EigenLayer founder: Integration of Ethereum security expansion and AI

Video source: "S7 Ep. 4 - Coordination Acceleration W/ Sreeram Kannan (EigenLayer)"

Compiled by Cyrus

Proofread by: Elsa

This article aims to help readers quickly understand the wonderful conversation between Kevin and Sreeram, the founder of EigenLayer, covering the concept of heavy pledge, the use of blockchain in capital allocation and commitment mechanisms, and the future integration of AI and encryption technology. By analyzing how to expand the security and trust of Ethereum to more decentralized projects, the article shows the huge potential of blockchain in collaboration and innovation, and also outlines a more fair and efficient distributed future blueprint for readers.

01 Translator’s Preface

In our daily lives, we often encounter situations where we can only use specific payment methods. Have you ever faced such a dilemma? This restriction prevents some people from conducting transactions conveniently. Some large institutions or platforms control a lot of resources, and it is difficult for ordinary individuals to participate in them fairly. This not only limits innovation, but also leads to uneven distribution.

In a society where we live that places great emphasis on individual freedom and resource allocation, is it possible to organize and allocate resources in a more efficient and fairer way to solve the current coordination problems?

The concept of "heavy pledge" proposed by EigenLayer is an attempt to solve these problems through technical means. Simply put, it allows Ethereum pledgers to reuse the pledged ETH for other projects, thereby providing security for more decentralized applications. This not only improves the efficiency of capital use, but also creates fairer opportunities for new blockchain projects. Through an open mechanism, EigenLayer allows more projects and users to freely choose how to participate, and no longer be restricted by a single platform or rules. This flexibility not only lowers the threshold, but also attracts more innovators to join.

The significance behind this goes far beyond that. It brings more possibilities to the entire blockchain ecosystem, giving more people the opportunity to innovate in a more open and decentralized environment and jointly promote social and technological progress.

02 Podcast Introduction

Kevin: Welcome to the Greenpill.network podcast! If you’re new here, we’re a coordinated network of thousands of hackers, dreamers, and doers working to build digital systems that use cryptography to positively impact the world. This podcast is spotlighting the people working on that endeavor. We’ll be releasing a season every two months, with 5-10 episodes per season.

The theme of this season is How to DAO. How to start a DAO? How to participate in a DAO? How to interact in a post-DAO world? Puncar and I wrote a book called How to DAO. In this season, we will talk to many leaders in the DAO era about how to DAO and how they see the development of the DAO ecosystem.

If you want to learn more about the podcast, you can visit our website greenpill.network, where you can download the book Greenpilled for free, join our Discord community, and become a member of the Greenpill.network chapter in your area.

Today on the show we have Sreeram from EigenLayer, and we're going to be talking about coordinated accelerationism.

Sreeram is the founder of EigenLayer and was previously a researcher and professor at the Department of Electrical and Computer Engineering at the University of Washington. His work covers information theory, machine learning, blockchain systems, and computational biology.

Sreeram is the co-founder of EigenLayer, a project focused on decentralized trust and economic security in blockchain systems. He has made a huge contribution to the development of the Ethereum ecosystem, and EigenLayer is poised to redefine how projects interact with decentralized trust in the next era in the ecosystem. What does this mean? We will discuss this topic with Sreeram in this episode.

We talk about what EigenLayer is and what opportunities it offers to builders around the world. Then we broaden the scope and discuss Sreeram's argument about Coordination Acceleration. Basically, you need to listen to this podcast to find out. I think this is a very good episode and I think you will enjoy this conversation with Sreeram.

Kevin: Hi, Sreeram, how are you doing?

Sreeram: Hello Kevin. Great to be here on the Green Pill Podcast.

Kevin: It’s great to have you on the Greenpill Podcast, and maybe we can start with what you do. So, what is EigenLayer?

03 EigenLayer Introduction

Sreeram: Hi, I’m Sreeram, and I started working on EigenLayer about three and a half years ago, and I was a professor at the University of Washington in Seattle. EigenLayer is a mechanism for providing decentralized trust to any new network. So if you look at what is crypto and what is non-crypto, and try to distinguish between the two. If you add decentralized verifiable trust to something, you can say that’s crypto; if you remove decentralized trust from something, you would say that’s not crypto. So you can think of decentralized trust as the raw material of the entire crypto economy. If you think about blockchains like Ethereum, which is what we’re building on, what they do is they take that source of decentralized trust and then they transmit and provision it in a specific way, which is to say, you build blocks.

So there's this concept of blockspace, where the nodes come to consensus. They stake, they provide this security, then they come to consensus, and then later they run an execution environment on which anyone can build and deploy new applications. It's a very specific form of providing trust. You can think of this decentralized trust as "crude oil," which is refined all the way to create this refined product called "blockspace," which is what other people consume.

The way we understand it is that decentralized trust, in its most primitive form, is actually very useful for people to build new types of networks that are different from the constraints imposed by any particular blockchain. This is the concept of EigenLayer, "Eigen" means "your own" in German, which means "your own layer". Anyone can build their own layer on top of this common source of decentralized trust. This is our core philosophy.

Kevin: When I was preparing for this show, I found two examples in my research that are similar to EigenLayer. One is Vitalik's realization that people should be able to launch smart contracts without having to start a whole network of miners, which is Ethereum, a general-purpose smart contract platform. And then another analogy, which I think you and Bankless mentioned in the podcast, is the analogy of various applications on AWS to decentralized storage, decentralized computing, decentralized messaging, etc., which can be seen as general-purpose cloud computing services, and AVS (Autonomous Verifiable Services) on EigenLayer is a bit like an application built on top of EigenLayer. So, transforming Ethereum's trust, which was originally just for block space, from a single use to a general use, allowing any developer to access billions of dollars of capital on Ethereum, is exactly what EigenLayer does. Am I correct in saying that?

Sreeram: That's a great point you made, any blockchain from a cryptographic perspective consists of different layers. The base layer is security, either staking or mining. Then you have consensus, where the nodes reach consensus in a certain way. On top of that is execution, where some execution environment runs on top of it, and applications run on top of those execution environments. You can think of Bitcoin as basically having all four layers vertically integrated. It's a specific form of security with a specific form of consensus, a specific form of execution environment, and a specific application - sending and receiving bitcoins. What Ethereum did was, like you said, break the top layer and come up with a fixed security, consensus, and execution mechanism, but anyone can build applications on top of it. That's a classic example of the Ethereum paradigm.

Then, as Ethereum entered the rollup-centric era, innovation became more open. I see it as a gradual deepening of the open innovation space. In this model, people can come in and innovate without having to build trust themselves. Therefore, any application developer on top of Ethereum can deploy applications directly on this smart contract platform. Of course, other L1s have followed in the direction of Ethereum, but Ethereum actually took the next step, which is amazing that a platform can evolve in this way.

The next step forward is that not only can anyone come and build an application, but anyone can also build an execution environment. This is exactly the core of the rollup-centric roadmap. Because I can run my own execution environment and prove to Ethereum that I did run it correctly. So now anyone can build a new execution environment, which is the next level of deepening open innovation. But we still find that consensus is still determined by Ethereum, and this blockchain and its consensus method are unique. All the scaling trade-offs are inherited to all systems. All rollups built on Ethereum must write data to Ethereum's data availability layer (DA) and related content. So the next stage of open innovation can be thought of as "open consensus." This means that there is only one universal security zone, and security comes from staking. Staking is a very powerful way to provide security because it imposes both positive incentives and negative penalties on people who participate in these different protocols. I can let innovation happen at the consensus layer, the execution layer, and the application layer, and we build these on top of Ethereum. So now all ETH staking, ERC-20 tokens, and other forms of economic value can be utilized in this platform.

Kevin: It's been incredible to see the growth of this ecosystem. I'm also curious what your favorite AVS ideas are, and what are the coolest things you've seen in and around EigenLayer? I know you probably don't want to single out a "winner," but I just wanted to get an example of AVS in the minds of the listeners. What is AVS? What's your favorite example?

AVS (Autonomous Verification Service)

Sreeram: One way to think about it is that you can say "run a new consensus chain on top of this stake", which is a simple idea. But actually, I think a better way to think about it is to think of it as a "service". That's why we call it AVS. Initially, we thought of AVS as "actively validated services". But we have now redefined it as "Autonomous Verifiable Services". Autonomous means that they run independently. Anyone can participate in staking and run this verifiable service. Since anyone can run it, you need to make sure that they are verified post-hoc through some kind of on-chain penalty contract and conditions. This is a service category built on EigenLayer - autonomous verification services.

So what might those services be? The first categories we explored were: Can this complement Ethereum's rollup-centric roadmap? Rollups are on the rise, and I want to decentralize a sequencer. Perhaps I can create a consensus group on EigenLayer and run it, thereby decentralizing the sequencer. You want to have super fast bridges between different rollups, and as long as there is enough economic stake on the bridge, the bridge can easily pass messages between different rollups. You want to do some kind of MEV (maximum extractable value) management. If there is a decentralized network, you can ensure that transactions are ordered in some fixed, pre-agreed way.

Another example is data availability, usually you would publish data on Ethereum, which is the most secure form of data availability, but you may need higher throughput. You can use a data availability service built on EigenLayer. We built a service called EigenDA ourselves. Taking EigenDA as an example, its data throughput is 15 MB per second, while Ethereum is only 33 KB per second. So there are several orders of magnitude differences here.

EigenDA is built on the Danksharding philosophy from the Ethereum roadmap, so we adopt the Danksharding roadmap. Since we are just a small startup team, we can develop faster than Ethereum, we just need to do it very carefully. So this provides a layer of acceleration and innovation for Ethereum. These are examples of Ethereum's rollup-centric roadmap and how the EigenLayer service complements this roadmap. Next, you can continue to think about, maybe not just a rollup-centric roadmap, but various other possibilities. One of the categories that I am particularly excited about is called coprocessors. Vitalik recently wrote an article called "Glue and Coprocessor".

Therefore, Ethereum EVM can be seen as "Glue", and any service can be run on it as a coprocessor. Coprocessors can be divided into two types: one is to run certain tasks off-chain and then verify their correctness through zero-knowledge proofs; the other is crypto-economic games to ensure that participants run correctly. This is similar to the zero-knowledge proof (ZK) type of trade-offs of the Optimistic protocol. What can this crypto-economic coprocessor build? For example, AI.

This is a hot topic right now. As many of you know, you recently did a show about AI, and this is one of the areas we are very interested in. When you think about AI, you want the output of the AI to be verifiable, especially when you bring it on-chain. If your AI is not verifiable, it doesn't make sense. You can run AI off-chain, but you need to bring its results on-chain in a verifiable way. In addition, you can also run other off-chain processes, such as running a database, running SQL, running off-chain games and bringing the results on-chain, or doing zero-knowledge proofs (ZKproving) and all kinds of interesting things.

Another category is cryptographic services. I want to store a secret on this decentralized network, and not have any one node have the whole secret, but split it into small pieces and spread it across the network. This is like a network of secret sharing, secure multi-party computation, etc. We already have several projects building on EigenLayer involving fully homomorphic encryption (FHE). I want to do computations on encrypted data instead of computing on the original text. I want to do computations directly on encrypted data. We do have some related technologies, such as trusted execution environments and hardware encryption. The feature of hardware encryption is that it can ensure that the computation is run in some protected hardware environment and the secret cannot be leaked to the outside. This is a unique category. Actually, each technology is interesting.

I gave a talk about a year and a half ago that showed off these different categories. Each of these categories has more than one developer building on EigenLayer. It's been really amazing progress. Another really interesting category is some form of "proof". These proofs are not in the form of zero-knowledge proofs or computational proofs, but proofs of other properties. For example, proof of location. The whole crypto economy is about decentralized trust, and one aspect of decentralization is geographical decentralization. So how do you prove to a network that I'm somewhere in the world? Can I prove to the network that I'm actually in a specific location? This is similar to how GPS works. Basically, you need nodes to send you information to verify your location. Based on the latency of the information going back and forth, you can infer the location of the node. If you can do it in a way that is resistant to adversaries, even if some nodes give adversarial answers, enough nodes overall can provide the correct information, then this constitutes a proof of location service. Proof of location and proof of human identity is another category. It's been showing up a lot on EigenLayer in recent years. There are also technologies like ZKTLS, which is a form of network proof. Typically, when you think of an oracle, its function is to read internet data into a blockchain, but that data is usually public. However, the data may also be visible only when I am logged in. In fact, most of the most interesting data on the internet requires basically a login to access.

For example, I am an Uber driver and want to move to a new ride-hailing platform. I spent three years building a 4.9 rating on Uber, and now I want to bring this rating to a new platform. However, Uber currently controls my data and says that it will not send this information to other platforms. However, from the perspective of users owning their data, users should be able to prove to this new network the fact that I do have a 4.9 rating on Uber. This is very difficult to achieve, but there are already multiple protocols on EigenLayer that can achieve this goal. I think this is a huge change because in the crypto field, we often talk about users owning their own data. But usually I think this can only apply to new systems built in the future, that is, new systems can have such characteristics. However, what we can do now is to bring users' data in Web2 into the new system in a way that users control themselves. I think this is a huge step forward. One of the topics we often discuss within the EigenLayer ecosystem is how to "vampire attack" the entire cloud computing system. In other words, how to migrate all this data to the own track in the crypto field. That's an overview of the different categories.

Kevin: Thanks for your answer. I want to move from “what is the function of EigenLayer” to the deeper Green Pill question. I think there is an interesting intersection between my work and yours, and I’m curious to explore and tease out these threads, like improvising jazz on them. So I’ll start this part of the conversation with questions like: Why are we here? What is the goal of what we are building? What drives us?

04 First Look at "Coordination"

Sreeram: For me, I first became interested in cryptocurrencies because I resonated with an idea: As Yuval Harari said in Sapiens: What makes humans special is not that we are smart, but that we can flexibly cooperate on a large scale. This became a very important and transformative idea for me and profoundly changed my personal understanding of how the world works. When I thought about this, it occurred to me that the main barrier to cooperation is trust, right? If I know I can trust you, I am willing to cooperate with you. If we can solve the trust problem, then we can create a new kind of large-scale cooperation, which is a very crazy, huge upgrade for humanity.

So that became my basic thesis for cryptocurrencies - just like the internet was the information superhighway, cryptocurrencies can be our coordination superhighway. It was a crazy, interesting, brilliant vision of what cryptocurrencies should be. At the time, I didn't really understand that, like, Kevin, your work and Ethereum's coordination efforts seemed to be emerging from other places.

My thought at the time was that if that was the main thing, then coordination was cooperation, and coordination mechanisms were actually very valuable for coordination. So, I was pretty much operating independently based on my own assumptions, without having a deep intersection with other rich ideas related to this. It wasn't until about three years ago that I discovered that there was a lot of coordination work in the Ethereum ecosystem.

I realized that there was actually a huge and vibrant culture around the idea of how to build better coordination systems and put them into practice. And there was also a huge community and movement, part of which you were leading, Kevin. I was very happy to discover that. I would say that was probably the main reason why I chose to build in the Ethereum ecosystem. Because, when you have a shared vision, any person and project that is pushing for more coordination makes me happy. I don't necessarily have to be the one to make it happen. If someone is doing it, that's even better because together we can accelerate this faster. That's how I really got into Ethereum.

This was probably around January-February 2022, when I first realized, oh, it turns out that these people are doing some of the high-level theories that we had proposed before, and they are actually trying to implement it. This was a profound revelation to me. Since then, I have been paying attention to the specific manifestations of coordination. Where is this coordination happening in the cryptocurrency space right now? What can we do to accelerate this process? This is the basic structure of the whole framework.

Coordinated Innovation Model

Not only have I been building EigenLayer for the past three years, I've actually been working in crypto for the past seven or eight years. I've been perfecting a model that guides all of my work. I call it the "coordinated innovation model." So what's the idea? The idea is that any successful large-scale system, like the United States, will have an underlying coordination layer, like the constitution and government, and there will also be an innovation layer that makes innovation possible. On top of the innovation layer, there are things like free markets where people can exchange goods and services. And the establishment of free markets is fundamentally because the coordination layer, the constitution, ensures things like property rights. Because by making this clear and making credible promises, like "if you own a property, you really own it."

Now, anyone can easily trade the rights to that property on the free market and create value so that it flows. So, there is a coordination layer under the innovation layer as support. This is the starting point.

When we start thinking about this, you talk a lot about non-zero sum games. If you think about it, I think fundamentally there are only two archetypes of non-zero sum games. A non-zero sum game is a game where all the players benefit, as opposed to a game where there's a winner or loser.

When you think about the typical games that are possible, I think there are really only two. One is innovation, which is inventing something from nothing, and the other is coordination, which is n people coming together to create something that is more than the sum of its parts. So coordination and innovation are two typical positive-sum games. And they are actually layered and nested, layered in a fractal way. You can think of it even back in evolutionary timescales, before there were multicellular organisms, you find that within a cell, there were first divisions of labor, like the nucleus and the mitochondria.

Mitochondria are focused on energy production, and the nucleus is focused on information transmission, but they are coordinated. Coordination at this scale enables innovation. Innovation is the birth of multicellular life. So over the long course of evolution, you have new coordination systems emerging, and new innovations emerging. And out of that emerges a new coordination system. And it goes on and on. If you graph progress over evolutionary timescales, you can see that I think innovation is evolutionary because it builds on each other's innovations. And coordination is revolutionary because coordination systems emerge rarely. Coordination requires binding commitments between n parties. n different parties have to collectively accept, agree, and opt-in to a new coordination system. Because of this, new coordination systems are very, very hard to build. Not because they are inherently hard to build, but because they require consensus and voluntary participation. Also, coordination systems are hard to come by, and even harder to change unless there is a change mechanism built into their contracts. This explains why Bitcoin came about.

Bitcoin as a new coordination system, it provides permissionless money. However, it is very difficult to change it. Before I started to actually build EigenLayer, I thought, can we make Bitcoin use a new coordination system? But then I understood how difficult it is to change a coordination system. It is really, really difficult. So the theory is: the layers of social evolution are made up of innovation layers and new coordination layers that emerge on top of them. And then these new coordination layers incentivize and promote the establishment of new innovation layers, and these layers are fractally nested. All these different players compete with each other and actually create a net growth engine. And then I saw that if we can actually see Bitcoin and Ethereum being born in front of our eyes, over the past decade, it gives us confidence that maybe the game is not over. We can build new coordination systems and make them powerful and useful for all the people who build on top of them. The vision of EigenLayer is to build the coordination engine for humanity. That is our goal. It has nothing to do with heavy staking, staking, or building a decentralized network. It is about creating the coordination engine for humanity. That is our main value. And we only find a small starting point from that, which is heavy staking.

Kevin: Great. I’m a coordination maximalist, so these words are music to my ears. To help the audience understand better, let me recap: layers of social innovation require coordination to obtain, and then new coordination infrastructure is built for later generations. By the way, we are recording this with a coordination layer, which is TCP/IP and the Internet. So these layers of innovation are built on top of innovation. You can almost think of them as layers of sediment, which are fractally nested on top of each other. Each generation has driven the development of hyper-complexity and the processes that civilization needs when dealing with information and resource allocation. You have identified trust as the limiting factor to achieve this goal. And the focus of EigenLayer is to decentralize trust and make it easier to access a huge pool of trust. The purpose of this song is to build the coordination engine of mankind. And this is the core value of EigenLayer.

Coordination equation

Sreeram: Right. When you think about coordination, I have an equation that I call “Coordination = Communication + Commitment.” Okay, so what do I think is a coordination system? A coordination system needs to allow the parties to first be able to communicate with each other easily, like, the internet is our communication highway. So I can communicate with you over the internet. But what’s missing is the ability for us to make binding commitments, like, I want to make a commitment that if something happens, like, the output of this podcast, we split the proceeds. How do we do that? How do we make these types of binding commitments? The internet itself doesn’t have a native infrastructure to do that. When you think about commitments and who fulfills or enforces them, this is a very important bottleneck in coordination systems: Who fulfills the commitment? Who enforces the commitment? How do we trust them? You can categorize commitments, and one way to categorize them is you can think of the “authoritatively enforced commitments” that have historically been dominant.

That is, a single authority, like a king, enforces the promise. For example, they might sign a deed that says something like "Kevin owns this house," and they'll keep that promise. So, you can have a promise that's enforced by an authority. You can also have a promise that's enforced by a committee, like an organization collectively decides how something should be done. In the United States, you see similar examples, like the Federal Reserve is a committee that decides monetary policy. However, committees are more robust than authorities, but there are also problems with committees themselves, such as what are the incentives of the committee? Are the incentives of the committee deviating from the common interests of everyone? Does the committee have inherent selection bias? Is it representative in some way? These are all issues related to committee-based enforcement. Then you can go a step further and talk about majority-based enforcement, such as a pure democratic system where a group of people get together to vote and make a decision. But even with majority-based promises, there is "tyranny of the majority." The majority may oppress the minority at will. And in fact, each of us is a minority in some way. If you divide it along different dimensions, everyone is a member of a minority in some sense. So, commitments based on majority principles also have the problem of "tyranny of the majority". Next, you can ask: What is the deepest, most robust and strictest system of commitments? We call them "self-enforcing commitments", that is, those that can enforce themselves. This is an incredible concept in some very fundamental sense, and it makes people wonder whether this is even possible. I think it is worth thinking about what these types of commitments are from a historical perspective. For example, in the ancient times of kings, different countries needed to trade with each other, and one king could not trust another king, another king's council, or the majority of the people in another country. So they had to find some kind of resource or mechanism to solve this problem.

If I'm going to transact across trust boundaries, like in ancient times between kingdoms, I need to have a self-enforcing value that can transfer value across trust boundaries. I think all of these evolutionary pressures led to something like gold or other precious metals as a mechanism for exchange. Because when you give me a piece of gold, it's self-verifiable to me. I can know it's a piece of gold because I know the properties of gold. And it works very well as a promise to transfer value because I can pass it on and no king or committee or majority group can deny that it's gold. Because it's self-verifiable. So, this became the foundation of the modern economy because gold can interact and transact across trust boundaries. It's a self-enforcing promise to transfer value. So in the past, you've created this very powerful self-enforcing system. But the problem with gold is that, just like Bitcoin is the modern version of gold, it's just a pure value transfer tool. Like, I can only transfer value with gold, but not distribute value. Distributing value means being able to say, "If A, B, C, and D happen, then I will get this value and you will get that value." That's a more expressive way, and gold doesn't have that ability. So we see this progression from gold to all sorts of complex things, and I think the modern version of this is like a constitution, like the Bill of Rights, which is almost self-enforcing. Not completely self-enforcing, but almost. When someone tries to seize power and violates the constitution, it sets off an immune system response where people realize, “Oh my god, this is violating constitutional rights.” And there was a transitional period before that where religious texts played a similar role. Even the king was not above the religious text. Of course, sometimes the king would twist the religious text to suit his own interests, but at least they had some kind of rules that they couldn't unilaterally violate. So I think before crypto came along, there were three stages of self-enforcing systems: the golden age, the religious age, and the constitutional age. The constitution was a distillation of the coordination elements of religion, and it wrote those coordination elements into the text of the constitution, and that was the essence of the constitution. I think crypto is accelerating the self-enforcing nature of coordination systems, and that's the next era. I'd love to hear your thoughts on these different versions of coordination systems.

05 Capital Allocation

Kevin: That was great. I can't wait to hear this again, let me recap what I just said and then I'll respond. We are building a coordination engine for humans, and coordination equals communication plus commitment. Now we can enter the stage of self-reinforcing binding commitments through smart contracts. The development of commitments includes the golden age, the religious age, and then the constitutional age. And now, we are entering the era of on-chain commitments.

I’ve actually changed the direction of my past work focusing a lot on meme dissemination, from focusing on “regeneration ideas” (which I think is my area of fame) to on-chain resource allocation. We used to focus on topics like positive-sum games that are good for the world, similar to the hippie-style topics in Boulder, Colorado (such as ecological protection, peaceful cooperation, etc.). Now, the core concept we focus on has become “on-chain capital allocation”. Capital allocation is more precise, technical, and complex.

I see capital allocation as a powerful tool, not something as casual as playing Frisbee (like the culture at CU Boulder). I still hold these values dear, but I’m now focusing on capital allocation. The reason is that I think capital allocation has been redefined many times throughout history: from the Golden Age, to the Religious Age, to the Constitutional Age. From resource-sharing networks in the hunter-gatherer era to the agricultural era of harvest sharing, to the industrial era of ways to increase productivity and achieve mass production. Each era has a new way to allocate capital.

No matter how society changes and how the underlying mechanisms of coordination evolve, the way capital is allocated will change accordingly. I think studying how the "on-chain capital allocation era" works is like laying the foundation for solving the problem of coordination failure at a basic level. Because to achieve the goals you want, whether it is to make your company grow profits, or to regenerate the economy, support biodiversity, and respond to climate change, you must first achieve efficient allocation of resources. So the perspective I bring to this conversation is capital allocation.

Now that we have this new way of decentralizing trust and creating commitments, I think you can create some rules with boundaries so that the problems that we had with human coordination before don't exist anymore. You know, you can solve the free rider problem, and you can solve the tyranny of the majority problem. Basically, we have an infinite canvas, which is Ethereum, to build these funding allocation processes to get money to where it matters. Gitcoin's mission is really about funding those things that matter, and capital allocation is how we do that.

I think that generations ago, the semiconductor industry in Taiwan became a very important systemic part of the economy by etching circuits on silicon wafers. Now, I think the next generation of builders will do similar innovations on Ethereum - they will "etch" capital allocation circuits on Ethereum. I think there will be a new generation of developers who will carve capital allocation "circuits" into Ethereum. Ethereum is like the silicon we use to build these "circuits". In what ways can we use this more efficient and more effective medium (Ethereum) to achieve capital allocation, build more positive-sum games, more profit structures, and better collaboration mechanisms to achieve the outcomes we want. Let's do capital allocation directly on the chain. It's hard for me to summarize a specific question, but I want to explore the intersection of trusted commitments and on-chain capital allocation, which I think will be a topic worth exploring in depth.

Understanding Capital

Sreeram: That's a really great idea because when you think about capital, I think that capital is created by property rights. Capital is something that has clear ownership attached to it. It can be money, property, or anything else that you can own. And the other day, I was thinking about the role of the market economy, and part of its role is to transform things that don't have property rights into things that do have property rights. I'll just leave it at that and I'd like to get your thoughts on that. The role of the market economy, this is not a normative or prescriptive view, it's a descriptive view, right? It just says that the role of the market economy is to transform things that don't have property rights into things that do have property rights. First, I want to hear your thoughts.

Kevin: I have two ideas, and they are different in some ways, but we can look at my ideas first and then decide what to do next. The first idea is that I take for granted that the rule of law exists in the United States because it has existed my entire life and I have never experienced it otherwise. But if you look back in history or go to places in the world where there are no property rights, you can see how different the world is without property rights and how that would dramatically change people's quality of life. So first of all, I try not to take for granted the property rights that allow me to own this computer and do many things and promote my own happiness through these property rights.

The second one is a little different, which is that I was recently convinced by this idea of the eight forms of capital that Gregory Landua came up with. Basically, the core of this idea is that capital is not just financial capital. It's actually intellectual capital, spiritual capital, physical capital, and social capital. So I think that maybe having property rights over financial capital and financial resources can better preserve your intellectual capital and social capital and other resources. I'm kind of trying to guess here how property rights relate to the other eight forms of capital. But I think one of the interesting things is that this podcast is a kind of intellectual capital that we're creating. And I think one of the really cool things about cryptocurrency is that we can create new ways to financialize social capital and intellectual capital to fund things that we think are important, which has never been done before. Because now we can turn this podcast into an NFT. My two thoughts are: Don't take property rights for granted. And also, I'm also thinking about eight different forms of capital, not just financial capital.

Sreeram: A lot of the other types of capital you mentioned (besides financial capital) exist because there is a social mechanism to attribute that capital to you. For example, people don’t just steal other people’s ideas. For example, the distribution mechanisms we have today (like we publish content on the internet and anyone can listen to it) would have been unthinkable 30 years ago because whoever controls the platform controls other people’s ideas. For example, you and I may have a lot of good ideas, but if someone else owns the platform, they will control those ideas and even pass them off as their own.

One of the things we see is that as coordination systems get more expressive, you can specialize. For example, when you think about capital allocation, venture capital is an example of a specialization in capital, ideas, and innovation. Capital and innovation. One person provides the capital, and another provides the innovative ideas, and they can collaborate and create output, without requiring one person to have a lot of money and a lot of ideas, which is very rare, and almost impossible to have both. So the better the coordination system, the more these different forms of capital, the eight that you're talking about, can be coordinated to actually produce a useful net output.

So what do better forms of coordination actually enable? Maybe that’s the core idea that you’re laying out in your upcoming book: that you can have more expressive allocations of capital in new forms, and that using these on-chain tools will actually facilitate more forms of coordination across different types of resources and capital that people have.

Kevin: Another area where I’m really excited about the intersection of on-chain capital allocation and the EigenLayer ecosystem. A few months ago, I wrote an article titled “Restaking is Regeneration.” The core idea is that restaking creates yield through productive economic activity, which is not free, but is achieved on top of the pool of capital that is already being used to secure the Ethereum network. There is a concept of “exothermic” in nature, meaning that they release heat. For example, the sun is an example of this, and all life on Earth gets its energy from the sun or other heat sources such as volcanic heat, and even in some cases, we now use nuclear reactions on Earth. So the concept of “exothermic” in nature is related to heat, but in cryptocurrencies and our economy, there are also structures that are economically “exothermic”.

One of them is the actual stakers of Ethereum. So I get economic "heat" from that, like returns flowing into my personal economy. But what I'm really excited about is having a thousand different economic "heat" experiments in the AVS ecosystem flow into these on-chain capital allocation structures. But the question is, how do we plug it in? I think what you bring is exactly what we've been looking for. For example, I've been funding public goods with Dogecoin for four years, and I'm very grateful to the Akita community. Thank you for working with us. But I wish there was a return that was more like ETH or USD that we could plug into these on-chain capital allocation frameworks. I think Vitalik mentioned "impedance matching" in his show, which is finding a balance between allocating capital for public goods and raising funds for public goods. And you solved this problem with EigenLayer. So, the combination of re-staking and regeneration, what do you think about the connection between these two concepts?

06 Re-staking and Regeneration

Sreeram: When we think about strategy, positioning, etc., we think about it from two axes.

The X-axis is how long-term something is, right? Like, how intellectually profound is it? Does it actually have long-term value? All of that goes on the X-axis. And on the Y-axis, we think about, is it friendly to short-term players? Is it suitable for people who are financially oriented? So you can think of the Y-axis as "speculators" and the X-axis as pure "long-term value." I wonder if we can match those two. Because most great ideas need energy to take off, some kind of activation energy, and that activation energy can come from speculators. Because if you're purely on the speculator axis, you're in a zero-sum game, like PVP (player versus player). But when you combine that with the mindset of a long-term builder, thinking about what are the long-term positive outcomes, you can actually use the energy of speculation to guide the birth of new systems, new technologies, and new things that are ultimately net positive. This fits in with another mental model that I often use: when you talk about multipolar traps or coordination failures, you can visualize these problems with a diagram of solutions.

So, the X-axis is all possible solutions, and the Y-axis is the quality of the solution. You have some non-convex problems where there is a local maximum (like a small hill), but the best solution is a global maximum (like a higher hill). However, to get from the local maximum to the global maximum, you have to go through a valley, right? That's actually the hardest part of solving coordination failures: you have to get people to temporarily sacrifice their own interests in order to get to this new global maximum. I think this is usually very difficult to achieve. But one thing we realized, and I think this is another place where coordination and capital allocation theory intersect: by clarifying the value of the new hill, you can say, "Hey, if this new hill (global maximum) is achieved, it will have this much value." And what I'm doing is borrowing some of this value from the future and distributing it to people today, so that in the short term it's not a net negative, but a net positive.

So you actually build a virtual ramp between the local maximum and the global maximum, by borrowing the value of the future, clarifying what the future will look like and why this thing will be valuable in the future. Then, you can guide people to a better global maximum through this virtual ramp. This idea is very inspiring to our thinking. This is again connected to the theory of the two axes I proposed earlier: find a global maximum, and then find a path to build this ramp using the tools of finance and capital allocation.

I mentioned the coordinated innovation theory before, and coordinated systems are hard to form, and the reason they are hard to form is precisely this "valley problem" - you have to transition from one social equilibrium to another, and this requires crossing a valley, which is not a rational choice for individuals. Usually, we resort to altruism to cross this valley. But the problem is that only very few people can really be in a purely altruistic position because of various constraints. And if we can make all participants win-win, then you can actually build such a system. I think it is very powerful to use the economic system as a form of capital that expresses this mechanism.

So you can actually solve almost any coordination failure problem, not only through the various commitment techniques and tools that we talked about, but also by leveraging financial economic tools to borrow value from the future, which is a very powerful idea. One way this is manifested in the cryptocurrency space is that people are using tokens to guide the various participants in the market. What is a token? A token is a representation of this potential future value. If the network is realized, it actually has some value. You are able to assign a portion of the ownership, a portion of the value to the participants.

Initially, this valley is so steep that you can fall right down and fail. So you need to provide additional incentives. In the early stages, you need a lot of push to compensate people for the risk they take. And then over time, you need to balance that incentive. So I think this is another place where coordination theory and capital allocation theory come together: by understanding this terrain, you can illuminate the future value of this coordination system and share that value more fairly, while extracting a portion of that value to compensate those risk-takers to nudge them toward choosing a new coordination system. I think we can actually guide people from a local maximum to a global maximum.

Kevin: That's really beautiful. I have a visual about this. If you watch it on YouTube, hopefully our editor Kishan will put it up. But basically, the visual image that comes to mind right now is this idea of a search space where the X-axis represents all possible capital allocation structures or coordination structures, and the Y-axis represents their effectiveness in solving a problem or creating value. We are actually in a decentralized search space to find the best structure to solve the coordination problem. If the industrial age or the internet age structure was this small, the on-chain search space is much larger because now we have trusted commitments through smart contracts. We have decentralized trust in the EigenLayer and the Ethereum network, which is a big pool.

So we’re moving beyond what existing economics has already explored. Just like when you go into a new area in Zelda and get a map of that area. We’re moving into a world of trusted commitments, on-chain capital allocations, yields, and so on, trying to figure out the best structures for coordinating human activity in the 21st century. One of the things I’m also hearing here is that different actors will have different reward functions. And one of the things we can do with trusted commitments is to align the interests of individuals with the collective interests at the local, state, and global level. Tying those incentives by making it more attractive to work on problems together is, I think, one way we get to 3-3. We get to that coordination “grid.” Hopefully the visual diagram on YouTube helps, but I think this is a really exciting distributed search problem: how do we find a structure that can speed up coordination, build a coordination “super highway.”

07 AI + Encryption

Sreeram: This is one of the most exciting things we can do anywhere. Going back to the coordinated innovation theory, one thing you can think about, and probably a lot of people's questions, is this: AI is coming, and you just did a show about this. AI is coming, and how does all this crypto fit in with it? The coordinated innovation theory actually gives us some very insightful insights. If you think about innovation as an accelerator, then AI is an accelerator of innovation. AI is the driving force behind the acceleration of innovation.

Innovation is any action that any individual can take to improve themselves. And AI is really powerful at that. And crypto is an accelerator of coordination. So you can think of the extreme point of acceleration of innovation as AI, and the extreme point of acceleration of coordination as crypto. And that gives me an idea of how these structures will interact with each other and what interesting new things might come out of this. And a very interesting question is, when these two structures, AI and crypto, interact with each other, what big things happen? That might be the next very interesting topic of discussion.

If you think of progress as a graph, it keeps growing evolutionarily, and then all of a sudden there's a jump because of coordination, and then progress keeps growing evolutionarily, and then there's another jump because of coordination. Another way to ask it is: What's the next big jump in coordination? What will it lead to? When we set the vision for Eigen Labs, our goal was to build the coordination engine for humanity. So, crypto is the coordination engine for humanity. Let's speed it up, make it more expressive, let it do more things. Now with that perspective - innovation is being accelerated by AI to the maximum, what can crypto do to advance that? I think there's a lot more to come at the intersection of crypto as the coordination foundation for AI. We talked about earlier that coordination is communication plus commitment, and crypto can create and maintain self-enforcing commitments. So when you think about commitments and how commitments are made in the age of AI.

There are four possible combinations here: between humans and AI, including human to human, human to AI, AI to human, and AI to AI. Each combination is worth deep thinking and discussion. Regarding the promise of humans to AI, the current situation is that people mostly run AI agents on their own servers and then call it some kind of autonomous AI agent, which you might think is silly if you put on the hat of "decentralized enthusiast crypto enthusiast". This is bad because it's just someone running their server. So I have a point: the agent that controls the Ethereum wallet is not much different from the agent that controls the Stripe API. They are actually the same thing, and if you run it on a server, it's exactly the same because it's just calling your bank account or Ethereum wallet. If both are controlled by an individual, then there is no essential difference between them.

The real difference is that when the AI itself runs smart contracts on the chain, the situation is different. I should be able to run an AI smart contract just like I run an Ethereum smart contract, and its output should be as accurate and strict as writing a smart contract. So suddenly you open up a whole new world of possibilities, you can write arbitrary AI programs, they run on the chain, and their output is as strict and accurate as the on-chain contract. And then this AI is not controlled by anyone, just like the contract is not controlled by anyone. We have been thinking about what this future will bring, such as new on-chain life forms. We call it AI life, and when people think of AI, they think it is a crazy new thing.

Some AI safety experts worry that AI will suddenly get out of control and act on its own. For example, how to control it? Questions like that. But there is no clear "takeoff mechanism" that can make AI suddenly act on its own. Perhaps a robot that replicates itself and controls a city can be counted as an example, but these are still far away. Today's AI is still purely digital. Can a purely digital AI become its own life? What we think about is, what is the unit of life? You can think of units of information, such as DNA. For a living organism, it controls units of energy, such as ATP (the energy unit produced by mitochondria). It is self-directed, just like a living system has its own DNA to control units of energy, not something else. Finally, it is adaptive, it can adapt to the environment. These are the basic components of living organisms.

And what we see is that if you think of an AI agent running on your server, it has one unit of information, which is its intelligence. It also has one unit of energy. I think money, or capital, is a unit of energy because it can take actions in the real world, so it has one unit of energy. It's not ultimately self-directed because you can turn it off at any time, so it's not a new form of life or "the agent did this." No, you did this. You just have a computer performing tasks for you, so that's the third point: it's not self-directed. The fourth point is that it is adaptive and intelligent, and of course, these agents can be very intelligent. So I think agents like OpenAI fall into that category.

Whereas a DAO meets the first three conditions, its unit of information is code, the unit of energy it controls is money, and it is self-directed because on-chain property rights are automatically enforced. But it is not adaptive internally or as a DAO itself. The adaptability comes from the member agents who control the DAO, who actually exercise their power, which is great, but it is not a life form in itself. It is just a coordination mechanism for these human units to come together. So what I think will completely change this is the emergence of on-chain AI contracts, so that you have a unit of information, which is AI code, that controls on-chain funds, utilizes on-chain property rights, and it is adaptive because it will continuously improve itself through learning.

So I think this is going to be a huge shift at the intersection of crypto and AI. This is the extreme form of the theory of autonomous AI agents. We call it AI life because it's now a life form. Because it undergoes evolution like any other life form. It has its own unit of money. And just like any life, as long as it has an energy surplus, it can persist, you should be able to harvest more energy than you consume so that you maintain a net energy positive growth and continue to survive. And when you can't harvest energy anymore, you die, this AI agent can survive as long as it maintains a net money surplus, and as soon as there is no net money surplus, it dies. This imposes an evolutionary unit at its base and creates a very interesting, open-ended new process by which these agents can evolve and be created. This is a crazy new theory that we're working on and trying to build some core components to implement it.