A Reasonably Necessary World Computer

Transcript

[applause] >> Good morning, everyone. Um, thank you so much for being here. My name's Tim Beiko, and for the last 8 years, I worked on Ethereum core protocol R&D. Um, as Karl Floersch says, part of that work, I helped coordinate network upgrades that got features like EIP-1559, the merge, or Ethereum scaling roadmap from research into production. And this gave me a really deep appreciation for what makes Ethereum so special. It's permissionlessness, immutability, and 100% uptime make it a credible settlement layer for everything from money, markets, coordinations, and more.

Um, a few years ago, when we were studying protocols a bit more broadly, we came up with a term to coin what's so unique about them, unreasonable sufficiency. What we meant by this is that well-designed protocols tend to be deceptively simple, but enable complex interactions on top of them. Ethereum's a perfect example of this. Coming to consensus on a relatively small virtual machine has enabled trillions of dollars in on-chain economic activity. So, for a long time, the question that kept me up at night was whether Ethereum could become planetary-scale infrastructure without compromising on what made it special.

And today, it's clear to me that it can. Over the last few years, we've seen things like zero-knowledge proofs and data availability sampling go from far-fetched research ideas to production software that's actually powering the chain, helping it scale without compromising on security or decentralization. This means we now have a harder question to answer. If Ethereum can work at planetary scale, what should we use the protocol for? My thesis is that for Ethereum to truly succeed, it has to provide the world with things that it needs, but cannot get elsewhere, and those things also have to leverage Ethereum's unique properties.

If Ethereum doesn't provide things the world genuinely depends on, then at best, it'll just end up as a niche open source protocol. If uh, on the other hand, it does but those things don't also depend on Ethereum, then we risk our most valuable use cases being unbundled. So, the most ambitious path forward is for Ethereum to become the substrate for commitments the world can't coordinate on anywhere else. In other words, the question I'm trying to answer is how do we go from having an unreasonably sufficient world computer to a reasonably necessary one? That's what I want to talk about today. Um so, first let me just unpack why I think it's really important for Ethereum to provide these properties the world needs.

In short, it's because Ethereum's most fundamental properties are fundamentally in tension with existing power structures. Permissionless access, immutability, censorship resistance, and decentralization means anyone can hold assets, make durable agreements, coordinate without the approval of any trust board, any third party, and do so without being locked out of the system or have anyone able to shut Ethereum down as the last 10 years of constant uptime has shown us. This is Ethereum reshape domains that historically have been mediated by large institutions and governments. Things like markets, identity, settlement. It's what makes it incredibly powerful, but also what means that its tolerance in the world is not guaranteed. The more Ethereum ends up mattering in these domain, um the the more they'll inconvenience incumbents there.

The strongest path is therefore for Ethereum to become useful enough that suppressing it is just costlier than tolerating it. Sorry, costlier than tolerating it. Um and one way to kind of emphasize how important this is is to look at two different ways this could play out. The first is a cautionary tale. If you look at something like The Pirate Bay, even though file sharing was really help really useful and adopted by millions of people, torrenting never quite became infrastructure the whole world structurally depended on. Um the most popular use cases like movies and music were effectively moved in more centralized alternatives that were more convenient to users and more aligned with existing institutions.

It didn't mean that The The Bay or torrenting went away. It just means that they still exist, but kind of as niche tools rather than world critical technologies. My hope for Ethereum is we can achieve more than this. At the other end of the spectrum, you have something like TSMC. It's a company whose ability to manufacture advanced chips at scale is so critical to the computing supply chain that the entire geopolitical order reorients around it. Structurally, TSMC is the polar opposite of Ethereum. It relies on multi-billion dollar physical centralized infrastructure and extensive trade secrets.

But strategically, I think this is the level of ambition we should have. Ethereum should provide something so unique and valuable to the world that suppressing it becomes costly even for the actors it inconveniences. So, that said, it's not enough to create something that that's valuable for the world. Applications can run on Ethereum without necessarily depending on it. If the core value doesn't depend on Ethereum, then these applications will be rebuilt or migrated elsewhere, whether it's another blockchain, a private network, or a sorry, a traditional database. It doesn't mean those applications are bad, but it just means Ethereum is only a distribution channel for them.

Stablecoins and real-world assets are examples of this, right? They create on-chain activity and are obviously extremely useful to users, but they can be issued on many different ledgers and represented in many different ways in traditional databases. If what users want is just access to those assets, the assets will move to whichever rails are cheapest, fastest, and most convenient for them. Today, that might be Ethereum, but it's a shaky ground to build this long-term competitive foundation. The cleanest counter example to this is probably something like money, right? If you want a globally accessible, censorship-resistant, sovereign store of value, um then Ethereum's properties really do matter.

But if money is the only use case that requires all of this, then it's not clear that we need a full-on world computer. I don't think that's true, and I think there's examples that prove otherwise and have been for years. Um so, it's worth going over them. The first is DeFi, right? DeFi uses Ethereum's shared state and credible execution to let anyone build financial applications without needing trusted intermediaries. Um L2s are another huge fundamental building block. They allow us to scale Ethereum applications to many more users without compromising on security. And then the liquidity that we already have on Ethereum allows us to plug in capital to any new markets as soon as they come on chain.

But these are just foundations, not really final answers, right? Much of the DeFi activity is either constrained to the crypto ecosystem or only partially secured by Ethereum because the Sorry, because the issuers and asset the issuers of the assets remains the ultimate source of truth for many of them. Um and while L2s bring Ethereum's properties to a much broader set of of users, they don't really answer the question of what those people will eventually use Ethereum for. And lastly, you know, liquidity historically has been sticky on Ethereum, but it's only really a moat if the things that generate the returns for it are dependent on Ethereum's core properties.

And this is really important because Ethereum's core properties are very expensive to maintain. Permissionlessness immutability censorship resistance, and decentralization don't come for free, right? They cost us in speed, simplicity, and UX. But these are deeply important things. They're important for Ethereum, they're important for ETH the asset, for the crypto industry, but much beyond that. Like many of you, the reason I work on Ethereum is I think it has the best shot at being the long-term home for these things to exist in the world. But if our best use cases don't rely on this, then we're asking users to pay for protections that they don't need and asking the ecosystem to defend properties the world hasn't learned to value.

In that case, the best case for Ethereum is probably becoming a shelling point for application developers where people use it because other people are already using it, which is not nothing but a long way from being structurally necessary. So, the bar I want to hold us to is higher. First, would the world genuinely want what we're building if it wasn't built on Ethereum? And second, are Ethereum's costly guarantees actually necessary in powering the things we end up building. Only when both of these things are true, are we in that reasonably necessary territory? So, what actually exists in this intersection?

Um it's a very hard question to answer, right? Um one way I've I've I've reframed this to myself is what does Ethereum let users do that other systems don't? I spent a bit of time in the past few months thinking about this and there's a few things that kind of stood out to me. The first is that Ethereum lets you have commitments that survive the context that produced them. Um institutions obviously already provide durable commitments, but these tends to exist within fixed domains, right? If you have laws or court orders that exist within a specific legal system, bank balances and financial assets exist within a specific financial financial system, um but Ethereum offers us like a different kind of durability. A public neutral record of what is committed to and what are the rules under which this can change.

This really matters when claims have to be recognized far in the future by different actors or in different jurisdictions. And this gets even more important in an AI-mediated world. Agents are going to be able to negotiate commitments and agreements coming up with consensus algorithms on the fly, right? We don't expect that they'll need to touch a blockchain to do any of this. And this is fine for one-off agreements, but this but it does break down when these commitments need to outlive the context they were created in. So, Ethereum's role is then to make these commitments and the rules for changing them changing them durable and legible enough for future actors to rely on.

The other property is canonical uniqueness. Um as zero-knowledge proofs get cheaper, more and more claims will move off chain, right? You can prove that a credential is valid, that a construct was satisfied, that a computation was run correctly without revealing any of the data to the world and again without using a blockchain. But what a local proof doesn't show you is whether there's another claim existing somewhere else that conflicts with your proof. This is what blockchains uniquely provide, this canonical complete set. Double spend prevention is kind of the money version of this, right? If you hold coins in one address on chain, the blockchain prevents you from sending it to two different addresses at once.

But, the same pattern generalizes far beyond that. If you need to count all the votes in a DAO, all the clients in a market, or figure out all the rights attached to an asset, this is where a generalized virtual machine becomes critical, right? The same guarantees that you had for this money and double spend apply to all these use cases. And the third property is programmable trust. Um Outside of blockchains, trust is often categorical. You're either inside a platform or not, approved by a bank or rejected, and trusted enough for a a an interaction, or otherwise it won't happen.

Ethereum gives us a different starting point. Because you can you can inspect the commitments um and the rules ahead of time, you can trust in an interaction without necessarily trusting the counterparty. From there, trust becomes incremental, right? You can add reputation insurance credential or stake as you need it. But, having this trust is extreme is really what allows us to go from trust being this discrete binary gate to a programmable dial. And again, this matters a lot in an AI world, because AIs are going to create a huge number of one-off interactions that are either too small for lawyers to adjudicate, um too big to rely on existing social trust mechanisms, and too varied to rely on a single reputation score. So, each interaction needs a bespoke trust budget.

So, what does this look like in practice? Um one way to think about this is that Ethereum's affordances give builders new laws of settlement physics. It's like an environment for experimenting with structures that can't be built anywhere else. A way I like something I'd compare this to is space manufacturing, right? Um in microgravity, you're able to combine materials in ways that you simply can't on Earth, and produce things that just would not exist otherwise. Ethereum does something similar, but for finance and coordination. Flash loans are a really good example of this. Um in traditional finance, if you want to borrow some money, you need collateral, credit, legal agreements, relationship with your counterparties.

Um but on Ethereum you can borrow a huge amount of capital with none of that. The reason is that we have atomic execution, meaning that when you take out a loan within the context of a transaction, either you pay back the loan at the end of the transaction or the transaction reverted and it's like the loan never happened. This isn't just about making like a faster or bigger loan, it's about creating a completely new financial object that's not possible outside of Ethereum's settlement physics. I don't think flash loans are the thing that uh will make the world depend on Ethereum, but I think they're a valuable proof point to show that we can build things that are valuable and unique um that no one else can.

So, I think from here the frontier looks more like figuring out what are environments we can create that actually compose these three affordances in new and interesting ways. Um this is where Ethereum starts to become reasonably necessary. One kind of version this could look like is things like prediction markets becoming risk infrastructure for the world. A single prediction market doesn't actually need Ethereum, right? You can run it on a centralized exchange or a centralized platform and for many use cases this is good enough. But if you imagine markets as these composable primitives, something really changes, right? You can imagine a wildfire uh market becoming insurance collateral, a supply chain disruption prediction market becoming a hedge in a procurement contract. These forecasts effectively become building blocks that

you can build entire systems around them. And Ethereum's properties are really important here, right? You need to have a single canonical set of claims, have the commitments be durable enough for people to hold long-term exposure, and different levels of trust for whoever contributes data, pricing, or liquidity in the system. And this also gets back to agent coordination, right? On-chain agents will be contributing signals, pricing risk, routing capital, making markets, and just generally taking a ton of actions on behalf of actors in these systems. For each interaction, the system needs to know what what the agent is allowed to do, um you know, where that authority comes from, how much collateral is at risk, what history should be public versus private, and again, this is where Ethereum becomes really useful.

Agents are able to inspect these formal systems directly. They can read the rules, verify commitments, and decide exactly what's the level of trust they want to commit to each interaction at a completely new scale and speed than we could before. Because blockchain rules can feel opaque and hard to parse for humans, for agents though, they're really good interface as long as the system is credible. And this kind of ties back to what we already have on Ethereum today. So, even though kind of this last example was quite speculative, I think one thing is the Ethereum community is exceptionally good at turning these frontier research ideas into open permissionless systems with real value at stake.

Um and we already have a super solid foundation for this to exist in, right? If you take this this this example of these kind of risk markets, um all the claims that agents would settle and act on would would would happen through DeFi, which we already have on Ethereum. L2s give us the room that we need to scale this, and liquidity on chain can now be routed to a new type of Ethereum native alpha, which is risk priced at machine speed. Effectively, this frontier stack sits on something that's already proven to work and be secure. And all of this settles back to Ethereum, a neutral base layer where the parts that matter the most cannot be rewritten, permissioned, or ignored.

And so, after spending many years on the protocol side, this is what I want to spend my time on now. Finding the applications to that deserve to run on Ethereum at planetary scale and help bring them to the world. Because if we build things that merely just use Ethereum, then they will be unbundled and pushed to other chains, which won't give Ethereum much staying power. But if we build composable systems where Ethereum's guarantees are truly load-bearing, we start to have a compounding moat. Each useful primitive makes the shared system more valuable for the next one, and the network as a whole becomes harder to suppress or ignore. This is how Ethereum becomes reasonably necessary. Thank you.

>> [applause]