Epicenter - Learn about Crypto, Blockchain, Ethereum, Bitcoin and Distributed Technologies - Celestia: 1Gb Blocks, Rollup Interoperability & Lazy Bridging. Ismail Khoffi & Mustafa Al-Bassam

Episode Date: September 30, 2024

As Ethereum’s roadmap shifted to a rollup-centric approach, a plethora of rollups have been launched, to the point where L3s have been conceptualized. Celestia sits at the forefront of modularity as it aims to replace the monolithic blockchain architecture with specialized layers that are more suited for scalability and customization. As a result, Celestia strictly focuses on data availability to accommodate the recent rollup expansion, as data storage represents the largest portion of fees on L2s. The Lemongrass upgrade lays the foundation for further use cases being enabled through Celestia, mainly revolving around interoperability and zk proof enabled ‘lazy bridging’. Topics covered in this episode:The vision behind CelestiaRollup architectureCentralised sequencers and the role of fraud proofsCelestia’s market shareRollkit & sovereign rollupsGovernance & light clientsThe importance of decentralisation: L1s vs. rollupsCelestia’s data availability capacityLatency & Celestia block timesInteroperability & ‘lazy bridging’The Lemongrass upgradeOn-chain economics & value accrualInterchain accounts on CelestiaCrypto’s mainstream adoptionFuture roadmapEpisode links:Mustafa Al-Bassam on TwitterIsmail Khoffi on TwitterCelestia on TwitterSponsors:Gnosis: Gnosis builds decentralized infrastructure for the Ethereum ecosystem, since 2015. This year marks the launch of Gnosis Pay— the world's first Decentralized Payment Network. Get started today at - gnosis.ioChorus One: Chorus One is one of the largest node operators worldwide, supporting more than 100,000 delegators, across 45 networks. The recently launched OPUS allows staking up to 8,000 ETH in a single transaction. Enjoy the highest yields and institutional grade security at - chorus.oneThis episode is hosted by Brian Fabian Crain.

Celestia is a very basic base layer that's optimized to make it very easy for anyone to deploy their own blockchain on top of it. You have a specialized layer that is only made for one purpose that doesn't have a state machine for like general compute. That makes it easier to scale it up for the demand of like roll-ups and layer twos. So there's no sort of on-chain smart contracting environment for developers to use. There's a layer one application. It's all in layers view. because the whole point of this architecture, if you just want to customize one part of your stack,

you can just deploy a robot within minutes and change something about your robot. You can already launch multiple visa-scale networks on top of the Celestia, and even like the whole Solana chain would fit, so to say, into Celestial, even with 8 megabyte blocks, right? If we get to a gigabyte that would already serve most use cases.

Welcome to EPSN. The show, which talks about the technologies, projects and people driving decentralization and the blockchain revolution i'm brian crane and today i'm speaking with ismail coffee and moustaf al-bassam who are the co-founders and uh you know c t-o ceo cecilability of celestia labs this is the second time we're having a mom actually i was listening to the first episode earlier today uh was exactly four years ago so it's a lot of has happened since then i'm excited to have them on again but you know before we talk with them, we'd like to briefly let you know about our sponsors this week.

If you're looking to stake your crypto with confidence, look no further than Corse 1. More than 150,000 delegators, including institutions like BitGo, Pintera Capital and Ledger trust Coros 1 with their acids. They support over 50 blockchains and are leaders in governance or networks like Cosmos ensuring your stake is responsibly managed. Thanks to their advanced MEV research, you can also enjoy the highest staking rewards. You can stake directly for, from your preferred wallet, set up a white label note, restake your assets on eigeneyer or symbiotic, or use their SDK for multi-chain staking in your app. Learn more at chorus.1 and start staking today. This episode is proudly brought to by NOSIS, a collective dedicated to advancing a decentralized future.

NOSIS leads innovation with circles, NOSIS pay, and Metri, reshaping, open banking, and money. With Hashi and NOSIS VPN, they're building a more resilient, privacy-focused internet. If you're looking for an L1 to launch your project, Nosis chain offers the same development environment as Ethereum with lower transaction fees. It's supported by over 200,000 validators making NOSIS chain a reliable and credibly neutral foundation for your applications. NOSISDAO drives NOSIS governance where every voice matters. Join the NOSIS community in the NOSISDAO forum today.

deploy on the EVM compatible NOSIS chain or secure the network with just one GNO and affordable hardware. Start your decentralization journey today at NOSIS. I.O. Cool. Thank you so much, Bo, for coming back on again. Yeah, it was a long time ago we had you guys on for the first time. It was still called Lazy Letter at the time. I think it was just sort of at the beginning. I think roll-ups. Actually, I remember you guys always. mentioned roll-ups there, but I think that was something that kind of, yeah, you were also starting to think about. So a lot has happened since then, right? You guys have come a long way. And then I think in the, I'm sure most listeners or probably pretty much all listeners have heard

of Celestia, right? They have like some vague idea. But I think at the same time, it's not always easy to wrap your head around like what Celestia is and what it does. So maybe just to, to kind of introduce people. Can you explain, like, what's the, what was the vision for Celestia and what are you trying to create? Sure. So in a nutshell, Celestia is a very basic base layer that's optimized to make it very easy for anyone to deploy their own blockchain on top of it, using things like roll-ups and layer two's. So, because obviously we have all these new roll-up and their physics technology, but no one really tries to build a layer one that is optimized just for layer two. and can think of Celestia as basically just layer one that is early optimized and early has layer two.

So there's no sort of on-chain smart contracting environment for developers to use, like all the applications. There's no layer one applications. It's all in layer two. And when we pressed slash mainnet last year in November, there was this huge need because a lot of people were deploying wallops on Ethereum, but they were extremely expensive to deploy because this. because there was an update availability capacity, which is called Roll-Up's Need. So this year, with Sleicester launch,

it was the first kind of like specialized data available layer for roll-ups. And since then, many roll-ups, there's been, many roll-ups have started to use to last year to effectively make the transaction fees a lot cheaper. And so we basically solved this kind of free-depressing needed to market at the time

that no one else was solving at a time, which was cheap, which was scalable and cheap DA for roll-ups. So, of course, Ethereum, right, since, like, years ago, has also chosen to kind of go into the direction of, you know, trying to be this layer one that primarily scales and through layer twos and serves layer twos. Is the, what is the biggest benefit of having a layer one that's, you know, specifically focused around like serving layer two is it just around cheaper cost or like what other advantages are there obviously there's cheaper costs but you can you can think of it as like a layer that is

highly optimized for one use case only and that itself brings like the main benefit which is like you have a specialized layer that is only made for one purpose that doesn't have a state machine for like general compute and um yeah that that makes it easier to scale it up for the demand of like roll-ups and layer twos i would say and also like another main i would say another key benefit is that they like you have more like the focus for celestia is also verifiability so we we put light clients and end users first or like users are like key players in the system. So I would even go back and say

like Celestia is not only like a layer one, which is basic, but which is right. But I would also say that it's an architecture for like abundant block space where you can like have massive scale. And, but this doesn't mean

you just deploy on a centralized cloud or something where you could have like EWS which would give you more scale, right? but instead you have like verifiability by end users and by roll-ups that they can verify the chain even with low resource requirements. I think that's also a key benefit. So let's let us maybe recap briefly.

So the way with roll-ups, right? So with roll-ups, we generally have a bunch of different roles, right? So we have like a sequencer, we have the data availability, then generally there is some kind of like execution thing or L1. Can you explain again like, you know, how do these different components work together? And what is the architecture? I mean, in a nutshell, a roll-up is simply just a blockchain that posts its blocks, not a blockchain. So I think the easiest analogy is a... that I use is a roll-up is basically like a virtual blockchain

that you deploy on some other blockchain. The same way, like, you might deploy a virtual machine using AWS on a bigger machine. So there isn't any specific, like, set of hard components that every roll-up needs. You can literally take a 10-minute chain and purchase blocks on Celestia or some other data and it becomes a roll-up, in theory.

So there isn't, like, some specific set of specific key components. Every roll-up have different components. But of course, in a common component, a roll-ups have a sequencer. So you might have a... So like all roll-ups do the way to sequence transactions.

So most commonly, current roll-ups, they use a centralized sequencer. So this is a like, like, Arbishop 1 or Optimism Mainnet with works, for example. It's a single sequencer that takes in transactions from users

and generates the shawlop blocks and process blocks to the base layer, like a theory or celestial year. However, the next thing about roll-ups is that even if the sequencer is centralized, in theory, if the roll-up has constructed correctly, you don't have to trust that sequencer, because the roll-up is inheriting security

than the base layer, because if the sequencer misbehaves, then because the data is available, you can generate a fault-proof or a ZK-proof of the behavior or the behavior or the misbehavior. And if the sequencer is censoring transactions, then the user can go back to the base layer

and force the sequencer to include those transactions by posting the transactions directly on the base layer. But not all roll-ups have sequences. Some roll-ups, there's a category of base roll-ups. Orlops called base roll-ups that don't have any sequences, and they just rely on the base layer for sequencing. So there's no sequencer at all, and the user just posts the data to the base roll-ups.

base layer. So there's many different configurations for roll-ups and there isn't, like, necessarily, all roll-ups don't necessarily share the same component or have the same architecture, but that's also the beauty of roll-ups and modularity in general, because the whole point of

this architecture where developers use roll-ups for applications is that the whole point is that it's extremely customizable. That developers have full-stike customization over how they're deploying applications. And because historically, if you want to have full stack

customized ability over your application, you would have had to deploy a new layer one using something like the Cosmos SDK or starting from scratch. And that's a lot of overhead. But now, if you just want to customize one part of your stack, you can just deploy a roll-off within minutes and change something about your roll-up.

You can add a new upcodes to EVM. You can choose to have a centralized sequence so that captures the EVV and FEVAV-V-V-Rew rather than leaking into the base layer, for example. and there's certainly customized ability options and I think that's a big reason

why developers deploy role-ups. So you mentioned the fraud proofs and at the same time Celestia is this sort of like very minimal layer one where you where you basically

just put the data for for availability. So like let's say that does happen and you have a roll-up that uses Celestia and now the sequencer does something and like you know I have a fraud proof like how does

like do you need another chain as well then the function or like how what would you do with that fraud proof so if assuming it's optimistic roll-up if the sequencer tries to include the entire transaction in the chain or change the state of that chain of the roll-up in that way

let's say that I try to steal people's money for example then because the sequencer is posting those blocks to Selesia that everyone can see the misbehavior and then they can generate a fraud proof of that misbehavior. And then that that fraudulent, depending on how the rollout is set up, can be, for example, distributed to their nodes of that roll-up, and that would allow, or like clients of that roll-up, and that would allow the users of that roll-up to reject that block and treat it as a valid. Or in the case of, in the

case of a bridge, for example, you can imagine a roll-up has a bridge to some other chain, for example, like an optimism roll-up has a bridge sure Ethereum, for example, then what would happen is that you would press the fraud proof to the bridge contract and then because the bridge, in the case of optimistic role, the bridge has a seven-day challenge period. And so like every block can be challenged within seven days. And if so, so then someone can purse a fraud-proof within the seven-day challenge period and then that would cause that roll-up block to be rejected by the bridge

and then they would have been able to steal money out of the bridge. So in the one example, you basically have, okay, if all the clients run a light client, then they will just basically recognize, okay, the thing that is

being sent from the sequencer is not legitimate, so then sort of the blockchain doesn't advance. Yeah, exactly. And the bridge, it's just like an on-chain light state. Yeah.

And then you would recover from this because like now you still like you then I guess would want to have some kind of process for maybe you switch out the sequencer or yeah I mean it would depend on how the sequencing works for the roll up

like if it's a centralized sequencer that presumably you would need some way to I mean actually I mean in theory you don't have to switch out the sequencer because nothing like the sequence that could just continue because the sequencer does not have to be trusted like even if you can't replace the sequencer the sequencer yeah

he generated any valid block, but it could still continue generating valid blocks. But obviously, either way, it's a little desirable for a roll-up to be with I-1 sequencer, because if that sequence that goes down, then that roll-up loses liveness.

So at least, like, roll-ups still need the way to recover if the sequence that goes down. And there's kind of like, right, there's kind of different ways that can be achieved. Like, many roll-ups have a, have, like, a governance, system and where you can have like control parameters of the roll up or more commonly what arbitrave does for example is that if the sequencer goes down the end users have the ability to

post transactions to this on-chain inbox which forces which effectively forces the transactions to be executed on the L2 even if the sequencer might be this malicious or offline actually i'm glad you bringing up this thing, right? Because one of the, one of the criticisms that you hear a lot of roll-ups is actually that, you know, the centralized sequencer, which is, you know, still basically pretty much all the roll-ups, right, have a centralized sequencer. And at the same time, I think many of them are working on some kind of, you know, decentralization of sequencing. Do you think this is essential or do you think, like, the kind of guarantees that roll-ups can produce such that, like, you know, a decentralized sequencer isn't absolutely

necessary, or, like, how do you kind of view that? So, technically, if a roll-up is constructed correctly, it's not, it's actually not necessary, or it's not mandatory to decentralize the sequencer. Because the whole point of a role-up in the first place is that you're inheriting security from the base layer, which means you're inheriting state validity and censorship-resistant for the base layer. So ideally, that if you construct the rollout correctly, you can get away with using a centralized sequencer

and still have the roll-up, completely search and should be resistant because users can force transactions to be included with an untrained inbox. And because you have fraud proofs and ZK proofs, the sequencer cannot include bad transactions. In my view, like, it's perfectly fine.

Like, it's perfectly valid that roll-up the centralized sequester because the roll-off is still such-sure-resistant, if it's constructed correctly. That's actually a massive advantage of roll-ups. You can get away, you can get away with very little infrastructure while still having

full sensitive resistant and security. But that being said, there are still advantages to decentralizing the sequencer because if you have a single sequencer, the assistive-resistance guarantees you get are slow in the sense that, for example, my understanding with the arbitrary on-train inbox is that

if users use this on-train inbox to force transaction inclusion the transaction is only included in 24 hours so it's like this it's like this slow censorship resistance that takes 24 hours to get around so I think there's still advantages to have the decentralized sequencers but the main advantage of that it would give you immediate censorship assistance rather than this kind of like slower form of censorship resistance where you have to wait 24 hours and that's what some protocols like for example

astria do so ashtria for example, which builds on Sylvester is creating and this shared decentralized sequencer. So it's like it's a decentralized sequencer that many roll-ups can share effectively that post the data to celestialia.

And I think other projects like Espresso are also building something like XVI. So I recently saw some statistic, you know, that seemed very positive about Celestial. It's basically that the market share, right? In this

blob space, right? has reached something like 40% that is Celestius being used for can you talk a little bit about like you know what are the main is it primarily Ethereum layer 2s that are using Celestia and is there

anything like what's noteworthy about the like the current usage pattern that you see of the Celestia chain so yeah currently it's the majority of chains deployed on Celestia are actually Ethereum L2s like or L3s So like optimism, arbitrium chains.

That said, I think it's called Eclipse recently launched, which is like a bit, it's also an Ethereum chain in one sense, but it's also different in a sense that it uses the SVM for execution. So it's more like Solana execution, but Ethereum for Settlement and Celestia for DA. So I think that also uses up quite a bit of Celestia block space. But there are a few sovereign chains, like Forma

and there's more in the pipeline so there will be more sovereign chains built on Celestia 2 which are not necessarily there won't be necessarily EVM chains or Ethereum chains See maybe we can go into

that a little bit right because I know there's like one project right that is a part of the kind of Celestia core team my understanding is Roll Kit and then there's this notion of like the sovereign roll-up. Can you tell us a little bit about like, you know, what is the sovereign roll-up

and what are the kind of different assumptions that you have in a sovereign roll-up versus one that's non-sovereign? Like, we started roll-kate because effectively no one else was building a general purpose smart contract framework. So at the time, like, OPE stack didn't exist. You had like, you had optimism, but there was no OPE stack. You had that you didn't have Albatrim Nitro. There wasn't like no general framework for people just to deploy their own roll-up.

So you have to use other people's role-ups, like optimism, Arbitram. But now we do have a lot of frameworks. Now we have IP stack, Nitro, and so on and so forth. And that's why we started building Rollkit about three, four years ago. And Roll-Kit is a sovereign roll-up framework. So, yeah, what is the sovereign roll-up? So if you look at Ethereum's Roll-Century and Word, App. Initially, they were, originally they were seen as these L2s to extend or skeletal.

And now also we have this concept of network extensions on Solana, where people are building L2s of Solana and they're calling them network extensions. But Svetia takes a very different philosophy to roll-ups. We don't see roll-ups as just something to scale some other chain, or just something to extend from other trains. We see roll-ups as a way to deploy your own chain. it's independent in its own right. So, like, it's the same way that you might, you might deploy your L1 chain.

Like, why do people, like, why do people deploy cosbo chains? People deploy, people deploy coswell chains because they want to have their own independent chain, not because they're trying to scale cosmos. It's because, like, it's not scaling technology.

It's just like they're trying to build a chain and trying to build a good product. So we see roll-ups as just, like, a way to deploy your own chain way more easily. Not just a way, not just as a way to scale and some other chain. So whereas like roll-offs on Ethereum

are there to scale Ethereum, so that year does not exist for roll-ups to scale to last year. So that's just to scale roll-ups. And so that's where the concept of a sovereign role that comes from. And because volus on Ethereum,

they are kind of like enshrined Ethereum as a cell material. They kind of enshrined Ethereum as like, okay, like, I have a bridge to Ethereum and that bridge defines my chain, right? But a sovereign roll-up is just like,

you have a chain as a roll-up, but your chain isn't, like, defined by some other chain execution. It's like, it's its own chain, it's own right, and it's sovereign because the upgrade part of the roll-up or the user is, uses hard-forking, or users directly choosing, which is the correct chain for the roll-up.

So that lets you like how a hard-fork in cosmos or any other L1 would work, it's not some other cosmos chain deciding and what is the canonical chain. It's the users of that roll-up deciding it. And so that's why it's called sovereign because it's a roll-up that has sovereignty the same way that a cosmos chain

or any other one has sovereignty. And so when the upgrade process, this hard-for-upgrade process, like for example, cosmos chains, right, generally use some sort of governance thing, right, where you have like the stake and validators, especially, you know, they vote on like, okay, let's make this upgrade and then the validates coordinate. So in the case of a sovereign roll-up, this could basically just be done in the roll-up itself,

right? That maybe you have like some roll-up native tokens and then maybe the owners of this token say like, okay, hey, I approve this upgrade. And then it upgrades like the clients that that people use to verify the state. Is that how it will work? Yeah, kind of. So, like, one of the core values of Sylvester is that off-chain governance is more important than on-chain token holder governance.

Because, like, if you look at what the whole point for blockchain in the first state, the whole point for blockchain is to create this, it's to kind of like a shared computer or a shared environment for people to use such that you don't have to trust any middlemen

or you don't have to trust and that includes you should not have to trust a majority of token holders and that's why like we're very against token holder relevance for upgrades because to me that that defease the whole point of blockchain

like if you're saying that 51% of the token holder can just rewrite the rules of the chain, then how is that, that's not really a blockchain, it's a blockchain, but that if he's the whole point of a blockchain, like that's just like,

the shareholder governance, like, similar to a normal cooperation. Whereas the whole point of a blockchain is that it's not, like, no single, a majority of, majority of token holders

should be able to arbitrarily change the rules of the chain. It's like, what if, like, if you want to say that people vote to do something bad or like, I don't know, they'll come, compelled by some court or some government, like what if, especially the validators, for example.

So if you look at like how, for example, upgrades work on Bitcoin or Ethereum, there's, they happen on this like very fluid off-chain process. So every time, for example, Ethereum is a good example, right? And Ethereum, that does not have on-chain governance upgrades. It has this off-chain process where it has EIPs and stuff. It's very improved proposals. And then there's kind of like this rough consensus. And then there's a lot of, and then there's a a hard fork. And if people don't agree with a hard fork, then they can kind of like focus to the road chain. And that's basically what happening with Ethereum and Ethereum classic, right? That was an on-chain governance. So, um, and well, so part of the reason why I think sovereign

role-offs are important. And it's because if you look at how carbon roll-ups work on Ethereum, for example, the devil is always in detail because the, the upgrade part of the role, up is kind of like most, it's almost a most important thing from a security perspective. And in many cases, it kind of defeats the whole point of the roll-up. Because if you look at the upgrade path of most of the theory roll-ups, for example, the upgrade of the roll-up is basically decided by some multi-sig or some governance process. The kind of like almost defeats the point of a roll-up because the whole point of a roll-up is that you're inheriting security from base layer.

But then if you're saying some multi-sick can change your entire rules of the roll-up to at Rung you basically then that's not really that that kind of is very bad from a security perspective so that's why for Slesia for example for software to roll-ups we kind of see an upgrade path

where we're not enshriding the bridge into some committee and then the committee decides what the upgrade should be instead we're saying the bridges are secondary to roll up and then if you'll upgrade that roll up it's the users that roll up and the nodes of that

roll up that decides what the canonical chain is. So the upper grid process would be the same as like any other one, right? Or like Ethereum or Bitcoin. The developers will propose a new version of the Node software, and then people can choose to adopt it or not. And if people don't adopt it, it won't happen. And if it's contentious, then there'll be a fork. And so you have this, you basically have this right to fork.

And the right to fork is kind of like very important to blockchain. First of all, it assumes, right? And I know there's like a big focus for you guys. It's that you have these kind of like light clients and that you're not just querying some RPC node for the state of the chain, but that you're actually verifying it. I mean, that helps for sure for like users that do not run full notes, right? But like it is not like it's what Mustafa said about the right to fork. like is independent of if users are running light nodes or not.

Literally the nodes of the network, right? Like that can be validators, but also full nodes, can the community as a whole can decide which fork to follow. It's more like a social contract. So that is, yes, the light nodes,

like the more users run nodes, the more direct impact they can have to choose on which fork to follow, for instance. but it's not a hard requirement that everyone runs a live note for that. Yeah. I mean, I guess I can imagine that to the degree to which this sort of works will depend a lot also on, you know, how are these applications created, right?

Because in the end, if you have some app that's being run as a roll-up, and, you know, there's like one team that's just building like the application and underneath it uses roll up and they just upgrade the application and everyone sort of automatically upgrades it and I guess it's going to be pretty hard

to coordinate some fort. Yeah, I mean, it's definitely it's also a true, for example, like in the theory of Bitcoin is the core developers, the core developers have a lot of power over kind of like

what upgrades get approved. But it's also the case that even though they do have a lot of power deciding what gets get approved it's also the a big point of it is that it'll be very difficult for them to propose an extremely contentious upgrade

so if you look at like these the contentious upgrades of Ethereum like imagine imagine like some imagine the core devs for like increasing the supply by 10X was a good idea that would actually be a contentious upgrade that nodes

or the community you probably would not adopt So it's kind of like more that the ability for nodes to choose which fork they follow basically provides some level accountability for the developers because it makes it a way they're selected for them to try to deploy a contentious upgrade. And that's almost what happened in the Dowth, in the Dow hack, right?

Like when they say I'm in the Dowdhack, the developers forked chain to undo the hack and that was extremely contentious that eventually it was fine right because eventually the fork happened and the ETH ticker is now on that fork that the developer shows

but it was extremely contentious and it wasn't for free like there was it did end up somewhat splitting the Ethereum community and you had the Ethereum Classic which was big for a while now I was not found dead but it wasn't it wasn't for free affected. It basically makes it that there's a big social cost to proposing contentious upgrades.

Do you think that in terms of, you know, creating sort of decentralized networks, you know, that are really like robust censorship resistant, can be shut down, you know, sort of realize the ultimate goals that, you know, I think Bitcoin and Ethereum and all pursued, Do you think that, are there good arguments for building, like, layer ones versus roll-ups? Or are they basically sort of, like, equivalent in, or maybe are they even advantages of roll-ups because of, like, this, that maybe force are a bit easier? Like, how do you see that? I see a lot of advantages for change to build roll-ups.

And I think the main advantages that you inherit, like what Mustafa said in, beginning is that you inherit the security of an existing layer one. So that means that you don't have to bootstrap your own security in your own validator set. I think that's the biggest advantage in the sense, especially in the bootstrapping phase of the early stages of the project. That's, I think, like, the key advantage. I think other than that, you don't, you have very little, like, downsides and no downsides, basically it's like

especially in the case of a sovereign roll-up it is more or less the same sovereignty that you get as running your own layer one like if a community launches a layer one or if a community launches a sovereign roll-up is not that different from

the actual like say amount of sovereignty they get right like they don't have complete control over the consensus or the data availability part but they can very the data availability part completely. And then over the state machine,

that's what people usually care most or the communities care most about, which contains the logic over the token and the more of the social contract of that community is enshrined in that. And then I think that's the more important part for these community or community computers, so to say. And like there's literally no disadvantage

unless you want to, for the sake of itself, deploy a different consensus, right? Like, that's the only reason I can see where it just makes a lot more sense to deploy a new layer one instead, right? You want to try out a new consensus algorithm directly, and you don't want, you want it bootstrapped from scratch. Like, that's the main reason I can see. I'm curious a little bit when you look sort of at, you know, this case and to what extent can Celestia like satisfy, you know, all the needs that exist for data availability? And I'm especially curious. I mean, I know there is this, I think, path to go to one gigabyte blocks. So like, is that, you know, how much is that?

Is that enough for like, you know, 10,000 roll-ups, all the roll-ups? Oh, do we need more? I would assume that, like, one gigabyte would be sufficient for all existing roll-ups, for sure, but also beyond that, like, needs that will arise beyond that. If you, that depends on how you calculate it. But if you use similar compression techniques as like Vitalik mentioned in his blog post for roll-ups, I think already with like four or eight megabyte, you can, you can already launch multiple like PayPal or like visa scale networks on top of the Celestia. And even like the whole Solana chain like would fit, so to say, into Celestia, even with eight megabyte blocks.

Right. So and then with like a gigabyte. So you can have like many of these, right? Like multiple Solanas, multiple visas. and visa scale networks and so on so forth. So yes, I think that would, if we get to a gigabyte, that would already serve most use cases.

I mean, I know it saw some projects that are basically arguing that, oh, but there's like some sort of AI use cases or some use cases that are maybe not roll-ups that actually have like much larger data requirements like do you think there are examples like that? Definitely there is. So for example, like we've talked to teams

who are doing like AI applications where they want to deploy like entire AI models on chain for each user. Like each user might have like a 5-bikabyte AI model that they want to upload a chain. And that requires vast amount of their people. And I think to use an analogy

is like, it's very similar to development of internet bandwidth where I started use cases like streaming weren't possible with dialogue for example there's a lot of internet use cases

that were only possible once we had once we had greater internet speed and I think we'll see a very similar story with blockchains like once we have much greater data throughput I think it's a matter of induced demand like we'll see

like use cases a lot that simply went also in the past. And I guess connected with that, so does latency matter? I mean, currently like the celestial block time in, you know, like 12 seconds,

is that the limitation and does it cause problems for particular types of use cases? I mean, so the, I mean, the core setting of bus slashia is that it's the only kind of like DA solution with a single slot finality,

or finality on every plus because we use 10. So for example, like Solana's finality term is four seconds. So like we have the same finality time of Solana. And we want to, we plan to decrease that even further. For example, like there's a

there's a network upgrade coming to half the final time for six seconds. And in the future, we want to decrease that even further. But we actually decrease it carefully because if we just decrease it

like naively then it will make it very expensive for light nodes because light nodes have to download a lot more block headers so what we're trying to do

is trying to decrease it but and having this idea of microblocks so like we have like very frequent microblocks but we have less frequent blocks and likewise

we have to download blocks with the microblocks so it doesn't affect how much data they have to download but I think Celestia is positioned by having this kind of like this fast finality because it makes the user experience a lot better, especially if you're trying to cross-roll-up transactions.

Like if you have two roll-ups of Celestia and you're like bridge between them, especially the ZK roll-ups, you can do with Ethereum, for example, you have to wait 12 minutes. 12 minutes is the minimum is the lowest possible time. It's really that you're posting ZK proofs every block. And 12 minutes is a lower as possible time because the Ethereum's finality time is 12 minutes. but because celestiality time is in seconds, that means you can have a very, you have a much faster,

bridging experience between these problems. Yeah, I mean, I think that actually is a great segue into a topic. Yeah, we wanted to bring up as well. I think, like, Ismail also, like, talked about this a little bit. So, yeah, let's talk about, I guess, a bunch of things, but bridging, interoperability, and I'm curious where,

the role is of ZK. I think you mentioned it sometimes. I saw I mentioned in that, you know, Celestia roadmap. So, yeah, I mean, maybe just start here where you think makes more sense. Sure. So, like, one of the core values of Celestia is that we want to have a very minimal

execution environment. And because we want to minimize these on-chained state. Because one of the things that limit of the constraints get blurt is state bloat. So if you've got Ethereum, like, one of the bottlenecks for Ethereum is the share of out of state there is, they have to store a RAM to one node. So, and that's why we, that's why there's a concrete design decision that we don't want to have a smart contract environment last year.

We're only focusing on data valid prosy and scaling that. But the problem with that is that, like, if a role or wanted to use a T.S. token in a trust, minimized or native way, they would have to reuse third-party solutions or third-party bridges. So, like, for example, if we have to transfer that TIA token over IBC to some other chain,

and then that other chain will have to bridge to them first. So right now, there is a really, like, a native way to bridge assets like Tia from the Slesia base layer to each roadups. They have to use some third-party bridge, basically.

But in all that, so in order to make that possible, you basically need some kind of, like, expressive way to have programs on the base layer. And the naive way to do that would be to basically have a small contract environment in the base layer. But it turns out that you don't have to do that. You can just add, suppose you keep proofs the base layer. And then what happens is that instead of programs being directly executed on the base layer,

the programs are executed off-chain and then prove it on the base layer. and that's basically how ZK roll-ups work. So if you have a ZK roll-up, that means you can now bridge directly to Z-Seltier using by verifying Z-K-proof directly on the base layer. This is kind of really powerful because it's basically like the Slesi will be the first base layer

to have this thing called functional escape velocity, which means that you can extend the functionality to the base layer arbitrarily. but without us leading to support on-chain smart contracts we just have to support ZUK proofs and basically the end game if it's effectively that Slesha becomes a base layer

that only does DA and verifies proofs and that's it and I think that's also kind of like the end game for Ethereum right but Ethereum is stuck with this on-chain smart contract environment that it's kind of inherited and then has become has a kind of a baggage that limits its capability. So

and the kind of cool thing is that so these roll-ups will be able to have to bridge directly to the last year. And we call this lazy bridging because it's lazy in three ways. It's like it's lazy for the user because the idea is to create an end-user experience

where everything feels like one chain. So if you're interacting with multiple role-ops, ideally we should have a wallet where interacting with all of those different roll-ups to feel like

we're attracting with one chain and that's possible with things like chain attraction for example but it's also uniquely enabled by the fact that Sylvester has an immediate finality

on every block so because we have like our finalities a few seconds instead of 12 minutes these actions that users can take they will feel like

they're happening on one chain so if say you have Tia on roll up A you can go and mint an FSC using that TIA and roll up B without having to bridge it first, right? And that's already possible, for example,

with 4x chain using Astria. So it allows the users be lazy, but it also allows the developer to be lazy because the developer can have access to all kinds of assets within not just the Selesia X system, but outside of the Seleshii X system, because Selesha supports IBC. And the kind of cool thing about that is that means role officer,

SELESHA have a related way to access assets outside the third-eco system without relying on these kind of non-neutral third-party bridges. So if you look at like, for example, like the bridge between Ethereum and Solana, it relies on Wemhole, right? So it doesn't feel credible, right? It feels like you have to trust some third-party to use that bridge. But Senexia, like Rolos and Sinesia will have assets outside the Lubs-Syssinic system because Celestia has

supports IBC. So for example, you can go on Mint USDC on Noble and then you can bridge that USDC to the ROLOP the RELOPS. This roll-ups will have native

access to UDC for example. And not just UTC, but any other assets on any other IBC support chain. And so that kind of like makes SELESHia very uniquely suited to being this base layer that allows users

or developers have access to assets as it's kind of like a bridging app. And finally, it's also lazy from a base layer perspective because it's still keeping the base layer lazy because the base layer is bernly responsible for verifying proofs. Like it doesn't actually execute the transactions under the show ops. It's just verifying proofs. We're not adding smart contracts to the base layer.

It's almost a year, right, that Celeste. launched and now I think yesterday right the first upgrade with lemongrass came through what were the biggest changes in that upgrade I mean the biggest changes were also bridging related one is like or or interoperability related one was adding interchain the support for interchain accounts and the other was the packet forwarding middleware such that you can like basically you can forward iBC packages through celestia or or basically that. These two features mainly

were requested by many liquid staking provider teams but also from teams that want to build better UX for bridging on top of Celestia. So I think that was the that are the biggest changes

but there were also many smaller improvements I think one of them which is a like a precursor for feet burning is adding a min gas price, right? Like, which is hard-coded into the software

instead of like having validators only choosing their own local gas price. Actually, I think that would be interesting to talk a little bit about here as the economics. I mean, I mean, one thing that's been interesting, right, to see on Ethereum is that, you know, Ethereum used to make a lot of money from, the fees, right, that were paid on the layer one. And now they've had this kind of strategy to scale using roll-ups. And it has led to dramatic decrease in revenue, right? For the theorem layer one, I mean, I've also seen, you know, numbers for Celestia.

I don't know what the numbers are exactly right now. But, you know, basically that, like, you know, the amount of, fees that are being paid for the durability on Celeste are still like very, very low. How do you see that kind of developing

and what are your expectations about the kind of economic dynamics we're going to see here? Yeah, I mean, it's kind of interesting because every few years like someone comes up with some different value of cruel story for why L1 is strictly value.

Like a few years ago, it was like the main value of course story for EAS was like it was used as a monetary asset for as and defy collateral to the defy ecosystem. Now people are fixating on the revenue. So it's like every few years, like it's like every few years people have this opinion, this opinion about how L1 token should accrue value. Ultimately, like revenue, fee revenue in all L1s is relatively, relatively, relatively, relatively, relatively, low, like even on Solana and Bitcoin, for example. But the way that we can't take is kind of like, it's kind of twofold.

So firstly, effectively, we believe that it's possible to put a base area to be sustainable using the economy of scale. So historically, the way that L1s have accrued value by fees is by kind of like this artificial scarcity. but so like if they have used to have a lot of fees because it had very limited in lock space and people were paying $20 transaction fees or $100 transaction fees for a transaction to get access to liquidity and like that's just fundamentally not sustainable like even like maybe you'll maybe that's successful in the short term but someone that that that the quixity is just going to move to some other chain with the transaction fees are that expensive so parliament that

believe like the only way to make a chain sustainable is by offering like by selling many transactions at scale to billions or billions of users. So for example if you even work with eight megabytes, net and one gigabyte of workspace, if each user is paying a tenth of a cent a transaction, then that can still equate to hundreds of millions or billions of dollars of protocol revenue depending if it's a, what half all of it is, whether it's a ZK or a probabilistic model. And fundamentally, and I think that's the only way to scale revenue if that's what, if the revenue is the important part story.

But ultimately, I think that's a, that's a theory of value accrual as it been proven. Like, there's no proven value accrual mechanism for, for L1 tokens. As I said, it's like, it's still very early. and no L1 really has that many fees. Even like Salonah, for example, compared to this valuation. Like, Ethereum had a lot of fees for a few years until they had roll-ups. So there's no kind of like proven theory of that year.

But the other kind of like value-acool mechanism or the other use case for the Sleicius Open is that it can be used as a token to boost-strap new roll-ups. So just like Ethereum can be used to pay for gas, for roll-ups or pay for and a cardful for the defy or as a bond for sequencers. Similarly, Tia can be used to effectively as a way to pre-strap new roll-ups, whether that roll-up needs a gas token or that roll-up needs some way to bond some mistake to operate the sequence set. And that's already what's happening today, for example, with the former chain and

Astria, change on Astria, use Tia for the pay for gas for the roll-ups. And also, for example, you can buy NFC's informer using Tia. We expect that kind of music to continue, especially as we roll out lazy bridging, that makes it much more native and much more frictionness for developers to gain access to Tia for the roll-up and other assets via Celestia. So it's like accessing Celestia with lazy bridging or Tia. going to be easier than other assets that are not native to Celestia? So, no, I mean, the beauty of it is that it will be as easy to access other assets as well,

as long as they come to IBC, because, like, you can bridge a USDC from Nepal to the and then you can go and access that for your role up, because it's all just IBC, right? Like, the way that the lazy version will also be using IBC. It's just like a different type of IBC client that verifies ZK proofs instead of about A signatures. Isma, you mentioned before ICA or like interchangeing accounts being added. What are the kind of ways you hope to see it being used? Literally, whatever you can do with it.

So it's not like this upgrade is basically like a precursor to lazy bridging, right? Like lazy bridging requires ZK accounts. There's still like ongoing research work or like very concrete. research work is very applied, but like it's it's it's still requires some development work there until zK accounts and zK iBc are ready. But like we don't want to block the ecosystem so to say on on using existing interop solutions and that's why pfm and i see interchained accounts are already in the upgrade. But I mean I think given that the teams that requested

these features are mostly liquid staking providers. I would assume at least that these are the major use cases like for ICA at least, for interchain accounts at least. But there's no opinion

from like I don't I don't want to say like this is what I what it should be used for because like literally whatever it can be used for it should be used for that said I think for the

for interchain accounts I think the set of messages like SDK messages are somewhat limited for instance you cannot do at least that's what the

CIP proposed right like you cannot use out CLE messages via interchain accounts so there is some opinion and it's like the minimum set of things are allowed such that it is useful.

So when I said whatever can be built with it, it's already a bit limited but only a very little amount. Maybe zooming out a little bit I think in general it seems like in the crypto space there's a lot of kind of concern

that there's been all this investment in infrastructure scalability, interoperability, and at the same time, we don't see a lot of user growth and adoption. Is that something you guys worry about? Like, where do you guys think we are

in this sort of crypto adoption cycle? Yeah, I mean, that's the is this essential question, right? Like, where are the apps? But I do think, I do think like we all seeing of like cooler and cooler apps plumbing online.

I think the original application of like crypto is ultimately payments. So I think that's not like the biggest problem. How can we actually make payments, cryptos of payments machine?

Because to me that's that should be clear application. But with roll-ups, like we are seeing like cool experiments with payments that are possible without roll-ups. So for example, there's a project called pay, like P-A,

that is a roll-up that uses Sylvesterquadia that is kind of like a privacy-preserving way to transfer stable coins and do payments. And so I just think we kind of like need a lot more experiments in payments to see kind of like which payment application can actually achieve this breakout eruption. And so far it's been like, it's been USDA on Toronto. USDC on Toronto is used a lot. So for some...

USDT. Yeah, USBT, sorry. But I think we're also seeing, I thought that payments were seeing a lot of other cool applications. Like, on-chain gaming is getting a lot more advanced. Like, you know, like, a few years ago,

you just be like, on-chain gaming was very basic. It was, like, mainly trade, like, NFC-based on-chain gaming. But now we're kind of like seeing people develop much more advanced on-chain games.

This cool project called B-3, which is kind of like a game studio, a new game studio, is launching of three games. I mean, also, ultimately, I think we'll see what kind of uses it unlocked by like raider and raided it to port. As I mentioned, for example,

there's a lot of AI use cases that people are building. That involve users uploading AI models that I review on chain. And especially if you sayes like ZK machine learning as well. I don't think, like, it should be one or the other.

Like, there might be a lot of emphasis on infrastructure but I think that's a necessity to actually have cool applications. Like Mustafa mentioned, for instance, these AI models, if no one was building something like Celestia, for instance, it wouldn't be possible. So I'm pretty sure that we will see more and more cool applications, but also I'm also not, I also don't think we're done with the infrastructure yet, right?

like it's not we're not fully like all the infrastructure does need to change anymore there still needs to go a lot of like engineering effort to actually scale the infrastructure that we have and then also to the upper layers a bit right like there's this infrastructure that makes things possible from like a like a base layer perspective but like for end users I think there's a lot of infrastructure that is still missing to make their user experience as smooth as possible. And then I think once that is all done, well, we'll never be fully done. It's like it's tech, it's engineering, so it's never fully done. But like once that improved significantly, we will see more and more applications.

I do agree with Mustafa that like payments, like even just payments would be just payments would be like the killer applications. I'm also very, I'm also excited about all these gaming roll-ups coming coming to existence more and more.

I don't have time to play them myself but I always believe that games and on blockchains or crypto are like a perfect fit. So I think like AI gaming

and payments will see more of that and also applications that we cannot foresee yet. And maybe we talked about a few things. We talked about some consensus improvements and the ZK accounts. What are the main other technical things that are, you know, like on the roadmap for the next two years? Yeah.

So, I mean, the main thing is like abundant block space, right? And that basically has two components to it. One is on the consensus layer. like the block producing layer. So basically the 10-dmen network, peer-to-peer network and like the mempool, like there's a lot of optimizations that will happen there to make it way more efficient than it currently is

and optimize it for throughput. And similarly, the sampling and the DA network site needs to improve a lot. And besides like this goal, like this north star of like one gigabyte blocks, it's abundant block space, We want the block space to be verifiable by anyone, and we want it to be frictionous. So I think, like, for the verifiability, basically the idea is that you can, like, run a node, like, a Celestial night node on every device.

Like, you can everywhere, wherever you are, whatever device you have, you can, like, verify the correctness of Celestia, so to say. one of the main features that is required for that is like a proper mainly rust implementation because it's easier to have it compiled into wasom that exists today already you can already go on lumina r s right and and run a celestial light node what i find even cooler is that this celestial light node is now integrated into selenium which is like a block explorer so what you what you ideally want is actually that you instead of using like Kepler's infrastructure

for IPC to submit transactions you'd actually have these light claims running in your browser in your wallet essentially and you communicate directly with them so you can like even have less trust assumptions for end users and having them instead of using some like

centralized API and RPC you'd have them oh I run this thing I've verified it like it's it's not something they have to do consciously, but like it is something that we are pushing forward such that the like end user gets the same experience that they have today or even better, but at the same time are actually running their own node. So that's like the the verifiability part. And then we want the block space to be frictionless that that obviously includes like your, like developer UX improvements or

developer experience improvements, like API improvements for the, for the, for the, for the, for the for the for the, for the, for the, for the, for the, for the, developers. But it also includes things that like Mustafa mentioned, um, with like ZK accounts and, and, and, and, um, what we call lazy bridging essentially. Um, such that you can like more easily stream like any asset in and out, uh, through Celestia. And I think the, the, the mental model here is that while you interact with, um, uh, a, an ecosystem of very heterogeneous chains on top of Celestia, like EVM chains, Solana, like SBM chains, like SBMs environments that don't even exist yet, and rocket chains,

and so on and so forth. Like while you interact with in this, like while you move in that ecosystem, it should feel like you're on one chain, right? Like that's the idea is like as if you were like on Ethereum or on Solana, like one layer one and you would interact like with these smart contracts. it should be at least as good as this and yeah I think that's the high level summary of the roadmap

cool well thanks so much guys for coming on I think this was super great to catch up on Celestia and yeah it feels like it's it's really sort of getting to the place where we're going to hopefully start see this the consequences

right of these abundant cheap high-performing block space. And, you know, I really hope that Mustafa is going to be right that there's all of this demand and use cases that are just sort of like waiting to happen when, you know, when we have that infrastructure, when we have that like cheap, fast, scalable, interoperable block space. So I'm super excited to see what's going to happen in next year and two years for Celestia in the ecosystem.

Likewise. Thank you. Cool. Thanks so much for coming on. Thank you. Thanks.