[Music] And for this one, I'd like to bring on Gregori to tell you about something that is going to be pretty fun. He's going to tell you that the feature of web 3 is not blockchain. So to have that full discussion, please welcome him on stage. Thank you. Thank you.
All right. Okay. I am Gregor Roshu. I'm um the co and founder of uh Pi Squared. And today I want to talk about an idea for the future of web 3.
A bit about myself. Uh I'm um um I got my PhD at University of California, San Diego and computer science. Then uh I joined NASA as a research scientist. Um then I started as a professor at the University of Illinois Urbana Champagne where together with my research lab designed and implemented the K framework which some of you may have heard it's being used regularly in a form of verification of smart contracts. Then in um um 2010 I started my first company uh runtime verification and then in uh and now pi squared.
I want to start by posing a controversial question. What if web 3 doesn't need blockchains? I know the idea that decentralization and progress could exist without blockchains might sound absurd in our industry. But given the blockchain's well document well doumented scaling limits, there is an argument to be made that web 3 may need something better than blockchains to thrive. What is clear and not negotiable is that web 3 requires payment systems, fast payment systems and verifiable settlement, ideally decentralized.
But blockchains are just one way to achieve those and not the only way. As we'll see, there is another way. Let's talk a bit about total ordering. So total ordering was definitely useful for payments, but we believe that for web 3 it's a limitation now because it slows everything down and in locks in architectural uh choices. While blockchains solved the double spending problem, it introduced its own architectural challenges.
the rigid fixation on the total ordering dictating that every transaction has to stay in line in a queue to be scheduled for execution by a monolithic consensus protocol. So this made a lot of sense initially in the context of payments where security and simplicity was very were very important. But now in the context of web 3 where we need complex applications with fast uh settlement that that itself this total ordering becomes a constraint. However, a few years ago, five years, six years ago, uh an fascinating protocol was proposed. Fast Pay.
Anybody here heard of Fast Pay? The Fast Pay protocol? Okay. It's literally one of the most amazing protocols that I've seen in the last several years. So it shows that you can have a payment system like Bitcoin but without any total ordering on transactions and essentially the speed of light and also without any blocks any grouping of transactions.
They all it's a simply an embarrassingly parallel protocol that allows you to settle payments at the speed of light and it inspired several systems lately. Linera in particular is built on on on this philosophy but also optimizations in sui and and and other and other chains. So what we've done fascinated by this protocol we generalized it and pushed it to go all the way to the end not only for payments but any computations of any kind. But let's debunk this common myth that total ordering is needed because I think this is a drag. This slows us down overall.
So what we really need is verifiability. Verifiability is more important than order. We trust that something happens because we can verify that it happened. And that's what matters. A payment happened.
Maybe a 100,000 payments happen in a second somewhere, but their exact order is not that important. And we can feel it in blockchain's growing pains these days where because of the total ordering we try to find other solutions to patch fix them and I don't want to go into any particular details but here I listed some of the recent amazing innovations but all of them still not addressing what in our view is the main problem the total ordering that is uh required Also, we see signs in the modularity movement. There have been a growing adoption of modular blockchains uh blockchain frameworks like Celestia essentially emphasizing the point that classical blockchains are too rigid, too inflexible. We also see data availability layers, execution shards or off-chain uh proofs or chain offchain computations. All of them trying hard to decouple blockchain's trusted validation from sequentiality from see the sequencing model because those can be advanced the other less so.
So does that mean that blockchain will disappear? No, I don't think so. But what I believe, what we believe is that blockchains will be just contracts, smart contracts if you want, sequencing contracts, okay, on a massively parallel network where you can do things at a unlimited speed, horizontal scalability. You can have some contracts that enforce a certain order, right? So those will be the equivalent of blockchains today.
And they will not enforce a total order on the entire universe. They'll only enforce a total order on those applications. They need a total order. Like unis swap, yes, it needs a total order. But many other applications, they don't need a total order.
Payments, transfers of assets don't need a total order. Voting doesn't need a total order. Auctions don't need a total order. More than 90% of the transactions on Ethereum do not need a total order. Then why have a total order?
So we believe that we are at a critical inflection point and also of course venture funds and defy protocols and Ethereum killers they are all financially and reputationally incentivized to keep the blockchains central to web 3. But um history has little mercy for technical incubants that cling to yesterday's models. And just as the internet outgrew its early um world gardens, web 3 is poised to move beyond the rigidity of the of the block based sequencing. And and and we believe that fortune will favor those who understand and capitalize on this inflection point. In case it was not clear so far, at Pi Squared, we are not building the next blockchain, what we are building is a new foundation for web 3 infrastructure for the next thousands and future chains and applications and AI agents, maybe hundreds of thousands, millions of them, um, where they could they they will be able to settle claims computational claims, payments, any claims that come with proofs quickly, verifiably and massively parallel in massive with massive parallelism without the bottlenecks of the traditional sequencing models.
So what pi squared unlocks and blockchains cannot. So first it will allow us to have parallelized trustless coordination between various agents because all the claims happen in parallel and very very efficiently. It will also allow fast verifiable execution including execution across chains because whatever happens on chains become claims that can be settled as well on the protocol because chains become as I said particular smart contracts applications on on the on the network. We are going to have instant claims no blocks. There will be no blocks.
Each claim will be validated and settled completely independently in parallel with the others. And these claims will be programmable. They are programmable. Actually, everything I'm talking about exists already. It's not hypothetical.
So these claims are programmable and through them developers can securely coordinate assets, data, logic across uh trust boundaries and uh without centralized intermediaries. the whole network is still decentralized and verifiable and also agents and AI native um infrastructure enabling hopefully soon not hopefully that definitely will happen soon we'll have billions of agents all of them wanting to do things at the same time they cannot settle all of them on any existing blockchains these days and with this I would like to introduce our um most recent paper and protocol, the fast set protocol that we just released this week, which is an improvement, a fast pay, as I said, with arbitrary claims. Let me tell you how it works because it's really interesting and completely different from blockchains as I said. So you see there are steps 1 2 3 4 seven steps. Okay.
So step number one and number two they are very related. You have to produce a claim and to verify it. Okay. A claim can be a transfer of an asset from Alice to Bob. Can be a computation in some programming language.
Can be a mathematical result. Pythagoras theorem. can be literally anything that you are Joe Smith for example anything that has a proof a proof which is verifiable okay that's what a claim is now that proof has to be verified by somebody okay so there is a network of verifiers these verifiers are specialized for that kind of proofs for that kind of claims that the application accepts those can be for example um verifiers of ZK proofs, right? If the claim comes with a zero knowledge proof and this is also something we do in our company actually pi squared comes from proof of proof. We do zero knowledge proofs of mathematical proofs.
We have our own way to generate zero knowledge proofs for computations in any programming languages uniformly with one circuit without compilers without translators without anything. But this talk is not about that. This talk is about having claims that come with proofs and those proofs can be using any technology proving technology including executions of programs re-exection of programs as it happens on blockchains. So now you have a claim and a proof and that has been verified by a quorum of verifiers. Now what you want next is to settle that claim.
Again keep in mind that claim can change the state can be an asset transfer. Okay, can be a program execution that changes the state. Now we have to settle that claim. There is a decentralized network of validators and sync millions of them because their job is not very hard. Now they will take these claims and their job is to validate the claims.
They will check signatures. Say yeah the right verifiers verify the right sender sent the claim. They verify, they validate the claims and now in theory they would have to settle to change their state but they are not ready because they don't know if you are cheater or not. So what you need to do is to get quorum from these validators right. So you validators check the claim send it back to you or your proxy then you aggregate signatures and once you have a quorum you send it back to the validators.
All right. Now validators received this certificate that says yes this claim is true is valid has been validated by a majority of us and we trust each other bizantly speaking and now it's time to settle. Now you can settle. However, you cannot settle right away because there could be some causal dependencies between certain claims. Okay.
And because claims can be there are messages ultimately they can be reordered as they are being sent then you may not be able to settle certain claims yet you may need to wait for other claims to come first. Okay you wait. So that's why there is a seven a six step presettle and then seven you just settle the claim finally and that's it. And look what we got here. We've got a decentralized network of validators.
You can validate any claim and settle any claim which means in particular any program in any programming language for which you have a proof. So it's not no longer an EVM or a wasome. It's not even a chain, right? There are no blocks, no chaining. It's just a set of claims.
All of them being settled in parallel at the speed of light. All right. So this is not just a concept. This is real. It is live and ready.
We just released our we just launched our devet last week. So feel free to scan this uh QR code and uh join our devet play with it. Of course, we do not yet support all programming languages and everything, but there are several applications that you can already uh play with. And I would like to take this opportunity to announce that uh we plan to launch our mainet by the end of the year together with a flagship product that will be essentially an asset transfer like a wallet but with any assets which will also be connected to different blockchains. Thanks to our um friends at wormhole, we have a collaboration with them.
We'll be able to bring liquidity on the network, not on a chain, on the set, on network. And um this wallet will allow transfers um at the speed of light. That's the hope by the end of the year. Thank you.
Thank you. [Music]
Automatic transcript — names and jargon may be misspelled.