Future Finality [LIVE]: The Scalability End Game with Ansgar Dietrichs, Illia Polosukhin, Amos Zhang

hello hello everyone uh so now um is the panel uh called the scalability and game uh which is uh the French the future finality series uh that we're recording live for East Denver uh and uh I'm having great pleasure uh to moderate this far side chat between uh nearest founder Elia and ethereum is core researcher uh esker and today we're going to talk about how we actually you know scale ethereum scale near and scale you know uh the whole crypto industry to get more and more adoption uh but first let's start a quick intro so my name is Amos I'm the founder of metal web Ventures uh where a 30 million dollars early stage uh incubator style fund uh we're also very research driven so there are the two primary ecosystems that we invest are actually near at ethereum uh hence the reason why I'm doing this uh fireside chat today uh and asgar will do the pleasure to introduce yourself as well sure uh have one um I'm Ansgar I'm a member of the ethereum foundation research team has been on the team for roughly a year now before that I was at a similar ethereum research team over at consensus called quilt and um um I have been working on several I'm not sure how I've been working on several things we have a feedback problem I don't know um several different things in the past including a kind of abstraction um scaling actually used to be involved as well in the research before ethereum pivoted to the roll-up Centric roadmap we had this these plans for what we call execution environments so basically the idea of sharding execution in an enshrined way on ethereum as well was involved with that research I was actually one of the first things I worked on when joining the team so um it's actually kind of it fits the connection to Nia basically because yeah you can maybe talk about how that's related yeah thanks uh great to be here my name is Ilya I'm co-founder of near protocol uh actually previously I was AI researcher which is funny that now remember um but yeah we started here in 2018 been pretty much focused on how do we scale blockchains in a usable way and actually our research started by interviewing metallic and Justin Drake and Danny Ryan and kind of other core researchers to understand where ethereum was thinking about scaling in 2018 and that's why a lot of the kind of discussion ideas started from that and then over time diverged that's kind of we were building out things cool cool uh so let's start our first question actually just right on the topic uh so motorization uh versus you know uh modelistic you know blockchain having a very trending topic uh for the past year or so uh so I'm just curious like in your opinion what are like the philosophical differences between you know uh modular and uh blockchain and like a monolithic blockchain uh will motivate you know ethereum to move from execution sharding to tank sharding uh and you know what keeps near essentially adherent to its original starting road map manifest so um I think basically the first Insight with um scaling blockchains in general and I think that's basically where we still agree and that's kind of I think the key Insight is just that you can't just keep linearly increasing the throughput of a single um monolithic chain right at some point you just run into limitations so the question is what can you do and so um you basically pretty naturally come to this concept of sharding where you basically say instead of having it having execution in one place you have like basically multiple different um places where you have execution and then you'll be basically all kind of bundle them together in onto some root system um so that basically that high level description I think fits describes both the the ethereum and the knee approach and polka dot and yeah a lot of it's not a very it's not a very unique kind of insight um and so then the question is just how do you actually do it and so um where ethereum also originally started was um basically trying to take the existing chain that we have with the evm and just basically copying that spinning up multiple versions um maybe even giving them a more modern like look maybe introducing some awesome or something right like that we had all these plans but they still basically it's fundamentally designing these charts in-house right having having like um a very opinionated approach and then just spinning them up over time and and increasing throughput that way where we at some point decided to go a different path is that um just because we it going down that path we discovered that there were so many different flavors of how you could do this right like that a billion different questions what kind of execution system evm was them some something completely different how how exactly what kind of rules what trade-offs in this game like how much scaling versus how easy it is for users to run the notes there's a billion different questions do you couple finality do you not couple finality like they there are many different questions where you would have to make opinionated choices and we decided that instead of making these choices and going one specific route we would instead open it up for permissionless Innovation and permissionless experimentation this is where you basically move from execution charts to roll ups so what we what ethereum's kind of understanding of the space these days is we only provide infrastructure for Roll-Ups to scale and we are not opinionated right it might be that a optimistic roll up a ZK roll-up system is the way to go it might be evm-based not in um based we don't care we just provide we provide the data and we provide the settlement and we let other people's make the other people make the opinionated choices um but I don't think so so in the end we end up in a pretty different kind of space but I don't think it's something where we fundamentally disagree on what is feasible technologically it's more that maybe ethereum just basically went the less opinionated route would you agree with that oh yeah exactly so when we started our our core perspective was how do we make this technology is easier to use and to build on and so sharding you know scalability is a instrument to achieve that goal it's not the goal in itself right if you're building on something and it's out of capacity because some other app is doing something that's not like a really good environment to build on and so kind of we work from that perspective and that perspective usually means you need to make some trade-offs and kind of build some opinions on how things should be to make it easier for developers and so we chose webassembly as a core because it's a very kind of generic technology it runs on in every browser right now it's supported by kind of a huge network of you know big and small companies it's actually growing like in in adoption extremely fast and it has a lot of programming language and a lot of tooling available already so that allowed us to have like a very solid foundation and then that kind of other opinions like evm like everything else can be built on top of it so we actually have for example Aurora which is an evm that runs as a smart contract on top of near so like you can't build all of these environments on top using this kind of a very in a way opinionated but very generic framework that we built right and so we're kind of making decisions like that a lot to how do we create something that is simple to use but generic announced that developers can build whatever they need on top and and so that's kind of where how we designing it and the complexity of sharding of you know of multiple different environments running is way too complicated like for any user or developer like thinking about like where should I deploy how the like you know accounts tokens fees you know finality propagation will be happening as a developer that's way too complicated and right now people don't even think about it half of the times right and they should and so kind of what we try to do is create an environment where it looks like it's a single blockchain even though underneath we Shard you know the blocks we Shard the state we start the transaction processing but at least from a developer perspective and from user perspective they they see it as kind of one thing and then hide a lot of this complexity under the hood and again it's true we're making some decisions while doing it but trying to make them as generic like enablers as possible just one question if I could briefly ask this so um to what extent do you think about Nia is like one monolithic system where you would versus say a more modulus like like would you ever consider say taking a specific layer of your stack and saying like maybe you know in addition to running things on our own chain we also we are perfectly willing to also say take our um wasm system and deploy it on other chains as well or we would um or maybe the other way around you like instead of only having our own wasn't based system you basically you you would be interested in kind of bringing other people's system in like like to what extent basically do you think it is like one monolithic and and kind of system versus kind of a more layered so I the way I think about it right and obviously there's a big team contributing now so there's a lot like few different opinions but the way I think of it as uh like you know PC versus MAC right BC is like a lot of different manufacturers making different pieces right and they're trying to like all agree on kind of some standard bus and PCI things and like there's a lot of issues with that but it allowed this permissionless Innovation you know anybody can make a thing and then there's mac which is inside it's still components and there's a lot of people working on different components but it's all kind of integrated like at the end is one product and so that's kind of how I think about is like at the end we want to see one product for the users and developers and although inside like we have actually peer-to-peer layer is separate you can use it actually for other use cases execution environment I have plugged our execution environment into tender mint you can run our Quantum Quantum near chain for example but like that's all like you know cool experiments it's not going to be a product and so what we're trying to pull it all together into one cohesive experience uh but yeah I mean inside it's all componentized you know you can work on every component independently and like they all have like their own apis and interfaces right because just as a point there because uh you were asking about like the differences in like mindset between monolithic where there's kind of modular and I think interestingly I would even say kind of like this both are both ethereum and Nia are on the Spectrum like I wouldn't say ethereum is the extreme if you go to a very stream of modular you end up with something like Celestia right where basically you really only take one layer and one layer alone and nothing else and if you go to the Very extreme of the monolithic case you go to something like Bitcoin or Solana or something right where there's basically no no kind of like no multi-layer scaling Approach at all and I think Nia's basically just like a little closer to the monolithic side and ethereum is a little closer to the to the modular side but it's still both kind of on that Spectrum not on the on the extremes exactly that's a great way to put it uh so let's actually uh switch here a little bit and talk about you know value capture because that's always like a very important part of you know any blockchain uh system uh so what does you know value capture like look like once again in your opinion in like a more modular modularized you know uh production uh design uh how would the fee market look like are we making the token you know less um the native token actually less valuable uh like will ease for example will yeast validator actually learn less after you know EIP 4844 but maybe do you want to go first for like yeah so I mean for near the idea is again to hide a lot of this complexity and so ideally like as a user I don't want to even pay in in a native token right I want fees to ideally be hidden completely and I just want to pay a subscription fee like that's really where and so Aurora have been prototyping that uh as an example where you just you know UI just take some token or you um you maybe pay subscription fee and you get some amount of transactions per month from that right and like those can be different tiers and so you don't need to think about what's the complexity of the underlying again infrastructure how much infrastructure costs and that's kind of how right now you're paying for Netflix you're not thinking about how many bytes have you received from Netflix of Ingress from AWS right that's not something that like any user needs to think about and that's kind of where we are right now is like how much gas is trans this transaction gonna take and so I think a lot of it for us at least like how do we move toward that direction and then underneath is built like robust systems for different players to for example for priority fees to kind of be able to uh like still provide some service to people who need kind of a higher priority and by to buying out some blog space in in the in the ecosystem right yeah it's interesting because um in a way you can think about transaction fees right they they live in this interface between the native protocol and kind of the apps that that are that are that sit on top and uh in a way the kind of like the near and the term design they're not that different because uh I assume as you're saying like Aurora or something because it's kind of these pickups they almost look like Roll-Ups in a way right so they basically then um they would also be user facing and hide hide a lot of the complexity in the same way that a Roll-Ups say can can make opinionated choices right if I run a roll up I can charge my users in Eid for the transaction but I can also hide that away I can have a subscription service I can have I can let them pay in usdc I can I can make any of these trade-offs and the the the place where the fees actually come into place then the interface um with the protocol and the interface is just in a slightly different place so for us basically you pay for slightly lower level kind of resources you pay for the settlement you pay for data availability as a roll up um and so these things we do charge in eth um and so so with um we were kind of making one first step in that direction and now with the introduction of Erp 4844 so we we introduced a separate fee Market specifically for this new resource we create like the data um and that's all complexity that's abstracted away on O'Neill that basically that that is more kind of lower level Exposed on ethereum um but I think as far as users go I think at the end of the day we are still very early in ethereum and if we project five to ten years in the Future No Matter What systems system the user will be on what what the underlying stack looks like I think basically there's there's no future in which this is not all abstracted away like the you the user does not want to think about any of these low-level concerns um so so I think as far as users are concerned these systems will end up looking and feeling pretty similar I'm actually curious if we if you project like let's say we take one of the Roll-Ups and let's say that roll-up become extremely successful they will need to have multiple sequencers which means they'll need to have some consensus they will need to have sharding because one one computer will not be able to process everything right so they will have sharding we don't need to figure out asynchronous communication between shards they'll need to have consensus multiple sequencers they'll have their own token why would they not decouple from ethereum at that point ah okay but that's that's really interesting uh interesting question actually so um in a way ethereum is kind of like as I was saying we are basically deliberately restricted in what we provide but what we provide we provide at a very high level so we Pro we basically the only things we will provide is um consensus on the execution change so so you can you can you can do settlement on top of if the ethereum evm and you get very very strong security guarantees and then in the future we will provide more and more data availability and again it's basically it's a very simple service it's just you give us raw bytes and we just give you a this was available like that's all like it's it's a very primitive service but we will do this at like a very very strong level of security so really basically what so so basically what that means is a as a as a roll-up if it's even if it's multi um like multi-sequence or something I don't actually have to solve for full consensus anymore because I already get full ordering from the base chain right that is part of the service we provide the only thing is that you have to have rules for like who is allowed to to provide to create the next block or something it's it's a smaller set of problems so it's like you specifically don't have touch the classic blockchain consensus part of it and then also more importantly you don't have to deal with security basically like everything is out of the box at the highest possible like ethereum basically there's no chain I mean arguably maybe Bitcoin but Bitcoin Vitality limited functionality there's no other chain in the blockchain ecosystem that has as high as security level as ethereum and I would argue after proof of stake now we probably even beat Bitcoin not on Raw dollars spent on security but the actual benefits you get out of it um so yeah I think basically like these are these are like very in a way primitive very trivial Services we provide but they are crucial like crucially important and they're really hard to recreate on your own so that's why I think basically I I think it's the other way around what I would expect is over time more of these you know full monolithic chains to realize that what they're doing is not sustainable and to move on top of not necessarily ethereum but some sort of maybe even on top of knee on top of ethereum on top of some sort of celestia-based system like I don't expect in the long run we will we will see any of these kind of roll your own consensus chains to survive I mean let's see it's an interesting question yeah so uh let's actually uh talk about something else because once in the both of you guys actually touch upon is actually the user experience uh which uh to me uh is what like scalably should serve like in the end uh at first you know back in the days you know back in Bitcoin in the early stream days you know uh blockchain are not very scalable right that's why you need all those like different solutions because let's say if you want to buy a cup of coffee in startup uh you don't weigh you know a minute or two for the transaction to confirm right obviously you know blockchain uh have came a long way to where it is right now uh both ethereum and near our launching bunch of you know features I'll have you know uh stuff planned in the room map they're gonna significantly increase the the user experience uh so would you guys mind talk a bit about uh what you what you think are the most important you know features that can help you know uh ethereum or near improve the user experience uh kind of really on board start uploading like a large uh brawler set of user you know from web to uh who are currently not on the street yeah so as I mentioned like we always focused on kind of how do we make it simple to use and so uh from the start we actually implemented account instruction system which was designed to allow pretty much a very flexible model on how people managing their keys how the applications are deployed on Smart contracts and so so that's been running since mainnet for two years now and kind of there's a few things we're actually going to be launching in the future one of them we call account extensions so instead of just account obstruction it allows you to pretty much enable different modules on your account uh similar like you know similar how you would enable features you know I want a multi-sig I want a Deadman switch I want a proxy you know I want a Twitter like authenticator and so you can just enable those features all of them are independent smart contracts and they will be plugged in into your account enabling different features and so it kind of allows to have a composable account model but all of that kind of is serving a bigger purpose and so we actually yesterday released what we call blockchain operating system and that and this is actually a fully decentralized composable front ends that are built on top of the near Network that allow us to pretty much build front-end applications kind of interacting with all this including interacting with ethereum and any other AVM apps and so all of that is actually tying all of those things we've been building together to create a user experience on top of the blockchain and in a way that's like fully flexible fully composable kind of allows you to explore the what's happening on a blockchain without really thinking about the blockchain experience itself and this is where like hiding the fees hiding the complexity hiding even the blockchain you're interacting with and like maybe like we're working on allowing you to build an application that uses multiple blockchain at the same time like a front end that allows you to interact with multiple blockchains so that's kind of the you know experience we want to achieve is users just on board start using like login you know with zema login visit face ID and start interacting with applications done thinking about where they're coming from how they you know what's these things are bridging between you know where's the Visa coming from and kind of you know subscription fee type of experience on top of that to really enable them these interactions actually this is kind of this ux questions topic is a really good topic to highlight where kind of even though we have somewhat similar approaches actually where there is like a clear difference just because now with our kind of limiting ourselves on kind of the infrastructure side and providing the tooling for olaps this is like the point where we basically do not want to be opinionated right so um things like account abstraction um this is something where we have already seen in the past individual Roll-Ups and launched with account obstruction so start net had a native um solution for that the ZK sync launched with a native solution for account obstruction um but we basically specifically wanted this kind of this free experimentation and it's a little bit of a maybe a cop-out though to say it always say because I can always you can always present a specific solution that could say oh we just let anyone experiment right and like oh that sounds great and it sounds like you know sure if open market will find a better solution than a centralized player but of course I mean in real let's see if you actually kind of look under the hood there's there's some asterisk here so for example um a lot of the Roll-Ups today what they actually end up doing is what they call evm equivalents so they actually specifically Target the exact same kind of specification as the underlying ethereum evm system and of course that means that all of a sudden they end up with the same user experience right and so a lot of the the time in in what we end up having actually having to do is that a lot of these solutions for things like account obstruction and of course now hopefully you all have heard about erp4337 which is the ethereum native version of account obstruction but where that ended up coming from was not the layer twos it was actually kind of from the base layer ecosystem it's not an in protocol solution it's a it's a kind of pure like external standard but basically it's something where just you know LED layers to it layer 2's experiment was not actually in practice how this worked right in practice it was still the the base layer Community coming together designing such a solution and then now up have having it be like an opt-in thing where most most layer twos have already signaled that they will support this and over time will implement it natively um but it does mean that in a way at the end of the day for some of these questions we are still in the same position where even though we don't want this responsibility it's still like a standardization question and sometimes even of course it's it's a trade-off space so I think because with like ethereum does move a little bit slower on these things I mean you have you've had a gun obstruction live for quite a while already we like 437 is just launching this now um but it's a trade-off right at the same time because it's a standardization process we bring a lot of different players in we have a lot of different perspectives so as always the reality is a bit more nuanced than this that black and white but but it is it is I think philosophically it is a meaningful difference of kind of like focusing on providing really good solutions for users directly versus just leaving it up for in a modular in a modular mindset really living it up for individual specialized players to come in and and design solutions that we don't even want to to be opinionated about yeah I think the question as you mentioned right because they kind of still need a standard to be interoperable with each other it ends up being a bottleneck that none of them can design their own independent solution otherwise they're kind of gonna get like out of outside of this common common group and so yeah I think it's kind of it's good to open up but like you still need a strong standardization system so we actually been running uh standard work groups specifically for this that invite our community and kind of invite actually members of other communities as well to join to to run this process kind of effectively and and be able to push new standards and align kind of the whole ecosystem around this because I think that's actually from infrastructure perspective in blockchain this is the hardest part it's not building the Building Systems actually everybody agreeing that this is a standard and everybody's like integrating it ahead of time pretty much ahead of launch so it actually works achieving human consensus even consensus cool I think that's actually a very good way to always like take us back home because it definitely showed that you know uh near inferior magic to very different approach uh when it comes to whether it's like uh scalability you know standardization and obviously usability and I think actually makes sense um because uh it's really hard to tell like what is wrong and what's right and I think it definitely makes sense uh for you know the two different blockchains especially at their different you know stage of girls obviously near is a much you know younger blockchain it has more flexibility uh ethereum is like a much older blockchain with much larger Community around the world right so I think both approach actually makes sense uh and let's see what's gonna happen in the next couple years uh so we're almost time uh any against any last war you want the audience to take home with them well I would just say that um kind of ethereum has only like it's not been that long ago that we kind of decided to actually go down the roll up Centric roadmap it's only been maybe two three years um and I think we're only now starting to really see a lot of the experimentation so while I think I try to in the conversation caveat it a little bit and show that it's not just always you can just always throw shout experimentation and then everything is systematically solved so it is a hard process but I do think that like we are only now starting to be to get to the point where we'll hopefully see more and more of kind of the actual the fruits of of all the different projects trying different things and there's a very very broad design space and a lot of projects I mean even just like base uh the latest kind of entrance here announcing a couple days ago so I think we are pretty happy with kind of seeing all the different kind of experiments playing out and I I do think that um most of most of the payoff of this approach is what we will see over the next few years um but I do think basically on the other side that means that kind of uh basically what Nia has today already what basically ethereum is is this lagging for now is just like basically like a harmonized approach to scaling um I don't know if if you would agree with that sentiment but yeah I mean kind of on the blockchain side it like we focus on usability and kind of alignment across all of those pieces but I think to leave everyone with one more pieces like the the thing we did launch yesterday it works across near it works ethereum layer twos actually folks already working on Cosmos integration as well and so what we're actually harmonizing now is the front end is the access point it's the entrance for the user and really what we're trying to do any wall any blockchain can support and use the crosses and what we want actually to achieve is a standard for the front ends and so if we have that actually users will have way less kind of uh you know like onboarding into this and be able to use this across and so at the end I think that's the layer that really matters right now like I think we have enough of the infrastructure Tech I mean the amount of like layer ones Layer Two that's just present on this is kind of insane but the amount of applications that are really usable is very small and so I think like it's time now to focus on that and that's what we right now focus on and that's what blockchain operator system is and so if you leave with one thing is that near is the boss blockchain operating system all right our time uh thank you again Elia and ask her for this amazing conversation uh if you guys like to see more of those content I will be recording more and more of those series uh online and physical events like this so just follow us at medwap VC on Twitter and we'll see each other again thank you guys thanks [Applause]