Realizing the Original Web3 Vision: Hybridizing On and Offchain - TOKEN2049 Singapore 2024

hi um I'm Sam casset um and I want to tell you about a new company called layer but I first I want to tell you about why we built this thing um layer is kind of a funny name right there's uh layer and I layer like layer layer whatever uh but we're kind of a new kind of layer there's really nothing like us that exists it's a new onchain primitive and of chain primitive as well um but first I want to talk about sort of why that is um hybridizing on an offchain compute for True de centralization is sort of the why when we first started thinking about the internet uh even the web one Pioneers when they thought about it uh one of them uh this guy John Gilmore was quoted as saying the internet interprets censorship as damage and routes around it uh we kind of lost that Vision they weren't really ready for state level actors trying to tell them uh what to what to stop and what traffic to to redirect um and web 3 sort of had that that Vision that let's let's correct that let's make a truly decentralized system uh but thus far we haven't really done that but I think it's finally time it's finally uh a place in time where we're capable of doing that and layer allows you to do that so uh around 2015 you know I was around for the beginning of ethereum um I helped create a company called consensus we funded basically all the ethereum ecosystem for the first few years uh and basically we just kind of had this handwavy let's just decentralize everything let's just you know let's make a dow that's run by AI we don't really know how we're going to do that but you know we thought we could do it but there wasn't really a treatment of how we would realize this Vision um so you know that was around 2015 uh this is an example of what we thought it was going to look like this is the Miss browser so in the very early days of ethereum uh you know this was the stack this is what we thought was going to happen that there would be specialized front ends that weren't integrated into your browsers metamask wasn't a thing yet uh we would have like form whisper evm all alongside each other and that would be how we would achieve this decentralization but obviously that's not what happened we ended up with a bunch of centralized interfaces we ended up with kind of decentralization theater where uh you know really you're running something on the side that's not decentralized your front end is not decentralized especially um but as we move forward uh We've sort of pushed that boundary a little bit but I think we're finally getting to a place where we can this is kind of a rough history of my standout moments of of where we where we thought about decentralization crypto kitties I don't know if anyone recommends this recognizes this this is the first time that uh we actually clogged the ethereum network and that's great if nfts are cool but where are these things stored right they're actually mostly stored on this thing called pinata which is a centralized actually now they have some decentralized parts but underneath uh when you know when that company goes out of business or something happens the the image can actually go away right that's that's not very decentralized 2019 uh you know this is when Ops uh first test net went live this is the sequencer diagram of op right so even though it's great that we can have scaling we have you know l2s that have higher transactions per second they have higher higher throughput uh it's much cheaper those things are amazing and I'm glad that our industry has progressed in that direction such that we can you know we can move forward but at the same time there's a centralized server sitting somewhere that's actually sequencing all the transactions at the centralized point of failure U and we haven't actually we haven't actually stayed True to this and we or or just haven't achieved it right it's still at the periphery it's still at the edges so around 2020 uh you know you could argue exactly when but this is when ZK first started becoming a thing so this is a diagram of the uh part of the system of Stark de actually and you know you have the M poool you have sequencers whatever but this thing on the right the prover um this is actually just an AWS node or you know maybe they're using Google Cloud I don't know uh but this is you know this is a centralized part of the system and you know when we look at how to do that some people tried to do it Alo has tried to build a big incentivized decentralization uh engine for ZK uh Mina did it as well but at the end of the day right there's still there's there's still parts of this machine that are not on chain they're not verifiable on chain they have no cryptoeconomic security about them it literally is just a trust relationship trust in the bad way not in trust in the good way um so like what can we do about this um now we're sitting here it's 2024 we still have our front ends set centralized right this is a this is an article about Unis swap having some trouble about their front end they had to go and censor a bunch of uh types of assets on there this is from 2022 you know they had problems as well but the you know the goal not just being making uncensorable things uh the goal being things that are truly resilient truly in the design philosophy of how we thought the internet would work um things that are self- sustaining things that don't require any centralized party to give them permission to exist and maybe things that can exist uh in in a way that that pushes the boundaries of uh what a uh almost like a life form might be something that's self- sustained something that doesn't require uh the centralized control that our society uh I think has has suffered from and that we're all playing with in this industry that's sort of the why of this industry but so what does this look like in the next wave right uh obviously there have been a bunch of different attempts not only to achieve decentralization but to achieve scaling to achieve different kinds of architectures and I think like sham who's the founder of aan layer it really well there are kind of three you know Choose Your Own Adventure paths that you can do ZK and you have to pay for these Pro costs and they're all centralized or mostly centralized at the moment uh you can do optimistic rollups and you have to wait for settlement lag um a lot of people are building in that direction which is great uh and I think it's you know it's done great things for our industry and you know it not to knock it but there's a third path and this is sort of what aan layer enables uh is this cry cryptoeconomic path right where you are paying capital cost cost but at the same time it's instant but what it allows you to do is put something at stake put your ether at stake in their case but put something at stake to guarantee the results of some transaction uh or some behavior of some system uh and that's another way to scale it's another way to it's another way to achieve different types of Primitives uh in web 3 architecture um and so we've we've chosen to go down that route um and we're you know we're building on top of Ian lir specifically we're building on top of the thing that has the most Economic Security other than Bitcoin but the most thing that is you know the most that's programmable Economic Security um aan L kind of lets you do that but there's this huge Market gap between the services and the things that might be built on top of Anan layer so Igan layer basically is able to coordinate and provide an economic uh incentive for all the all the validators of ethereum or most of the validators of ethereum to run something else right the core Insight is something else can run on the same computers and you end up with a sort of decentralized AWS right that's their vision but what you get when you try to build something on igen layer or any of the other reaking systems is you basically get an empty computer right and then what you do with that empty computer you don't have a chain you don't have anything that you can really you don't have a way to take payments uh you don't have a way to orchestrate the type of code that runs on those systems right now basically what you get is a Docker container so a Docker container is a virtualized Linux system right but it contains an entire Linux distribution in it and and then then you're on your own okay now you have a computer and that computer has slashing conditions then what right uh how do I take payments how do I if I want to make a derivative or I want to set rules about how this thing behaves uh if I want to you know compute ZK and then send it somewhere how do I do that if I want to compute something on AI and I might need three of five consensus on the results to make sure the AI is doing the right thing what how do I do that I don't know so a lot of people have tried to solve these problems but there's something there's something missing here right there's there's there's something that we we you don't know how it's going to work uh I think it starts to look a little bit like this I don't know if anyone's seen this video but I find it very hilarious this is someone trying to turn on a a van but the solutions look a little bit like this to me um so some people say okay well we need something looks like a stack you can't really use op stack that doesn't really makes sense uh maybe we could use some other you know polygon has some stuff that maybe you could put in there but no one's really successfully done that thus far that I that I know of um you know but you need you need something in there right maybe but a lot of people want to build a chain they want all the benefits of a chain but building a chain is hard right so one of the ways you can do that this is the cosmos logo for people who don't know Cosmos is basically the only game in town for how you can build a whole blockchain with a pre-existing stack a pre-existing framework there are a number of companies doing this like a lined layer for instance um is using this for their ZK prover Network it allows you to build a chain but it has a lot of problems it's hard to upgrade uh you know the modules aren't aren't actually modular um you know it has a lot of issues um and so a lot of people find a lot of difficulty with this and it's not also it's not fundamentally built for a layer it's not built for restak but what do we put here layer is what you put here so layer layer is a novel stack you could build a uh you could build an L1 with it um we decided to build it from the ground up so that we could have this reusable modular stack that is benefits from the last 10 years of building blockchain Stacks um some of my co-founders are amazing people that have built uh things before they uh one of them Ethan built a lot of the core Technologies of like IBC tender km wasm which is the virtual machine that's uh very ubiquitous in that in that domain um so layer is a full stack but what it also is is is a chain it's a chain that sits on top of igen layer so it's running on all the IG layer validators and it has normal smart contracts everything you think about with the chain but it also has something else this actively validated layer architecture um and what that really is is it allows containerized avss so instead of just this empty computer that you get with slashing instead you get a chain that's already running it's already recruited all of the proof of uh all of the the lrts all of the stake that you would need all the economic security that you would need it's already recruited all the operators that you would have to go convince to run your code um but instead of doing all that and instead of starting with an empty container and making making your own chain or doing some clui offchain thing instead you just ship a compiled code bundle or just basically a computer program you can think of it that's really your business right if you're a ZK company if you're an AI company you should spend all your time thinking about that you shouldn't spend your time thinking about maintaining and upgrading a chain or doing weird things that that are not really secure um in order to take payments or to make derivatives or interact with defi Etc you get all the benefits of a chain but you can just ship your ZK code send it to us you can tell us you know I want to run it on this many machines and I want the result back um I want to orchestrate it I want this thing to run at this time it's all done for you so basically it dramatically reduces by an order of magnitude the uh time to Market and the cost to market for anyone who's trying to build in this restake security model um to tie it back though what this really does is it it lets us achieve that web3 Vision because now you don't have to touch any web2 service at all to deploy a fully functional application you could deploy front ends to this if you wanted to um it's a fully onchain fully crypto economically secure system that lets you build whatever you want in a way that was sort of in that original vision and I don't think that's ever really been the case before um so containerized AVS are kind of uh one of the centralized one of the central core insights of this technology um so slashing conditions payments Etc it also is a reaking protocol because if you think about it an ABS something that would have been built on igen layer it needs Economic Security to be lent to it to be given to it to function so uh how do you get that you get it from us it's sort of restak to us and then it's restak again right so there's a reaking protocol that's built inside of this as well so containerized ABS is built on layer or really anything that speaks IBC protocol can restake to each other restake to us um or restake to other chains theoretically so it's also its own reaking protocol as well um obviously it's running on the igen layer validators and then underneath ethereum as a settlement layer we actually have some really special sauce um as to how we uh we're able to read and write etherium as well so you can directly access the ethereum chain from anything that's running on layer so uh I think this is like something completely novel it maintains sovereignty it allows you to build with the entire stock if you want if you want to make your own chain if you want to make your own layer your own actively validated layer you can um it dramatically simplifies development as I was saying uh people that go to do this that go to build an ABS basically the simplest example that they have is it just squares two numbers and it's 20,000 lines of code and it's 100 roughly 100 files and it has no chain no connection into the three other things and then you're kind of on your own and what it also do does is it allows a completely new type of system to be built so if you think about this there are things that we we haven't even anticipated that I'm sure people will build with this when you can make a fully onchain system and you have persistent things like AI agents that can be running on this system and those things are composable with each other and they can call other ones so for instance say there's an AI system that's reading financial data and it's reading sentiment analysis and it's you know maybe be triggering a trade in a def platform and then that trigger that trade that triggers something in a defi platform updates a bunch of oracles and those oracles then you know maybe underneath are using some kind other kind of analysis as well some kind of AI analysis and it's all proven by ZK without ever touching any centralized system that's now possible and as those things composed it creates a new architecture uh for all of web 3 and I hope people build on it I hope they use it we just came out of stealth today so today is the first time anyone's ever heard of this company in public um um layer on E is our Twitter handle layer. XYZ is our website please check it out follow me on Twitter thank you very much