Fast Ethereum

Transcript

[music] [music] Thanks. All right. Hey everyone. Great to be in Mumbai. Good to see old friends and new friends as well. Uh I want to talk about Fast Ethereum today. Um tell you a bit about our vision at TF Labs, why we're pushing on on this um specific topic. [snorts] Um so yeah what are we trying to do here? So when we started EFlabs uh five months ago we said our mission is to make Ethereum the settlement layer of the global economy with EH at its center. So E is a very special asset powering this

entire economy on top. But what does this mean exactly? So I just want to unpack a little bit uh what we mean by that. To me really it comes down to looking at the blockchain as a place where we can securely hold uh property rights or assets. And why is it secure is because it's secured by Ethereum which is this network that never goes down that has validators all around the world that has full nodes all around the world uh and that is always accessible by anyone. And it's not enough to just like hold assets and property rights on there. The thing that gives it the the settlement quality is that you can do operations on these assets. So you can transfer them to your friends or people who buy assets

from you. Uh you can swap them. Uh you can do securityization, you can create new assets out of the base assets that already exist. Uh so there's a lot of things you can do on top of that programmable um settlement layer. And all of this runs with code which is verifiable. Um so you know that the operations are just going to run according to the way that they were committed. So it's a credible uh computer credible settlement layer. Not all assets are equal. This not an Ethereum thing. It's just how it is in the in the world. We make the distinction often between bearer assets which are the ones where just holding the asset it makes you the the owner

just like the the possession of it means that yeah the deed is yours and the asset is yours. So if is is such an asset you can think of it as not your keys not your coins. So if you have the keys the if on there is yours. Um and you have also registered assets. So like a house which means that these assets you hold them but they are registered in some kind of database elsewhere and this is what really holds the the ground truth of um of the ownership and so Ethereum utility as a settlement layer I think it's quite useful in fact for both types of assets it's maybe more obvious for the bearer assets because there the chain itself is the ledger of record and so that's really where um you

are going to have the operation s and the balances of a users. um in particular for if um it's in my opinion also quite um quite useful even for registered assets to have this access to the to a settlement layer because even if they live perhaps on another place they're registered like in some government entity or some private corporation having access to an open programmable blockchain that never goes down uh gives them access to composability interoperability livveness so Yeah, just having uptime.

And so all of that is is is useful. So the vision is to to get to this world where Ethereum is really rooting the world's assets um on its network. So how do we get there? How do we make Ethereum the best uh settlement layer? I think first we really need Ethereum to be useful. Like there's no moral reason why people should move to Ethereum. I think there's a good business case for it. uh and the business case is rooted in the fact that Ethereum is very permissionless, that Ethereum is the most trustless platform that we that we have. But the useful um applications of it is there's a there's a couple. I'm just listing some here. One of them is

self- custody. I think self- custody is inherently useful. As a user, I want to have um yeah access unmediated access to to my assets or my data, my information. Um, having access to more assets and markets is also useful. There's still a lot of um, gates that prevent people from owning certain types of assets depending on where they are on Earth. So, opening access, I think, is is a really high value thing that that Ethereum offers. And financial freedom is also useful. I think being able to invest in what you want, when you want.

I think these are all things that are very yeah give users a lot of of utility. So utility to be a bit precise you can think of it as it's the value you get from a service minus the cost uh you're paying for that service. So the value could just be like returns or some kind of yield or just your inherent happiness from using that service. And then costs look like counterparty risk. So if you use more trusted platforms, perhaps the rug you can be counterparty risk. Uh you have duration cost. Sometimes your capital is locked up. You have to pay some yield on that. Uh that's a cost. And then you also just have fees on the on the networks when you move assets around. And so that's

also a cost. And so the mission to me should be to increase the utility of the network and make it um yeah give our users more more value. Um and so yeah, if I synthesize a little bit the the three uh points before, what we're really trying to get is fast, cheap and secure settlement. And what I want to focus on in this presentation is the first one which is uh being fast. So as a as a platform, as a network uh and moving at the speed of the of the assets.

So why does it matter to be to be fast? Um I think when you're slower than alternatives, um users have more of an incentive to use these alternatives which are often less secure or third party more mediated alternatives than Ethereum itself. So blockchains are built to be essentially confirmation engines. you give them data, they confirm that the data was received, that it's part of some kind of canonical history and that it will never be reverted. And so if you're not using the blockchain, the Ethereum blockchain uh for the confirmation service, it's kind of missing the point, right? Like we we're building this really complex

system that's like highly secure, but if people find that, well, it's not fast enough, it's too expensive, and they're using something that is less secure, I think this is a this is a problem. And you can see that for instance when you when you look at bridges so the way to move assets around let's say from L1 to L2s or from to centralized exchanges um of the time they don't really depend on the actual Ethereum confirmation they just have a quorum of reporters that tell them their view of the chain and it's essentially a big multisig and these multisigs we've seen we've seen hacks and these hacks are I think pretty bad press for for our users we can do better I think we should be aiming to offer

Ethereum confirmation which is finality of a proven valid state using ZK proof. So we're building towards that future. It's going to come but we really need to accelerate it. Um and if we are slow I think this may be moot. We can build the most amazing infinite machine but I think people might still not depend on it as much as they as they should. And so to me, if we can be faster without breaking the protocol, obviously uh this is a user-friendly choice and we lower the cost of of choosing the secure option. And I think this is really something that we should have in mind doing this work that it's not just being fast for the sake of it, but it's actually being fast because we want to give more value to the security that we offer.

So yeah, how do we how do we do that? Uh [snorts] there's a couple of things we're working on at if LabBabs. Uh the first one is fast slots. So you know that on L1 slots today are 12 seconds long. So you have a new block every 12 seconds. We want to make that shorter. Uh there's many benefits to that. The first one is just improved UX. When you are a user sending a transaction, often your wallet kind of confirms it when it sees the transaction being included in the block. So that is faster, just nicer UX, very relevant. Um, also much better markets. If the chain can produce blocks more often, the prices on chain, all the oracles, they refresh faster as well, which means that you're leaking less

value to the outside world to arbitrageers that pick off from the stale quotes. And in fact, you even have a security argument. If you are not going so fast that you break the chain, um, you're just adding like more proposals per unit of time. And so, you increase the censorship resistance, you increase the opportunities for users to be included on chain. And I think people underrate a little bit the benefits of having uh more blocks for censorship resistance. So the aim for us is to push for 10 seconds from 12 to 10 in in 2027 as a as a first reduction and then gradually reduce. Um my aim would be all the way until 2 seconds less if we can but I think 2 seconds would already be like a massive improvement.

Another threat of work is fast finality. This may be like a little less um something that the end user can directly see but many systems for instance real world assets rely on finality as a confirmation to to take action. So they just wait for the ultimate Ethereum guarantee that the chain has stabilized and that they can commit the state of the chain on their own offchain ledgers. And so many of the systems that Ethereum interoperates with are blocked on the on the finality latency. Today finality is about uh 15 minutes. The aim I would say for 2028 is to get that number down to one to two minutes and then keep pushing until we get to 10 to 20 seconds um max.

So that will take time and in the meantime uh something that we've been pushing quite a bit uh something that was worked on in EF with with the core developers of Ethereum 2 is the fast confirmation rule. So it's live today. It's essentially a slightly weaker confirmation than finality, but one that still has very very high security guarantees and one that we think is a good option for many systems that don't rely on finality, but want quite uh strong confirmation before they process um some kind of action. And the upside is that even though you weaken the security a little bit, you in you decrease the latency by a lot. So this is 30 times faster than Fidelity. You can get this confirmation in 13 to 24

seconds and yeah many systems can use that. It's already been deployed on Gnosis chain to decrease the the latency from Ethereum to Nosis chain. Um it's also going to be part of chain links packages of of bridges. So yeah we we'll continue the roll out and we'll aim to have like more um yeah more systems in Ethereum adopting the fast confirmation rule so they decrease their um their latency. Maybe a quick point because I just said well this confirmation rule decreases deposits out of Ethereum. Why is that good? Like why is that good that the money is flowing out of the of the L1?

And the way I would put it is is this. So L1 is to me the premier venue for the world's asset. It's the home of EF in particular. Um every other venue than L1 has higher counterparty risk. doesn't have like the same trustless guarantees as Ethereum L1 does. And so the when you issue assets on Ethereum, when the root of trust is on Ethereum, uh there that's the only place where they can maximally enjoy the properties of Ethereum, which is censorship resistance, uptime um and safety. So no double spend.

And so whenever you move assets from L1, you you never really lower your risk. But it might be useful like there's plenty of places in the world where your assets can enjoy like more utility outside of the L1. And so that that may be a good reason to move them. But that move will always like increase the the risk. And so for assets that are deployed on Ethereum that are issued on Ethereum, it's not really that you're deploying them out of the network, but it's more that you're deploying them from Ethereum. So they're still rooted in the L1. They increase the risk when they travel outside. Um but that increases the utility of the asset itself and so that means the upside to me uh the L1 just becomes a more attractive place to issue hold and use

assets. So in particular it fulfills possibly many functions. The first one it's a bit like DTCC. Um it becomes the record keeper let's say of tokenized assets. I think this is a role for V1 that's really uh relevant. You can also think of it as a savings account. It's a place where you just like hold value longterm and you acrew some yield on it. Maybe not like the highest yield but a strong yield. It's also a place for wholesale money market. So you have like all the L2s that are connected to it, they all need liquidity. So that means that they need some kind of facility for for swapping. And so you do have very deep pools uh on the L1 like very high financial activity there uh and and deep liquidity. So it fulfills let's say this

um triple function. So yeah, [snorts] these are some arguments for why being fast is good, how being fast like improves the quality of Ethereum as a settlement layer, which is one of the goals that we that we want to achieve. Um but I wanted to yeah as I was making these slides talking about fast ethereum I was also like considering a little bit where the zeitgeist and the conversation was heading and since the last few days there have been a lot of discussions around well we have to talk about postquantum like why are we talking about fast Ethereum when we are thinking about like this very existential threat to the nature itself of of Ethereum and so I was thinking let me do like a little bit switch and change a bit the

tone and the vibe here and move to maybe something a little a little like a u modification to to fast ethereum and call it instead urgent Ethereum. And so here [snorts] what I want to say is um security is like still the fun the fundamental the base of a masslo pyramid of needs of Ethereum. And I took this screenshot from a talk that ya gave recently. She was saying, well, [snorts] people need to understand why certain decisions are being made, why we build Ethereum the way that we build it, and in particular, why Ethereum can't offer the fastest transactions. And I think it's it's relevant that she's using this this

example because it's clear we're not going to say, you know, we have to be fast because we have to compete head-on with like all of these other chains that offer like down to millisecond uh speed. We made choices. We made choices that are relevant for the value proposition that we have, which is the most um secure platform. And [snorts] if we didn't make this these choices, we just would not have this value proposition. And so we don't need to be the fastest. But in my opinion, we still want to be fast enough for our settlement to win and for our users to use our settlement and to to benefit from the the properties of of Ethereum. And so yeah, so I think we are aligned on on security still being um the most important thing

that we that we need to preserve. That doesn't mean we can't be fast. Um at the same time, there are yeah two things that agitate the community quite a lot at the moment which is AI and postquantum. It's clearly like a moment that we have to to meet and we do have a plan. I think the plan is is very well thought out. We have a lot of the pieces. I first maybe if you haven't followed I think this was quite all over the place but why is this a threat to Ethereum so postquantum if quantum computers come in they can break the cryptography that we use that's obviously like existential uh AI has become kind of a new risk like we've seen AI just becoming more and more

capable and so now the question is can very smart AI algorithm find um let's say um fault lines in the in the cryptography that we use and and exploit them and possibly break them. And so it's really existential that we that we meet this moment and that we make the network resistant to these um two threats. But to me, the only risk on that path is limiting our ambitions to only meeting u to only like being safe from the PQ and the AI threats. I think we can meet them like we we we're doing a really good job here. Um but we should also be clear that other chains are prepared. Um they incur a much lower

cost to migrate chains like Solana, like Monad, I think it's pretty clear that they have strong plans as well and they will keep shipping in the meantime. And so I think even though it's ambitious to meet this moment, it's kind of table stakes and we should really make sure that we keep pushing uh at the same time. And and so whatever we do, I think it's important that we meet the moment with urgency. And so yeah, perhaps my statement here is moving with urgency but without panic. To me, that's maybe fast Ethereum more than the core properties of the of the protocol. It's more of a a state of mind for the social layer like feeling that we can really take on this

challenge that we can take it on with with urgency and be fast to deliver like all the things that our users need which is both security but also utility a couple days ago. [snorts] Sorry that was yesterday actually time goes really slow because there are so many conversations. Uh yesterday Justin went on stage to to present mission 2028 which I think is a really strong statement to say that um we will have an AI and PQ secure L1 by by 2028. When I look at the view of let's say core developers, people I'm working with the community uh people who are building on Ethereum as well I see that yeah we

basically have two years to meet three moments perhaps. One is tokenization. The other one is AI. The third one is postquantum. And I'm seeing honestly like AI just moving at breakneck speed like making research and engineering challenges feel like child's play. Things that we didn't think was possible uh maybe 3 months ago. Making progress on well specified problems but also on not so well specified problems. Like I think it's been like [snorts] a revelation to me just how fast we can move um with this technology. So our new consensus design has been formally verified like extremely quickly in a speed that would have been unthinkable I think just six months ago. And I think the the magic here is that we still have

humans at the wheel like we still have these extremely talented developers and researchers. But really like getting them to to move with that urgency to me is like mission or at least priority number one. It's not a silver bullet like people also assign a lot of hopes to AI that maybe don't fully pan out but I think it is our chance to ship both a more useful and a secure Ethereum at the same time and so to mission 2028 I would add uh let's also ensure that we ship both an AI and PQ secure L1 we can do this I think we have a very strong plan but let's also make sure that we ship all of the other things at the same time so faster slot moving from let's say 12 to 8 seconds by 2028. That seems doable.

Faster finality from 15 minutes to 1 to 2 minutes. That seems doable. Scalable Ethereum scaling 5x from where we are today. I think that seems doable as well. [snorts] And privacy, which I think is also a huge um feature for for our users. We can do this. So, let's get down to it. Yep. Thank you, [applause] Thank you so much sir. Amazing highv value session. Uh obviously like we discussed we do have quantum questions.

Uh but I'll take it up later. So the first question that I want to ask is in a world where this happens we move very fast and L1 itself is fast. What happens to roll up? >> Yeah. Um I mean as I was saying in the talk my ambitious target for like how fast L1 can move is still above let's say 1 second and most rollups that are considered fast are already like at 100 millisecond if not less. So I don't think anyone will ever catch up to the speed that rollup can move at. I don't think it should either. I think it's fine to leave space for rollup anyways.

It's not space that we can claim. But speed isn't the only value proposition that roll-ups can offer that the L1 doesn't have. Like roll-ups are also excellent go-to market machines. Like they are often pegged to some kind of company that launches them. That company has distribution that makes the roll-up more useful. So I think there's still like many many ways that roll-ups, L2s, L1 apps can innovate in ways that the L1 really cannot. So I don't see a future where we are going to be able to do everything we want to do on VL1. It's going to be multi-chain. It's going to be heterogeneous and we should just embrace it.

>> Perfect. Thank you. I was actually pushing the postquantum question down but people are interested in it. So I have to ask uh postquantum signatures are far larger than today's BLS signature which means more data to move every slot. Does quantum migration put a ceiling on how fast Ethereum can get? >> Yeah, that's okay. That's like the crazy thing. So quantum signatures are 10x larger and so yeah, naively you would think you know that means we cannot move 10x faster. We should maybe even move 10x slower from where we are today. So there's there's a couple things there.

Um first I think we are just continuously making progress on our networking which means that we are able to push more data through the network. I think that's really great. We should be doing a lot more of that. The cryptography itself is making a lot of progress and postquantum signatures have properties that BLS signatures don't have that make them also a little bit more flexible. So we're leveraging that. That's also great. But yeah, [snorts] beyond that we're also like even thinking about changing the architecture. So that's decoupled consensus. Yan was giving a talk about this yesterday. um we're able to think of ways to make the consensus move at the speed that it wants for the slot

time. So produce blocks as fast as possible while the postquantum cryptography does its own thing on its own separate thread. So we're able to like optimize independently these two things. Like it's a really ambitious upgrade, but it's something that now we have like pretty good specs formally verified protocol. We're just like yeah rushing to to deploy it. So we can really have our cake and eat it too which I think is like wonderful. Yeah. >> Perfect. Thank you so much. Uh less time to build each block favors the more sophisticated builders. How do you stop speed from centralizing block building?

It's sort of connected to the next question as well. I mean uh if we are racing towards faster and faster Ethereum, does it normally lead to a less decentralized network? Yeah, I think all LSQL obviously like the faster you go the more tight the deadlines are, the more tight the constraints are. There's some arguments you know you can make you can reduce the gas limit which we would if we reduce the slot time you would reduce the gas limit per block. You would keep the same throughput but you would just like make the blocks uh individually smaller. [snorts] I think again we we can do it. So we're that's the reason to be iterative with this to try and push the slot time like lower and lower over time so that we can

also test and understand the effects of a network like whenever you do something I think also the network just adapts like it's a very emergent uh organism almost this network so when when you push it a bit more to its limits people will figure out ways to be smarter with their nodes to be smarter with their resources um so yeah I'm not so worried about this besides I don't actually think speed of building a block is really what makes the difference for block builders today. I think it's more access to order flow and other proprietary um maybe algorithms, but speed to me doesn't feel like the most relevant input here.

>> Cool. Uh just short question. I mean are precons dead? >> Um I don't think they're dead. I think they're just really hard to do on VL1. So I'm not surprised that yeah, they haven't really found traction there. I often have to explain why I don't think precons are the answer instead of let's say shortening the slot time. I think they're still just like very difficult system to to to make work and the validator of Ethereum is the valator set is designed to be permissionless. So it could be anyone on the other side and I think building a precon system on top of that permissionlessness just feels to me like very difficult in principle. So yeah thanks >> [music]