On The Brink with Castle Island - Nick White (Celestia) on building the first modular blockchain network (EP.299)

Episode Date: March 21, 2022

Celestia Labs COO, Nick White, joins us on the show to kick off our modular blockchain mini-series. Celestia is the first modular blockchain network, which decouples consensus and data availability from execution and settlement. Celestia implements a novel design to unlock key properties such as scalability, interoperability, and sovereignty while preserving decentralization. We cover:   Why scaling and decentralization have historically been at odds How Celestia creates a modular system to reimagine traditional monolithic blockchain architecture Leveraging data availability sampling to unlock scalability How Celestia intends to be the engine that powers sovereign communities Learn more about Celestia here. 

Welcome back to On the Brink. This is Ria from Castle Island Ventures. In this episode, we had the pleasure of sitting down with Nick White, the COO of Celestia Labs. Nick breaks down so much in this episode, including why skilling and decentralization at the base layer have historically been at odds, how Celestia reimagines traditional blockchain architecture from being monolithic to specializing and creating modular layers. that plug into one another, the importance of data availability and how Celestia leverages something called data availability sampling to unlock scalability while preserving decentralization, and how Celestia intends to be the engine that powers sovereign communities and much more. I hope you enjoy learning about the paradigm shift that Celestia is enabling as much as I have. So with that, let's turn to

our conversation. Welcome to another episode of On the Brink. This is Ria from Castle Island Ventures, and I'm joined today by Nick White, who's the CEO of Celestia Labs. Celestia is the first modular blockchain network, and at a super high level, this means that it decouples consensus and data availability from execution and settlement to unlock features like scalability, interoperability, flexibility, while preserving decentralization. We'll get into exactly what that means, but I'm very excited to have you on, Nick, because you have an incredible ability to break down really technical concepts into layman's terms and with a lot of patience. Before we get started on modular systems and Celestia, could you tell us a little bit about yourself, how did you get into the industry and your journey

and crypto so far? Sure. Well, first I just want to say thank you, Ria, and the whole Castle Island Ventures team for having me on. It's a pleasure. I've listened to this podcast a lot and drawn a lot of insights from it, especially on, you know, the parts, the non-technical parts, like the regulation and just like the broader analysis of the space.

So I'm really excited to be here. And I want to congratulate you guys on your new fundraise and really excited to see where you guys go in the coming years. So about me, I got into crypto in 2017. I originally studied AI like neural networks, but crypto just kind of grabbed my interest. And so my first, I fell down the rabbit hole. And then in 2018, I co-founded a project called Harmony, which was essentially building a protocol very similar to the ETH two designs at the time. So we were implementing sharding and proof of stake. And that project was, you know, learned a lot and the project did really well.

But eventually through my research into scalability over those two and a half years at Harmony, I encountered this white paper called Lazy Ledger. And I had read a previous paper or skimmed it, you know, to be totally honest, called fraud and data availability proofs, which is kind of prep my mind for it. When I read Lazy Ledger, it just completely floored me. And basically for two weeks, I practically couldn't sleep because I was so intrigued by this idea. It was, and I remember the first message, so after I reading the paper, I was so taken by this idea that I googled the project and I went into the Telegram channel and I wrote this message. and the core part of the message was

this is the most innovative idea in blockchain since Ethereum. And essentially, I mean, we can get more into it, but basically there was just a massive leap forward in this new movement that we're now calling modular blockchains. And so once that happened, I just couldn't go back and I was so obsessed that eventually I ended up leaving harmony and joining Celestia full-time. So Celestia is the name of the project that was formerly

called Lazy Ledger. You were really early in recognizing this paradigm shift that Celestia and modular systems could enable. And now a lot more people are stumbling upon the work that you're doing and quickly realizing how groundbreaking this architecture could be. So I guess just taking a step back, It'd be great to know how your experience at work at Harmony informed your understanding of an excitement about Celestia. What did you learn there that made you feel like this could be something that is extremely groundbreaking? That's a really good question.

So I was very deep into the literature around scaling and had thought about these problems a lot because that's what, that's the problem that Harmony was trying to solve. but the angle that we took was sharding. And there were still a lot of sort of unsolved problems in charting. One of them, namely being data availability. And there was also just a lot of drawbacks to the sharding design. It's pretty brittle and complex. And so it never really felt like it was the end game, so to speak.

And so I think having a lot of that background knowledge made me, prepared to recognize a brilliant idea when I saw it. And essentially, you know, as I read the Lazy Ledger white paper, I realized that, you know, what the paper was describing was something that could achieve all the benefits of sharding and then even more benefits, while actually doing so in a more elegant, simplified architecture. And so basically, like, you can actually, there's actually a lot of parallels between what Celesteia does and sharding, except that, as I said, sharding is a little more complicated and therefore less secure. And it's also less flexible.

So Celestia has just like, it's sharding plus plus in my mind. And it really feels like, it felt to me like, wow, this is the answer that to scaling and so many other problems, interoperable. interoperability included that the whole sort of like research space around blockchain have been waiting for. So I think, and especially at that time, the reason that I think it's taken so long for people to wake up to this is just that there wasn't that much information out there. Like when I was trying to learn about LaserLedger back then, I had to dig through all these forum posts on ETH research and piece together all these different white papers.

and like it was really not easy. And so you had to like dig and dig and dig. And luckily like I just caught the bug. I dug in and I found I found something really, really compelling. So among other things, sharding and Celestia's modular system are a response to limitations in scalability and congestion and transaction throughput. But skillability has been hard to solve for because certain approaches that have been tested, compromise on core characteristics of public blockchain networks, things like decentralization.

Why have scalability and decentralization historically been at odds with one another? So many people are probably familiar with the idea of this scalability trilemma. And I think it was popularized by Vitalik probably around 2017. And it was kind of the motivation for the ETH II sharding roadmap. And what it says is that you can, out of three properties, you can only have two. And those three properties are scalability, security, and decentralization. And so essentially, like, kind of, I would say the root problem here is that in every blockchain, the nodes, the full nodes have to basically validate the entire chain.

And so what that means is that every time there's a new block that's published and there are new transactions in that block, each full node has to download the full block and re-execute all the transactions. And so what that means is that you can only publish, you can only basically run as many transactions through the system as however fast the full node is. So in this case, so with Ethereum, you know, their target sort of, and by the way, every protocol has a sort of target spec for a full node, meaning this is how much, you know, memory and storage and bandwidth that you expect a full node to have, to,

be able to participate in the network. And so basically, that spec kind of defines how many transactions you will be able to process because the slowest person in the network is kind of like the ultimate bottleneck. If you start processing too many transactions, they're just not going to be able to keep up. So essentially, you're kind of stuck in this thing where either you try to, if you try to increase the transaction throughput, then you're going to leave some people behind. and that causes centralization, right? So you can start to see how scalability

and decentralization are kind of at odds with each other because in a decentralized network, you want to have as many people as possible be able to verify the chain themselves. And so that's broadly, I think, in a very simplified way, what the scaling trilemma is. And I can go next into why Celessia is like a different approach than other chains or other scaling approaches that

had been tried before. I definitely want to get into that. But before we do, I quickly think it'd be helpful to define skillability and decentralization. So if I'm understanding correctly, you define scalability as an increase in transaction throughput with a sublinear increase in the costs and the resources required to do and user verification. And decentralization means that it optimizes for anyone with a consumer-grade device to be able to perform this verification in a trust-minimized way. Is that the correct way to think about it? Absolutely. So you brought up a really good point, which is, you know, how do you actually define

scalability? And the way that we define it in Celestia is it's not just, the raw throughput of the network because what that number can hide is the cost associated with actually verifying the chain so like if you

scale the chain by just increasing the cost to validate in other words you're increasing the node requirements then you're not really scaling right so scaling has to be done in a way that does not increase

the node requirements and increase the cost. So like, you know, where we can sort of see the difference in approaches is, you know, project like Solano where the way in which they increase the scaling, I mean, they do some clever optimizations as well. So there are, there are optimizations and ways to speed things up, but those have a limited sort of like scope and how much of a difference they can make in the grand scheme of things. But really, I think one of the big things that Sala has done is they've said, you know, we're okay with increasing the node requirements. So all of a sudden, instead of having a laptop, we're going to require you to have a really, really fast computer with a big GPU and a, like, you know, gigabit internet connection.

And by doing so, we can process a bunch more transactions. But again, if you think of, and but to call that scalability is a little bit misleading because, again, you've increased the no requirements and the cost. So you've centralized. You've caused the network to be more centralized. So less people can be first class citizens of the network and verify it directly themselves. And then I think another example to give here that's not Celeste, but also I would say more akin to true scaling. is sharding. Although sharding also falls short, and I can explain why.

So in sharding, what you do is you break up the network into smaller blockchains. But those blockchains share security. And so each node in the network only has to verify one chain. So that chain, that subchain, that shard can be smaller and have less throughput. But when you, and so that node, that full node can actually keep up. But the whole network as a whole has a bunch of these shards, so those add up to a total, much larger throughput

than any one particular given shard. The thing is that sharding also doesn't actually increase scalability because if you wanted to verify all of the shards, you still, you need to run a node or have the equivalent resources of a full node for every single shard. So charting just decentralized block production. You can participate as a validated

network with less resources, but you still actually can't validate the entire network without also increasing the total amount of node requirements. So Celestia and modular blockchains are the first scaling solution that I'm aware of that actually allows you to increase the throughput of the chain in a way that increases the node requirements slower than the amount

of scaling increase. So it is actually delivering for the first time true scalability. Those are both really important distinction. So thank you for bringing up both of those examples and framing them in the context of how you define these things. So let's talk about how Celestia is actually realizing the ability to scale and preserve decentralization by creating this modular system. What exactly does it mean to be modular and how does that compare to traditional monolithic architectures? Yeah. So yeah, let's talk a bit about modular versus monolithic. So Mustafa, who is the co-founder of Celestia Labs and the project, he is author of Lazy Ledger and fraud and data availability proofs of the paper that I mentioned earlier, he coined this term modular blockchains.

And what a modular blockchain is is a chain that takes a fundamentally different approach than every chain that's been built before since Bitcoin, since the origin of blockchain itself. So Bitcoin, Ethereum, Solana, even Harmony, every chain that exists so far, I would I would classify them as monolithic. And so what a monolithic chain does is every time it proposes a new block, they're not just coming to consensus on that block. And consensus is where you just basically say, hey, this is the next block in the chain and it's not going to be reverted.

Like this is now mined into history. But they don't just say, they don't just do that. They don't just put the block into the as the next slot. and the chain, they also do what's called execution on that block. So they take the transactions and they process them and they say, okay, all these transactions are valid. This is the new output, like the new state of the chain. Like these are new balances and all the addresses and smart contracts.

And they say this is a valid block. So they do both consensus and execution at the same time. And so what I'm like and and just to sort of maybe put a teaser in there, you can you can see how, you know, given how we talked about scalability earlier, when you're, when whenever you're trying to validate a block as a full node in a monolithic paradigm, you have to download the full block and you have to re-execute every transaction. That's the only way that it can work. Right. Now, a modular blockchain comes from this very, very deep insight, which says, hey, you know what, actually, we could split apart. There's no reason why consensus and execution have to happen at the same time. There can actually be independent processes that are handled by two different networks or two different layers, right?

And so, and this was the insight that Mustafa had, he's like, we can actually build a really, really simple blockchain. And that's where the name Lazy Ledger came from. It's a blockchain that does no execution at all. So all it does is it takes transaction data and then it puts it into a block and it comes to consensus on that transaction data, meaning that it prints that data into history so that it will not be changed. And then it also does something which we can get into a little bit later called data availability, which is it makes it possible for nodes in the network to know that all that transaction data is public information that's been released without having to download all of it. But anyway, so Celestia takes care of the consensus part. And then the execution piece happens on a layer above using layer two construct.

like roll-ups. So the end result of this is that we've broken apart the typical like blockchain architecture into separate pieces. And each one of these pieces are like modules that take care of a certain function, either consensus and data availability or execution. And there are a few others like settlement, for example. and they specialize in that one function. And then builders, users, developers can recombine those functions to build the stack that they want.

So you can also see, so aside from the fact that it adds scalability, it also adds flexibility. Because all of a sudden you can pick and choose, like you have the freedom. to recombine, like, different components to achieve the results that you're seeking. So, like, in a monolithic chain, you are stuck in, like, everything, all the choices have already been made for you by default. Like, you're, you're kind of locked into whatever the original choices were by the, the core devs of that chain. So those are just, I mean, we can go a lot deeper into all the different benefits, but that's,

that's at a high level, the difference would be monolum. analytic and modular. An analogy that you guys have used in the past that has really helped me contextualize the power of specialization and talk about this flexibility that Celestia can unlock is the evolution of web hosting from having to host things on your own server initially to eventually the adoption of cloud services to unlock greater efficiency. and flexibility. Could you walk us through that and talk about how you draw parallels to the evolution of

public blockchain networks? Absolutely. So it's kind of in retrospect, and all credit goes to Mustafa for drawing this parallel. But you can think of the evolution of blockchain architectures as very similar to the evolution of the traditional web. So in the early days of the web, it used to be the case that if you wanted to host a website, you had to go out and buy your own physical server and set it up in your office or your home, like plug it in and then configure it and build the website.

Like it was very hands-on and it required you're basically like, you know, building it all yourself from scratch. Right. And then all of a sudden the services like shared web hosting. providers like GeoCities came out and they said, hey, you know what, like we're going to make it way easier for you to build a website. Like you, we will have the servers and you just have to, you know, write some code. And then you can deploy your chain or sorry, not your chain, your website. But the thing was that you were stuck within sort of a very limited framework that GeoCity is sort of like, like there's only certain supported languages, essentially.

So as a developer, it definitely made it easier, but you still didn't have the ability to fully customize. And so you're kind of limited by sort of that platform. Then you had virtual machines and the rise of the cloud. And what that enabled was like the modern internet infrastructure that we all use today, services like AWS, Google Cloud, Azure. And essentially, what those are. is you can have this really scalable, cost-effective server farm in, like, located somewhere, and it has a bunch of compute just sitting there for anyone to tap into, and they can optimize it just, you know, to be super cost-efficient and scalable.

And it's very, like, it's very flexible resource. And anyone can go there and deploy their own web app or their own website. and because of like this virtualization, you can actually have a lot more flexibility in defining what kind of operating system or programming language or configuration you want. So all of a sudden, it kind of gave birth to this much more scalable, flexible internet architecture.

So now if we switch gears to blockchain, what started, you know, the origin, of sort of the blockchain architectures was Bitcoin, right? So, and in the Bitcoin era, it was the case that anyone who wanted to build a new D-app or a new, it had to build a new chain.

So, you know, Bitcoin was the first decentralized application that was basically just payments. But other people are like, oh, you know, we can do other things with this. So people had ideas for either iterations on Bitcoin, like Like Bitcoin or ideas like math, like MasterCoin or Namecoin, but every time that they wanted to build something new,

they had to deploy an entirely new chain. So this is just like having to buy your physical server and build it and configure it and plug it in, right? Then Ethereum came along, and it was revolutionary because it was like GeoCities and these shared web hosting providers that said, hey, actually, you don't need to build your own blockchain if you want to build a decentralized application.

Actually, we have this general purpose blockchain that allows you to basically build an application using what we call smart contracts. And you can deploy that smart contract onto our sort of like server, let's say, by analogy. But the downside, just like GeoCities, was that you were limited to the EVM. You could only really build within,

the sort of like structure that they provided. So it limited your sort of flexibility. And as we know, it didn't have very good scaling properties. So finally, that leads us to what I think is the next evolution in blockchain architecture that is very much like the cloud, and that is modular blockchains like Celestia. So Celestia is like this, it's like, sort of like AWS or

Google Cloud and that it is this resource for block space and it's very flexible, very scalable, very optimized and anyone can tap into it

and deploy their own blockchain or their own D app. But because it's modular, just like virtual machines in the cloud are super flexible and give developers the ability to deploy, you know, whatever kind of operating system and have, like, full control. Celestia also gives you that level of customization. And, like, each roll-up is sort of like its own virtual machine, its own virtual blockchain,

in which developers can choose the execution environment, the programming language that they want to build in. So just like in the cloud for the traditional web, modular blockchains open up this new, frontier of scalability and flexibility for blockchain infrastructure. And it's going to enable anyone, just like it's super easy for any developer anywhere in the world to deploy a new web app or website. Celestia makes it that easy for developers to build and deploy new blockchains. So I think we're going to really see an explosion of innovation and experimentation and scaling in the space as modular blockchains become mature.

Another thing that I think a lot of people get stuck on or confused about when they are learning about Celestia is the nomenclature that we're used to. So by decoupling the component parts of what a blockchain does, the idea of something being an L1 or an L2 becomes less applicable. What's the appropriate nomenclature when looking at something through the lens of a modular stack? Yeah, that's a good subject. So the idea of layer two, I think, has been around for a really long time. In fact, even in the Bitcoin days, there were, I think, proposals on various forums about, like, building, you know, applications that, leverage Bitcoin as sort of like this, they didn't have, I don't know if they had the, the, like, they use this terminology, but basically using Bitcoin as this consensus and data availability

layer to build other more sort of like functional or interesting applications. And obviously, you know, Bitcoin has a very popular layer to system called Lightning. And then obviously with Ethereum, there was plasma and a bunch of different proposals. that's kind of where the origin of this term layer two comes from. And it is sort of the idea that you can build a decentralized system without having to actually build an entire blockchain. A layer two leverages another blockchain to basically secure it or to provide certain functionality that are critical for its security.

But I think as time has... gone on and the research and thinking around these concepts has matured, we've realized that there's sort of just a handful of key functions that compose that are critical to the composition of every sort of decentralized system. And so I think now what we'll see is that the terminology like layer one versus layer two may become less popular. And we might start using terms that just refer to those key functions. And those key functions, in my mind, are consensus, which is the ordering of transaction data and blocks.

Right. So that's how you, as I said, let's how you agree on the history of a chain, right? That's how everyone on the network knows what the ground truth is of like what has happened. The second function is what we call data availability. And this is a more nuanced point. But essentially, for blockchains to be secure, everyone in the network has to, not everyone, but you need to know that the data that everyone is using to run this chain is public information.

Because if it isn't, essentially the block producers can do what's called a data withholding attack, which is a really, really, really bad attack that breaks the security of these systems in a big way. It's worse than double spends, for example. And the third one that I would say is execution. And that is how you take all the transactions in this history and you verify that they are valid. And then you also compute what the final state is. Like, what is the result of all these transactions? Right.

And that's really important because that's actually where that's, applications live. So those are the three things. So I think we'll start talking more and more about consensus layers, data availability layers, execution layers, rather than layer one, layer two. And I will also add one other one, but this one is, in my opinion, optional and just a sort of special case of execution layers, and that is a settlement layer. And a settlement layer is basically a common place where a bunch of execution layers post their state updates. And to me, it's sort of like a bridging function. But we can talk about that more later if that's of interest.

I definitely want to talk about settlement, especially as it relates to and gives rise to sovereignty on Celestia. But before that, let's talk about something you've mentioned a few times, which is data availability. So what is it and why is it important? Absolutely. This is a really, really big topic. And it's a very nuanced concept. And a lot of people get confused about it. In fact, I found it very confusing when I first learned about it.

But I will do my best to try to distill it down to be as simple as possible. So to understand data availability, you have to understand the data withholding attack a little bit. So essentially, there's this problem in blockchains, which is that especially as you scale, right, If you want to build a system where not everyone has to re-compute or re-execute every transaction, which is fundamental to being able to scale a blockchain because everyone has to just redo everyone else's transactions, then you can only do that. You can only add more transactions by increasing the node size. So now, if you want to scale it in a way where not everyone re-executes every transaction,

you have to have some way of verifying that the execution was valid without actually redoing the execution yourself. And essentially, that's what roll-ups do. So in a roll-up, the roll-up operator generates a validity proof that proves, hey, this block, I can prove to you, this block of transactions is valid and I can prove it to you. Or, and that's a zero-knowledge roll-up case. or in a optimistic roll-up case, they publish a block, and then if that block is invalid, any of the execution there is invalid, someone will generate what's called a fraud proof, proving that, hey, this block producer of this roll-up is trying to pull a fast one on you, and he's trying to basically make an invalid state transition, meaning he's like either printing himself money or he's taking money from someone else. He's doing something that he's not allowed to do by the rules of the protocol.

So now you have roll-ups and they enable that. Like now all of a sudden, people don't have to re-execute every transaction, but this is where the data availability problem comes in. And that is that what this roll-up, the last sort of hole that you have to patch to make this system secure is you have to make sure that all the data is publicly available. And so the reason for that is that if a, let's take the optimistic roll-up case, if I'm a roll-up block producer and I propose a block, right? But then I actually don't reveal the transaction data behind that block.

Like what are the actual inputs that I use to generate that block? Then no one else in the network can look at that and say, hey, there's fraud. the rules because they can't actually see what I've done. So all of a sudden, I could propose a block and no one would be able to generate a fraud proof, right? And so all of a sudden, I could basically commit fraud and no one would know. And so basically I could, like the whole system breaks at that point. So what you need is you need a way for someone to verify that all the data behind the block that I published is public information. And that way, I can know that someone who's watching would be able to generate a fraud

proof. So in typical blockchain systems, the reason why you don't hear about data availability in monolithic chains is that they solve this data availability problem in an implicit way, which is that they require every full node to download the full block. Right. But it doesn't matter because in the, in the design of a monolithic chain, every full node has to re-execute every transaction. So they're going to download the full block anyway, right? But again, we're trying to build a different system in which not everyone has to re-execute every transaction, right? And we also want to make it so that not everyone has to re-download all the transaction data.

Because again, if now if we had to download all the transaction data, then the only way to increase the block size would be to increase the bandwidth requirements of all the nodes in the network. So what we need is a way for nodes to know that the data behind the given block was published and is available without downloading all the data. data because the naive way for me to know that data is there is for me to download it all and be like, yep, it's here, check. But again, because it doesn't scale, we need a different way to do it. And so that's where this concept of data availability sampling comes in. And essentially, Mustafa invented this scheme where you take block data and you use something called erasure coding to add redundancy to the data. And you basically, build the block using that redundancy and it makes it so that I can, rather than downloading

all of the block data, I can just sample a very, very, very small subset of that data and know with statistical certainty that all the full data behind that block is available. And what this enables, again, is that you can now increase the block size, but you can have The block size basically is no longer limited by the resources of the nodes in the network. Because you don't have to actually download everything. And what it also enables is that unlike other chains, where each additional node in the network actually does not increase the scalability. In Celestia's case, with data availability sampling, the more nodes there are sampling,

the block, the bigger the block can be safely. So essentially, you have this really interesting scaling property where the more nodes are on the network, the more throughput the chain can have. So one of the critical things about modular chains is they don't have a finite block size. So a lot of people popularize this idea of, oh, well, blockchains are like cities where there's a finite amount of space and everyone's competing over that space. and that's just that's just how it is like end of story but ultimately that's a really bad property for blockchains to have because then you're you know you're always trapped in this

thing where more users and more applications means higher costs for everyone and so modular chains are the first uh architecture to break free from that because like the reason that the quote like land in the city is limited is because the block size is finite and cannot expand. But in a modular chain, that block size can expand. So I like to think of Celesteia and modular chains as more like, rather than like a city with finite space, it's more like a frontier. It's like a new continent. And like people settle, you know, it's kind of like the settling of America.

They might settle on the east coast. And then gradually over time, as things get more populated, they expand west, you know, when they start settling new areas. So there's always more frontier for the block space to expand into. So you have, it's like fundamentally, a completely different model. And so instead of cost increasing

with the number of users, the cost can stay constant. Because each new user, each new person running a node increases the amount of space for everyone else. So that's why data availability is such a fundamental problem. and why this breakthrough of data availability sampling completely changes the game and will, you know, it breaks free for the first time from the scaling trilemma. This concept of data availability sampling allowing for block size to increase as the number

of nodes in the system increases is kind of mind-blowing. I guess on that point, though, is there a limit to how big blocks can get? That's a really good question. So there, the, the, so for people who are into computer science, you would know about like big O notation or like sort of like, I guess it's like a complexity theory. But basically, there's the, there is, so in a typical blockchain system, you would say that it scales linearly with the number of. or with the amount of resources of that node. So what I mean by that is, again, that like the, what that says in simple terms is that the only way to increase

the amount of throughput in the chain is to increase the amount of resources of that node by the same amount, right? And in Celestius case, it's not that the, if you increase the block size, it's not that the node requirements don't increase. They increase, but they increase very, very, very gradually. And the factor is rather than being linear, which would be O of N,

it is O of square root of N. So each time, let's say you increase the size of the block by four times, rather than having to increase the node requirements by four times, it's only by the square root of four, which would be two times. But I'm oversimplifying.

It's actually like, better than that, but that's that's like a really high level way to think about it. And then if you think about it, as a block gets really, really big, like if you can think, if you can visualize in your mind, sort of the square root of N curve, right, it, it flattens out. So as you, as a scale increases more and more, like you get more, more block size with less and less increase in the resources required for nodes. So it's like, you know, it's, it is just fundamentally, like, hugely more scalable than existing systems. So to put a little bit more, I guess, like concrete terms around this, I remember doing some back of the napkin sort of calculations a while back.

And so to compare this to Ethereum, so Ethereum has, let's say, on average, an 80 kilobyte block every 15 seconds. So every node has to download 80 kilobytes of data, right? You could have a Celestian node downloading 80 kilobytes of data, but the block size could be, I think, because this is a while ago, I think the block size could be 250 megabytes, meaning that with the same node resources as Ethereum nodes today,

you could have how many thousand X, like maybe 5,000 X? I can't think off the top of my head, but around like, or maybe 2.5,000 X the amount of throughput. And so, yeah, that's kind of like a high level way to think about it. So it's not that the, that the, you can scale without increasing overhead. There is a slight increase, but it really is quite small by comparison.

It gets smaller and smaller, the bigger the block is. So this just goes back to our definition that node requirements may increase, but at a rate that's sublinear compared to how much throughput increases. Exactly. Yes. I want to talk about something that you brought up earlier, and that's this idea that Celestea facilitates innovation in the layers that deploy on Celestia. And if I understand correctly, you're agnostic to the types of execution layers that build on top. And roll-ups are just one example, but maybe can you talk about, you know, what other types of execution layers you are seeing or expect to see deployed on Celestia?

So Celestia is sort of this general purpose resource that provides consensus and data availability, almost as a service. So anyone is a permissionless network that anyone can tap into to post data and that data will be ordered and made available. And so this is a primitive that anyone can build on. And while I think for the time being and for the foreseeable future, the most compelling use of this primitive are roll-ups, I can foresee a future where maybe there are new kinds of layer two constructions that get invented or even, yeah, fundamentally just new kinds of applications that are built that that may blur the line of like, you know, being a blockchain versus being, I don't know, something, something else. And so I think we're still very early,

but right now, roll-ups are the primary use case. But there are some cool things like Mustafa has an idea for a way to implement, for example, like a domain registry service that is not really like a roll-up. It's just sort of like a public posting board where like, you know, you just, you post a message there to register your name and, and basically there is no like centralized operator like in a roll-up. So there are some ideas like that, and I think we'll see more of those coming out as time goes on. But for now, I think roll-ups are going to be the main, like use case that plugs into Celestia as this consensus and data availability service. Now, the interesting thing is that Slesia also opens up the design space for different kinds of roll-ups, which we call sovereign roll-ups.

So in the Ethereum model of roll-ups, these roll-ups always sort of refer back to the Ethereum L-1 and post their state updates there. So the Ethereum L1 acts as something that we call a settlement. layer. So the, and this has a lot of benefits, right? Because that roll up, that means that that roll up has basically a bridge to Ethereum, where all these assets are, all these users are. And also, the Ethereum L1 can act as this conduit and sort of like bridge between all the different roll-ups. So now if I'm in optimism and I want to go over to Arbitrum or let's say Starknet, I take my assets and go down back to Ethereum, and then I can go from Ethereum back up.

And finally, importantly, Ethereum can act as sort of like this exit mechanism where if something bad happens on the roll-up, I can exit my funds back to the layer one. So it's kind of like this fail-sape escape hatch. But there's actually... And pretty much every Ethereum roll-up so far has been built in this model

where the roll-up is kind of a baby, chain to Ethereum, which is like the parent chain. And we think that this is really selling roll-ups short. I mean, I think it has a lot of advantages to it, but there's also a lot more design space to explore, namely this idea of sovereign roll-ups where a sovereign roll-up is a roll-up that does not actually have a settlement layer. So it just, it's a lot more like a just a standalone blockchain. in the way that it functions.

So it just posts its data to Celestia, and rather than settling on another execution layer or L1, it settles on the peer-to-peer layer. So, you know, in the Ethereum case, the validity proofs or fraud proofs go to Ethereum to get verified. In this case, they would be gossiped to the nodes on the roll-up directly. And what this allows for is much more, well, sovereignty essentially. So that roll-up is basically like its own chain.

Whereas in Ethereum, it's tied in and bound to like the Ethereum L-1. So whatever happens on the Ethereum L-1 kind of defines what they can and cannot do. Like they can't, for example, fork or upgrade their roll-up without also interacting with either forking or upgrading the contract on the Ethel 1. So it has a lot of limitations. Also, the fraud proof or validity proof has to be verified on that Ethel 1. And that could limit the kinds of things that you verify because you have to do it in a way that is the EVM is able to process. So in a sovereign roll-up, it increases sort of the scope of flexibility and expressivity.

And it makes that roll-up like its own chain. so that the community, the developers, can actually fork and upgrade and do, the heads have way, way, way more freedom. And why might this be important? Well, first of all, it allows for faster innovation and iteration. So if you want to experiment on the execution layer part, you can start to like upgrade and make changes.

Like you'll notice that like the EVM has, even though there's tons of EIPs to, you know, change things in the EVM, none of them have really gotten through because too many things would break if they make those changes. But in sovereign roll-ups, that is fundamentally different. And you have the ability, your community is much tighter and you're able to make those changes. And they only affect you, not like the broader system. So that is one of the things we're really excited about.

And I think just generally on the topic of like accelerating, innovation on the execution layer the advent of modular blockchains is sort of, there's a real parallel to the sort of Ethereum and that Ethereum made it really easy to experiment and deploy new decentralized applications

and so we had this explosion of different ideas that brought us all the things that we love like NFTs and defy and dows and all that. But it wouldn't have happened. It wouldn't have had so much innovation if it weren't for Ethereum lowering the sort of bootstrapping costs of deploying a new decentralized application. Instead of having to build a new blockchain, you can just write some lines of code and click deploy. And then boom, you're off to the races. But Ethereum and monolithic chains have not opened up the ability to experiment of the execution.

layer with low bootstrapping costs. Because again, if you wanted to change execution, you had to launch a new blockchain. So Celestia and modular blockchains open up for the first time the ability to, hey, you want to try modify the EVM or you want to just try a fundamentally new approach. You know, you can just roll up to, roll up and like rock up to, you know, a chain like Celestia and deploy it for very little costs. It's just like write some code and click deploy. So I think it's going to really open up a whole new era of experimentation and innovation

of the execution layer. I think we're still super early in the whole exploration of that space. I'm really glad that you brought up sovereign roll-ups because that was going to be my next question and something that I've really been grappling with is how where settlement takes place, really informs this property. Another observation is that, you know, we've historically been taught that all aspects of these systems are meant to be rigid and immutable, but sovereign roll-ups really flip this concept on its head. Yes. And they say, you know, a community should have recourse, a community should have flexibility, a community should have the ability to fork if there is something

that's harmful to them and it really gives them a way forward to be able to do that. Yeah, let's let's go a little bit deeper there because I don't think I went deep enough on the sovereignty piece because this is really, it's another really subtle point, but I think is extremely powerful and should not be overlooked. So as you said, we've kind of, I don't know necessarily where this started maybe um i don't know it seems like like kind of an old philosophy with them launching but this idea of code is law right and it's the idea that you know whatever was written in the contract or in the the implementation of the chain is you know that's that's the way it's meant to like if there's a bug or something bad happens like hey well code is law like you know that's that like it's over

right um and and i think you know that was really kind of demonstration demonstrated in the Ethereum, original Ethereum hard fork, where the Dow was hacked and all this money was being drained. And even though, you know, hey, it was written in the code as a bug, like anyone could have exploited it. The Ethereum community decided, hey, well, this is actually not really, we don't agree with this. Like, sure, you know, this is quote, like within the rules of the system in a very limited sense. but it wasn't actually the intention of the code was not to have it so that someone could just drain all the funds. Right. So the Ethereum community got together and they forked the chain to basically undo that.

Right. And I think that was a great thing that they did because it restored faith in Ethereum. It also saved people by now like billions and billions of dollars. but that also caused this fork, right, between Ethereum and Ethereum Classic. And so Ethereum Classic really stood by this idea like Code is Law. Now, as Ethereum has evolved and it's become, A, a much more valuable network, B, there's way more users and way more applications on it, it's now pretty much, I think Ethereum has kind of become more and more immutable. And I think the likelihood of any or the ability of anyone to actually do a similar kind of hard fork is pretty much impossible.

I think it would have to be the scale of the attack would have to be astronomical, right, to galvanize everyone. And this is kind of an inherent flaw in a monolithic chain design is that you have everyone all on one shared state space. and all the apps are all like running on the same chain at the same time. And so any fork or change that you want to implement for one of them affects every single other application and user. And so what that means is that the more and more people you have, the ability to socially coordinate basically becomes impossible because now every change you want to make or every fork you want to,

to make is either going to break another app or it's going to screw someone else over. Like if you do a rollback to undo some kind of hack, well, it's like, well, hey, this guy, he made a trade and he made a bunch of money. So I don't care about this other defy app that got hacked. Like, you know, they're SOL. Like I want to, you know, like I don't agree to this. And plus, on top of that, the sheer coordination required to do, to sort of roll back the chain or make a fork is massive

because you have so many stakeholders, so many different nodes, so many different exchanges and interfaces to this chain that like to coordinate something like that is just huge. Like let alone the, the ability to get everyone on the same page to agree that you should do it,

then on top of that, like actually executing it is almost equally as difficult. So where modular chains and roll-ups come in is that they enable people to build their own chains and their chain can be sovereign and isolated from others. So all of a sudden you can have a defy app

that lives on its own chain and if there's an attack, all the users and stakeholders within that chain are likely going to feel like the early Ethereum people did. And they're going to be like, hey, this is messed up. Like we didn't sign up for this.

We're going to roll this back and take our money back, right? And the reason that they'll be able to do that is because they will not be, they're not going to be embedded into a much larger monolithic platform. They're going to have sovereignty over their own spot, their own application. And so all of a sudden, like code, like we don't think, we think that social consensus and like forking is a, is a feature, not a bug, right? So code is law, true, but only insofar as it actually upholds the sort of

like social contract of what that application was supposed to do. Right. And so like at the end of the day, it all comes down to social consensus, like the layer zero. Like people, people often call, try to call Celestia a layer zero, but really layer zero is people, right? It's like what gives Bitcoin value, what gives Ethereum value is the people that believe in it. And so you can't escape that fact. And I think that modular chains enable. us to get back to a place where we can enable the power of people and social consensus again.

And there's like so much more that I can talk about this. And I guess the last thing is that aside from the fact that the social piece becomes easier in a modular blockchain context, because the communities are sort of more tightly knit and separated into their own sort of like spaces. but on top of that, the infrastructure piece becomes easier because, you know, when you do a fork or an upgrade in a monolithic chain, you actually have to do that. You actually have to start like regenesis the network. You have to actually like, you know, coordinate all the nodes and get everyone to be like, boom,

okay, we're all going to fork at this time. Okay, ready, set, go, right? So it's a huge undertaking. And then also, as you're probably aware, you know, when Bitcoin and Bitcoin cash forked or when Ethereum and Ethereum Classic forked, it splits the security of the network, right? So the mining power or the staking power, whatever, gets split up. And in Celestia's case, it fixes those two problems.

So essentially, in a modular blockchain, the security is defined by Celestia, which is like the base layer that is forming the consensus and data availability. So now when you fork, you can fork a roll-up, and both roll-ups still have the same level of security because they both post their block data to Celestia. So obviously you can fork without diluting security. And then the last thing is because you don't run your own consensus layer, when you are rolling out the fork, you're not having to coordinate like a regenerative and like actually creation of a new blockchain.

It is strictly a software thing. So when you want to fork and upgrade or make a change or rollback or whatever it is, it's as easy as the developers of that rollup writing a new sort of like node release and saying, okay, everyone, we're switching to this. And so, you know, you choose, you vote by basically saying, hey, I'm going to run this new node release for or not. And that basically defines the fork. So it makes it like so much easier to actually execute a fork and like less,

less overhead, less waste. And so I'm really, and I'm really excited about that feature because forking is what makes blockchains as, as powerful as they are, and anti-fragile. And that's also how we're going to, like, be

closer, that people are going to be more empowered to exit or change systems I don't agree with. So that's, yeah, that's sovereignty. That's a long rant, but it's really, really important in my mind. no thank you for taking us down that rabbit hole of what sovereignty means and what it unlocks and why you believe it's so important i also want to call out that you're now empowered to make

these changes based on social consensus without in theory impacting the security of the underlying layer and i also didn't appreciate that celestia could potentially allow both roll-up to maintain the same level of security that they had pre-fork, but I guess based on the previous statement, it makes intuitive sense. Yeah, and I want to give hats off to John, spelled C-A-N of Delphi, for articulating this really well in the Delphi report and even adding a lot of things that we ourselves as the core team, like hadn't thought through before.

So, yeah, like I know it's behind a paper, well, but the report is absolutely fabulous. I hope that they will release it to the public sometime. Yeah, John is brilliant, and we're huge fans of Delphi here and all the work that they do in breaking down really complex topics. So shout out Delphi. Another topic I want to touch upon that I think is misunderstood or maybe sometimes considered in a vacuum is composability. So how should builders and users be thinking about composability between applications building on Celestia or or between applications building on layers that are building on Celestia? Composability, I think we should define it before sort of diving deeper. So composability is this property where you want to be able to compose different applications together, meaning you want to

to sort of layer them on top of each other like Legos. So you hear a lot in the Ethereum community about like Money Legos, right? That's their term for DFI. And it's really appropriate because in DFI, you kind of, it is actually modular. I think that the good DFI protocols are more modular, meaning that each application sort of serves one function, right? And then you have, let's say you have a lending function, and then you have a trading or swapping function. And then you may have a flash loan function or there's all these different like applications that do a specific thing. And the beauty of Ethereum is that you have composability, meaning that I can in one single transaction call all of those functions in a specific order that ends up resulting in

something that is even more powerful. Right. So I can take out a flash loan and then, you know, use that fashion to buy some things and then arb some or borrow something else and then arb it and go back and then I make money. So composability is a really powerful property of a blockchain. And in a monolithic chain like Ethereum, you get composability for free because every application lives within the same execution space. But there's kind of this problem, which is that as soon as you start separating out the execution into separate chains or roll-ups. So whether that's through sharding or whether that's in an ecosystem like Cosmos or Avalanche where you have either different zones, so different blockchains or different subnets,

those chains are no longer able to compose those transactions together in an atomic way. And again, atomic means that it all happens altogether at once. It's just one shot. Right. And so this is, but the thing is, I mean, the way that I see it is two things. First of all, the only way to maintain composability is for everyone, every transaction to be on the same chain. And fundamentally, that means that that chain needs to be monolithic. and everyone has to re-execute every transaction.

So you fall back into this same problem, right? Where you can't scale the same degree. You can't be, you can't sort of get these nice properties that we've been talking about. Also, when we talk about sovereignty, if every application is on the same chain, you lose all those benefits of sovereignty that we were just talking about. So there's a lot of reasons why you actually don't want every application by default to live on a single chain, even though composability is great. And then the last thing is that in, so in a modular context, you can still have composability.

So if I have applications, applications that need to live together in the same execution space can. So there can be like a D5 focused roll-up where, you know, lending and borrowing and swapping and all those. an options protocol or whatever, they're all there and they can interact with each other in the same way as they would in a monolithic chain. But then, you know, all these NFT apps or I don't know, some kind of Dow or whatever that doesn't need composability can live somewhere else, right? And they would want to do that for sovereignty reasons, also for the fact that they will be less congested, like they're basically the amount of like transaction throughput going through reach roll up will be sort of like only dependent on that one application rather than all the others.

So in the same way that like you don't want when there's a big NFT mint going on, all of a sudden the, you know, Solana network breaks, right? And then it breaks every other application. So there's a lot of reasons why you want to be on your own chain. So that like, you know, that other chain breaks, you're fine. And then the last thing that I would say on this subject is that there's ways of having, I don't want to call it composability. Well, it is composability without atomicity. So, atomicity is this idea of like, okay, when I send this transaction, it all needs to execute

a one shot, one go, and then revert and be done. But you can actually still compose applications in a cross-chain way. So like in the Ethereum ecosystem, like there's nothing stopping in the cosmos ecosystem, for example, like there could be osmosis, which is like a swapping function, and then there could be another chain

that's like a lending chain, right? So I could take my assets and I could deposit them in the lending thing, get out the sort of the wrapped token that represents the fact that I deposit those tokens. Then I could bridge those tokens over to osmosis and then trade them and swap them. So in a sense,

those assets are, applications are still able to be composable. Like, I'm still stacking those Legos on top of each other. But I, in between each step, I'm bridging the tokens. And so it's not what we call atomic. Like, it doesn't happen all at once. Like, there's still, there's more steps involved. But you still, I think, like, you don't lose that much, that much functionality.

And that's exactly how it will work in a roll-up context. Except that, and we didn't get into this, but in, you know, the last benefit of modular chains that's really powerful is this idea of trust, minimized bridging. So unlike in an avalanche or cosmos ecosystem, the separate chains or roll-ups can interoperate with like top, top-level security. And so like the, the argument against, like, the insecurity of cross-chain bridging falls away as less, is almost irrelevant in a modular their context. So that composability between applications to me becomes even more strong. What's notable as it relates to sovereignty and composability is that Celestia gives rise to all

of these properties that exist on a spectrum, but you as the builder have the power to decide what tradeoffs, for lack of a better term, that you want to make to do what better. suits your purpose, which I think is really interesting. Okay, last question. And with this one, I'm going to pull us out of this amazing rabbit hole that we went down. But I would love to understand Celestia's grander vision and the purpose in trying to enable all of these critical properties that we've talked about in this whole conversation. So Celestia's core vision really comes from.

from Mustafa in his background. So he has been a hacktivist since the time that he was a teenager. And he's always been motivated by the idea that, you know, software and the internet and, like, decentralized systems can equalize the power and level the playing field and make the world a more fair place. And, and sort of like, yeah, allow us to, you know, have more say in how things are done and recreate. And now I'm kind of adding my own flavor here.

But I think like blockchains that will enable us to recreate and redefine a lot of the institutions that are critical to sort of the functioning of society. And we architect them in a digitally native way, in a way that will sort of eliminate the ways in which those institutions have been captured. But so on a high level, what Celestius's vision is, is that we want, we think that blockchains are community computers, right? So we don't, you know, a lot of people in crypto have read or may have read the book, The Sovereign Individual. I read it, and it's one of my favorite books. And it basically talks about how in the Internet era, it, like, and it was written before the advent of crypto, but it says, the internet and basically digital money and digital like systems like crypto are going to

free individuals to sort of become on the same playing field as sovereignty like countries so each person can almost be like a country and um i don't think there's a really powerful idea but it actually misses the core point and this is where Mustafa's vision comes in and then is that for us blockchain and is not about sovereign individuals, it is about sovereign communities. Because blockchains fundamentally are our community computers. They're only, what they're used for is for people, groups of people with shared goals to unite around a common system and be able to coordinate amongst each other in a more effective way. And so sort of like reshape and redefine like what they, like the world around them essentially.

And so you can think of a blockchain as almost like this tool, this very, very fundamental tool that will enable people around the world to sort of create their own digital sovereignties, their own digital countries. And so it's not about individuals becoming sovereign. It's about communities becoming sovereign to us. And so that's why Celestia's goal is to be able to enable anyone in the world, any community in the world to build their own blockchain. And I think, you know, that's the future that we want to see. And we think that like the world will be a fundamentally better place with something like Celestia enabling that. The concept of sovereign communities is such a powerful vision. And that was really the perfect way.

to wrap up this conversation. Celestia is such a revolutionary and elegant system, but it introduces these new concepts that can be really hard to grasp. So, Nick, I really cannot thank you enough for all of your time, spending so much time with us, and breaking all of these things down so eloquently and so concisely. Maria, it's been an absolute pleasure. Thank you for letting me rant on all these things.

and if people want to learn more about Celestia, our website is Celestia.org, and our Twitter handle is Celestiaorg, and we'd love to see you in our Discord. We have a DevNet, we have a test net coming up soon, and fingers crossed, a mainnet launch at the end of this year or early next year. So stay tuned. We have a lot of exciting stuff coming.

And if anything, any of these, things resonate with you. We'd love to have, love to build with you, and we'd love to talk to you. So don't be shy and get involved. And Nick, you also post amazing resources and threads on your Twitter. So where can people find you? Yeah, of course. So my Twitter handle is, unfortunately, I have a very common name. It is Nick White. So my Twitter handle is Nick White, but the second I in the white is an eight. So N-I-C-K-W-H-8-E. Amazing.

Thank you again, Nick. Thank you, Ria. Take care.