Scaling the L1… or Not | Erin Grasmick - Parity x Polkadot

talk about scaling the L1 um this is a pretty hot but button topic for some people so uh I'll just get into it um I am Aaron grmi uh I've been working at parody for more than six years and I've been in the blockchain space for more than that um I have a computer science degree from W Madison uh my focus was distributed systems and job schedulers which is you know a a way to get into blockchain I guess um I've done a lot of test net work um thousands of no deployments um but now I've been I've been thinking like higher up the sack and more about um like how blockchains work like systems Engineering Systems design and things like that in this talk I'd like to cover understanding each other like we should be we should be like friendly to other blockchains I think um like look at other people's approaches take learnings from other chains and um like look really look take a deeper look at technologies that we're building and take the best bits of all of them and put them together I'd like also to address like premium block space so so this is something interesting um that people talk about sometimes when referring to the ethereum L1 um and I'd also like to talk about new approaches that we can take to uh to maybe combine some some technologies okay so let's start why do we even care about this um people congregate towards like believing a certain set of trade-offs is correct and there's like a a huge tribalism about this like in blockchain um it's been like a hugely hot button topic um not only in the ethereum community but also in like Solana whatever pokot why do we care so much about scaling why do we care so much about how scaling is implemented like why are people so divided over this like the block space wars are still ongoing um like there's a belief that like a huge supply of block space can like revolutionize the apps and the use cases that can be built on blockchain and like choosing one of these directions means like you have to commit to that and you have to really um undertake these massive engineering problems that arise by like the choice of scaling uh that you that you have so like the modern scaling landscape um today like monolithic scales with Hardware it's fast which is good it's synchronous which is like what a lot of people want for defi but like it's potentially centralizing because of these these Hardware um uh uh sorry um like Hardware requirements and it has like definite limits because of like the internet like you know the speed of light the speed of whatever speed of CPUs um and it gets expensive so like it's it's definitely a centralizing factor but the rollup road map um it's expressive programmable l2s can be fast uh scales out to like the capacity that the L1 can actually take um like blobs and things like that um but there's so many multi now for for l2s and like this is also centralizing and there's like you know single companies running uh running rollups and you don't really know how that's going to end up like if they're going to decentralize or not and you also have fragmented liquidity and and users and this is kind of like a bad state so why would you scale vertically in the first place like obviously ethereum is not doing this um so but I want to like help us like understand a little bit like why someone would choose this this approach Hardware is always improving so like you know it might be more expensive to run validators but like you know that Hardware is always going to get better so if you write software that improves with the hardware you know that you're going to keep scaling and you might not even have to change your code um a single- shared State makes mental load for developers and users easy um like you don't have to really think about what chain you're on like where you're going like what uh what chain you're connecting to anything like this um it really simplifies the tooling and interfaces uh for these types of chains because you just have one chain um and there's one chain to rule them all um the bad part is it does really introduce a different set of challenges um like hardware and physical limitations are insurmountable like the speed of light like I said um and one machine is faster than many so this is something that we're seeing more and more today with like Solon forks and things like that where um like people just want to collocate nodes together and actually recentralize previously decentralized change like I think there's like a fork of Solana right now that's actually trying to collocate validators in data centers just to make it faster and that's kind of ridiculous it's like going backwards right um so again one chain to rule them all but in a bad way so why would you why would we scale uh via rollups um It's like a design you know like ethereum has been going for a very long time it's worked um there's a ton of engineering and research going into this um it allows developers to build exactly what they want it's very expressive programmable and it scales out a lot you can have you know ethereum has you know 100 plus rollups now um and it's you know working pretty well um and there's fewer centralizing forces do due to a lower Hardware requirements so you can have like pretty low powerered validators and still um scale in this way but but um it introduces a massive amount of user and liquidity fragmentation and there's a ton of again time engineering money and effort trying to fix this issue today um it's a difficult Paradigm for interface development we've seen this in polka dot quite a bit where we want to um present the chain as like one network and to the user but it's really difficult to help developers understand understand this and I think like this is also happening in ethereum now where you just want to have like one thing and the user might not even need to know what chain they're on with like you know tents and things like that so um interoperability is a core problem of this um and it it continues to be like even in Chains with like polka dot where interoperability is built in from the start it's still like a little more bit more painful it does introduce friction um and like another issue is economics so I'll get into this like a little bit more um maybe later but like the gist of it is that the economics of the relay chain or the relay chain or the beacon chain I should say um needs to be balanced with like the l2's growth so we're seeing this as an issue where people are worried about the value cruel of the of the host chain token like the L1 token um because the L1 needs to have a strong token in order to back all of the rollups because of Economic Security and but the but the value acral of the L1 token is maybe not straightforward if you have a bunch of rollups that are not using the l1's token as a gas token or aren't really paying the L1 that much for Block space and security so this is kind of an unsolved problem and I think one of the most interesting things um about the rollup approach um and each rollup doesn't have visibility into larger system like there's no big shared uh layer that uh they can connect to and um interoperate within does it even matter this is this is a good question um maybe monolithic is enough for today like we we see that both of these approaches are viable like we have monolithic chains doing well we have uh Ro of chains doing well but we do want world scale futureproof blockchains and and like do we want to scale with Hardware or do we want to scale in spite of Hardware like we don't necessarily want to rely on Hardware scaling forever because we will hit these limitations that I talked about and more supply of block space likely equals more demand over time so if we see an explosion of demand for Block space it's likely that we're going to run out if we don't continue to scale and I don't think that monolithic scaling can get us there so really what we need is an additive system where larger networks is more capacity instead of having smaller networks that are just faster because they're small so I want to talk about blobs a little bit as well um more is better but more expensive so uh rollups on ethereum post blobs in differing frequencies um and like this uh basically affects how it it affects like how much of the rollup you could roll back um and this is kind of like weird to me as a polka dot person if you're if you're interested in how the polka dot model works I I I think you should look into it um but like certain rollups on ethereum only post like once a day or something like that so their state rots only go there once a day um and blobs are pretty cheap most of the time obviously there's a a scaling um uh like a re like a blob fee scaling mechanism um but a lot of people when the fees go up they actually post a blobs less often um just because they want controlled costs and this is uh this is probably going to continue like I don't really see a way out of this um where where uh chains will actually willingly pay more there definitely are some that will willingly pay more but I don't think all of them will do this um and it's like it's hard to know how that's GNA how that's going to go um so there's also this concept of the L1 having premium block space which means that uh it can't really compete against these rollups because they're always going to be cheaper um due to their due to the way that they're they're they're architected right so right now ethereum is more expensive but people still like using the ethereum L1 because it it has this sort of premiumness to it um because ethereum doesn't have fully shared security and so if you want the security of the ethereum L1 you use the L1 and you do pay more for that but if it did have shared security and all things were equal why would you pay more for the L1 right like you would just use an L2 so this is kind of another issue but let's try something completely different let's combine the best bits of all of this so let's combine native Roll-Ups which inherit the full full L1 sec with uh like stf verification um use decentralized sequencers um shift the L1 execution to a rollup casual users wouldn't even know about this if you did it properly um and make sure you have to completely decentralized governance so there's no further trust assumption so we're saying that you know all of this like all of the all of the rollups are going to be exactly as secure as the L one and pokot does this um this this is just for illustration polka dot basically has native rollups thank you Sean for the slides um shared security as well so polka dot has uh completely shared security and that's that's pretty unique so like every single rollup is just as secure as the L1 on polka dot and so this is why we can do something like this which is a radical Vision which actually um balic mentioned um in a recent blog of his where an extreme version of of like taking the execution out of the L1 is you could actually move the evm to a rollup in order to um get this level of performance increased performance and I find this really interesting because this is something that uh polka dot is actually doing today so this kind of thing is going to be live this year on polka dot we are actually moving all of the execution and the user level stuff like B balances staking Etc onto a rollup and so then we get the um we get the we get all the performance benefits of um of a rollup but we can just like we can just move it and then the then the relay chain or the beacon chain just does nothing all it does is consensus and secures all of the other rollups but we get we get we get security and stuff for free so I think this is like a really good way to do it um especially in a world where like when does all this stuff end like it's it's like a a horrible um like Jenga sack full of stuff and like all of the stuff where you have to like put different things together in order to make things scale and interoperate and uh make them secure I don't think is a good world and like it's not a developer friendly world and it's not a user friendly world so uh yeah I think uh I think I think that's about it [Applause]