Josh Stark I Atoms, Institutions, Blockchains I Ethereum Foundation I Pragma Denver 2024

The floor is yours. Please welcome Josh from the Ethereum Foundation to talk about atoms, institutions, and blockchains. Okay. Hi everybody. Thank you, Kartik.

Thank you all for being here at Pragma in Denver. Uh I didn't come to Denver last year. Uh I missed it. I'm very happy to be back. So, my talk is going to be maybe pragmatic in the spirit of things, but in a very galaxy-brain kind of way.

Um but we'll we'll get there. My name is Josh. Uh I work for Ethereum at the Ethereum Foundation. Um that means lots of different things. I've worked with the EF for about 5 years now.

Uh I started in 2019 working on some special projects. Uh and in 2020, I joined the leadership team. Um I like to joke that my job is figuring out what my job should be, and that changes every 6 months. Although, hopefully it's settling down a little bit now. Some of the things I'm thinking about these days are what are the longest-term challenges that are facing Ethereum?

What institutions could help address those challenges if they existed? How can the EF help both with the challenges, but also building those institutions? And how should the EF change and evolve over time reacting to the ecosystem that all of you people and many others are building? But that's not what the talk is about. I've done lots of different things uh in my career uh in crypto.

I worked on an early Bitcoin wallet. I ran a Bitcoin ATM in Toronto for about a year in 2015. In 2017, I co-founded EthGlobal, as well as a company called L4 working on early L2 scaling solutions. And over the years, I've written a lot of blog posts, although I write them less frequently these days. And through all of this, I still couldn't really explain what we were doing or why I had this feeling that this was very important, that it was going to change the world.

And don't get me wrong, I can, you know, talk about individual applications. You know, Bitcoin, we can explain. NFTs, we can explain. Uniswap, ENS, all these things in their little silos are not that hard to explain to somebody. But when we start trying to talk about the big picture, about what it means that we can do all these things, what it means that there's one technology that enables all of these things and many more, we start to stumble a little bit.

We're doing a lot of this, a lot of these mental gymnastics as we throw buzzwords at people and try to explain things. But we really need to get to here, and I don't think we're that close now. Two years ago, I wrote a blog post really as first as a personal project to try and make sense of this for myself. Um, enough people have told me that it was valuable and interesting to them, um, but I think I should speak more and write more about these ideas. I hope it can be a contribution to just trying to make sense of this, to helping the world understand what blockchains are for.

Cuz it's not enough that blockchains work and that they be reliable. People need to understand them. They need to be legible. We are asking the world to rely on this new technology for mission-critical civilization infrastructure, and that's a very high standard. You can think about it this way.

Society has a kind of sense-making social software that we use to understand technology, what it's for, how we relate to it. And I think blockchains have revealed an edge case in that social software. It requires an update. So, consider this a pull request, uh, a suggestion for how we can update that social software. We're adding some new ideas for sure, but we're also deleting a lot of legacy code, and that's always a nice PR to see.

So, let me tell you a story. In the 1930s and '40s, Claude Shannon was surrounded by the beginnings of a new age. At Bell Labs, he worked on fire control systems and cryptography during the war especially, and he started to think about a more general general approach to to information, to the things that he worked with day in day out. But, he didn't call it information at first. In 1939, he wrote to a colleague that he was thinking about the transmission of intelligence.

The word information had a different meaning then. It didn't have the meaning that it has today as this abstract property common across things. He published in 1948 The Mathematical Theory of Communications, a foundational piece that laid the way for the information age and the development of everything that we touch today. But, most importantly for us, it introduced for the first time an abstract idea of information, a definition of information that was not tied to to music, to speech, to literature, to codes, to something else, but something that was completely abstract. This is the paper that introduced the idea of the bit, the irreducible unit of information that you could measure in any context.

It didn't matter what it was, it was all information. And when you look back, uh you know, it's it's hard to kind of I guess make these historical claims, but before this moment, no one really had this concept of information as a universal general thing. When you look back at older definitions of the word, they're always tied to specific contexts. You know, intelligence meant one thing, message meant another, communication meant something, information meant something else. There was no concept that unified all these things together as radical as that seems now.

This has always struck me as insane. Like, we've been using information technology, as we understand it now, for thousands of years. It's really inextricably linked up with what it means to be human, to use speech, to use a language. But, we didn't name the underlying property or category that was common across all these things until very recently. Uh one of the outcomes of this is that people took Shannon's ideas and began to apply them to other contexts.

Uh uh uh chemistry professor named Homer Jacobson in the 1950s started to think, well, if we can just measure bits in information, then we can measure the informational capacity of the human body, of our ears and our eyes. And he published papers on this. And of course, the greatest example is DNA itself, uh which we would not have been able to understand if it were not for some of the ideas introduced by Shannon himself on error correction, um which were found to actually be, you know, properties of our own bodies. All these things now we understand as being just types of information, but that wasn't always the case. So, what I want you to take from this is that there was a time before we had the idea of information, and there was a time after.

And what if we are similarly missing something so fundamental? So, as I'm jumping around, struggling, doing these mental gymnastics to explain blockchains, I keep running into these weird things and how we talk about them that I think are clues towards something bigger. Number one is that we describe blockchains as both trustless and trustworthy. That's odd. Uh in Satoshi's white paper, we talk about, you know, not relying on trust.

In the Ethereum white paper, we talk about cryptocurrency having the goal to eliminate the need for trust. But, later in the Ethereum white paper, we talk about using Ethereum to make applications more trustworthy. Uh this famous Economist cover from 2015 talked about blockchains as a trust machine. And it's not just limited to the Ethereum ecosystem. Here's the Rootstock roadmap, which talks about making the network secure and trustworthy.

I'm not trying to dunk on anybody here. We mean something real when we say blockchains are trustless, and we mean something real when we say that they are trustworthy. Our language just hasn't caught up with they know. Contradictions or absurdities like this are always worth paying attention to. Sometimes they're just absurdities, but sometimes they're revealing a gap in our abstractions.

They're each pointing towards a concept that hasn't yet been mapped in between into our language, and I think that's what's going on here. The second clue. We talk a lot about how blockchains are different from centralized institutions. Bitcoin versus central banks, ENS versus DNS, but we very rarely talk about what they have in common. Even saying that might sound like anathema to some of you.

Isn't the whole point to be as different as possible from the centralized institutions of our past? But oh. Yeah, this is our our our normal story, right? That blockchains are decentralized and institutions are centralized on that side. But they must have something in common.

There's only a few things that can be used to make a global money, and institutions and blockchains are two of them. Those are the only two things we can use to affect contracts, and that's institutions and blockchains. They can be substitutes for one another in many contexts. If you've ever traded fiat money for Bitcoin or ETH, you have substituted them for each other. If you've ever sent USDC instead of a wire transfer, you've substituted them for each other.

They must have something in common for that substitution to be taking place so regularly. So, this is our normal story, and there must be some category or property that they both share. And it's pretty normal with new technologies, right, to contrast them to what came before. We talked about horseless carriages when cars were invented, dividing the world into horseless and horseful transportation modes. But at least we could name what they were.

They were vehicles. They had the property of helping us move stuff around. It wasn't complicated. When we introduced digital records in the information age, we talked about paperless and I suppose paperful mediums of information. But we knew what they were.

We had a category for them. They were both mediums of information. So, what's going on here? It It It seems like we've invented a technology before we've invented the category to which it belongs, causing a lot of confusion. And it gets even weirder because whatever category we ascribe to institutions and blockchains also needs to apply to rocks some of the time because the other thing we make global money out of is just stuff we find lying around in the ground like gold or gems or silver.

So, what is the thing that that, you know, atoms that that nature has sometimes that institutions have and that blockchains have that make these three things so fundamental for so many of our social infrastructure like money? What's going on here? My diagnosis is that the problem isn't that we don't understand blockchains. The problem is we don't understand everything else. We don't have the category to place blockchains into in order to relate them to everything else.

So, here's my answer to what goes in that box and I call it hardness. Here's another story in five simple steps and then we'll go into more depth. First, our civilization depends on this social infrastructure like money and law and so many other things. And they need to be reliable, right? They need to behave like we expect them to behave at least most of the time for them to be useful to us.

Otherwise, we wouldn't rely on them. They wouldn't become a money. Second, it's very difficult to achieve that necessary level of reliability. So far, there's really only three ways that I can see that we do it. One is atoms, one is institutions, and now the third is blockchains.

Each of these things have this special property I call hardness. They have the capacity that the power to let us make the future more predictable in the really specific ways we require for these complex coordination games. Gold's inflation rate is reliable because of just the properties of of nature, of our planet, of how much gold actually exists. It's atom hard. A contract is reliable because there's a whole institution or multiple institutions of people out there who will come and take your stuff and throw you in jail if you don't follow your commitments.

Your smart contract will operate because it's secured by a cryptoeconomic protocol with billions of dollars on the line. These are the things that are making these claims hard or reliable. And you can think of these things, atoms, institutions, blockchains, as like building materials, like wood or concrete or steel. They're different, sure. They have different properties, but they're part of a shared category, and we can use them together and even substitute them for one another uh when we need to.

And we can compare them, importantly. We can measure them in similar ways, their tensile strength, their compressive strength, their durability, etc. And we use these things not to build buildings, of course, but to build a civilization. And maybe with better materials and the right mixture of materials, we can build a bigger, better, taller, stronger civilization than the one we have now. So, what's in the box?

I think hardness is in the box. If you want to call it something else, fine, but I think there's an idea there that underlies all these things, and this is my contribution. So, I think this is easier. Once you have this notion, this idea of hardness, it becomes trivial to explain what blockchains are for. We need, you know, these reliable or certain or hard things.

We already do this with atoms and institutions as we're all familiar with, and blockchains are just a new way to do that. It's only confusing because it's been like literally thousands of years since we've invented a new way to do this. So, that's the big picture. Let's go into more detail and flesh this out a bit and offer a definition. So, here's my working definition now.

It might change by the next time I do this presentation. It Hardness we talk about in two ways. One is the capability to make these specific outcomes very likely to be true in the future for a certain purpose of enabling social coordination. And secondly, as a as a measure of the certainty or security or hardness of that specific outcome. And hardness comes from, as I said, one of these three sources.

Let's walk through each part of this. So, hardness is a capability. We use it to create really specific outcomes that we want. And so, you might say it's about programmable hardness. We need to be able to program the future to be reliable in ways we want.

And so, we're not talking about just everything in the universe that is certain. Many things are certain. The sun setting, the sun rising, these things are certain, but we can't hang an economy on top of that fact. You can think about it like this. Most of the time when we talk about the Earth, we're talking about this extremely thin skin where everything that ever matters has ever happened, and not the 99.

9% of the mass of the Earth that, you know, hopefully doesn't matter to us most of the time. It It's similar with hardness. There's a thin layer of expectation of future reliability that matters a lot to can you pay your rent, to what happens in the future, to is an economy strong. That's the stuff that hardness is for, and not just any arbitrary thing that is reliable. Hardness is about these specific outcomes.

And what is that outcome? I'm unfortunately introducing another new word here, which I'll call the cast. Um and the cast is just any possible future state of the world that is made certain or secure using hardness. I apologize for the jargon, but the reason to have a word here is that I don't think we have one that is generalizable across all sources of hardness. It's maybe like the bit.

We need a concept that we can talk about in many different contexts and switch between the sources without being tied to one of them. A cast we might have related to a loan would be that if Alice does not pay Bob back, then legal institutions will use increasingly severe threats and actions to force her to do that. This cast is hardened using institutional hardness. A cast about gold might be that a certain amount of gold will enter the market each year for the next 20 years. It's made reliable by the physical properties of our Earth.

And a cast about Ethereum might be a claim about that assets can only be transferred if you hold the private key corresponding to a certain public key. Hardened by blockchain hardness. In practice, we're usually interacting with bundles of these things all woven together, right? If you own gold and hold it in a bank, a lot of things matter to you. Casts about gold supply in the future, casts about the strength of the bank's vault, casts about the strength of the legal agreement between you and your bank and their insurance, casts about the reliability of the legal system in your country that would enforce those rules if something went wrong.

Yeah, as I said, we need a new term to generalize across types of hardness, and cast has some useful connotations. It's things we throw ahead of us, things that we make to harden, protect, or shape, and casts can ultimately sometimes be broken. So, hardness is also for this purpose of of social coordination, as I said. Um remember, it's not about just everything that is reliable. It's reli- it's about reliability about really specifically shaped things in the future.

And hardness is powerful cuz it lets us make casts of that specific kind. You know, institutions can do a lot of really complex things. They can enforce, you know, paternity agreements. They can enforce really complex financial agreements that really matter to human beings. And that particular shape is what we're looking for here.

Secondly, hardness can be talked about as a measure of security or hardness, right? Um and it's always measurable in theory even if it's hard to do in practice. As an example, how hard is this cast that a certain amount of gold will enter the market each year for the next 20 years? Well, one way you could look at it is to do a probability measurement and look at all the data and try to guess about, you know, what is the likelihood that a certain amount will be mined each year. And this is something that that we do, right?

That many institutional investors and so forth know a lot about. Or you could look at it from a cost perspective. What would it cost someone to break that cast? To make a world where it turns out not to be true. Well, one way, if you're a nation state, is to use the powers of war and maybe international regulation to bring it below a certain level.

That's certainly something could do if they wanted to. Or you could go the other way and go get an asteroid from space with lots of gold in it, uh you know, circumventing the physical limitations of Earth to radically change the supply. There is a price to break almost any cast. And lastly, hardness comes from certain sources, atoms, institutions, and blockchains. This is the reason that the cast is hard.

And each has different properties that make them useful in different contexts. And that's where we go next. So, one thing that I like about this framework is it lets us ask these deeper questions and not just talk about, uh you know, specific properties of blockchains, but to really compare all of these different things and think about where they're appropriate and where we use them and in what combinations. Uh you know, we're being like Homer Jacobson here. We are kind of maybe taking a concept and applying it outward into the world to try and compare things on a fair basis.

What are the properties of each source? How do they make their hardness? What are their strengths and weakness? And what sources are good for which purposes? So, atom hardness, as we've said, is about when we find reliability in just nature around us.

Literal physical atoms, but also other naturally occurring properties. Uh you know, we do this when we use gold or beads for money, when we use physical structures to define property rights, or record uh property rights in a physical object like a deed. It has many advantages. Uh you know, automatic enforcement, shared state, a universal rule set. It is very convenient for human civilization that the rules of physics apply everywhere equally, at least at the macroscopic scales that matter to us most.

Uh but it has many weaknesses. One is that we're limited to this what we can find out in the world. Atom hardness is kind of like an architect who wants to build a rock face into their home. You kind of have to find one that works. You can't just make a rock face.

You can change it a little bit, but you're just relying on finding the property with the naturally occurring feature that fits the particular need that you have. Uh another problem with it is that we can't just give it new rules, right? Like we have gold, we can't just ask the universe to give us a new kind of gold with lower inflation, you know, more fair geographic distribution, and maybe like fix the weight issue, which is really a barrier for some people. We can't do this. Another problem and they stem from that is it has very limited programmability.

There's only certain kinds of hardness things you can make out of atom hardness, mainly monies. You can't make a a marriage agreement out of, you know, atoms. You need something more complex like an institution to do that. And the cast are often undermined by our increasing human control over nature. Using shells for money is fine until you're part of a global economy that might radically upset your expectations about what the inflation of shells ought to be, and suddenly your economy is wiped out.

Uh using gold as a as as a value of exchange might have the same problem someday if and when we can obtain that and change our assumptions about its supply. But, it's even more subtle than that, right? Sometimes we have casts we don't really realize exist, but then they're gone because of something changed and we have to remake them. So, there was a hard cast about the speed of trading in financial markets for a long time that it could only be done at a certain pace. You know, maybe the pace that someone can show to each other on the phone the floor of a trading operation and, you know, write down those commitments.

And this cast was Adam hard. We just couldn't communicate faster than that. But, of course, new technology completely undermined those assumptions and we realized that we actually can trade it radical speeds using computers. We realized that we actually liked a version of that Adam hard cast and we remade it out of institutions. We introduced new regulations that limit the speed of trading and enforce things like circuit breakers, etc.

Institutional hardness is a very wide category. It covers most things we might think of when we think of civilization. Our legal systems, our legislatures, our police forces, corporations, everything, right? It includes all the things on this screen and more. All institutions that provide hardness of some kind to an area.

We created casts that gave order to our societies, punishing antisocial behavior. We created hardness as a platform by letting anyone create their own casts made hard by institutions if you follow certain rules. We created casts that spawn new assets and provided sources of credit to growing economies. Institutional hardness has many advantages. It's very programmable.

You know, human beings grouped into organizations can take really complex or subtle instructions. This is a very large design space of possible casts and they're made of people and people are good. Maybe it's good that sometimes someone can step in and say that I'm not going to enforce that because I think it's wrong. It's good that maybe sometimes there's a break in the system for someone to be a whistleblower or a rebel. But, it also many weaknesses.

It's limited by borders. Only in certain countries you really have access to institutions that enforce the rule of law that are actually hard. Uh it's exposed to political or state failure. If your government just can't agree on things or you are invaded by a belligerent nation, certain institutions you rely on for money or for contracts might just fall apart. Uh they're often opaque.

It's hard to tell if an institution is really hard or not until something goes wrong. We don't always know and that's a really bad thing as well. And they have a high startup cost. We can't just easily make new institutions at the scale of the federal legal system to iterate on them. We're kind of stuck with the ones we have and there's a a really like strong default or lock-in.

And they're made of people and people are bad. The reality is in this country and many others that many people have not really had access to the hardness provided by institutions. They weren't able to get a mortgage. They weren't able to open a bank account uh because when you staff an institution full of people, it's subject to their evils, their prejudices, their ideologies. And our reliance on institutional hardness is only increasing, right?

The problem with software eating the world is that most software is really just made out of an institution behind the screen and we're giving them more and more power as a result. Blockchain hardness. Satoshi's invention, of course, was more than just Bitcoin. It was the kernel of this general-purpose technique for creating digital hardness in a digital environment. It has many strengths.

Universal global access. It's made of software and anyone can write software. The degree of hardness can be transparent and auditable. It has a low startup cost and it's easy to iterate. And they're secured by its market incentives and markets are rational.

But it also has weaknesses. It requires a technological civilization. We we couldn't have had blockchains before now because of the requirements that it had. And in the civilization in the future that doesn't have what we have now, won't be able to use them either. Uh it's made of software and software can be written poorly.

The scope of CASTs are limited to on-chain environments and they're secured by market incentives and markets are irrational. So, what does this mean? What does this give us? Why is this more than just a an academic interest for me? Well, a lot of things start to make a lot more sense when viewed through this lens.

One is the question we started with. Why do we say that blockchains are both trustless and trustworthy? Well, the reason the explanation is this. When we say that blockchains are trustless, what we really mean, I think, is that their hardness doesn't depend on a person or institution. And when we say that they're trustworthy, we just mean that they do have hardness, just as if a different kind.

Our inability to make that distinction is what causes this confused language. It explains why private or centralized blockchains are not interesting, right? A blockchain that is not decentralized just collapses back into being an institution. You know, if it's controlled controlled by three banks or a handful of validators all funded by the same organization, then it's just an EVM secured by institutional hardness. And the most interesting thing about blockchains is not the EVM, it is that there's a different source of hardness that is not correlated or subject to the same failures and limitations as institutions.

That's why it's different. That's why it matters. It also helps understand like the spectrum of possibilities and the kind of default ideologies people fall into in the blockchain space. Many people are very focused on using blockchain hardness to compete with or replace institutional hardness. This is what a lot of the Bitcoin community is about, what a lot of DeFi is about, even ENS is, you know, just trying to replace DNS or complement it in some way.

But then there's also people that look at this as, you know, there's probably things that blockchain hardness can do that institutional hardness can't. There are ideas that no one's ever tried before cuz we never had this capacity, this certain flavor of hardness before, and now we can explore those things. Maybe NFTs are there or games like Dark Forest or the movement around autonomous worlds. Most importantly, I think for me at least, this framework, it raises our ambitions because personally this is what matters to me and maybe it resonates with you that I'm not just here for these applications. You know, I'm not someone who's just really all about Bitcoin or all about DeFi or all about NFTs and maybe that's you, too.

That there's something bigger going on here and we can honestly set our sights higher than money. We can set our sights higher than finance. There's a much bigger picture. I think this actually helps define for me a vision that feels adequate in scale to the challenges we face and to the opportunities that blockchains offer. The mission is not just to replace the Fed.

The mission is to improve and expand the very materials that we've used to build our civilization. To give to to lower the cost of these tools so that everyone on Earth has access to them. To allow for more change to happen. Oh, and by the way, that'll that cost be getting lower soon. To help humanity keep playing this infinite game by letting more people change the rules.

That very few people can enact a law, but anyone can write a smart contract. We are expanding that capacity. I think a lot of people in many different countries, many ideologies, feel like we're stuck. That the rules of the game aren't what they should be anymore, but we are powerless to change them. We're stuck in so many ways in this local maxima and we intuit that that's wrong.

And blockchains don't fix that, but I think they can help. They open up a new space for experimentation. They let more people change the rules, to write new rules, to contribute to that infinite game. Yeah, we can't write laws, but we can write a smart contract. And I want to end on this note that, you know, if you've seen talks by people at the EF before, you know, we're fans of the book Finite and Infinite Games.

And one of the maximums from this book is that only that which can change can continue. We can't stay stuck in this local maxima. We have to change things, and I think the blockchains help us do that. Thank you very much.