Scaling Ethereum with based rollups | Piergiacomo Palmisani | ETHWarsaw [4]

Um very excited to be here. Actually my first time at if Warso. I gotta say I love this format. Uh way better than like the big conferences. Everybody goes there.

Nobody listens to the talks. So very excited to see a good group here. Um I am Pier Jakom. I'm head of ecosystem at Tao. I go by PG.

It's way easier, way shorter. Um, so today I'm going to dive a little bit into what base rollups are. Um, some of the cool features that uh we shipped recently, uh, including pre-confirmations, actually base preconfirmations. I'm going to expand on what that means. Um, and also one piece of R&D that we're very excited about is called Gwennneth that is going to bring synchronous composibility on Ethereum.

Um, so without further ado, uh, let's get into it. Um, this is the agenda. Um, why does it matter? Some definitions, some some jargon, I guess. I'm going to dive deeper into the protocol, pre-MPS, and then Gwennneth.

Um, so why does this matter? Um, we've seen pretty much all layer 2 so far have been running uh a single sequencer. Uh, some may call it a centralized sequencer. Um, it's without any doubt uh a single point of failure. Um, I want to start by saying that it's okay uh to make compromises.

Uh, blockchains are all about trade-offs. Uh, there's no way around that. Uh, the important thing is realizing uh that this cannot be the status quo forever. Uh, and that something can be can be done and needs to be done. um so that we can actually uh preserve the core properties of this technology that we all love uh namely censorship resistance, livveness, security uh and so on.

Um you cannot do that with a centralized sequencer uh unfortunately. Um and arguably also uh with Sure sequencers. Uh there are some projects that are tackling that uh namely espresso. Um that can be one approach. Uh although there are some uh tradeoffs there as well uh namely livveness but also you got to trust uh a new set of infrastructure.

Um so why not trusting the L1? Um what uh pretty much a base rollup does uh is it offsets uh uh sequencing uh to Ethereum L1 validators directly. Um this of course brings all the great properties that we all love about Ethereum uh to layer 2 um while preserving again like uh decentralization, security and livveness. Um why single sequencer can be bad you may ask? Uh well there have been uh um some practical uh situations uh for example uh the sequencer running out of gas hence the whole chain alting uh and then developers having to manually top off gas so that the chain can start again um I guess it's okay in the short term uh but if we're talking about like bringing trillions of ch on chain uh bringing like mass adoption uh hello welcome welcome um this cannot be the status quo, right?

Uh because it's just too bad. Um single sequencer can also censor transaction. Um some have force inclusion um it baked into but not all of them. Uh and this is of course a problem because uh uh if uh if the sequencer censors your transaction then you do not have any guarantee that uh your transaction will actually be included in the block which is of course a big problem. Um and last but not least um centralized sequencers cannibalize L1.

Um what that means is that um they get majority of the fees uh without giving back much to Ethereum. Um I have a pretty telling slide at the end that shows how much Tao paid to if uh unfortunately some of that was also because we had to optimize the protocol. Uh but um nevertheless um you cannot just like get all the good stuff from Ethereum without giving back. Um also um we have so many L2s. Uh liquidity fragmentation is obviously a problem that we need to solve.

Um what we're doing especially on the Gwenf side is enabling synchronous composibility. Uh this of course is uh um something we can achieve by having all L2s becoming based. Um I'm going to dive deeper more on uh on Gwenf later on. Um there's also other solutions I guess that have been explored uh a layer super chains where you know they're aiming to try to solve this liquidity fragmentation issue. Uh but why not using the L1 right like we have it it's battle tested we just celebrated 10 years anniversary 100% uptime uh all the great properties of Ethereum why not using that right why do we have to like build a whole new overlayer of complex infrastructure that we need to trust um so yeah this kind of like why matters and I have a spicy slide uh this is a tweet from one of the co-founder of Celestia that argues that uh it only that single sequencers only sacrifice censorship resistance.

Um I just want to like make sure that we are on the same page that um we're doing this because all these properties actually matter um because if we just say that censorship resistant is not important um then why not using servers and just you know doing web two. Um so yeah I think uh we should you know hold oursel accountable to the higher standards in the industry because uh otherwise uh again we just go back to web two and just use that right um so I want to make sure that you know this message uh uh comes through very uh very strongly um cool so um just some highlevel definitions uh on base rollups um and roll-ups in general in general um so if you settle on Ethereum, you're a layer two. Uh if you use Ethereum uh as DA, uh you're a rollup. Um if you use Ethereum uh as a sequencing layer, uh you are a based rollup. Uh and if you use Ethereum uh as execution layer, uh you're a native rollup.

Um the concept of base rollup was introduced first by Vitalik uh in uh uh I think it was 2021 uh in a paper called total anarchy. Uh it was picked up and developed further by Justin Drake uh which is you know also one of our advisor uh and has been uh a key person in for us to develop the protocol and all the features that we've shipped so far. Um so yeah we kind of like took all those concepts uh and built them uh and so actually Tao is the first base rollup uh that is live today fully functional uh and we've been so for a little over a year. Um so what are the superpower based rollups? Uh well as I said um we inherit all the core properties of Ethereum um the good and the bad.

Uh here are the good. I'm going to get to the bad later. Uh but you might guess what they are. Um so of course decentralization. Ethereum is the most decentralized L1 that we have today.

Um for obvious reasons. Um livveness. Um livveness is uh definitely one of the most important properties of a blockchain. I would argue even more than decentralization which is something that we cannot really quantify in mathematical terms. Uh whereas livveness is definitely something that we can quantify.

Um security is another one. Uh Ethereum has is again uh the most secure L1 uh as of today. Um it has never gone down. Um apps on Ethereum have been hacked. Uh we've seen the Dow hack.

Uh but since then we can argue that uh if is fairly uh secure and strong. Um neutrality and censorship resistance. As I said, um there's no way um to censor uh on Ethereum. Um and by again using L1 validators as sequencing layer, this is a property that base rollups inherit overall simplicity. Um we uh tao itself is what is called a type one or aum equivalent.

uh which means uh um we get all all the if um properties infrastructures and smart contracts out of the box. So it's very easy for developers uh to deploy. Uh there's no workarounds, there's no up codes that you need to like tweak. Uh it works just straight out of the box. Uh and as I said, synchronous composability, this is a very important point.

um as you use L1 um you can like synchronously compose uh between different uh uh base rollups uh and again I'm going to expand more uh later on on the Gwennneth side of things. Now the drawbacks uh well we all know that uh Ethereum is low. Um unfortunately we um we we had to chew a little bit of glass for for a year or so uh because Tao inherited Ethereum block time which is 12 seconds and actually on Tao uh finality was sometimes even longer. Uh we solved that with base preconfirmation which is something that we shipped uh last month. So very excited which brought finality down to 2 3 seconds and we're actually aiming to go uh further down to ideally u hundreds of milliseconds that still TBD and still a piece of R&D.

Um base rollups are inherently more expensive. Why? Because we don't run uh again a single sequencer which makes operation way cheaper and easier. um we have to like make it up with uh L1 gas plus uh priority fees which of course get uh everything a little bit more expensive but it's also an economic incentive for L1 validators to opt in and become pre-com um so this is actually something that uh is a drawback a little bit now uh but is actually something that can be a plus uh in the future uh and yeah we're less profitable in the sense that we do not capture um the fees uh we give it back to to EH for the majority. Um so yeah, we don't make tens of millions of dollars per year.

Uh but again like we again this is why we're Ethereum aligned and we you know we want to give back to what is making us what what we are. Um oh another thing to mention uh we do not uh capture me. Um so that's also why uh we're less profitable because all me goes back to L1. Um okay so what is Taco Alifia? Alifia is actually our latest implementation of the Tao protocol.

Uh basically is a set of smart contracts deployed on on EEL1 uh that enforce execions rule manage proofs and facilitate rollup operations. Um in simple terms uh we don't run any consensus. Uh we're again just a set of of smart contracts on L1. Um we use Ethereum. Um and this makes everything fairly easier to to kind of like optimize and and build on top.

Again, as I said, full Ethereum equivalent type one. Uh very easy to build. If you have an app on on Ethereum or any other uh L2 that I guess is Ethereum equivalent, you can just like redeploy it with like minimal overhead on Tao uh which is fairly convenient. Um no centralized sequencer. I've talked about that uh a lot.

Uh multiproving architecture uh ZKPs and TE's. Um we're very eager to see realtime proving coming out soon. uh very excited about uh some of the work that it's been going on in in in zero knowledge uh with NARX but also namely like ZKVMs u very excited to see some of the teams there and we've actually been working with them since since day zero um some of them are uh risk zero um succinct we've been using their infrastructure for for quite some time so um very excited to see what they've been shipping uh configurable proof system uh what that means H you can actually like switch between ZK uh and optimistic uh as you may although we are aiming for um full ZK coverage at the moment on tode ZK coverage is about 40% mainly for costs uh but also uh because again like there's still a little bit uh of optimization that needs to be done uh on on on the zero knowledge proving side um decentralized proving and proposing anybody can become a pro uh a proposer or approver on Tao. Um it's pretty trivial. Uh if you're interested in doing so, we have full documentation on our docs.

Um the hardware requirement is not crazy. Um also in terms of like uh um stake, we're at 25 tao at the moment. So it's pretty pretty straightforward. Uh and again like that's why it's it's decentralized and and permissionless because pretty much anybody arguably can can opt in and and participate. Okay.

So, uh this is kind of like a little bit of uh of tao uh economics how how the protocol works. So, uh we start from a user uh that wants to submit a transaction. Uh user pays for a transaction. Um the uh an L2 transaction fee uh that gets split in two priority fees. 75% of the base fee goes to the block proposer on L2.

uh 25% goes uh uh to the Tao uh Dow treasury. Uh this is also to uh make sure that the projects is sustainable uh that the community can have the upside uh from all the activity going on on Tao. Um the block proposer uh then uh pays L1 fee so that the block can be submitted to Ethereum L1 uh and also pays approver fee uh to the block prover. uh block provers the block, pays um fee to to generate the proof uh sends back the proof uh and submits the proof to to the L1. As I said, livveness bond is 25 tao uh and current zero coverage is about 40% but again we're we're aiming for like full ZK coverage.

Hopefully again in the next few months again this is not just depending on us but on our ZK partners but uh ideally this will come soon and again we all know ZK is the endgame uh also for scaling Ethereum and and the L1 itself. Um so again very excited about like all the all the development that is going on there uh and feeling lucky that we can leverage this this amazing technology. So yeah, this what I was saying. Uh we're a great tenant on Ethereum. Since inception, we spent uh $12 million.

Spent is a bad word. We gave back to Ethereum uh $12.5 million. Um so as you can see, actually this is uh pretty telling. Um you can see the nonoptimization of the protocol that then gets optimized.

Uh and then we ship pre-confirmation and that's why you know we pay less fees. Um before we were kind of like uh when especially like before blobs and then after blobs were implemented uh we were submitting some empty blocks uh some things that were not super optimized. Um so that was kind of like not ideal but again like blockchains are trade-off. You have to chew some glass uh to get there. uh but our our choice was to again go the right route in the sense of like uh uh keeping all the the properties alive while spending and sacrificing on UX and so on.

And actually currently even though we're definitely not even close to like all the activity that are going on that is going on like maybe some of the other like bigger chains of course B is is blowing. So, um they're they're the first one. Um but you can see that we're actually second right now in terms of like um rent paid to to L1. I don't know if that's a bragging thing or not, but yeah. Um yeah.

So, a little bit I guess less technical stuff, but still like very important because sometimes, you know, there's a lot of like jargon and technical gibberish and then you say, "Okay, but who the handles that and who's behind the protocol?" Um so you know we're we have a community um these are people that start discussion groups uh we recently launched a DAO so uh kind of like traditional model start a discussion going that becomes a proposal uh and then gets passed or not uh we have taolabs that develops and maintains the protocol and does all the R&D stuff uh we have tao foundation that starts with like ecosystem growth um and then uh we actually um recently finalized uh our security console uh which is going to get us closer to be a stage one rollup. Uh very excited. We have some great folks uh from Nethermind Aragon and some other um Ethereum Foundation uh some great developers in the ecosystem. So very excited for that.

Uh and you know looking forward to become stage one pretty soon. Okay. So let's get more into pre-confirmations. Um so on the right side you can see uh the glass chewing I was mentioning before. Uh that was uh what Tao felt before uh pre-confirmations.

Uh and on the left side uh you can see what Tao feels like with pre-confirmations. Um these are mascot we call it blobby. Um but yeah that's what it is. Um, so just I don't know if you guys are familiar with pre-confirmations. This is actually uh something that uh is uh pretty widely uh used uh in our day-to-day is not at all a crypto uh thing.

Uh it's actually something that you experience every day whenever you go, you know, pay for a coffee uh your favorite Starbucks. Um the transaction doesn't actually get finalized right on the on spot. Uh what happens is that um your credit card connects to your bank and then your bank uh kind of sends you a pre-confirmation that you're actually you have the money to spend. You spend the money and then uh sends um that to to the to Starbucks bank. Uh but the actual transactions get confirmed you know hours later sometimes days later in web two word.

Um so this is exactly what goes on in the background. we just don't feel it because the web two experience is just so easy, right? And so seamless. Um so the goal is to get there uh with again all the properties that we care and love about. Um actually um there's also a lot of like uh um chains that today use pre-confirmation.

Uh the difference is that they don't use base pre-confirmation. So um you still have to trust somebody in the background uh some middleman that um your transaction is actually going to go through. Uh with base preconfirmation you kind of get the best of both world. You get UX as I said right now uh we were able to achieve uh between to two and 3 seconds finality uh which is pretty cool. um trustlessness um as I said like L1 validators can become pre-comers um so you know you can trust them and which means there's no trust assumption to begin with uh profitability uh precoms can be uh a great economic incentives uh for validators uh which again is an additional revenue stream for them uh which again of like ideally will make a lot of validators opt in uh and choose to provide this service.

Uh backwards compatibility um as we ship these pre-confirmations uh there was like uh no overhead, no workaround that needed to be done um by developers applications. Uh it just worked right out of the box. Um as I said, censorship resistance. There's no one that can censor transactions. um and future proven.

Uh we're building this for for days to come which is uh again something that cannot be changed or or hacked or um yeah again uh best of both world decentralization as decentralized as Ethereum performance as fast as centralized chains. Uh we are aiming again to get to the finality of Solana. uh which by the way uh this is something I should have mentioned earlier uh pre-confirmation is something that uh is baked into certain protocols uh certain L1's uh for example Solana um why because the hardware requirements is much higher uh hence it's way easier uh uh for validators um to offer this service um not as easy on Ethereum uh but that's also why uh Ethereum is as decentralized and as permissionless because it's so easy to run validators and infrastructure on EF. Uh but again with with this upgrade we're bringing the best of both worlds and ideally and hopefully we can get to Solana performance uh uh fairly soon. Uh yeah, I guess uh I went over this um trustlessness uh and economic security uh economic security uh because um you are putting uh stake uh to become a pre-confer um and if you don't uphold um the pre-confirmation guarantee that you give to the user then you get slashed uh which of course kind of like uh sets the rules uh to avoid malicious actors um while again like preserving all all the other properties.

Um this is a little bit of our road map uh on precoms. Um again as I started saying there are trade-offs in blockchains. The phase we are right now is a permission precom phase. Um I'm don't want to hide it. Uh why?

Because we need to test the software. Uh we need to test the mechanism. We need to be sure that this works. Uh, and we actually have some great partners doing that with us. Uh, namely Chambon, Garaka, and Nethermind.

Uh, I don't know if you guys are familiar, but these are people that have been coreum developers uh, pretty much since day zero. Um, some have built applications, protocols, some are block proposers. Um but yeah, these are kind of like our free um main partners that are offering uh the pre-confirmation service for now. Uh we also need to wait for uh the Fuzaka fork um on L1 uh which ideally will happen in November in Buenosirus. If you guys are going to be there, we might be celebrating that all together.

Uh but yeah, that's something that we need to wait for um to get to the second phase. uh which is kind of like the endgame uh where pre-confers are going to be stake based uh L1 validators will be able to choose to opt in or to delegate their pre-confer right to some of of uh the service providers um to become a pre-conference you have to have a little bit more um you know requirements rather than like any normal validator uh but it's something that you know it's fairly accessible to most especially if you're you know running already infrastructure on Ethereum. Um and then it's a matter of like you know you choosing whether to opt in directly uh or delegating that power to to a third party. All right. Then uh uh as I was mentioning in the beginning uh this is a piece of R&D that uh we're currently developing uh and which we're very excited about.

Um I guess kudos to our CTO Bre uh for doing this. Um and it's called Gwennneth. Uh what is Gwennneth? It's a horizontal scaling solution for Ethereum uh which aims to bring synchronous composibility. Uh welcome guys.

Um between L2s and between L2s and L1. Um, in practical terms, L2s are going to be able to read state uh between each other and also do the same with between L2 and L1. Um, what this enables uh is pretty much to uh cancel all the fragmentation uh that we've been seeing so far um especially among the the L2s um given that these L2s will be coming based. Uh but we're actually very excited. Um I don't know if you guys saw it, but uh the rollup put out uh pretty pretty cool infographic.

I think it was yesterday uh about kind of like the base rollup landscape. Um there are some projects like Rise like Surge that are a little bit um more advanced in terms of like development. Um and then all the other L2s are kind of like exploring. uh we kind of like paved the way and again as I said to a lot of glass uh but um ideally that's what we want to see uh we want to see a world where like all the L2s are based uh and they can synchronously compose with each other uh each L2 is going to find their own product market fit uh their own use cases whatever that might be uh but we cannot uh live in a world where you have to use and rely on bridges um you have to like bend the knee to terrible UX just because um this will feel like a unified single chain uh where you can just like move assets around um you can just like use your wallet to use different networks uh and this again ideally where we're going to be um I don't want to like make promises my CTO tell me not to make promises in terms of like when we're going to ship this uh nevertheless hopefully in six months but uh that's uh that's something that we shall see. If you're interested, definitely uh keep an eye on this because this is uh uh something that has been debated uh fairly often in in in the treum ecosystem.

Uh is definitely a big issue. Um and this our uh our approach and and our proposal into solving this. Um Gwenf network has theoretically unbunded throughput. Um ideally even scaling to like a million TPS. of course is yet to be seen and tested in production.

Uh but this is ideally where we want to be especially if you know a L1 can get to like 10k TPS U then of course this will make everything way easier for for L2s overall. Um as I said booster functionality enables L1 state being accessible accessible by any L2. uh and also one missing piece as I said before uh is real-time proving which seems to be very close to be achieved uh and right around the corner. Um so um ideally when that will happen uh Gwennneth will you know be a reality of course given all the other things that uh we need to do. All right.

And this is the last slide, I guess. Uh the future of Ethereum, uh cannot be only Coinbase. Uh but it can definitely be based. Um of course, uh I guess kudos to the base team for all they've done in terms of adoption uh in terms of like development and uh also getting like I guess more common users on chain. Um but again like we this kind of like an intermediary step.

it cannot be the endgame at all. Uh because again otherwise we all go back to L2 or we use Binance Smart Chain or we use Solana, right? Uh if we want to use Ethereum and if you want to use Ethereum in a cheap and and fast way, uh it has to be in a way where all the core properties of Ethereum are preserved. Uh and base rollups exactly do that. Um so yeah, if uh you want to learn more, join our discord.

Uh, you want to follow me on Twitter, I guess. Uh, that's my account. Um, and that's all I have for you guys. I'm happy to answer questions. I hope you like the presentation.

Thank you very much for the presentation. I was curious suited for app chains. And when I say upchains, I pro I think of like gaming. uh I think of like high frequency trading per uh things that like require very very very very fast finality which we're not there yet. Um maybe in the future yes if we see the the need and the adoption.

Uh one extra thing I would say that is for sure uh important uh which goes beyond technology. I think it's more of an ecosystem thing. Um we do not have the firepower of like some of like the biggest chains that raise like hundreds of millions of dollars. Uh we cannot afford to give out like you know multi-million dollar grants for apps that might or probably will not work. Um so one I guess strategic decision that we made internally is to build some stuff internally.

Um especially on the DeFi side. Why? Uh because uh also on DeFi one of the things that you have to have is incentives unfortunately or I guess it's part of the game. uh and the issue there that we've seen and I guess overall in crypto is that if you give incentives to third parties then they're going to use it and they're you know everything is fine and you know happy place for some time and then the incentives dry up and then they just go chase some other grants from another chain right um so we are building some D5 primitives uh inhouse uh if we see enough adoption those might become upchains on their own of course leveraging Tao uh but yeah for now I would say not so much. Uh but why not in the future I guess?

Um ideally yes. Hope that answers the question.

Um we don't really have much time for the next question. So

Oh, sorry.

Catch up with

uh PG.

Holy What?

Yeah, that's a hell of a name. All right. Okay. Uh

thank you guys. Thank you. Thank you.