Permissionless Interoperability for rollups and the modular blockchain stack by Yorke Rhodes

hey guys uh welcome welcome to this talk on permissionless modular roll-up interoperability uh it's a bit of a word soup so I'm gonna kind of talk through some of the definitions there but uh my name's York I work on a project called hyperlane it's an interoperability project uh kind of embracing some of the ideas that I'm going to talk about in this talk um and if you have any further questions about hyperlane or anything I talk about here definitely reach out either on one of these links or after the talk you can come up and ask some questions okay so I'm gonna get started um so first I just wanted to Define what uh interoperability is so we can have kind of a cogent discussion uh my definition of interoperability is composability amongst dapps that are on different infrastructure uh so that's like a pretty high level definition um but we've observed in crypto and in like the blockchain ecosystem in the past I don't know you know 10 20 years is that uh you know daps that are enabled by this new technology are on increasingly Divergent infrastructure and this is because they want to take advantage of uh new Innovations new uh basically new new advances in research that are only being deployed on kind of like new infrastructure layers because the ability to either include those properties in existing networks like Bitcoin or ethereum or even Cosmos or one of these kind of uh established networks with network effects it just becomes a political battle you have to lobby and you have to you know get your pre-compiles included and you have to you know ask for a hard Fork to get the feature your application wants and at some critical uh point in a I guess networks life cycle it stops being able to just include everything that people are asking for because there's obviously trade-offs here so the idea behind the modular blockchain thesis is we want to allow applications to express the types of trade-offs that they make at every layer of their stack um and I think a lot of people you know the modular blockchain thesis is a very like uh you know loaded term I think it comes with a lot of uh different associations and assumptions um because we're at an ethereum conference I'm going to talk about kind of the ethereum roll-up Centric roadmap and what that means in like this modular blockchain thesis context um yeah this graphic over here is taken from zksync's website this is kind of like what they think the future of I guess their roll-up architecture is going to look like for application Developers um you know you can see that it's very complicated there's oh yeah it looks pretty complicated um so we need to come up with a way for developers to interface with this stuff that is you know not so overwhelming for people to understand but yeah just just to summarize I guess uh you know as we increase the [Music] um you know complexity of the infrastructure layers and and their implementations uh becomes harder and harder for adapt developers to like rock this stuff and to like evaluate the trade-off space of like oh should I be using Celestia as my data availability layer or should I be using ethereum or should I be using eigen da right like these things are like very nuanced um so we have to just like come up with the right interfaces for people to express this so like I said I'm going to focus on kind of the ethereum roll-up Centric view of this like modular blockchain thesis um and you know in case you guys aren't aware uh ethereum and the ethereum research community and ecosystem has kind of galvanized around roll-up Centric scaling roadmap this has been kind of the what the core devs I've been working on for the past couple of years uh basically transitioning ethereum to be this base layer for uh this roll-up Centric future for the ability for us to get increased scalability properties so basically like transaction throughput right uh we don't want to make the L1 prohibitively hard for people to run we wanted to like maintain some of these like Network effects that it already has so we can't make super drastic changes uh but we can make a change like uh eip4844 which enables some Innovation to happen on a layer above without requiring basically like a lot of controversial protocol changes which might have a have a big impact on the existing Network effects and stakeholders in the ethereum L1 um so this is kind of where we are today uh you know this 4844 you've probably seen it on Twitter um yeah basically makes Roll-Ups uh able to use ethereum as their data availability layer for way cheaper that's like kind of the super high level takeaway and what we're seeing as ethereum kind of uh moves in the direction of this roll-up Centric roadmap is adapt app developers are actually adopting these new architectures um so both of these Graphics are taken from a website called l2beat uh basically like visualizes roll up activity definitely check it out but um you know you'll see pretty recently actually the aggregate amount of activity on these l2s has kind of superseded the activity on the ethereum L1 if if we're using kind of transaction throughput as a metric for activity um and if we look at kind of d52vl uh basically like options protocols have exploded on arbitrum uh you guys have probably heard of GMX uh they're they're killing it right now on arbitrim [Music] um and yeah application developers are starting to adopt these Roll-Ups they see the benefits of lower feed and the lower fee environment for their users and they see the performance benefits of having these like more centralized sequencers and you know they they realize it's a way for them to differentiate from their competition and so they're they're migrating to these new player twos um which is exciting right we're actually seeing a lot of innovation happen in real time but there's kind of a problem that comes with this Innovation and that is that the ability for all these applications which are now deploying on these new Roll-Ups uh no longer have the kind of like Atomic composability that they had on ethereum L1 um and and if you think about kind of like defy summer and like the ethos of ethereum as this like generalized execution environment a lot of people believe that like this property of atomic composability is the reason that ethereum is like such a valuable uh developer primitive uh and um that we can deploy smart contracts that leverage each other's Network effects rather than having to like bootstrap our own network effects within kind of some like siled application uh this is like the idea behind lens right instead of like if you want to compete with Facebook you you know have to go and build your own social media company and like get people to come on board with lens or with any of these like smart contract platforms you can leverage these existing Network effects um without uh being competitive you can be cooperative but when these applications are on these l2s there's kind of this fragmentation that's been happening uh and there's all these different types of interoperability now there's no longer just this like simple atomic composability there's deposits and withdrawals from L1 to L2 there's this like asset bridging that's happening between Roll-Ups to share liquidity and to kind of like balance demand across the different roll-up environments and there's also General message passing between Roll-Ups so this is like you know asynchronous uh composability uh basically having like smart contracts interact with each other in an asynchronous manner there are some attempts some recent attempts and like working groups to address some of this fragmentation um I I noted one of the eip's here uh oops but generally it's really hard for as an application developer to look at this stuff and like know what you should be doing um so yeah just kind of highlighting that um with this you know new roll-up future dap developers have them have a much harder task to to understand what what interfaces they should be using um this uh interoperability especially between these like heterogeneous environments that is kind of being enabled by this roll up future uh unlocks like new forms of innovation that actually weren't possible on ethereum L1 so not only are we getting the scaling benefits but we're also seeing new forms of applications that are only possible because we have an application that's leveraging properties of some infrastructure uh like multiple infrastructure layers so there was actually this really cool article that uh a16z uh crypto's research team just released they've been doing a bunch of like Research into auction theory on chain auctions like verifiable Fair auctions and they built this cross chain application which leverages Aztec connect which is like a privacy L2 to do a sealed bid auction which is kind of like this thing that a bunch of people have been trying to implement but it's like really hard to implement and like maybe it's a solution to Mev uh but we're not sure um but yeah just wanted again to highlight like not only are we getting scaling and we're getting uh you know applications can now basically like do things that they couldn't before which is also really exciting um a lot of people are like light client maximalists and this is an idea that like somewhat comes from a cosmos ecosystem so they believe that like interoperability between these heterogeneous environments should only use what's called native verification uh which is basically like you have this thing called a light client and it's basically uh syncing on chain with some other chains consensus so it's consuming the headers that are being produced by some other chain and it's verifying them locally and so this is called like native verification [Music] but the problem with this in our like you know super fragmented roll-up worlds is there that first of all there's not like client implementations for most of these things uh but second of all like the implementations are very specific to the state model and so in this like modular blockchain thesis we want to allow applications to express their preferences at every layer with something like Cosmos there's actually a lot of baked in assumptions that enable IBC uh and some of those are basically like you know you have to use the same consensus model you have to have the same state model and this allows you to have this I guess Universal like client interoperability but where we are in ethereum uh this basically is Impractical um especially with the size of the eth2 validator set a light client for the E2 validator set is like extremely extremely expensive to run on chain um basically like prohibitively expensive um so we need to turn to alternatives people are doing some really cool research stuff on like DK on the ZK side of things I'm a little further away from that but like people are working towards solving this like eth2 like client problem but it's like uh still very much in the research stages Okay so we want modular interoperability for application developers in this like roll-up Centric world we need to separate the transport layer and the application layer um so this is kind of where we depart from maybe like uh what the cosmos model is where there's kind of like a bit of a verticalized stack um and yeah the basic premise is like dap developers should just understand the interface to the application layer this is going to be a simple like send and receive or in this case A dispatch and handle interface that should be consistent across all these Roll-Ups all these like different execution environments we can have a consistent interface that's possible and if we do this we don't need like to maintain all these different implementations uh we can just have a spec uh I mean maybe we want like some diversity of implementation but generally um this makes the DAP developer's life a little easier because now they don't need to understand okay L1 to L2 withdrawal versus L2 to L2 now generic message passing versus asset bridging between L2 and uh you know some Cosmos chain like all those right now are completely different applications with their own uh set of interfaces so you like the complexity of some of these code bases is just like crazy because you have to special case basically like every single pair of chains that you're connected to you want this interoperability um we also want to allow dap developers or maybe some delegate that app developers select to express their security preferences and their risk tolerances so as a recipient application when you're getting some inbound uh intent for interoperability between some other chain or from some other chain or some other roll-up you want to basically specify through some consistent interface like what is the security model we can have like basically the equivalent of token lists like uniswap has a list of token lists so we can have like a community governed list of security models for all these different channels that developers can just opt into have a consistent interface too but they don't need to all Implement themselves these verification mechanisms from all these I guess heterogeneous environments there will be some kind of like uh you know canonical set of verification mechanisms maybe some are like clients maybe some are like these optimistic verification mechanisms some are proof of stake uh sort of systems and application developers don't need to like necessarily understand all of the transport layer stuff that's happening in between they just have this like configure configuration layer for security and a simple like dispatch and handle interface I get this like operability across all these environments so at hyperlane we call it Sovereign consensus it's kind of like a funny name um but yeah the the cool thing about this is like applications whenever these new verification mechanisms that are coming out of the research Community become available they can just like change their configuration they don't need to go in and like re-architect their entire system to accommodate some new uh light client implementation or something like that they just say oh I'm going to set my interchange security module to this new uh this new light client module so the way this is uh I guess modular uh in the hyperlane architecture is we basically allow anyone to deploy this uh so it's a protocol but it's also basically just this interface of like dispatch and handle configuration layer um and you know if you're spinning up your own roll up which is kind of the idea of a bunch of these like roll up as a service projects like uh you've probably seen the optimism op stack uh Celestia has their like roll mint roll kit thing arbitrum now has something like this a lot of new projects kind of competing on on this like roll up as a service model uh people want like app specific Roll-Ups because they get all these scaling benefits and like they get more free Revenue um so you don't want to have to go out and like Lobby some validator set or some uh Bridge like you don't want to have to go to Wormhole and say like hey please support my chain and Bridge here so that I can like Leverage existing Network effects from these other chains you can just deploy this protocol yourself and you get all the benefits of inbound security if you want uh you know other applications to basically like compose with you on your outbound communication you know they need to opt into that so they need to kind of configure some interchange security module which knows how to verify your roll up or your chain you can go and deploy that for them um but yeah this is the basic idea uh you you don't have to go and do BD to like get interoperability anymore we now have these modular systems uh where you can just as an application developer like be chain agnostic uh not bacon so many assumptions when you're doing your like initial architecture or like design or deployment but rather be like super extensible to all these new Roll-Ups new verification models um and yeah I encourage you guys if you want to like spin up your own roll up this week you can try deploying hyperlane on it and sending messages between our existing deployments you should hop in our Discord tell us about what you deploy we're super curious what people are building uh in these like architectures I think my time's up but yeah definitely come down and ask questions if you have any um and yeah thanks so much for my time foreign [Music]