobody and still build systems that are highly trusted that only JP Morgan could build in yesterday. But that is not in our view enough open innovation because what's happening is you can anybody can build applications that can fit on top

of the Ethereum structure which is the security model the consensus model the execution model the EVM and then build on top. Ethereum itself took one more step further to do the roll-up era where said anybody can now build new execution environments that can go on top of this common security and so now like a rollup is basically an execution environment you build that you prove to the Ethereum blockchain that you're running correctly but now people can build different kinds of execution environments not just solidity and EVM but you can build more you can build all kinds of other things web assembly and rust and everything else but even that has a consensus bottleneck because the way nodes come into consensus and then create your final solutions are all determined by

the Ethereum blockchain. So you have the consensus layer is shared layer is designed so that you it's just raw decentralized trust. You have a group of nodes which stake and these nodes stake and then you create an arbitrary service and for this let's say the service is I want to get consensus on AI inference. So these nodes run this AI inference software and then arrive at whatever consensus protocol you write and then like give you the outputs. So I think of this as the progressive deepening of the amount of permissionless innovation that's available. So from Bitcoin where you know anybody can participate in the network but nobody can program it to Ethereum where anybody can write smart

contracts to the rollupcentric era where anybody can write execution environments to layer where you can write any crypto service you want on top of a common decentralized trust network. That's a beautiful way to explain it. Well to summarize it so like Bitcoin is this vertically integrated blockchain that's almost entirely application specific. It has one purpose and that is to serve as a kind of store of value. It originally intended to be a blockchain for payments. In the end, it ended up being more like a kind of digital gold and that is the one thing that it does very well. Ethereum modularized a little bit further by decoupling u sort of the security aspect uh from the application.

So it introduced a programming language. Now you can write all sorts of different kinds of applications as long as you conform to the one specific virtual machine and execution like the runtime that Ethereum puts forward. like it has to be EVM and that's kind of the constraint the set of constraints that you have to operate under and you also just have to take the consensus uh implementation that Ethereum provides and so I layer goes even further and decouples like the security aspect the security that the validators provide to the network from consensus such that you can now build your own uh consensus layer and therefore also build your own execution environment have your own virtual machine uh which really kind of unlocks the design base and makes it much broader for certain things that

maybe don't fit into the EVM. You can now you can now sort of build something that would normally would would require you to build an entire layer 1 blockchain from scratch. Now you can kind of more easily tap into the security that Ethereum might offer and and build it in a way that's much much more much easier using layer. Is that like a good that's a that's a great that's a great summary of what we're trying to do. So services built on top of a layer we call them as actively validated services and it's chosen carefully like it's a service that is built on top of a decentralized trust network and the mental model that we want to create is not like hey whenever you want to build a new thing you have to build a chain. A chain is basically an ordered ledger which has some kind of

execution attached to it. But a service could be like a co-processor. You're sitting on a blockchain and then you make a call and then it returns you an answer. It could run it based on computation. It could run it based on storing data. It could run a service based on fetching data from the internet. But an actively validated service, it could be a chain, but it doesn't have to be a chain. It's just a service that provides something narrow. And you know, one analogy is to the inb 2 and cloud. You have the idea of software as a service, right? So you build a service, you say, hey, you know, here's a piece of code. I throw it on top of AWS and it's running this code and anybody who wants to use it just have subscribe to this fee model and

then use this particular service and when you're building an application today on cloud you just go and usually concatenate a bunch of software as a service solutions and then only write the things that are not available you know for your particular application. So you don't have to write your own database. You don't have to write your own login. You don't have to write your own payment solution. You use stripe. You use MongoDB. You use Oath. And then you put them together to like build whatever applications you want. That's how we got so much open innovation in in the web world. Anybody could now like spin spin out like a website like Shopify makes it easy for you to run your own e-commerce store. All this kind of stuff. We want to bring the same amount of permissionless innovation to crypto. Yeah. Another metaphor which I'm

curious I'm curious if you would agree with this this way of framing uh what I layer is just sort of historically in the world of computing there was this whole trend of virtualization where you you had like a virtual layer that allowed uh developers to essentially write code that was decoupled from the hardware the actual sort of physical bare metal on which those applications would run. Uh and that layer of virtualization would allow for much more portability of code. It allowed for more flexibility. you didn't have to know as much about the hardware on which your your application would ultimately run. And then maybe here in the world of blockchains, if blockchains are computers, kind of the consensus, the specific consensus implementation plus it runtime is kind of like the bare metal. Um and and maybe I en layer by kind of decoupling that from the

underlying security that a blockchain provides is kind of like a virtualization layer where now as a developer you you can kind of develop um an application without having to know or having to care as much about the specific implementation of the underlying consensus and you have so much more flexibility of the kind of underlying quote unquote hardware to pick from. Uh and so maybe kind of layer is like the natural next step in the same way that in traditional computing we we went from like targeting very specific physical hardware to eventually moving to a much more virtualized environment to the point at which you then have containers maybe AVS are somewhat analogous to containers and the way that containers can interact with one another may be similar to the way that AVS can be composed with one

another as well. uh and it's dangerous to think by analogy but I think it maybe is a useful framework just to understand uh like like the implications and the significance of something like layer but I'm curious do you agree with this I I think it's a great great analogy somewhat different but related analogy we use is layer as verifiable cloud right so you have the cloud and you write an application and then you deploy it and usually cloud is made by virtualization right so the idea is like you you you have an arbitrary piece of code and then earlier you have to build your own server to actually run it and and then the cloud said, "Hey, I'm going to abstract all of that out." You just run a piece of code, throw it on top of the cloud and then let it run there.

It's a cloud for decentralized services which are very fable. So you you write a service and then you say, "Hey, you know, end nodes need to put up so much stake. They need to come to consensus in this way all of the conditions and then just throw it on top of a layer and then you have this running service." So the the other thing you know uh Ali talked about containers and composability, right? So one of the things that we do want to do here is one of the p counter patterns we observed in blockchain is because every new like breakthrough needed building its own blockchain. They all fragmented the ecosystem and and each and to build a layer 1 blockchain you need to do so many things well. You need to get your wallets well. You need to make it real money. You need to get a

community and all of these things as opposed to specializing right like SAS worked very well. In fact turned out to be one of the most profitable sectors to invest because it's hyper specialized. build something which is really really focused on what what you're really good at. You know, it's much easier to figure out who will bet build the best distributed database than who's going to build the next social network. And so that's exactly what I layer allows us to do is everybody specialize and build like super specialized services and these services can now compose and work together on a common settlement layer which is a good makes makes 100% sense.

uh and we'll come back to a lot of the technical details and some of the implications of uh of what you're building and then also to the broader kind of the sort of trends in the space. Uh but I also want to I wanted to touch on your uh sort of journey um and how you came to want to solve this set of problems and then also your history as as a professor and then kind of your journey from being in academia to becoming a startup founder. So what was that story? How how did uh what's the genesis story for for a layer? Before starting a layer, I was a professor at the University of Washington Seattle where I ran the W blockchain research lab. Before that, I used to work on computational genomics. So maybe I can start from why why that happened. And my

PhD itself was in actually peer-to-peer wireless networks. I got into my PhD, you know, doing research out of curiosity, right? So you work on things that draw your interest and attention rather than out of a specific goal for impact, right? It's it's like you're wandering and you find something interesting and you go like, you know, dig dig through what that is. And that's how I got into know the space of network information theory, which is what I did my PhD in. And at the end of my PhD, like we solved some long-standing open problems in that space. And I was expecting something to happen, but I didn't even know like what I was expecting. It's like oh my god we solve this major problem in network information theory and it's going to be like something and I look around and other than the 20 people who are working

on that sub field nobody gives a damn and you know it was a little bit disappointing you spent six years trying to solve a problem and you solve it which you which in my view like I shouldn't have been able to solve but I did it and I'm like okay this is going to be amazing and it turns out like nobody cares then I'm I sit around I'm sitting and trying to like root cost this problem and then I find out that if you did not optimize for impact, you optimize for curiosity and like you know doing something which has beauty and depth and so on rather than something that has impact, you should not expect impact as a kind of like a random outcome because you're not thoughtful about it. So I drew this two axis like X-axis is like you know technical depth

you know and Y-axis is impact then I was working on problems which had a lot of technical depth but not necessarily high impact. So I said okay like let me just try to figure out what does this other axis impact and I was looking around and I went to a talk on genomics by one of the founders of the human genome project Craig Went and I'm like wow crazy stuff going this is 20 2012 crazy stuff going on here guys like inventing new species you know talking about why should robots be mechanical we should just make biological robots to do whatever we want all kinds of crazy stuff I'm like why are we like trying to get 10x download speeds here let's go and figure this thing out. So I switched all over to computational genomics which is what I

did my postto at Berkeley and Stanford and you know I was working on DNA sequencing and after I finished my post talk I was looking for faculty positions and one of the places I interviewed was at Caltech and at Caltech they asked me this question hey you know you're working on this DNA sequencing algorithms what if somebody invented like a new machine that just like solves this problem end to end what would you do and then I'm like Oh, somebody invents a new machine that just asked this problem end to end. What would I do? Said like, hey, I didn't know about DNA sequencing two years back. I figured this thing out. I'll figure something else out. And they said, okay, okay, let me ask you this question in a different way.

Suppose I give you infinite amount of resources, infinite amount of talent, the best people you want in the world. This is part part of your interview. This is part of my interview at Caltech. A great interview. Yeah, this was one of the best lessons I got in my life. And suppose I give you like infinite amount of resources in infinite um you know talent and you got 30 years. So you just go shut yourself like a hermit in your lab and after 30 years you have to come and tell me what you invented. What would you do? I was so baffled by the scale of this question because it requires the degree of unconstrained thinking that I was not used to because we're always in our own I have X I can only do X plus delta like that's our nominal way of thinking.

question was unconstrained in every sense like resources, time, talent, everything and then know what are you going to do and I was baffled and I'm like okay I I I didn't get the job like no wonder I did get a job at the University of Washington which I'm thankful for where I started working on genomics we had all kinds of awesome collaborators but within like sometime I was like I need to have a 30-year project otherwise I'm not working something so I formulated my first 30-year project as you know something in genomics basically decode how information is stored in living systems. So that was my third year project. I started I had like great collaborators that was working on it until 2017 2018 when I got into

blockchains. And so that was like a big big deal because I had a 30-year project and I had great collaborators including some Nobel Prize winners, some like breakthrough prize winners, all kinds of really good people. And my adviser calls me from my PhD and says, "Hey, you know there's this thing called blockchain know Bitcoin. Bitcoin has throughput and latency problems. this is what you used to work on your PhD. Was this David say promote Dishna? Promote. Yeah. Yeah. Who is the professor now at Princeton here? Um and um I'm like okay yeah I'm trying to figure out how information is stored in living systems. Why would I work on some random speculative BS?

took me several months like you know it was a very good fit technically for what what we knew but it was so off from my long-term view of what I should be working on for 30 years randomly one of the days I was reviewing uh rereading uh sapiens from Ulari and he makes the humans are special because we cooperate flexibly in large numbers and so I take this to mean essentially that we are special not because we innovate not because we are intelligent but because we coordinate and the usual friction to coordination is trust. So if you can solve the trust problems, it's kind of like a significant upgrade to humanity.

So that's that's that was my thesis for getting into crypto like just like the internet is our communication superighway. Crypto is our coordination super highway. So I got into crypto based on this 30-year thesis to even now like I labs our vision not to build restaking or verifable cloud or anything. is to build humanity's coordination engine. Yeah, that's our goal. And so we backtracked from like this crazy 30-year unconstrained problem to what we can actually build today. I love this because in an environment that is saturated with with memecoins often, especially on Twitter, uh this is like just a good reminder of what what crypto is actually for. It's a technology for enabling greater scale cooperation. That is truly the most impactful uh sort of

thesis behind what what crypto is and what it's about. Uh one question have uh did the ideas from your PhD and then from your work in um in genomics influence uh kind of the ideas behind layer or do you see them as completely distinct? Definitely the the set of tools that I learned from my PhD. I worked on this field called network information theory which is you know information theory is about quantifying the fundamental limits of information propagation information communication inference all of this stuff and network information theory is when you have a network of nodes what are the limits of like information propagation there and I would say absolutely like the set of work we did on blockchains is built on

that less so from genomics but you know what happens when you kind of like move fields like three four times in your life is you pick up a set of like meta heruristics which are really really really important. For example, one of the things I learned from genomics is I used to work on DNA sequencing and this question that I was asked at the Scaltech interview which was that hey what happens if you somebody invents some new technology one of the huristics I picked up from there is you never work on a problem for a time scale larger than it's stable for right so then the problem is stable the then the problem is stable for like if if some technology is changing every two years trying to work on a theoretical model that's

that'll last 30 years is actually like a that bad. So you want to make sure that your the time scale of your problem formulation and the the stability of that particular problem needs to be matched. And I find lots of places where it's you know academics work on problems which are extraordinarily mismatched. So this these kinds of like metahistics inform taste in like how you how you figure out like what sets of things to work on. But most directly it was tools from network information theory that actually played a role. For example, this question, suppose you have nodes in a distributed network and you want to arrive at consensus and each of them have some bandwidth, you know, C. What is the maximum rate at which you can

arrive at cons consensus with a very high degree of security? And if you look at all blockchains today, they're all built on full replication, right? Like even Salana, which is the fastest blockchain today, is built on full replication. Every node downloads and receives all the data. And so if each node's bandwidth is C, the system can never run at more than C. Whereas the theoretical limits we inferred was actually if there are nodes, each of them have bandwidth C. You should be able to run the system at bandwidth N* C. That's like an insane thing. And especially in Ethereum, for example, N is the number of validators is 1 million. So you can actually scale these systems 1 millionx. I'm like, wow. I I used to work on wireless, right? I

mentioned before. and wireless if you could scale a system 5% you'll be the king of 5G right if you just make wireless better 5% you'll be the king of 5G here you have a system you can improve like a millionx nobody seemed to be doing it I'm like this can't be true we should do it so that that's an example of what kinds of fundamental principle or first principle thinking that you learn and you apply it and actually you can find and build systems that actually do it so in some ways even though you weren't thinking along the lines of impact when you were doing your PhD maybe indirectly like some of these ideas made their way into what you're doing now and we'll have we'll have great impact. Yeah, thank you for saying that. Hopefully it'll have great impact.

We're just getting started here. Um the the thing that does compound is knowledge always compounds, right? Especially if you build huristics that build on top of each other. You have a lattice work of mental models that like kind of you're successively refining over time. So absolutely you find that you can use what you know. Well, maybe outside of the sort of intellectual questions or the intellectual lessons uh from your time in academia, are there any lessons or anti-lesssons uh that you learned from that time for being a startup founder and running an organization, hiring people, building a product, uh running an engineering team, etc. Absolutely. So one of the things I used to notice among um academics

especially those who work on like theoretical and abstract stuff is I noticed this pattern which was a little bit weird which is that over time a large fraction of them kind of become bitter and I I used to see this and I'm like why is this happening? It's the bizarre thing and um then I used to then you know on probing and trying to find out what what is going on here and you mean like throughout the arc of an academic's career or even just within like a PhD some people some people learn fast I guess other people like you know take time but um you know over like a 10 20 year period uh es the more abstract and

disconnected with like real world things are the more likely this used to happen and I was wondering why is this because these guys are working on some of the coolest things in my view at that time and the thing I realized is the following the more abstract so imagine you have an idea for how to solve some global warming let's say by moving earth okay like fantasy land idea right now you you're looking at your friend and he has an idea how to move a buffalo jobs and you find like your friends actually like way more successful in getting these buffaloos to move around than you are able to move earth around and you know in one sense

it's obvious that it's easier to move a buffalo than it's more it is to move earth around but it feels unfair to the person thinking like this that hey I have such awesome ideas but I'm actually like not able to do anything with this and then I kind of like solve this puzzle and the solution is basically Like when you're working on more abstract, more powerful ideas, you need way more activation energy. To more Earth, you need like a million X activation energy. Then you move to move a buffalo, you need like 1x. You can just go and move it. And so when you when you need 1 millionx activation energy to actually do something, nobody's going to award you that kind of a leverage to actually go

and do it. So you have to earn it. And the only word way to earn it is by compounding your activation energy. So you have to go and move a buffalo and then like you know you get a 10x output and then you take the 10x and then you go and you can now move a car and then like you take the 100x and you can know move a building and then you can move a city and then maybe you can move earth. And it's not unfair that life works like this. It's actually really really good that life works like this. And the reason is, you know, you give somebody who has never moved a buffalo the ability to move earth and then maybe the whole thing explodes. It's a very very fragile thing to actually do that. It's very anti-fragile to give increasing compounding amounts of leverage to

somebody to actually do more and more expanded scope. So now, if you really had a 30-year vision that actually says, you know, X, let's say make humans multilanetary. Let's to to take Elon Musk's goal, you can't just go do that. You have to first build zip 2, which is just put hello pages on the internet, which is like nothing to do with make humans multilipu and then you have to build PayPal, which is let's build payment system for humanity. And then you have to go from PayPal to build Tesla and SpaceX. Like so this is an you know a arc of increasing like leverage to actually get to your 30-year goal and so I would say this is informs like a lot

of how we think about you can have like crazy land ideas about what what you want to do but you have to back project them into concrete things that are incentive compatible today otherwise you're not going to get even a single user and we learned this in our first startup so we built before this called trifecta So you know uh I started with coordination is the thesis for crypto and then I was looking around like you know blockchains have low throughput let's just go build a high throughput blockchain. I pitched to Ali among other people uh you know it's my fifth time pitching to Ali that he funded me so don't uh don't uh assume that every time you pitch you have to get it. The first time was you know trifecta and trifecta was this idea that hey let's just build like a super fast blockchain and there

three professors me my professor promote uh at Princeton and David Shay who's at Stanford we're building this thing and we just said let's build like a super fast blockchain and this is you know bad market of 2019 I didn't even know that something called a bull market or a bad market at that time so far out theory land and come in I'm like hey you know this blockchain runs at 100,000 transactions per again like why why is nobody funding this? And um what we did not realize is everything that you build has to have clear you know customers from all you know if especially things like a blockchain actually need many different types of customers to come

together and it's extraordinarily difficult to actually orchestrate it and so and then like that didn't go anywhere and then we went back to academics and did a lot of work and then after some time I was like okay you know it didn't work because we started with just technology not understanding the market incentives different participants in the market I said okay let's start from market first and we actually started an NFD marketplace very few people know this we started an NFD marketplace uh called Arctics and uh this was basically like I got my tenure at the University of Washington I'm like hey I'm I have to build this thing I have this 30-year goal but I have to get started somewhere

I was an adviser to Dapper Labs which was and NBA top at the time said, "Okay, this is cool. Like there's something going on here. Let me try to build something on this marketplace. I started building this stuff with the first time I'm actually touching applications, right? I was far away in like infrand building uh core, you know, blockchains before that. Uh, and you know, as I'm touching applications, I'm realizing, oh, how do you get an NFT prize oracle? And I don't know how to get like a good bridge to move NFTs around. I don't know how to do virtualization. I want to run like a new execution engine on top of Ethereum. I don't know each one. And I find look around and people are just like doing

band-aid solutions for each of these problems. And none of them like make sense. Okay, for each of these problems, the meta problem that I was finding is, oh, you need a trust network to run a bridge, you need a trust network to run an oracle, you need a trust network to run a run a virtualization. And then I was then experimenting with should I build an oracle? Should I build a bridge? And then I figured out, oh actually you should build the meta thing that solves all of these problems, which is build the trust network that actually solves this problem. Actually, as I was doing this NFD thing, one thing I realized is I really don't know how an NFD trader thinks. So, it's very hard for me to build an NFD marketplace, which actually

makes sense because if you don't know how your customer thinks, if you're not yourself a customer or if you're not very proximate to your customer, you're not going to be able to build a product that actually makes sense for them. But to build a trust network for somebody who who builds an oracle, I can immediately figure out how that works because I am the kind of person who will build an oracle. So that actually made sense that now I am I can be the customer of my own product. So actually building something on for that customer makes much much more sense because the gradients that you're taking on your product are much sharper. Mhm. When you're you're the customer of your own product. So let's just go build this like meta trust network. We didn't even

know how it's going to work. We called it layer already said anybody should be able to build arbitrary layers on top of a common trust network. So this trust network is our product came up with many many iterations of it before we settled on initially restaking and later this concept called uh interjector staking organ staking which I can talk about but the idea that you you have this common trust network and anybody should be able to build services on top. So that's how we actually landed up with that. That's an important lesson there because I feel like so innovation so often happens through the exploration of something that's very specific and very concrete first and then that leads to a set of insights that's adjacent that maybe then generalizes to something broader. Uh where you actually realize that the

specific thing you're solving is kind of very similar or has adjacencies to a much much bigger set of problems. And that then allows you to move up the the layer of abstra a layer of abstraction and build something that is much more powerful and that feels especially in computing I feel like that is like a very common pattern and the approach of just trying to start from an ivory tower and trying to start general or start from just like big level like highle ideas uh and build something that's all-encompassing or extremely impactful. It's just it's just it's much harder to go down that path than than to kind of be grounded and connected to the real problems um that are sort of more concrete in nature. Uh that's absolutely what we found because the the previous approach was this ivory tower approach

like to to build traffic as hey you know let's just start this protocol and this later approach was much more narrow and when you start with something narrow you may find something broader and you have this view of what the long-term horizon is but you can you have to take gradients of val from where you are well and now that you're running Ian layer as an organization how do you balance uh research open-ended research on one hand uh with product engineering go to market uh on the other hand I think each of us uh come up have natural biases and we you know which are also natural strengths but want to make when you're building a company you have to make sure

that you're not just building based on your natural biases. So one of the things we did when we started trifecta was all everybody who we hired and brought into the initial team were all like PhDs thinking the same way. So this was the error that we made in trifecta. Then I realized this when starting layer and we said basically no researchers. So we we started a company like we are researchers. I'm a researcher myself and some of the early like you know founding developers were all researchers but there's no job called researcher at labs uh till very recently. So we said because that's our failure mode like our

failure mode would have been just go off into like crazy lands invent new crazy stuff but which have no tethering to reality. So we we wanted to retune it, bias it in the opposite direction which is what can we build which actually you know matters today and I still always used to have this longer term thinking but I used to basically shield the team from it because you know if you just like drag everybody into like things that are long-term which don't have any material way of manifesting today it's not useful. So we built a team which is you know on this theory which is you know now common called the zone of genius theory which is that you want to

bring different people in their zones of genius not build a homogeneous team. You know some people have this idea that like as a founder your job is to build and scale yourself. I have an exactly opposite theory which is that you want to find people who are in their zone of genius potentially very distinct from your own. But the hard part when you do that is first calibrating. You know, if I'm going to try bringing a sales or a marketing guy, how do I know what's zone of genius there? It's actually really hard. So, it takes time and like sharp models to actually calibrate that. You take help of you know your investors, other people to actually help calibrate you there. But still, it's really important that you develop the calibration. That's number one. Number

two is how do you coordinate people who are all in very different zones? Now you get the best guy in legal, best guy in like marketing, best guy in like you know writing code. In in our own team for example we have like crypton natives, we have like you know people who are like veterans in the cloud. We have people who are successfully done like previous startups, academic researcher types. We have all of these types and how do you make build coherence into this? So that those were the problems I had to solve to actually make this work. Yeah, that's beautiful. What do you see as the as the endgame for blockchain architectures say 10 years from now? We could also even do the 30 year the 30-year version and then of course like where does layer fit in

that in that vision regarding blockchain like the first thing I heard was blockchains are truth machines and I'm I didn't like connect at all. Truth is a very high bar right like who has access to truth? was kind of crazy to say something's a truth machine. And so I was thinking about it and I think the sharpest formulation I have is blockchains are commitment engines. They let you make and keep commitments. That's all that I I think pretty much all what we're doing in blockchain is just make and keep commitments. You know, I started this like you know long-term thesis with humanity's coordination engine. If you think about what is coordination, coordination is communication plus commitment, right?

You know, you put 30 people together and we should be able to communicate and work together, but we should all be able to commit to a common cause to how we share the outputs of the work that we are engaged in and so on. So coordination is just communication plus commitment and internet was our communication superhighway and crypto specifically is our commitment superighway. So this is our specific thesis of blockchains. So if crypto is commitment super highways what what does it mean? So when you think about commitments you should be able to make commitments and then keep commitments.

So you think of like how do you enforce commitments as a kind of like a very fundamental question. And you think about it like you know in in the large arc of history in fact you can even say like civilization is just a system of commitments we make to each other right so the constitution is a commitment law is a commitment like you know when you marry somebody it's a commitment like you know civilization is just a system of commitments we make to each other and when you think about like mechanisms for commitment enforcement you see like you can have an authority like a king enforce commitments you can have a committee like the CCP enforce commitments ments. You can have a majority like a pure democracy like where a majority can vote and and decide what happens. That's a mechanism for

enforcing commitments. But even in a pure democracy, the problem that you have is the tyranny of majority. Like a majority can dictate terms on the minority. And if you look at it, all of us are a minority in one way or the other. You can always slice and dice in a way that everybody anybody's a minority and anybody's in the majority. So there's a tyranny of majority. And so I think of the most robust systems are self- enforcing commitment systems. Systems of commitment that don't require an authority that don't require a committee that don't require a majority to enforce but they enforce themselves. That's really what crypto is self- enforcing commitment systems. And so that's what we think about at layer is you know we we have this long-term view like humanity coordination engine. What

we can actually like manifest is how do you maximize the volume of commitments made and kept? How do you maximize the variety of commitments made and kept and how do you maximize the verifiability of commitments made and kept? So these are the three dimensions we think about when you think about volume of commitments you think of throughput right. So when you say humanities coordination engine how many bits of coordination can you get bits per second of coordination that you get. So that's our like northstar for volume. So we think of coordination engine coordination bandwidth. What is humanity's coordination bandwidth?

Actually if you think about it like even 30 years back how many bits of information can humanity agree per year right and say oh this guy's the president of that country like something like that is all you can agree it's a few bits per year of coordination bandwidth we had as humanity there's also the bandwidth and I think also latency highly limited back like 30 50 years ago or so uh just like for example um one very good good uh example of this might be just like the cost of setting up a corporation and how maybe that has gotten marginally faster with with some technology and with the internet and email and things like that and you communicate with your lawyers more efficiently and then you still have to pay $50,000 and you set up an LLC and then that's a structure that

can enable cooperation uh between a group of people. But if you could actually do that with a smart contract, not only is it the number of bits that that of coordination that you're able to effect through the blockchain, but it's also the fact that you can do that in a split sec like you know seconds in seconds and at really low cost. you know, it's not $50,000, it's $5. I'll actually speak to the setting of the company in a bit. The volume, the rate at which you can actually communicate. So, that's something that we specifically optimize. We're building a product called IDA, which is basically this data availability system. When you think of bandwidth, what does it mean?

Like how many bytes per second of information can you write into a common ledger that's what like volume of commitments is. And so, we recently did this. So, Ida is on mainet and you can write data to it. So it writes data to this decentralized network and then commits these on on Ethereum and the throughput that we have today is enough to fit all blockchains. So we we had this experiment called all chains run on IDA. So you can take crawl Salana you can crawl like every chain in the world that we could get our hands on and then you just write them in real time into IDA. So this is an example of taking this big picture like large you know commitment systems and then maximize volume of commitments. So it's not just

volume of commitments that you are able to make you also actually want people to be making these commitments. So that's that's something that we'll be working on next. So the next one is variety like you know layer itself is actually a mechanism for creating a variety of new commitments that you can make. What stakers are doing in layer is they're staking and then saying that hey I'm going to run your service correctly. Now that service itself can provide a service level commitment which is hey you have this AI inference man when you call this AI inference service you will get the right answer because of all the cryptoconomics underneath it and you will anybody can call it permissionlessly and censorship resistant. So you can think of like each of the services having a service level

commitment and so variety of services can be built you know on a common substrate. So you can actually get many service level commitments that that can be built. Now applications actually the way we think of applications is they concatenate service level commitments. So I use a DA layer to write data. I use an AI inference to kind of get verifiable inference. I use a ZK layer to get like verifiable proofs. I concatenate all of these and then build an application and the application layer commitments what really matters to the user. If you think of like all of like the enduser journey of crypto today, it's basically apps. I think of it as apps with commitments. That's our thesis for all of crypto is as an end user I'm coming in. Why should I buy and hold

Bitcoin? Because Bitcoin makes me two commitments that I don't get anywhere else. You own your Bitcoin. There are only 21 million Bitcoin. just two commitments. Look at there's endless number of like normie podcasts on Bitcoin, right? They all talk about only these two things. You own your Bitcoin, there's 21 million Bitcoin. And these two commitments have trillion dollars of value. Isn't it crazy? So now think of each application that you're using, whether it's Facebook or Uber, you know, if you're building a new Uber, you should be able to make a commitment that says I can never take more than 5% takeway. This should be a perpetually enforcable commitment. It's not just that like I'm taking 5% today and then

tomorrow I can just like rug you by taking 30% after I get all the people to come together. I should be able to make perpetual credible commitments. Commitments that I as a creator of an application I cannot violate. And the way we think about it is each application by concatenating a set of services that make different kinds of commitments actually can make these commitments. Just like in the internet you can make you can build an application by concatenating a set of SAS services in on layer you should be able to concatenate a set of like services each of which make service level commitments to make app level commitments and the way we envision it is like you go to an app and then you'll see these claims in English it'll say can't take more than 5% thing ever you know you go to a Facebook a new crypto

Facebook and then it'll say hey I can't change the privacy policy without a majority of users voting on it many many commitments that actually matter to us as you know as more of our life moves digital it's absolutely imperative that we have systems like this so that's that's how we view like the kind of long-term thesis of coordination leading to commitments and then we ourselves working on maximizing the volume of commitments the variety of commitments and the verifiability verifiability is like can you verify these commitments on chain one of the things we did with iigen the the token is actually make any commitment that is human verifiable to transport that onchain. So that's a whole mechanic I won't get into

but that's really how we think about it. So you want to build the maximum volume variety and verifiability of commitment. Awesome. All right. Well let's open it up to questions. So I had a question on marketing. So I think for better or worse narratives drive crypto to a large extent and many projects want to you know set a narrative or form a narrative that other people latch on to and very few projects actually succeed and I layer of course has done a very good job of getting the narrative of restaking and everything else to catch on as well.

From your point of view, what has worked in terms of getting people to catch on to this narrative and what advice would you give to any of us who are thinking of having a narrative catch on as well? Amazing. Yeah, I think this is a really important question. So going back to my like this um millionx activation energy problem which is essentially the way we think about it is like you have these um when you this picture is x-axis is landscape of solutions and y-axis is how fit they are. When you think about it, what happens is the solution space is very non-convex, which means basically there's a lot of local maxima. And what happens is you usually get stuck, not

you as in a person, as a community, we all get stuck in local maxima because there's no incentive for like moving out of the local maxima. And the only way I've figured out the only way to move out of or one of the big powerful ways to move out of local maxima is actually narrative and speculation. Mhm. Because what speculation is is basically trying to get people understand a vivid picture of the future so you can bring it forward today and then people can impute some value into it and then you can borrow from the future to distribute to the people who have to climb down the value with you. Right? Right? Like this is what venture capital is as an example, right? Speculate on okay, you know, this can be a big thing and then

borrow it today and then you use that to pay your employees and then you can move together. But you can actually do this much much more powerfully in crypto because you can bring all your all the sides of your community. For example, I can is a four-sided marketplace. The first time I explained this to Chris who later became our COO, he's like you must be insane to try to build a four-sided marketplace. We have stakers, we have operators, we have AS developers, and then we have AS consumers. It's impossible, right? So you bootstrap this one side at a time. So we went sters first, then we went to operators, then we went to AS developers. We haven't even started with consumers. So only three that's why I say it's not yet done. It's not even started because we haven't even closed the circuit here. So

that's how hard it is. Like these things take many many years. And the longer the time scale of like actualization, the more you have to be able to bring the future to today. So yes, narrative is important. And it's not unfortunate, it's actually like fortunate that there is something called narrative to actually bring the future to the present so that you can climb out of local maxima as a society, right? Make humans multiplanetary is a narrative. And it's very powerful. It motivates people. It lets you go on the journey with Elon on this like long-term vision. Okay. Okay. So, how do you do it? I think there's lots of huristics, lots of things I had to kind of figure out. I'm so far away from this world. Like, when I enter, I see I need to do it, but I'm so far away. So, I have to kind of figure out

all the huristics for how to actually make this work. Never been on Twitter before, none of this stuff. So, it's it's like a new thing. But this is one of the things I learned is when you have when you commit to like a 30-year project, you have to do what it takes. You can't do what you're good at. You have to do what you're good at and what it takes. So, you know, recruit people, do whatever, make it work. Okay. So, how do we do it? One thing we figured out is if you have a vision, one of the best ways to like recruit more people into the vision is to find people who already think similarly, right? If coordination like acceleration is our thesis, you first find who already has that thesis.

I found Ethereum already has this thesis embedded. there is like a latent like you know collective unconscious which is coordination is very important as part of the Ethereum psyche. So and I'm like wow if somebody's already has that I should not replicate their effort and go and fight with them I should actually ride on top of it and help them achieve their own coordination goals and build on top of it. So this is very important. So first decoding the kind of like so when you think about narrative narrative is basically like belongs to the collective unconscious like you're speaking to what people already like want and and believe in like otherwise you're not going to find like resonance and usually what you want to do is you

want to speak to a subset of people who are already like motivated and aligned on that direction right say make humans multilanetary there'll be people say like what's wrong with earth you don't want those people you only want those crazies who want to go with you to like mocks and so that's the same thing like you want to speak to the people who actually like have the same belief structure and so but for that you have to articulate your belief structure very clearly so you say that oh coordination open innovation like permissionlessness credible neutrality all the things that you believe in I found like the substructure of Ethereum that actually believes in the same thing so we can align together on that goal and when you're aligning on a on a common goal it's important to like not compete with

the same people who are building the same thing. You can't align on a common goal and then just say, "Hey, you're doing this. I'm going to do the same thing but a little bit different." It makes no sense. You actually figure out how to empower them to get to their goals better. So we come in, we look at Ethereum. Ethereum has the roll-up era and then but roll-ups are expensive. So give some ammunition like hey you know DA like you can actually solve it and actually get farther by having a data availability system which is higher throughput and so on. So make things incrementally compatible. So it's not just narrative it's the game theory of how you structure your system needs to align with the structure of who you're trying to recruit to actually go on this journey with you because you know it's hard to find out people who will stick

with you for a 30-year journey. So that's all I'll say. There's lots of huristics that I can talk about offline. Yeah, I I would say that I mean I think maybe more than 50% of uh narrative and marketing is actually upstream. It's it's coming up with that crisp like vision of the future that is extremely long-term that actually stands for what you're working on that can serve as like the thing that everyone can get aligned on. And then I think the other part is to articulate a master plan which I think is what Elon Musk does so well. articulate a master plan that is sort of very long term for how you actually get there. And the more clearly you articulate both of those things, then the easier it is to actually then do the tactical sort of more blocking and tackling that actually goes into telling

a story and marketing what it is that you're doing. And I think you you've done a really good job at both of those things, articulating the 30-year long-term vision, but then also talking about how I layer today begins to actually make progress toward that end. I I think for example the first thing that we did is I I was called the re Ethereum restaking collective. It's not called the coordination engine for humanity which nobody will be like what the hell is this guy talking about right? It's reststaking. It's one side of the four-sided marketplace because that's all that like you can actually send through the matic sphere restaking just one word right and so after you build that you can say oh as a as a new category and we had to work very hard to

create as a category and now like people building on Salana elsewhere they're saying oh as I'm very happy because you know it's really hard to create categories and then we came up with this other crazy thing called inter subjectivity like basically the information bandwidth that you can communicate like it's kind of like limited to like a few bits per year the broader that you want to reach the fewer you have to distill the message into fewer bits. So one thing that it seems like you you're one way you're different from other founders uh is that I I think you have a very clear and you have your your vision is is not the technical. So it seems like there are three levels of

vision right there is one technical vision restaking and then higher level vision is you're you're you're creating a difference in in developers lives by making it easy for people to build uh chains and there is an even higher uh beyond life vision which is open innovation coordination engine. What how do you how do you build that muscle in your brain to think that way? I I think I'm stuck at the level two where uh make ZK the substrate of uh computation um default substrate of computation. That seems like a level two, but I'm I'm struggling to get to level three. How do

you build that muscle uh to are there any actionable tips you can give us to get to level level three? That's a great really great question. I think the the things that I would say there is that's why I told you the Caltech story the that question of suppose I give you infinite time infinite resources uh infinite uh talent what would you do is I think really really like you know it opens up like your brain to like say what is it that you really want to like get to so suppose you you had this magic box what would you do so and I think the other thing is it goes deep into like what you

believe in like because you know I'm I'm a very like uh nonzero sum type thinker if I were trying to build something which is like you know let's compete with somebody else I would not be very successful at it because it doesn't fit is like the founder market fit like to build a coordination subset you need somebody who thinks like me because that's that's the nature like I think about everything in non-zero sum like I don't want to compete with anybody if you're doing X if you're running I'll figure out a fly or something else like you know I don't I don't want to do the same thing that somebody else for example when we had like reststaking we could have said everybody thought we're going to build another liquid staking token or something like that said hey you know these guys already are good at

it like why should I do it and we only do the things that we think nobody else is doing well and so that's that's part of like our sub you know framework of thinking and find out what is it that like motivates you each of us uniquely and then figure out from there what is it that speaks to you as a person not as a technologist and that is what will get you at level three. Yeah. Well, I think uh just from the investment side one um kind of observation is that there are two approaches for coming up with a vision and uh I think maybe it's it's a good model to think of it in three layers and I think you could approach it either top down or bottom up. You could start with like a cool technology and

then try to sort of extrapolate what the technology might be good for and then then sort of essentially uh rise to level two and then continue doing that to get to level one. But then you can also start top down and think about like what is the very large problem that you want to go after and solve and then work your way down and figure out what technologies you'd have to build to solve it. And I think the best founders do both where you kind of have to approach the problem from both ends. You have to lock yourself in a room. Think about this 30-year question with no constraints. Essentially, you have all the resources in the world, all the time in the world, all of the talent. What would you do? But then you also have to lock yourself in a room again and then

do it differently. You have to then ask yourself, what are the resources that you actually do have? What what is the talent that you do have access to? What are your interests? What are the what are the things that you're uniquely well suited? And then finding kind of the intersection between those two and being I think that the failure modes are are often ones where where founders approach it from only one direction. Whereas like if it's just top down, it's kind of pie in the sky. It's not grounded. It's too abstract. It doesn't land. It's very ivory tower. If it's too bottoms bottoms up, then maybe there isn't like a big problem that the technology is actually ultimately going to solve. It's more just like a sort of a uh like a curiosity as opposed to something that ends up having impact. So I think maybe

um yeah like I think that thinking about it in both ways and exercising that muscle. I think this question from Caltech is incredible. uh really good question to ask oneself. Yeah, I I I I I love what Ali said. I think it's the two sides like you have to figure out where the kind of like far away hill is, but you have to figure out how to take gradients from here to there. I think the academic failure mode is not finding the gradient. Just stuck in some maxima and you're just sitting there and you're wondering why the hell nobody's giving you leverage to go like move earth and the uh like move the buffalo approach is the problem is you don't know what once you move the buffalo somebody else moves the buffalo faster than you you don't know what to do. So this is actually

like a really important thing that I learned from the Caltech question is when if somebody solves restaking somebody solves like decentralized cloud better than me I know what to do on top of it. That's why I say like we welcome Ethereum to like enshrine a people think I'm just joking but I'm not. If you you're going to build that because my whole goal is to build humanities coordination engine it's not done unless you can you can run education on chain unless you can run healthcare on chain unless AJ is tethered on chain we're not done. We're nowhere near done. I know what the 10 next things to do. Somebody does the thing I'm doing today. And I think that that is the alpha that it gives you. But if you can't do anything today with your long-term vision, then it's no good either. One question that I

wanted to ask uh was posted you five times. Um I was so curious as to like what you saw from the investor side chain with Nathan each time he came um you know for those five times that he pitched you. Um so that's my question to you and uh the other thing that you said Shirram that I was like really surprised by was that you often shield a lot of the team from this sort of grand vision that you have and that's surprising to me because um I am perennally on crypto Twitter and some of the some of the messiahs that exist in crypto today that have been doing a fantastic job of um evangelizing what I layer can do has been the team that

you've got and so that that's surprising to me And so wondering like how you balance shielding versus what you're willing to reveal and Ali to you like how do you assess that level of change cuz you made a sizable commitment to uh the layer vision. So, so the four times I pitched are uh first time for trifecta, the second time seed round for uh layer, third time was series A for layer and then series B for that's four not five. Uh maybe I'll just tell an anecdote before handing it to Ali is uh the time when I pitched for seed for labs. So one of the things I decided when I built both like the NFT marketplace and I labs is I'm building what I think is interesting and what I

think can get customers not what VCs want. So I have very kind of like clear thesis because there is a overfitting to VC narrative which I think is very very risky. So I was like okay you know and I put and the other thing I did which is crazy is I put all my money in. I said okay you know nobody needs to fund this. I'm going to fund it myself because if if I if I'm not willing to put my money in, there's no like good reason for me to go ask somebody else for that money. And so I was doing that and I was at that time know the idea of back layer first. We we were like virtualization was actually our first idea. You can run a different VM on a different chain.

This is 2021 and we are like okay what what VM are you going to run on what chain? Maybe you can run EVM on other chains. EVM seems to be like picking up. So why not run it on other chains? you know the chain I looked around and some people were trying to do it on Salana so we were like let's do it on Cardano let's build is from so I know her from there um uh the why uh you know because it has so much of liquidity like $80 billion at that time and there's no functioning smart contract systems even today so I was like okay let's go build this and I pitched to Ali and Guy uh and they were actually like call they took

the call from outside I think the Salana conference and they came and they're like I said all this is interesting why the hell are you on so and uh you know but I said we don't go by VC narrative but I was we take inputs very seriously so we thought a lot about why is he saying it like what does it mean that like why you're building on Cardano. And one of the things we one of the models that I built was when you're um you know people say this a lot when you're building a startup you need to have incessent focus and I didn't understand why people say it like in and usually if I don't understand it I don't

pay much heat to it because it doesn't fit into my model and the the thing I learned about focus was if you think of a startup a startup is basically an exponential game. What's an exponential game? You want to translate linear amount of work into exponential impact, right? And exponential games are extremely rare, right? Who's going to take a linear amount of work and translate into exponential output? It's like magic, right? And it's very rare. Exponential games are very rare. So, if you think you have three ideas of exponential games, you probably have none. That's the first one. Second one is maybe you're a genius and you have three. Then how should you allocate energy? How you

should allocate energy is find the one which has the highest exponent and allocate all your energy to it. It's like you know exponential is like 2 to the x right like x is the energy. Then instead of having 2x2 + 2 x2 you should just have 2 to the x just bigger than 2x2x2 plus 2 x2. So this insight basically explained to me why startups can absolutely not afford to be split. And initially we were split. We were saying, "Oh, we're going to build this on Cardano, but then we're going to also build something on Salana and something on Ethereum and this and that." And it was super clear to me like after this pitch came back and we were like, "Hey, why the hell does A6Z pass on us?" And

we had this meeting and we were like uh discussing it and it seems like if you're building on Ethereum, they would say yes. And we don't know if that's true. I'm going to listen to Ali first time about it. But let's figure out like what it would take for us to build on Ethereum. This became a driving question for us because otherwise you're splitting the exponential. You're having a product here, a product there, a product there. And as a startup, you absolutely can't afford that. So this is one of the things I say to anybody who's building anything. People say, "Oh, I'm going to build on three Vstaking platforms, four chains." Like get your thing together on one thing. choose something like whichever you think has the highest chance of success and then like kill it before you expand it. It's interesting to hear that from from your

perspective. Uh it was really good to kind of do the retrospective. Uh well I would say that that now in retrospect it's pretty clear we we kind of made a mistake earlier on invested in every one of those rounds. Um I I think that I mean it comes it comes back to some of what we were talking before u with respect to approaching the problem from both sides. I think that we have historically had we've had a strong bias um to not invest in in professors in particular and I think that that's and actually probably a a misguided we were too biased in that direction. I think there's some truth to the the fact there's a risk that someone who comes from academia might uh might sort of steer towards approaching the

problem top down too much sort of thinking thinking too well either direction. And it could be like pie in the sky and ivory tower or it could be it's just a cool technology and not really being able to connect the dots between those two which I think is what informed our bias. And I think also back in like the 2017 um market like in that in that cycle in the peak of that cycle we made many investments uh in in professors who were building crypto professor coins professor coins infamously named uh and and I think we I think that it is often the case that that professors will ultimately not go the full length to to want to uh build out the company and to and to do the hard things that a founder must do to

really have impact and they'll be too focused on just the highle ideas and I I do think that we had a blind spot as a result of that experience. Uh I think that as we like it's it's really always great by the way uh to meet founders multiple times uh because that's like that's often the way that we invest. We'd like to have a a relationship with a founder over a long long period of time. So often times we do pass in the initial round because that's just the initial is the initial way for us to get connected and and we because we're a multi-stage fund there are many opportunities for us to work together. Uh and ideally we just have like a longstanding relationship at the time that we invest because this way there's more trust. This way we we just know the person better and it's not just about

the pitch. It's more about the the person the relationship etc. Um and then also I think that gives us a view not just of the snapshot of what the of what the progress of the company is at one moment in time but also of the velocity right like you have like now can you can uh interpolate between the various different data points that you have over time and then extrapolate like where is that going into the future. I think one of the things that was remarkable about every time that we met FURM is just that the velocity like the the gradient uh going from like wanting to build on Cardano to to like really refining exactly what I layer is and to being very connected and grounded uh with not just the critical ideas that will make

layer work like the academic ideas but also the market uh kind of the surrounding players the ecosystem the strategy the tactics all of those things that that um that ultimately will make ien layer successful. Uh we really saw that evolve very very quickly every time that we that we met uh with Shirum. I think also at the very the very last meeting when we when we actually invested it be it had become very very clear that Trium is a founder right like he he is also a professor and academic and and he was yeah I left the university right exactly and and but you still have like the DNA of a researcher but you were also a founder and you're and you're 100% committed to um um making the company successful and that became very clear which I think was the critical thing that made us really want

to work to work together. Uh I think another factor honestly was was the fact that some of the meetings that we had were were purely remote. Uh they happened via Zoom. And I've realized it's just so critical to meet people in person to really kind of again back to that relationship component. It's much much harder to to build that if it's just a remote a remote meeting. Uh in the final time like I think we met in person. We we had more more time to really kind of get to know each other which I think was also a very important important factor. Well, one thing I really loved about uh Ali and A16Z uh in the entire period is it's not like hey we're not we've not invested in you and therefore like we are not going to be helpful. We try to be helpful all the

time and you know if it works we'll work together otherwise it's okay like and I have the same attitude like some founders take it very personally and they're like oh you know these guys which have passed and I'm going to show them and so one of our core principle my core principle is win-win or no deal I think this is like Steven Kov's from seven habits of highly effective people and no deal is really really important to make things win-win because you work together when there is a win-win And no deal is the default. No deal is the default. Like most times you go to pitch to a customer, you go to pitch to an investor, go to pitch to anybody, you should expect a no deal because alignment, think of each of you as like vectors in multiple dimensions.

Alignment is rare, extraordinarily rare. So you know, you should not be assuming alignment. You should actually like but I I found ACZ to be like very very poster. So it fit my personality. I've not had the same experience necessarily with other funds. So, shout out to uh like that spirit. Even the the way you get the rejection is so nice. It's like hey ABCD FG but H JKL Mop. That's good. I'm very fascinated what you said like uh blockchain is a self- enforcing commitment but democracy is a majority but as I like I'm just trying to understand because I think uh blockchain also is underlying has a has

like a majority assumption like it can tolerate so I always think of blockchain as like a more efficient democracy machine while the reality democracy is like less efficient I'm not sure yeah let me comment on this I I I can talk about this for hours Okay. So we talk about the four levels of commitment enforcement uh authority committee majority and self- enforcing. And if you think of like before crypto what are self- enforcing commitment systems um the biggest one is gold. So think of the problem as like value transfer. I want to send somebody value.

And inside a kingdom a king can come and say hey here is your ledger. Here's how much you own. Here's how much you own. But when you want to transact across a trust boundary like I'm my own king and you're your king now you want to ex like exchange value how do you do it like this is a value exchange problem in society and the emergent solution to this was gold and gold is self- enforcing once we all agree that gold is kind of valuable now no king no committee no majority even can say like gold is not valuable as long as you and I like accept to trade in gold you can trade in And it's selfverifiable. Once you get gold, you can see it's gold.

It's really cool. So, gold was the first self-inforcing coordination system. And after that, I think the big but it it is only self- enforcing commitments for value transfer like I can send you value. In fact, you know, in Bitcoin, you have this is this culture of hodel, right? You keep holding bitcoin. I I wrote this recent article called godl. Gold has a similar longstanding culture. For example, in Indian culture, you never sell gold. It's like blasphemy to sell gold because people realize over like, you know, generations that if you had gold, a new ruler can't come and like just take it away. Like they can take your land away, but you can just you have a piece of gold, you can just

take it and run away wherever you want. You can pass it. So, it's one of those very few things which is self- enforcing across long like time horizons. The other one you know you pointed out democracy pure democracy is majoritarian right it's dependent on on the majority the self-inforcing nature of democracy comes from the constitution the constitution is self- enforcing constitution gives you each individual uniquely the the right to life liberty property and the pursuit of happiness and this is like one is to one and so the majority like a senate or a house voting on a bill is still regularized by what the constitution permits you to do.

So the constitution is the self. Now coming to like blockchains, aren't blockchains majority honest or majority coordination services? And this used to be my model actually like we did so much of academic research on majority honest assumptions and and the the day I realized that actually this is not true is you know back in July 2021 when I was working in in practice and stumbled upon one of Vitalik's article from 2014 like you know people say like you know anything you talk about today is already there in bitcoin talk it's kind of like that you can find something in ethereum research usually vital has written something about it so back back in 2020 2014 he had actually written about you know this concept called subjective

occracy which is essentially the idea is even if a majority of validators in a blockchain do something wrong you don't have to accept that block because you can verify yourself that that block is wrong. So as long as you can make things self-verifiable you will fork around majority false. So the blockchain is not correct based on a majority of nodes that would not be a great system. And so a lot of people a lot of blockchains even like major blockchains build on this assumption that blockchains are built on majority honest trust. It's not a really good system. It's Vitalic who decoded like this kind of like principle of forkability. You know even Bitcoiners have this thing forkability as the superpower of blockchains. And so that

is the self- enforcing nature of blockchains forkability because if something's wrong you can fork and everybody knows what. So you fork on things which are pre-aggreed and self-evident. So that's our thesis of forking. The IGEN token itself is built around leveraging the superpower of forking and then giving it to all applications. So there's a whole whole thing to talk about there. Amazing. Well, thank you so much. Thank you so much. [Music]