The Sovereign Rollup Thesis | Mustafa Al-Bassam (Celestia), Preston Evans (Sovereign)

Transcript

so sustainable if you assume like one synchronous blockchain will serve the entire web that's ridiculous like you can't assume it's like assuming like one server will will serve inside an internet this episode of Empire is brought to you by fails markets an options platform on arbitrim and optimism that gets you exposure to crypto in the simplest way possible they've made it as easy as simply choosing up or down you'll hear more about fails later in the show all right everyone welcome back to another episode of Empire we have Mustafa the co-founder of Celestia Preston co-founder of sovereign Labs uh Professor Mustafa welcome to Empire guys hello good to be here good to have you guys here uh fun fun time to be doing a

conversation on Roll-Ups uh there is for anyone who's been reading uh John Charbonneau uh there's a big big debate this week on uh basically just what defines a roll-up ignited by the two John charb pieces and I think my knowledge has basically if you've ever seen that chart where it's like you think you know everything and then you realize you know literally nothing and then you slowly build back up from there I think I'm at the point of realizing I know literally nothing so I think what would be helpful uh for me at least is like let's start at the very top before we go deep and so I think Preston I'll throw this first question to you just like not even what is a sovereign roll-up but what is a roll-up and then we can go deeper from there yeah okay this is a great question

um so James prestwich is the guy who I think has maybe the best definition here and his definition is that it's um it's another basically it's a virtual blockchain so the idea is you have some additional state which the main roll-up nodes don't agree on um so basically a roll-up is some subset of the full nodes of the regular blockchain the L1 keeping some additional state which the rest of the L1 nodes are not aware of um so really if you take a step back the idea is that you're running a new virtual blockchain on top of an existing platform so you're inheriting consensus from some L1 but you're not inheriting their execution rules so the great thing that gives you is that you can inherit what people typically call the security of the L1 meaning it's consensus

guarantees um without having the limitations that are imposed by the need for decentralization on the base layer two different types of Roll-Ups here well maybe maybe actually three of kind of the the two big ones right now are ZK Roll-Ups and optimistic Roll-Ups and then re like you guys have kind of introduced Sovereign Roll-Ups but really specifically that's probably tied into the ZK Roll-Ups I don't know how you how you bucket that but maybe take us one level deeper and like what are the two distinctions here sure yeah and Mustafa feel free to jump in here I think Mustafa is the one who coined the term Sovereign roll up um but broadly speaking you've got optimistic and ZK and that's about sort of technical implementation details of how the roll-up and state transition is

is proven to like clients so taking one step back you've got this blockchain right so it's a set of rules that has been socially agreed upon um that you're gonna run and you say whenever a transaction happens we're going to say like check signature and if the signature is valid then we'll transfer some money from person a to person B okay so you've got these Roll-Ups um with their their logical State transition function now anybody who's running a full node just implements those rules and that's the end of the story for them as soon as the transaction lands they execute it and they know what's going on the distinction between optimistic and ZK is okay let's say that you can't afford to run a full node so a good example is like you're an end user on a mobile phone or maybe you're another

blockchain that cares about the state of this roll-off chain how do you find out what the state is so broadly speaking there's two ways you can go about doing that one is that you can just ask somebody right so this is an optimistic roll-up somebody goes on the chain and they say hey I promise the state of the roll-up currently is X and I'm willing to guarantee that by putting up some money and if anybody else is able to prove that the state of the Rope wasn't X then they get my money so it's a way to get an economic guarantee or a crypto economic guarantee that the state of the role is correct all you need is the ability for one person to come in and basically fast check them so as long as there's at least one honest guy who's watching then they'll want to earn the money so

they'll keep the guy who's attesting to the state honest so that's an optimistic roll up the other thing you can do is a ZK roll-up which is just using fancy cryptography to prove that the state is something rather than doing this crypto economic game um so that's that's ZK versus optimistic that that's pretty orthogonal to the question of sovereign versus smart contract Roll-Ups um and this distinction is maybe more of a social distinction than a technical distinction again Mustafa feel free to jump in here but broadly speaking the idea is if you're a smart contract roll up you've got some enshrined contract on a particular later one so a good example is like arbitrum or optimism they have a special contract on the ethereum base

layer and this contract is responsible for validating the state of the roll-up and sort of telling ethereum about it and it's very deeply baked into the architecture um so most of the assets on optimism and arbitrum are actually bridged from the ethereum base layer in a sovereign roll-up the idea is that you don't have some special smart contract instead every end user sort of decides the state of the roll-up for themselves and that means that there's less focus on bridging Assets in and more focus on sort of native assets on the robot hmm I want to double click on this idea of uh it being a social distinction rather than a technical one I was reading a blog post on uh Celestia's site so Mustafa maybe I can throw this to you it

says to not enshrine a settlement layer is primarily a social distinction rather than a technical one which means that there is a social contract between the Roll-Ups community that the role of transaction validity rules are defined by the community rather than an immutable L1 contract you got to break this one down for me yeah so uh what you quoted there is actually the article uh at first defined what a serving Roll Up is and we're first introduced but it's not even Roll Up is um and as I said uh like what a sovereign a settlement layer or a bridge and what does it mean to not enshrine

something and I argue that it's mainly a social distinction not a technical distinction um which means that if a if you could if their roll-up Community should agrees that the upgrade process for a rollout is that they will agree they have a process where they can hard fall to grow up and they agree that it's not the bridge that defines what the state of the roll-up should be it's the community or or running the nodes for the roll-ups that should Define what the roll-up is and that's kind of like playing into the recent debate is the role of the bridge or is it the chain and that's what that's why and the fact

that I say successful civilization uh like one of my first um kind of published that post uh like Kelvin from optimism thought it was kind of weird because it's obvious like of course like yeah of course uh the the node defines the chain and he he himself sees to some extent that up to the optimism chains can be solving roll-ups and that kind of really plays into the fact that it is really a social distinction and it um more so than a technical decision am I understanding it correctly in that a sovereign roll-up is really technically not that different than it like your typical eth roll-up except

that there's this social distinction that it's um it's really uh what you're describing is it's about the community that decides on upgrades via Forks rather than the bridged am I understanding that correctly yes but there is also kind of like what we've been discussing recently this idea of social sovereignty and Technical sovereignty um so like in order to have that social distinction in the first days you also need some form of technical sovereignty which is idea like you need do you need um that you can't to be to to even have a a concept for something to be useful you need to have the ability for end users to run nodes that

follow the role of chain directly not just via the L1 uh because on the L1 you obviously have a bridge contract between to the roll up and that bridge contract basically operates as a light node or light fire color roller and in many cases people don't need to run alert for the roll up they can simply um run the L1 chain and inside the alarm chain is a live client as a smart contract for that roll up so you kind of like implicitly indirectly running a node flight roll up and if you do that it's not what we what I would describe as technical sovereignty that's what we described like as a smart contract roller but on the other hand if we talk about a server and roll up there needs to be the ability for users

to follow the state of the roll up and get roll up block headers not just by following the Unchained smart contract but actually by by collecting to some PHP Network directly and downloading the blocks and checking for themselves um directly that the robot blocks are valid whether it's for you optimistic or for food proofs or other key proofs and that's what I that's what we refer to as technical sovereignty and that's what Sovereign SDK allows for as well before we completely lose everyone who's not the like 120 people in this world who really cares about the distinction between Roll-Ups um why don't we why don't we actually zoom out even more and we typically don't do this to kick it off but maybe like let's go even more high level than

this and just talk about like your vision both of your like vision for the world and how crypto will look in maybe three to five years and like why the hell it matters that like like why are we talking about this and maybe actually this is a good time to plug like Sovereign and celestion like I think both of what you're building to me in to me when I understand both of what you're building it's not things that are like incredibly imperative to today's crypto world but will be in but will be very essential in the future crypto world so maybe Preston if I could ask you to first like describe your vision of the world in crypto in like three to five years and then tie that back into like why this whole conversation about Roll-Ups and Sovereign Labs really matters

yeah absolutely I'd be super happy to um so before we get too deep into this one thing that's super super important to understand is this notion of a trust Zone so broadly speaking when you have a blockchain you've got some set of minors or validators some people who are in charge of defining what data is on that chain and within one of those trust zones so for the base chain and for any Roll-Ups that are defined in terms of the data on that chain you uh you're able to circumvent some limitations that would that you'll run into if you tried to go across trust zones um so in technical terms there's this problem called the asynchronous fair exchange problem that's the problem we're we're interested in here if that's

if you want to look into the technical details but broadly speaking if you're trying to bridge across different genes you have this fundamental problem which is like let's say you've got tokens on Solana and you want to move them over to Ether right now what you want to do is you're going to lock them up on Solana and then the ethereum blockchain are some smart contract over there is get a look at the state route of the Solana chain and it'll say oh yeah those tokens have been locked up so it's safe for me to Mint them on the ethereum chain and then you can use them over there okay so the technical problem here is that those two things are in different trust zones which means that if one of them hard Forks or if there's just some problem and the chain rolls back then

maybe that transaction no longer happens so let's say that Solana somehow rolled back um suddenly you've minted these tokens on ethereum but they're no longer locked up on Solana and so you've just double spend your tokens okay so across precedence this is a fundamental problem there's no way around it if the chain hard Forks or rolls back then the other chain is just straight out of luck it's got these copies of the tokens that are not backed by anything anymore so what that means is that you want to have large trust zones right you want a lot of different activities all happening within the same trust Zone so where I see the world going in the long term is that we'll have a few base layers that dominate so um maybe three to five is it's a good

approximate number you know who knows but a very small number of base layers with their own validator sets and then on top of those base layers you want to be able to do all sorts of different things right so think of the internet um imagine if there were only two different programs that you were allowed to run on a computer right that would make no sense what you want instead is you want lots and lots of different computers running lots and lots of different programs that can be deployed in a permissionless way and then they can all talk to each other um so the technical way we describe that is that you've got synchrony within your local machine so you can you know you can access your hard drive you can do computation and then you've got asynchrony when you want to interact with somebody else

um so let's say that you want to download a nice uh cat gif from Pinterest right your computer is going to make a call over the network and that call might fail but that means that you're decoupled from Pinterest so if Pinterest goes down your computer doesn't stop right so I expect blockchains broadly to play out in this exact same way um you'll have lots and lots of different chains doing lots and lots of different things and they're all their separate little Islands right they're all just doing their own particular business logic they're not necessarily even aware of other Roll-Ups or other base layers but they're all on these broadly shared trust zones and that kind of gives you the best of all worlds right this way it can be totally permissionless to deploy a new chain so if you need something that's not being

offered by a current roll-up you can just make your own in the same way that if you want something that's not available on a current website you can just go and deploy your own website um but then you have you have that trust properties inherited from the da layer so again speaking very broadly we expect fpda layers with hundreds or thousands or ideally tens of thousands of Roll-Ups all running on top and I'll able to communicate but not communicating with each other all of the time hmm Mustafa how does this align with your vision of the world yeah so we share a very similar Vision uh like our vision kind of revolves around how we also saw the web develop so there's a lot of analogies with how we

see blockchains develop how we saw the early web develop so like in the when the web was first created to create a website you usually had to have your have your own physical server and that's very similar to the early days of Bitcoin it was like one application per blockchain if you wanted to create a new application as we create a new blockchain it was Bitcoin which was the first application it was money and there was other chains like name coin Litecoin those coin and so on and so forth there was no erc20s or anything like that and then there and then uh the web you had this idea of shared hosting providers like jio City uses early one but it also had like dreamhost Bluehost and stuff like that uh and the parallels of that in blockchains is ethereum ethereum came

about and said uh instead of creating your own chain for every single time let's just create like a shared blockchain with a shared exclusive environment just upload your code and it will run every run on the same chain we can share the same computational resources that's a very simple I'd like another analogous to the early web where we had shared hosting providers you upload your code to geocities dreamhurst or whatnot and your code runs on the same server as like hundreds of other websites who are sharing the same resources but more recently um we realized that that doesn't really scale well and it has very many limitations like your stock for the same execution environment as that shared host gives you or you're stuck with the

ethereum virtual machine like what if you want to experiment with new execution environments that are more scalable like you know so on this virtual machine or move or any other execution environment that supports parallel processing or do class or like this people want to create modify the evm to add new up codes that weren't possible before so then you had you have um certain idea is like the evolution of the shared Road passing providers was this idea of virtualization we had virtual servers we had virtual machines and the cloud so it became the case that he wanted to create and deploy a new application on the web

instead of sharing the same uh like operating system or execution environment as everyone else on a share the web host what you would instead do would go to Amazon AWS ec2 or digital version or any virtual machine provider and in five seconds you could deploy your own virtual server you have your own so now you can have your own execution environment as a virtual server within seconds so then you have the best of both worlds you had the you had the flexibility of your own server but the convenience of using a shared host and the Roll-Ups do the exact same thing that's why they described as virtual blockchains you can now instead of having to create your own chain from scratch using your own consensus or

proof of stake Network you can just create a virtual blockchain that inherits um some um applicable consensus and data layer like Celestia and that's kind of taste into our multi-chain webchain thesis where in the future we expect that there'll be millions of app chains or chains running or shared the Alias like Celestia just like how does it millions of virtual machines or asynchronously talking to each other on the web right now if I understand both of your visions of the world correctly you both think that right now we have like dydx has created their own app chain you both think that there will be millions of apps who have created their own app chains is that a fair thing to say yeah exactly

well yeah okay and will these might be really dumb questions will all of these app chains be built on different l2s like the optimisms and arbitrums or Preston that is actually what Sovereign Labs is building in that like they will hopefully be built on something like a sovereign Labs is you know SDK yeah absolutely that's that's our vision is that most of these chains will need something custom so the evm is of course great and it's there's a lot of reasons why you might want to use it but it's also very limited in some fundamental ways so one limitation is that it's extremely unfriendly to parallelism which puts a limit on how much

transaction throughput it can ever do so if you need a super scalable app evm is probably going to be the wrong choice another big limitation of the evm is that it's transaction model doesn't allow for batching so think of uniswap today right if you want to send a trade on the unit swap you are roughly guaranteed to get sandwiched if you don't use like a special Mev protection API right because the evm itself can't protect you from sandwiching the only reason for that is that the evm can't batch trades so you could trivially build a version of uniswap which didn't have sandwich attacks just by settling trades once per block instead of settling every transaction separately and then everybody would get the same

price so there's no intent at all the sandwich right that would mean suddenly your slippage tolerances could be way lower you know um basically it's just it's just a fundamentally better version of uniswell the problem is there's just no way to implement it in evm and so that's going to be broadly the case for any particular choice of VM right every virtual machine has to make different trade-offs and some of them will be good for different applications but for any particular application like any particular VM is not the right choice so you need flexibility and things like you know evm Stacks just can't give that to you in the same way that like it it doesn't really make sense to have one canonical website that everybody can customize a little bit to

like run the app on right like no you just need to be able to write your own apps and surf them via the web so that's what Sovereign Labs is trying to provide uh okay so like when I think about the Three core tasks of a blockchain it is executing transactions achieving consensus on the transaction ordering and then the da layer right like guaranteeing the availability of that transactional data what part of this in a world okay let's use uniswap for example uniswap goes out they build their own chain um what one of those three things is happening at like that app chain level what is happening like one layer below what is happening still at ethereum what is happening here on Celestia like maybe Mustafa you could you could take this one

yeah so are you talking about like a hypothetical example if like edu swap one thing to be an app chain roll up for example that's right yeah so I guess like the idx is also quite a similar example because dydx um was use stockx so Dy DivX itself was a roll-up on ethereum by the way it would roughly work is like uh you would you have the execution environment and the execution execution environment is the roll up itself so like it was a unicorn for example let's say you would create it in using ethereum virtual machine for example um and that would be that would be an excuse environment um and that would that's that roll up

with post it's block beta to a another consensus and dnaire likes last year and what that does is uh Celestia will make the roll-up blocks it would give ordering over the swallow blocks to see which ones came first so that this objective truth to what the actual transactions are and it would make them available the data the data behind them available so that if there's any fraud proofs they can be generated or if there's any misbehavior it can be challenged with fraud proofs for example hmm okay I'm just thinking I'm trying to almost visualize in my head like you've got

basically the data availability consensus settlement and execution and a monolithic blockchain all of that happens on ethereum with a roll-up the D.A consensus and settlement happens on ethereum but then the execution happens on your own chain um and then with something like a sovereign roll-up the da and the consensus would happen with Celestia tell me if this might all be wrong but like I think the da in consensus would happen with Celestia and then the settlement and execution would happen on your own chain it was yeah I saw everyone roll up or even if it wasn't servable up you could also have a different settlement layer and like if for example you could be a L3 through the L2 and one of the one of the advantages of that might be have um like you have a bridging you have access to a bridging

Hub which was more for trust memorized version I think when a lot of people still think app change they think Cosmos but my understanding here now like getting a better understanding of this is the cosmos SDK is great but the cosmos SDK is meant to build l1s which means that you then have to uh recruit your own validators you have to get your own set of full nodes Etc with a roll up you inherit the security of the L1 so like Preston with what you're building with Sovereign you're meant to build I think ZK roll up so like it just makes it it just makes everything a little easier to get off the ground

I guess that's a question yes no okay absolutely yeah imagine if every time you wanted to play a website like you had to go and build a data center right like sure there's lots of people who are interested in building data centers for you but it's just a pain to actually go out and do it um so just use an existing cross Zone this episode is brought to you by Fales a new frontier in simple on-chain options here's what you do you choose an asset a strike price and the market you want to participate in and that's it with its powerful and capital efficient amm based architecture thales is able to offer low fees automated liquidity and effective utilization of Leverage with

no funding rate and known payouts they just launched this new UI it's super clean step one you choose an asset step two you choose a strike date step three you pick a market and choose the USD amount getting exposure to crypto assets price action has never been easier because of fails so here's what you got to do you got to go to thalesmarket dot IO t-h-a-l-e-s Market dot IO failsmarket dot IO they're on arbitrum and optimism go play around choose a crypto asset choose a strike date pick a market choose the USD amount of bing bada boom check out sales Market let's okay good high level framing let's get back into the weeds a little bit um I think maybe the first place to talk

about is guarantees um I don't know who to ask this question to but like what guarantees are different with a roll-up where you don't have this baked in right or you don't have like this enshrined smart contract uh versus like an L1 where you actually have the enshrined smart contracts does that I don't know who to ask that question to but curious about curious about the guarantees here I mean in theory you should get the same guarantees uh like interestingly um like Celestia used to be called lazy Ledger it's called and the reason why I was called lace Ledger is because like it's um it's it trips back a layer one to only the core components that layer one needs to do to get all the guarantees that

blockchain provides which is you know uh provide a ordering over some set of data so that nodes can get what the final state of application is so it's like as long as you have ordering and consent add ordering a data availability over some data you get the exact same guarantees that any other chain provides it's just that is decoupled maybe tell me about like the the end benefit here uh to the user or something like I'm assuming there's some sort of benefit like cheaper fees or faster confirmations if you're if you're completely Sovereign but maybe tell me about like the why behind that and and if I'm right about like that being the end benefit to the user if you are completely sovereign

yeah so um going back to that conversation we had earlier this is again primarily like a social distinction not a technical distinction really the whole concept of a settlement layer is not fundamental in any way so the reason we have this concept of settlement is a historical accident where the first Roll-Ups were on ethereum and people were thinking of them as like a way to do more execution of ethereum things on a second layer right so then the idea is like okay if the second layer is really part of ethereum it has to settle to ethereum but fundamentally like there's there's nothing about settlement layers baked into Roll-Ups so a roll-up is just a blockchain of its own you can follow the rules of that blockchain completely independent from any other chain um and so you can imagine like having an

L3 which has its data on L1 and then it has quote settlement to two different l2s and there's no contradiction in terms of it right the settlement layer is not a fundamental part really what a settlement layer is like it's a social contract about we think this chain is sort of important to the role in some way so ethereum is sort of important to optimism and arbitrum because pretty much all the assets from optimism and our drum are bridged over from the basement um but really like a settlement layer isn't really a thing it's just a chain that you bridge to yeah so then the question that you're asking is really like what are the trade-offs involved in bridging from a roll-up and the answer is there are are fairly you know

implementation dependent but um broadly speaking like running a bridge smart contract is a bit expensive um and so especially if you're a smart contract roll-up which has to sort of by social contract has to be updating this bridge frequently then that can get quite expensive for end users so let me be very concrete here like imagine that you are a starkware stark X exchange or something your social contract is that you're supposed to settle your trades onto ethereum which means that you have to be posting ZK snarks onto ethereum sort of as frequently as you can afford to but verifying snaps on chain is very very expensive and so that translates into higher fees for end users so if you remove this social contract if

you say know the roll-up is Sovereign on its own and then we'll Bridge back to ethereum based on supply and demand you can translate that into lower fees so the way this works is you create your ZK snark and you do that completely off chain on your you know on your own peer-to-peer Network and you can send that snark to light clients right away so end users can get finality in the minimum amount of time that it takes to generate a snark which is usually like a couple of minutes and then when you want a bridge you can do that on the longer time frame that ethereum forces you to use because of how expensive verifying snarks are so I think concretely on ethereum today like starkler Roll-Ups settle about every eight hours or so um the reason for that is just gas

pricing right it doesn't take stock where eight hours to make this narc it's just that it's too expensive to verify so if you separate those two components out then you can reduce the time to finality for end users um but you can still have your Bridge contract your trust minimize Bridge with ethereum so really this is obviously some productive sort of abandonmental model coming from the history of Roll-Ups right like settlement which is not an inherent part of a role just to make sure I fully understand that one one of those pieces of question the the is that and is that because ZK like starboard for example Z is that because zks are so um like cost intensive basically they consume so much gas that this is sort of like of your sparkware or sarcastic or whatever you you won't actually post to

the block you won't actually uh post to the blockchain and like you know every minute or every 10 minutes because that would be like far too expensive so you'll do it every five hours or every 10 hours or something like that and that becomes a problem if I'm a user because if I'm looking at the chain I don't actually see my transaction for maybe even several hours after it's posted and a sovereign role-up would fix that is that correct yeah exactly exactly okay so to be clear struggler could do the same thing right they could make their proofs and they could send them to users directly um the trouble with structure is that they don't post their transaction data on chain uh they only post like State diffs if I recall correctly and so because the data is not settled on L1

until they post their proof you really don't get finality but if Starbucks were to post their transactions on chain then they could do the same trick where they make their proof soft chain and send them to users and then just you know settle or Bridge every few hours so Mustafa tell me about your prediction for just I guess what happens with optimistic Roll-Ups here then because it seems like uh with Optima optimistic ropes you need the fraud proofs and with ZK you don't so it seems like optimistic Roll-Ups eventually get run over well in theory uh like um theoretically like ZK roll-ups kind of like have bad properties and optimistic Roll-Ups in the sense that you don't have to wait you don't have to wait for four proofs but I guess the main reason

why people are using optimistic problems right now and is simply due to the high proving costs of proving certain computations Suzuki Roll-Ups like for example if you want to create something in the evm creating a zika evm is very expensive but like hopefully uh like I'm like I'm kind of agnostic I like I like both optimistic and ZK robots I'm very neutral and like uh but my hope is like of physically Roll-Ups of course to hopefully it will become more and more efficient in the future when it comes to proving these competitions Mustafa you were talking earlier I guess you both were about like comparisons and analogies to the web when you think about what you're building at Celestia are there comparisons for the smooth

brain folks like me like comparisons to early web or AWS or cloud like real like HTTP like any analogies that you can pull out of this of what maybe that you both are building uh to kind of help me visualize and get it get a better understanding of of really what we're doing here yeah I mean as I mentioned I think the best analogy is just like how the cloud lets you deploy virtual servers without having to maintain your own physical infrastructure Roll-Ups allow you to deploy your own virtual blockchains without having to maintain your own blockchain consensus infrastructure or the or maintain your own layer one

network I think I think that's the kind of like the best analogy um I could effectively find and also another analogy like when it comes to moisturizing the stack uh like there's this thing called like the the OSI stack was like computers today they're building a very modular way like you have like the hardware there you have the networking layer we have the application layer and so on and so forth and I think the key takeaway is like as any technical technology type matures it starts to mature by motorolaizing so imagine like we have TCP IP for example and that's like on the network layer uh like it's a protocol for computers to communicate with on top of that we have

HTTP which is how websites work but like imagine we had to imagine if we had to uh kind of like create a whole new network in there every time we wanted to create a create a new kind of like way to use the internet like imagine to create the web or like to create HTTP imagine so for example like you have HTTP and https okay https is like a more secure version of HTTP uses encryption by imagine imagine if to in order to deploy https we had to modify the entire networking layer of the internet and modify the actual routers and the actual like Wi-Fi chips and everything like that like it will take ages and

that that's exactly like what we've been doing with IPv6 ipv4 like we're running out of IP addresses and this IPv6 which Inc which um increases the number of IP addresses and that's basically taking like two decades to get massive option because you have to modify like every Wi-Fi chip every like all the hardware all the routers and so to set you can use the same analogy for like blockchains imagine if you had to every time you had to create and modify your execution environment in some way that you want to experiment with a new upgrade to the EDM imagine if you had to create a whole new layer one just to experiment with a new execution environment that it will be insane and that's basically how we've been operating over the past 10 years uh like we've been stuck in this like Loop what

I describe of like every um this one really thick layer one Loop like every cycle there's this new layer ones that have incrementally approved up on the all the other ones like on the first cycle you know 2015-16 we had ethereum and the next cycle we have things like EOS uh them I think and then like a bunch of others cardano in the previous cycle we had Solana Avalanche and now we have sui and atlos so it's like when does it end like we can't it's like it's not sustainable like the industry needs to mature a little bit like we can't have like this like the constant new day ones I just copy all the applications in the previous layer ones the same nft ecosystems the same application exactly the same applications the early selling

points like it's a new layer one just just for incremental Improvement so you okay so if you're looking at this OSI stack like you've got seven layers of the OSI stack right and like down here you've got like the fiber and the wireless and then you've got the ethernet TCP apis and sockets like SSL the presentation syntax layer and then you've got the end user layer like the application like https and things like that um yeah what you are saying is that what we're trying to do in like in in most of Krypton with ethereum specifically right now is when there's a change that should happen it just at the HTTP level like the end user the application level level we have to go all the way down sometimes to the like the packet level or the frame level and like modify ether like

modify TCP IP when really we should just be sitting up here modifying the syntax of the end user layer exactly yeah so what's the what's the bar so like in the OSI stack to keep this analogy running I might uh running out too far here but like the physical and the data layer like and even the network layer like really don't change like TCP like ethernet like the wireless cables like no one's messing with those things these days I at least that's my understanding um so your understanding here would be that like ethereum like what what's the what's the base of that layer like is it is it going to be ethereum are there like several several of these l1's that will win what what do

you guys both think here I mean yeah I mean it doesn't matter perfectly to the OS type because ultimately it's analogy but like the way I view this text in a boxing there is I I Define there zero as social consensus that's where all thoughts uh like you have for layer one to have meaning in layer zero social consensus to agree with the rules of the layer one is and I see the layer one as the kind of data consensus layer and then Layer Two is the execution layer and then on the layer two on the execution layer on the layer two execution layer you can have applications on top of it which is like the layer above

oh it will describe them as layer threes but I would describe them as a layer above you can like if you have an evm spot contract fall up on Layer Two you can deploy other you can deploy applications on top of it Preston would you agree with Mustafa on most of this yeah I think Mustafa and I broadly agree on this thing um you know in crypto we like to make distinctions of like L1 L2 and then now people are saying like 1.5 um all of these things right like a lot of this is sort of it's it's stuff that's fun to debate over but not super like inherently meaningful um I think that the point that most definitely I agree most strongly with is that social consensus is like layer zero

or like the most fundamental thing and everything on top of social consensus is a bit malleable um and then I also really like his point of L1 being like the trust zone so the validator set yeah um we'll just sort of where like computers and the social consensus interact and then on top of that you've got freedom and you can stack things as high as you want to go and it doesn't really matter right you're you're all within the same trust Zone if you're sharing l0 and L1 hmm when does so both of you share this this app train thesis belief when does all of this become a need not not a want is it in the next potential bull cycle where there's just a crazy amount of demand and like what yeah when does all of this

become a need not a want I think pretty much like yesterday when uh you know Bitcoin and ethereum feast every single bull market l1c Skyrocket and it's like okay whenever every single bull market is like okay crypto is not ready for Mass adoption um so I think ultimately the ultimate lead is we need scaling and that's why ethereum is now doing a roll-up Centric web map um so like scaling is obviously one of the one like the most important reason why we need um like this multi-chain or upchain thesis uh because I think everyone well not everyone like Solana isn't on the same page that so sustainable to assume like one

synchronous blockchain will serve the entire web that's ridiculous like you can't assume it's like assuming like one server will will serve the inside of Internet uh but other than that I think we need also need it it's very important to break it's very important to have up to enable app chains to break out of this layer one loop I described but every time we want to make it incremental Improvement to execution environment uh we lost a new A1 uh and it's not sustainable uh like it's not economically sustainable at some point like these layer ones are not going to be funded anymore um and so that's why I think it's also very important to have ROBLOX there's one more really interesting

point here which is that you frame this in terms of a bull market right like when the next bull market comes we're going to need all these app chains I think that framing is very much indicative of the state of crypto like right now you're sharing this single computer between the entire world and so the only thing you can run in on that computer is like the very highest value thing that you can think of so if there was only one Mainframe in the world we would probably use that Mainframe like to run NASDAQ or something right we would use this for something incredibly high value but nobody wants to live in a world where like nobody has computers except for Nasdaq right so what we're building out is the infrastructure where suddenly everybody can have a computer and it's too early to say exactly what

people are going to do with their computers like people didn't necessarily predict Friendster in Myspace and Facebook and then Tick Tock and Instagram right but it's obvious from even where we're standing today that like there are a ton of applications for useful blockchains in the same way that there are a ton of applications for home computers even if you don't know quite what they are yet and so I think you know 10 years from now we'll look back and we'll think it's kind of ridiculous that like activity was so tight the prices that's just an artifact of the fact that everything on chain is financial right now because chains can't support anything non-financial um so we're trying to build out that layer and then you know I think we'll see a lot of exciting things you'll probably see gaming and social as early applications

um but broadly speaking the reason we need app chains is not just to scale finances it's to enable interesting use cases that are just not possible with the limitations of blockchains today president Mustafa are there anything any things that stand out that you guys might disagree on well I can't think of any right but I think we've had various disagreements about how you see realize serialize certain data structures in like certain protocols but I don't think there's anything like super fundamental I think I think one thing that we disagree slightly on is where we see optimistic Roll-Ups fitting in in the future um so we're both broadly fans of ZK and optimistic but I'm I'm more bearish I

think on optimistic roles in the long term than Mustafa is um so I'll tell you sort of why and then Mustafa feel free to put us back but the reason I'm a little bit bearish on optimistic Roll-Ups is that I think we have this shared vision of a future where you have tons and tons of different app chains and they're not necessarily all aware of each other so the thing with optimistic Roll-Ups is that they have a very weak assumption right they have an assumption that there's at least one honest person per person watching the Roll-Up okay but in a future where there's a million different Roll-Ups maybe that assumption doesn't hold up anymore because you might have some like you know like I have a personal website that's the little draft that's available on the corner of the internet but I

don't think anybody has ever gone there via Google right like I don't think anybody but me has ever seen this website um and so but if your infrastructure assumes that like there's always going to be an honest person so it's always safe to bridge back and forth with any optimistic roll-up that assumption might break in the future when we have you know when blockchains are where the internet is today so for that reason I'm really bullish on ZK just because it gives you complete firm Trust nothing can ever go wrong and so you can build simpler and more robust systems now in the short term I think optimistic has a great future because we're you know a decade away from from where we see this ending up but I think that's one area where maybe we disagree slightly

yeah I mean I think ultimately it depends on the scale you're talking about like I would say uh like using having optimistic roll up app chains definitely allows you to get a lot more chains securely than like using your own layer one value assets as you have with Cosmos right now because of course like having a honest minority assumption scale is better than having an honest majority assumption but of course having a no uh a no honesty assumption I'm just verifying a proof they're actually uh scales even better they're even better but yeah I'm like I'm bullish both I'm optimistic and CK Roll-Ups um and I think you know I think ultimately 40K Roll-Ups it's just a matter of reducing improving costs and it also

kind of like pays a lot into the idea or Liberation discussion of like what optimism saying which is like the roll-up is not the bridge to fill out for example like optimism actually wants to also ZK approve their role up like you can have multiple Bridges one of them is optimistic bridge and not another of them like a ZK Bridge you can even have a pessimistic role that doesn't have either four proofs or optimistic or validity groups you just have to run a full node and interestingly that's actually like how uh like people build application applications on bitcoin before ethereum even existed so that you could say tether itself is a pessimistic roll-up that uses Bitcoin uh because it uses Bitcoin as a DA layer

that's interesting what is the uh in this world that you both Envision where does the value accrue here and maybe we can continue moving forward with the with the unit swap example here yes I mean my thesis is that there will be um fewer dnaires than there are execution layers and there'll be fewer execution layers in our application layers and my thesis is that the total value of the application layers will be very early in the terminal value of layers below it so for example like the total value of the execution layers will be greater than the total value of the data layers but um the value of the bigger Claire will

be bigger will be potentially be bigger than the value of the biggest application for example simply because of the because there's less DL errors or application layers on average potentially uh so it's kind of like similar to like how AWS provides infrastructure but like the the value of anybody else is much larger than uh the average application you know it wasn't available is but once you take all applications built on the lws and Aggregates they're probably greater in value uh than the value of AWS itself that's that's like my my theory um but I'm not like strongly related to

it Preston what do you think yeah I think that's a very reasonable thesis there um you know so one interesting thing to think about is exactly this where does the value accrue I think you generally accrue value whenever there's scarcity right so if you're providing something that not many other people can provide then you can accrue value from that so da layers will approve value because being on a particular da layer it makes you part of the same cross Zone which means that you can Bridge with other Roll-Ups in a trust minimized way so for that reason GL leaders have sort of natural Network effects which means you should expect them to accrue value you'll also see value accrual on app change specifically which are popular because uh any popular app is going to

have more demand than it can possibly Supply because there are limits to how much you can ZK prove how much you can run on a single computer and fundamentally like the idea of a single chain is that it needs to be able to run on one individual full node we don't currently have a good robust way of you know mapping a single machine to a bunch of computers so the peak throughput of an app is the peak throughput of a single laptop basically which means that any app is always going to have you know basically a few Market emerge because more people want to interact with it than it can support um so this is actually a great reason why people should build app chains if you're a token on ethereum today like if you're a popular app you can launch like a governance token but all of the free

revenue from your app accrues to the ethereum validators and the only way to make your token valuable is like either you have some you know like social pump or something right which can happen um or you have to like try to try to Wrangle some Innovative way to like steal away the gas freeze from ethereum and make them accrue to your rollup token but if you launch as an app train then like it just happens you you immediately start accruing the value of your app to your app token so it's a much nicer much more natural way to align communities um and to reward the people who are actually creating value so like you know uniswap is responsible for something like 80 of the fees on ethereum today and the uni token accumulates basically none of that right until they enable the free switch it does accumulate none of

that and even then it will only you know accumulate a small amount because much of unit swap is actually happening at the gas layer um so there was a bit of a tangent but broadly speaking we expect da layers and then application tokens to a group pretty much all of the value huh that's interesting because I was trying before this recording I was trying to think of both of your business models and I wasn't fully sure I was trying to think of the business models and the moat and I guess maybe I'll turn that into a question for Mustafa which is like how hard is it for an L2 or an L3 once they've they're using Celestia as the data as the da layer like how how tough is it to for one for an L3 to move from one da layer to another

um and maybe there's yeah yeah I mean we want to be as easy as possible even though that might seem contradictory to you know our goals uh like we uh like we once you see like common ba interfaces um kind of like being developed in the industry so for example that we we've even contributed like a modular DNA interface to the op stack uh because I got like the semester only makes sense in the world where there is a modular blockchain stack and to um kind of like get to that we actually need to cut me to make it possible to replace the a layer because I don't think anyone wants to build

a wall Garden I don't think anyone wants to build an application on a specific piece of infrastructure without knowing that they have the flexibility to move somewhere else if they wanted to and without doing a lot of work um and I think that that's a major advantage of a modular stack like if we create these common interfaces then developers can uh have like a lot more flexibility or freedom and confidence if they choose a particularly their execution layer uh they do because they're using common interfaces they can switch if in the future if they wanted to and they're not like building in a wall garden and stuck to a particular module yeah and when you when you guys I'm assuming you guys have done a bunch of

like user conversations and customer conversations like when you talk to those folks when you talk to potential apps uh apps who might go build their own their own L3 is there any conversation about like hey we we might actually just run our own da layer or we might just like we might just build this in-house basically is that a conversation or they all like this idea of Outsourcing it uh the generally or like uh the idea of Outsourcing Outsourcing it I think some Roll-Ups have like uh like have like I don't existing for example wants to develop DK Porter uh but like very like it really um kind of like comes to fruition at least the same extent that we're bowling like we're building a decision layer so

for example like Albus room has any Trust uh but that's like a like a centralized data availability committee a software has something similar uh like a sexualized data available committee and like we've been actually like talking to software projects and we're working we're trying to work with stockware to kind of like make it possible to deploy stockware Roll-Ups or stockx Roll-Ups using Celestial SD layer because when we focus these stockx Roll-Ups um they don't like the idea of having a centralized data availability committee that's being provided by default and so like if you want to do things properly and have a decent fights data layer you can't just like they wake up and decide to do it like it's actually very very difficult because

you have these actualized you need to have um decentralized valid assets but also you need to be incredibly neutral uh like if especially for application specific roll-up um they prefer building or are incredibly neutral the area like let's say like uh polygon had it's from the earlier like it's unlikely for example like arbitrary and we want to use polycon polygon avail and maybe that's what part of the reason why Avail uh spun out from polygon to its own independent project so it can be more currently neutral nice Preston let me just ask you then um I want to actually just learn a little more about this announcement so like

until now it's been very tough for a developer to deploy their own CK roll up um you guys just launched this uh this SDK to do this is this something that folks can use today is this something that's like here but coming soon tell me about the SDK that you guys just launched yeah absolutely um so Sovereign broadly speaking is building a toolkit for building Roll-Ups so we were talking about this analogy of like the OSI stack earlier imagine that every time you wanted to launch a new web app you had to like build your own HTTP server and then you had to like write your actual application logic in assembly by hand that's basically where we are with Roll-Ups today like every single roll-up team has their own full note

implementation and they've got a full-time set of Engineers who are just building this full node and then on top of that they've got like PhD cryptographers who are handwriting circuits to like vary in a very specialized way um to encode the application logic of the ZK world that's just never going to be sustainable or scalable right so what Sovereign is doing is we're building a toolkit where you get your full known implementation totally for free and you can write your business logic in a normal programming language that lots and lots of developers know how to write and then we generate all of the snark machinery for you uh so suddenly instead of having to like recruit a whole team of people you can launch your own roll up in just a few hundred lines of code

um so we have demos of this today now we just launched our Alpha and the alpha is basically like hey for the first time Sovereign has enough features that you could actually use it to build an application with um the alpha is definitely an alpha release like there are security problems that we know of there's lots of performance improvements that are coming down the pipeline but this means basically like for the first time you can deploy a roll up on Celestia with just a couple of lines of code and you can fully customize Its Behavior like you can use accounts you can use utxos if you want to deploy a you know a full virtual machine you can do that if you want to have an app specific role you can do that so that's our Alpha release it's the first time that Sovereign has

enough features that like it actually runs and you can use it um and we're headed towards beta which will be when it's like you know we've we've done some audits where we're very confident that it's reliable and secure and you can use it to build up it's just not quite fully feature complete and then we'll release a full version after that nice that's exciting and you guys are working with Mustafa yeah on the on and Celestia on the data available uh the da layer yeah so The Sovereign is is pretty interesting in that it's fully generic over basically every component so you can plug in different da layers you can plug in different ZK snark systems um but Celestia is one of the leaders in the da space um and so we have been uh we use them for our demos we've also got a

partnership with a bail the formerly polygon Avail and we're building some demos with them as well right now but yeah Celestia most of our design thinking was was initially based around Celestia because it's such an interesting environment to program for right because there is no smart Contracting layer it forces you to think very um very robustly about exactly what capabilities your roll up needs from its underlying VHA so so last year was our first integration and is where our downloads run today and we also expect it to be a great place to run in the long term nice Mustafa anything coming from you guys this year that we should be uh on the lookout for absolutely we have mainnet coming this year uh it's gonna be a combination of four years

of effort since laser white paper was first grade in 2019 um so yeah staying true to stay tuned for that amazing exciting all right cool guys for Preston Mustafa I will admit some of this was a little over my head but I definitely learned a lot and yeah I appreciate uh appreciate you guys coming on definitely giving me a better understanding of what you guys are both building but also just the the app chain thesis Roll-Ups l-3s in general so yeah really appreciate the time likewise [Music] foreign [Music]