Episode 268: A Rollup-Centric Future & Sovereign Chains with Mustafa Al-Bassam

Transcript

foreign [Music] welcome to zero knowledge I'm your host Anna Rose in this podcast we will be exploring the latest in zero knowledge research and the decentralized web as well as new paradigms that promise to change the way we interact and transact online [Music] this week I chat with Mustafa al-basam co-founder at Celestia we chat about where Celestia as an idea emerged from and explore how their roll-up Centric data availability system compares to that of ethereum we also touch on what concepts have emerged in the last year

within their ecosystem things like Sovereign chains and what we can expect from their upcoming launch now before we kick off I just want to direct you towards the ZK jobs board there you can find job posts from Top teams working in CK check it out and maybe find your next job opportunity working with the best in the field now Tanya will share a little bit about this week's sponsor [Music] Elio is a new layer 1 blockchain that achieves the programmability of ethereum the privacy of zcash and the scalability of a roll-up if you're interested in building private applications then check out Elio's programming language called Leo Leo enables non-cryptographers to harness the power of zkps to deploy decentralized exchanges hidden

information games regulated stable coins and more visit developer.alio.org to learn more you can also participate in Elio's incentivized testnet 3 by downloading and running a snark OS node no sign up is necessary to participate for questions join their Discord at elio.org forward slash Discord so thanks again Elio and now here's our episode [Music] today I'm here with Mustafa al-basam co-founder at Celestia welcome to the show Mustafa hi thanks for having me on so Mustafa in preparation for this interview I went back and listened to an earlier episode I did with your co-founder Ismail Coffey in that we described Celestia and what it was Data

availability but I think I really want in this episode to kind of like revisit it because I know you guys are much closer to launching I feel like you're going to have more insights into what what it actually looks like in production I think I'd also like to explore a little bit more what's happened since and what's what does the project look like going forward now before we do that though I think we should get to know you so when I talked to Ismail he had said that he was coming more from like a cosmos ecosystem and that you and John Adler the other two co-founders had come more from ethereum so yeah tell me were you an ethereum person what was your sort of start yeah I mean I definitely um

started using ethereum you know when it first came out but I actually first got into kind of blockchain networks and Technology in around 2011 um 2010 when Bitcoin was one or two years old and I got into Bitcoin because even before Bitcoin I was naturally interested in peer-to-peer networks like including like peer-to-peer foul sharing systems like BitTorrent were really interesting to me so oh generally kind of very interested in the concepts of like peer-to-peer technology and um I kind of got into the Bitcoin scaling debate and the the general area of Bitcoin scaling through the Bitcoin block size War back in around you know 2013 to 2016

and when bitcoiners were arguing about how to scale Bitcoin should we use lightning or should we increase the block size and when the when the Bitcoin cash Fork happened the reason why I found ethereum very interesting is because it was obviously a general purpose smart contract platform but the innovation in Bitcoin has started to die down the Bitcoin Community started to become very technologically conservative and uh they were more interested in Bitcoin as sound money which is very interesting and a good goal but for people like me I was also interested in the technological aspects of decentralization and peer-to-peer networks and so a lot of researchers and Innovation moved to ethereum and that's

kind of like how it actually ended up in the ethereum space and also eventually ended up in Cosmos what year was that that you kind of switched over to ethereum was that 2016 then kind of at the end of that block space kind of conversation in Bitcoin yeah this was around 2016 or 15 when I think this was just after ethereum had released this mainnet I was doing my last year undergraduate computer science project on developing ethereum application at the time it was like a decentralized sort of identity service oh very ahead of its time there's like a lot of those now that are kind of coming into the Forefront but so you were doing an ID

like a digital ID thing yeah but actually even though it might seem ahead of his time when ethereum was a few months old there was there was like 10 of those projects already oh because it was like it was like a very kind of like obvious or obvious idea for building something on ethereum and I remember going to ethereum Defcon 1 in London I think yeah I think it was Defcon one and that's kind of like when I realized that you know this thing was becoming real and people are actually building stuff cool you would sort of explored this uh decentralized ID thing but like did you build anything else in ethereum were you involved in any other projects that we might know so I didn't really build anything else you know that went into production ethereum at that time but

after 2016 I started a PhD on the topic of layer 1 scalability in 2016 at UCL in London and over there I was looking at sharding specifically or like layer one sharding and I became you know somewhat involved ethereum research Community around then because ethereum at the time had a very complicated ethereum 2.0 spec that also revolved around execution sharding yeah and so there was a lot of interesting questions around how to do sharding things like how do you do cross Shard transaction calls and how do you make Charlotte insecure and I will and I was hanging hanging around a lot on the

ethereum research forums and as part of my PhD research we created an academic project called chain space which was the first uh kind of like proposal to create a sharded smart contract platform and that was later commercialized into a company that was later acquired by Facebook oh yeah but I didn't join Facebook and I ended up working on Celestia instead which was formerly known as lazy Ledger yes and I think so like I actually have had John Adler your third co-founder also on the show long while ago when it was called Lacey Ledger who's talking about Fuel and lazy Ledger I haven't listened to that episode in a long time but we can dig it

up if listeners want to go back and sort of see the trajectory of this project so that idea of separating execution from consensus was that what you were doing like had you sort of formulated that or did that idea already exist I think there were some like General like early ideas like technically speaking the idea of separating consensus from execution actually existed before blockchains themselves okay there's this paper from I think the early 2000s that proposed a BFD Byzantine fall tournament protocol that was only responsible for ordering but execution happened in a different committee but in terms of the idea of separating consensus from execution in

blockchains there were some early discussions about it but no one really kind of formulated a a system like lazy Ledger where the only purpose of the system was ordering like people had always talked about it actually and even before ethereum if you think about before ethereum existed we had mastercoin and mastercoin which is the system that tether uses and how on bitcoin that's basically a a roll up on bitcoin technically that uses Bitcoin for ordering but not for execution in that case this is sort of a side question but like what part of the Bitcoin architecture is it using to actually do that is it like the memo field like what can you use in Bitcoin

to create a Roll-Up yeah so it uses this uh Bitcoin op code called up return and so like and I think in that op code you can basically have 40 bytes or I think it used to be 80 bytes of arbitrary data but there's many different ways of including random arbitrary data in Bitcoin transactions I think actually the way that mastercoin did it was you send it to a fake address and the address encodes the data that you want to include nowadays it's completely different and nowadays you have something called Taproot input scripts and you can include extremely big files like entire jpegs on the Bitcoin blockchain now cool so then you were you were kind of working on that the sharding like basically what ethereum became do you

feel like it was it influenced by your work at all or do you think that they kind of continued on their path and you just sort of like split off to form Celestia or do you think there was yeah was there some sort of takeaway it's hard to say if it was directly influenced but what I do know is that the ethereum roll up Centric web map as it is now was was basically what we're building at Lazy Ledger a year before ethereum switched to a roll-up Centric web map got it because yeah originally there was this idea there was there was going to be a consensus part there was going to be execution sharded but it was like in the actual client software all run by the exact same validators and the roll-up concept was introduced I actually don't remember exactly the year

2019 maybe I think 2019 or maybe late 2018 yeah yeah and um I do remember that blog post from vitalik when he kind of laid out the like roll-up Centric future I know in my last episode with Ismail we talked about how Celestia is this purposely made blockchain kind of L1 it's made very much for that roll-up world so it kind of cuts out a lot of the other stuff that maybe ethereum or other ethereum like blockchains have and just focuses on being kind of the perfect Center for that roll up like I guess you do you call it like a hub and spoke it's kind of like a hub and spoke right well we didn't say it's a hub because we

don't have a settlement layer so like the Roll-Ups don't bridge to it but they all use it as we call it a shared da layer so it's like all the Roll-Ups use it as a shared the a layer you can call it a hub in that sense but it's not a hub in the sense of it like the cosmos hub for example I see I see so yeah so but then let's think about that let's maybe explore it for our listeners so are you thinking about the other blockchains as Roll-Ups or are they actually Standalone blockchains what do they have and what do they not have in order to actually work with Celestia so they are Roll-Ups but they're not l2s I see Celestia why not tell me why because they don't have a bridge in Celestia but you can't transfer assets from the Celestia chain

itself I see you're in a trust minimize way to the Roll-Ups so in that sense like we don't call them l2s because they're not really like baby chains they're not they don't really extend the L1 I got it they inherit security from the L1 but they don't necessarily extend the l1's functionality are they just Outsourcing consensus then like they are just pure execution and Celestia is pure consensus yeah they Outsource they basically Outsource consensus and data availability to Celestia they have very similar properties to normal layer 1 blockchains the only difference is that they Outsource their consensus and data availability to Celestia and you did just say that like you're actually working on kind of ordering does celestia then act a little bit like the

sequencer of Roll-Ups it can do in the sense that Celestia definitely orders the Roll-Ups blocks but it does not mean that Roll-Ups can't also have their own sequencer I see if the roll-up wants to capture its own Mev but you can they roll up can choose so let's suggest you to order the transactions inside the blocks but then that means for this year will capture the nav oh it would there be like a Searcher community in something like Celestia or is this the validators it's automatically kind of built into the system I mean there's a lot of interesting questions around how Mev low would look like in a modular stack and it depends on where the mov is captured I think it's most likely that

Roll-Ups would want to capture their own Mev and not leak the Meb to Celestia so it's likely that there's going to be Mev within each roll up and and the sequencer available will capture its own EV but there could also be you know some Mev on the actual data layer itself if you constructed the roll up so that it has like a decentralized set of sequences and the roll up could order the blocks for example within the data layer so I'm sort of picturing like so Celestia is there it's doing the consensus and data availability for these different chains they are Standalone chains they look like regular blockchains but then how are they connected to each other would they be connected to each other like is that

something that's kind of encouraged do you have any tooling that helps that or is that sort of up to them they can use any bridging that they want yeah so when we talk about interoperability and bridging that's usually something that is usually a concern of the execution environment and so it's the Roll-Ups that can decide how they want to bridge with each other but what's less she enables them at least to do is to have some guarantee of shared security so like for example if you had two Roll-Ups and they wanted to communicate with each other in a trust minimized way and verify each other's for proofs or ZK proofs they would also need to verify the data availability of the respective

rollup and in order to do that they need a ideally they need a shared data availability layer and so that's what Celestia allows them to do if they wanted to they provide shared security and that share security without enable more seamless bridging if the Roll-Ups wanted to do that but like in the case of ethereum you are kind of doing the same action let's let's kind of dive even deeper into like the differences here like so what does ethereum actually provide that maybe Celestia doesn't and vice versa the main difference between ethereum and Celestia is that ethereum kind of started out with a non-roll-up Centric food map where they already had the ethereum virtual

machine enshrined into the chain itself so they already had like a smart contract environment on the chain and now that they want to go to a roll-up Centric roadmap you have the kind of existing baggage of all of the state of the ethereum virtual machine whereas this last year we don't have any small contract platform enshrined into the data available layer itself so what that allows is if you want to create a sovereign roll up for example which we can discuss later it allows roll ups to be more overhead minimized because the Roll-Ups no longer have to take an interest in their validity of everyone else's smart contract on this enshrined execution environment and that makes it more overhead minimized

but also the other big difference is that Celestia does not have a what we call a settlement layer because in ethereum Roll-Ups ethereum roll-ups post their fraud or validity proofs and their headers to the ethereum chain and that allows for the Roll-Ups to transfer assets to each other through the ethereum chain but in Celestia we don't have that we don't have that enshrined into the a layer instead what we expect is that we expect people to create their own settlement layers or networks even as Roll-Ups potentially on Celestia this I need to understand so in the previous example in the ethereum example you have Roll-Ups who are like I'm going to use

the zk1 because we talked about that more on the show but like so you've created a proof you have a verifier on chain as a smart contract that in a lot of ways ensures that like the data in the roll-up is also accurate but you're sort of saying that you wouldn't necessarily have you're like you don't have an execution layer where you can deploy a verifier but can they put some sort of verifier into it like yeah I want to explore this a little more yeah so the key thing to realize is that an ethereum roll-up contract or an ethereum roller Bridge fundamentally is just a blockchain like client because if that what it does it receives block headers for that roll-up chain and it can verify Decay proofs that assert

to the validity of that roll-up header but you don't have to run that like client as a smart contract on chain you can just run it as a normal like client locally on your machine like if you consider let's take Mina for example Mina is a ZK chain it's not a ZK roll up it's just SDK chain and it has light clients that are actual nodes that people can run on the devices yeah and the same thing could work with a as you can roll up on Celestia the ZK proofs can be distributed peer-to-peer through the actual nodes or that are run by the users of the rollup and they can be verified directly by those nodes rather than necessarily by an Unchained contract I'm trying to picture this so

like you're saying they don't need this verifier as contract but rather the light client sort of the compressed version and it's like where is the snark if you're using ZK in this case like are we using snark for recursion and compression and so that's what you the output is sort of like this nice thing that can sit in a light client and then it's never verified it's just sort of assumed to be correct I mean the smoke is distributed with the block header of that roll up yeah and then when any node receives a block header they have to first of all verify that stock before they accept that header as valid but who's looking at the like client maybe this is this is going

to help me like is it the chain itself checking its own like client or is it like external Pro like chains checking those like lines black clients are run by the users of the chain anyone for example running a wallet or any or for example service providers like exchanges or shops or or payment measures and so on and so forth all right so it would be like adapt user would basically be doing that or like adapt exactly developer okay so it's just like how an ethereum who's running the nodes for ethereum is the same answer as who's running the select clients for dck Roll-Ups okay but do the different chains that have like locked in and used this did they ever verify other chains like clients yeah they can do if they want to have a trust

minimize fridge with each other do you still call these Roll-Ups ZK Roll-Ups though in this case yeah why wouldn't why wouldn't they BCA Roblox ZK roll up to me always had this ethereum context of proverb verifier verifier contract but I guess it still uses snarks so I guess this is a good segue into Sovereign Roll-Ups or the different types of Roll-Ups okay so that's why as as I said earlier we don't call these Roll-Ups l2s of Celestia because they're not bridging with leisure uh we call them just Roll-Ups or roll-up chains but it's specifically starving roll up chains because they are not they're not Altus and what that means is that if you consider a ethereum roll up an ethereum

roll-up enshrines ethereum as its canonical Bridge or it enshrines ethereum as like a hard-coded settlement layer or bridge to it and what that means is that effectively is the ethereum chain that defines what is the valid roll-up chain it defines what is the valid Fork Choice rule of that roll-up chain based on what the contract says but in Celestia we we created this idea of sovereign Roll-Ups and a sovereign roll-up is Sovereign in the sense that it's not some Unchained smart contract that defines the correct chain but it's the community of that roll up that defines the correct chain

similar to how an L1 would work like when you bootstrapper L1 or when you hard 4K L1 is that community that defines the protocol rules or the correct chain of that check of that L1 and so it's similar with the roll up you can imagine a sovereign roll up to be just like an L1 chain the protocol rules are defined by the community and is enforced by the nodes of that roll up not by a smart contract and so that's why the fruit and Decay proofs are distributed peer-to-peer when you talk about that like in the case can we kind of walk through an example of one of these Fork Choice rules would look like on ethereum and then what it would look like on Celestia so like and I I think I think in the past we

have covered this once where it's like say you have one of these applications that are so important to the ecosystem and they have a lot of collateral and they have like they have everything that lives on it say those live on a roll-up and there's some sort of problem or Fork that needs to happen What would happen today how would they not have control today yeah so it will kind of boils down to what we consider the upgradability of a roll-up to be like in this traditional thinking for um an ethereum roll up it was considered that well if a roll-up has upgrade Keys then it's not really a true roll up from a security model perspective because the whole point of a ethereum roll up is that you have a trust minimized bridge

between ethereum and the roll-up and that means that there's no party or set of parties that could steal people's funds then you have a very difficult question which is um well how do you upgrade a roll up if there's a bug or if you want to add new features or something like that if that roll up did not have any upgrade mechanism then the only way to upgrade it would be if you hard forked ethereum itself whoa so like if totally impossible okay yeah but of course that's not like that's not practical so even ethereum roll ups today have some kind of upgrade path yeah the current thinking if I understand correctly is that some Roll-Ups on ethereum what they want to do is they want to have like

some upgrade committee like upgrade multi-sync that can upgrade the chain but I think that's what they plan to do is that the upgrade only happens after a month or something and that gives users a time a chance to exit the chain into the L1 like Bridge their funds back off it I guess exactly and so what we've said is you might as well just be a sovereign roll-up if you want your upgrade path to be community-led because uh I guess the difference is that many people including myself they we don't believe in token holder governance what they can hold a voting especially for upgrades it's like if you consider it ethereum for example

how is how are eips approved or are ethereum how is ethereum upgraded it's not some multi-sake or took into all the governments it's uh the community or the ecosystem like if if the ethereum developers or some way or try to push some controversial EIP that's still everyone's money even if the the developer of guests wanted to do that it's very likely that exchanges for example would not uh download or use that patch yeah yeah and so you can apply a similar kind of upgrade mechanism for Roll-Ups if it's a sovereign roll-up you don't rely on token hold the government so you don't rely on unchain governance or multi-sig you just rely on hard Forks as an upgrade mechanism and yet the hard Fork

here in this case like consensus of Celestia is not altered it's not like the chain is changed and like so tell me how you would actually hard fork in this case right I mean I'm assuming Celestia like the the validator set of Celestia is not like obeying one of these Roll-Ups Fork changes yeah so that's kind of like the the BTF Sovereign Roll-Ups you don't have to hard Fork Celestia or the d a layer to hard Fork your roll up to hot you roll up you just need your roll up Community to agree we're gonna change the protocol rules in this way and then everyone has to download the same patch and upgrade their node

software to change their protocol rules in the same way would you then have to create new like clients on would you just sort of like wipe the old one and make a new one it would just pick an upgrade to the code like let's say you wanted to introduce a new transaction type for example that would basically be changing the kind of state machine of the roll up and then that would just be like a you know it could be like a few lines of code that change in the consensus critical part of the roll up interesting and a good example is the Dow hack like it like back a few years ago when ethereum was starting out there was this project called the Dao and people invested 150 million dollars into it which at the time was five percent of the ethereum supply and then it got

hacked and some hackers stole stole all that money and then ethereum community decided that they're gonna they want a hard Fork to undo the hack yeah because five percent is way too much uh to be to be stolen in one go it's enough to wreck a new network exactly and the way they they did that is they hard forked ethereum chain and they said that the hard Fork basically said at certain height we're gonna move all the solvent funds into this new contract and so like my question is like why like I think Roll-Ups or roll-up chains should have the same power to do the same thing because they could also deploy things that have bugs or there could also be some sort of like application on top of

that that's gathered so much of the wealth and and value of that roll up that gets wrecked and needs to be upgraded yeah I mean it sounds like though the Roll-Ups do have mitigations you said most of them are upgradable they have committees when you say community like how how would the community of a sovereign chain look different from the community of a roll-up the main difference is between do you want the roll up to have a token holder base or multi-sig upgrade mechanism or do you want the upgrade mechanism to be based on more ad hoc social consensus there's many people including myself that are not in favor of token holder governance or consensus because to me the whole point of blockchains is that it's a trust minimized protocol

were given a certain set of rules no one can violate their rules even if there's a dishonest majority of people in that ecosystem in other words like the only way that you could change the rules or steal someone's money or seize assets is if everyone in the system agrees or if the vast majority of people in the ecosystem agrees to those new rules but if you have a system where that's based on token holder governance or a multi-tig you're basically saying that actually the state's transition function of that chain is not some specific rules but the state transition function of this chain is what that committee says is fine or what those these whales say is fine whereas to me the most important part of

a blockchain is actually it's social consensus like layer zero for blockchain is sexual consensus what gives a blockchain Authority why the Bitcoin chain has a BTC ticker and why the ethereum chain has the ETA Shaker is due to social consensus about the the social community of that chain agrees that is the rules of that chain what happens though in the example you gave a sovereign chain to the other part like the what do they call it the like the orphan chain the one that's left behind like does that still exist Does that cannot still interact with Celestia it can do like ethereum classic for example yeah like ethereum classic still existed after ethereum hard forked but it was

completely separated so it was just like its own thing but in this case because they're still using the Celestia consensus what happens to it pretty much it becomes its own kind of hard fork or or chain but like it's the same thing as ethereum classic like the stain still exists or at least the social consensus of the new chain does not recognize the old chain as the source of Truth the source of truth becomes the new chain or you actually just get two chains yeah you definitely get two chains but what chain do all the exchanges use what changes all the wallets use what chain gets the same ticker and so on and so forth I see I say I want to explore a little bit the not only the data availability part of this but also sort of the querying for

data because this was something where I understand there's like a very specific architecture so these Sovereign chains or roll-up chains actually do call all the roll-up Sovereign chains on Celestia oh so all the chains that other reckon Celestia are sovereign chains okay but you can you can have non-sovereign chains on top of sovereign chains like l3s kind of thing yeah like you can create you can create a sovereign chain that in transit I see okay but let's talk about like what like let's go a little further into what the data availability side of things are like one of the things I understand about Celestia is being unique is this ability to kind of query faster or easier or something so like yeah tell us a little

bit about that how does that work yeah so that relates to the sort of data application model of Celestia uh the whole point of Celestia is that it's supposed to be a very simplified data layer like the only thing it does is it lets developers post arbitrary data to it that's list your orders and includes in a block and Celestia commits that data in a medical tree and then you get a medical route and developers can use that metal root to authenticate certain data in Celestia but the way that we do that is we have this concept of namespaces yeah and namespaces are kind of like you can kind of think of them as like like radio channels like if you have if you've got like a walkie-talkie for example you might have like different

channels in that walkie-talkie or you can also think of them as like Twitter hashtags so once you when so when you create a roll up or you create an application you give the application a certain namespace is it like a field is it sort of like a field and it give you give that particular roll up a number or something that's like an identifier yeah it's just an eight byte integer right now okay the developer just chooses the arbitrary integer and they can post arbitrary data to that namespace and then all of the other people can query the selection blockchain to say hey can you please give me all the data can you stream to me all the data that has been posted to this namespace and

this is sort of a way to like identify this particular Chain versus other chains does this just make searching for data faster because you can still kind of do this on ethereum right it's just that it's like clunky and it's mixed up with everything else so what allows you to do is because we used what's called a namespaced medical tree it allows you to query a node in Celestia for a specific namespace and and then that node can give you a miracle proof to say here's the data in that namespace and you will know that that response is sound and complete in the sense that you will know that because it's a namespace medical tree you will know that that is

all the data in that block like they haven't emitted any data in that namespace ah I see okay so it's kind of like an authenticated or trust minimized way of querying other nodes to give you all the data in that namespace in a trust minimized way without you having to download the entire chain yourself like you can run a light node your roll up sequencer can just be a Celestial light node so does that mean that it's faster or more efficient like what's the comparison then the main difference that it's more trust minimized in the sense that if you were to do this on ethereum and you were running ethereum like client and you just ask another full note to give you all the data and for this smart contract for example you would have to trust that full node okay to give you

the complete set of the data or at least you would have to trust the validator set of ethereum to give you that complete set of data it's not like the main big thing of Celestia but it's like a little nice feature to make it easier for developers is it possible for the node to give you the wrong data like on back on the ethereum case like would that be it can't give you the wrong data but it might not give you the full set of there oh it might emit data does it also help with like looking at history of data the way that you've built this like namespace Merkle trees like does it have this is the thing I'm trying to figure out it's like does it actually enable something new or does it just make for more efficient uh like query like does it just make it

easier for for like clients to do it it's more of a just like a nice developer ux type of thing okay and I guess that means like building applications that rely on it all of a sudden become simpler because you're not trying to parse through all this stuff and Trust all of these nodes and yeah like um if you go on our docs we have a very straightforward data availability API you can submit data into the namespace and you can receive data from a namespace you can search for data in any space and those are just like two basic API endpoints but behind the scenes the developer does not have to worry about what what's happening behind the scenes but behind the scenes that you know that it's happening in a trust in Miles way it's not connecting to inferior or something like that it's not connecting

to some centralized some socialized like index or something like that yeah yeah it's connecting directly to the Celestia peer-to-peer Network and that's kind of like a very big aspect of what Celestia is about and it's like a very big value of Celestia that users should be first class citizens of the network they shouldn't have to connect to infuro or some centralized API to get data out of blockchain they should be able to just connect directly to the peer-to-peer network of that blockchain and get data about it in a trust minimized authenticated way when you mention sort of these like client constructions it definitely makes me think a little bit about some of these new um projects that have emerged they're sometimes acting as like I know they

don't they don't call themselves bridging projects they're not really Bridge projects but they are like the like client that could connect two chains to one another and like the projects that come to mind here like succinct or Herodotus ckibc Axiom there's a bunch of them so they're like they've found ways to create these like client constructions do you see yourselves working with projects like that are they competing at all with you so this projects are generally operating and they're a slightly different security model because if I understand correctly what people typically mean by azuki IBC or ZK Bridge is that what they're proving for in the forms in ZK what they're proving

is not the state validity of the counterparty chain but what they're proving is that the counterpart the consensus of the cotton party chain they're proving that some kind of a party chain was signed the headers of it were signed by two-thirds of the validator set for example that's kind of like a very helpful thing to do but it's not as secure or as trans minimized as a like a normal GK rollout Bridge because a typical like a normal ZK roller Bridge like a bridge between let's say ZK sync and ethereum it verifies not just that the header was signed by the correct sequencer it also verifies that every single transaction in that roll up was actually correct inside the ZK proofing system but what people talk

about things like ckibc does that mean that they typically mean proving the headers book signed by the correct value is set that's very different thing and it's helpful but it's not what we're trying to build because we don't consider committee based bridging to be scalable or secure because the way that kind of like Celestia started out as being in a cosmos ecosystem is that Cosmos is a very interesting idea because the idea of it is that everyone has their own chain with their own validator set and everyone communicates with each other using IPC and that's great but it's it's not scalably secure because there's no shared security it's like if we imagine a world where there's millions of Cosmos

chains which is very possible if you have millions of app chains it's very unlikely that all of those chains will have a secure evaluator set and IBC assumes that the chains that you're bridging to have an honest majority in their validator set because the validator set could lie and insert in the value stage transitions so that's not a scalable kind of shared security mechanism it's not shown in security at all to work so this year I tries to do is it tries to kind of scale the cosmos vision of having millions of chains but the key Insight of Celestia is to imagine a world that does not rely on committee based bridging assumptions or committee based security assumptions

because we don't think that committees are scalable to billions of or even thousands of chains and instead if you try to replace committee based assumptions with ZK or fraudproof based assumptions instead of relying on a committee to tell you that the state is correct you can actually check that the state is correct yourself using fraud proofs or ZK proofs well I do think some of these are actually doing aren't they doing something like that basically taking one Chain's full node and then compressing it into a like line on another chain yeah I think at least some projects are definitely looking at that um I know like Risk zero there's a lot of like projects that are associated or adjacent to reasons zero without looking

at that I think there's people trying to create a Bitcoin full node as a ZK program for example um in software for example actually using Cairo interesting and do you think you could actually use like one of those projects that I said before where they're like using they're taking a like a full node trying you know creating a like line out of it would they be able to use you as the data availability side like would that be sort of complementary in some way yeah definitely and that's something that we're actually looking at ourselves um like the idea that you could compile like any obtain into let's say risk zero or Cairo in the future you could like turn any chain into a GK roll up into in

theory and um yeah like one thing I was I was looking at for particularly for example is there's like a lot of efforts or some effort in the Bitcoin Community to try to zkify the Bitcoin State machine and then I think it was something like they wanted to if you had a ZK op code to bitcoin you could have like Bitcoin l2s for example and I was also interested to see like how you would like to see Celestia could be used as an off chain that ever bit layer to bitcoin Altus for example wild but would that ever happen I mean it's hard to say uh as I said like the Bitcoin community or the becoming developers are very conservative in terms of what they add but actually I do think they're becoming kind of recently more liberal like the

fact that you can add jpegs the Bitcoin blockchain now it's kind of wild that you would that was like that was like back in 2010 and um 13. people were discussing using Bitcoin for non-financial applications like DNS and the Bitcoin developers like Gregory Maxwell were extremely pissed off at that kind of idea like they were kind of like very um sort of philosophically against it almost eh yeah it was like it was very much like you should only use Bitcoin for financial transactions like some developers are still like that like like Luke Jr but uh the developers like roast beef who are looking at you know Bitcoin all twos so I do think there's a potential future where Bitcoin would add

a ZK op code wow it's possible but not like in the next year maybe in the next few years cool do you kind of imagine the case where like Celestia is the data availability Hub I know you're not like there's no token settlement but there's the data availability and the ability to query history of State transitions stuff like that could you also Imagine a separate Hub that would be for settlement instead of this current like mesh of bridges that would maybe even like work in tandem with Celestia yeah definitely and I see people potentially creating settlement layers on stop Celestia because as I said we don't enshrine a settlement

layer we don't we don't enshrine a bridging hop to Celestia yeah we try to mix this yet as minimal as possible but that doesn't mean that someone could create a bridging Hub on top of Celestia the other Roll-Ups reached to and for example there's one project doing that for example called Dimension and they're a kind of like a cosmos based settlement layer for Cosmos Roll-Ups that uses Celestia as data availability layer cool so it's sort of like in the works yeah for sure and there's kind of a lot of debate about whether Celestia should enshrine a salted material or not but I think my current thinking is that it's nice to have a data rapid layer that's neutral to settlement layers so that

people can't innovate with their installment layers so that's kind of my current thinking is that so this year should be credibly neutral but that doesn't mean that in the future of the celestial Community could decide to enshrine and settlement layer for example by using um interchange security and you could have like a shared security layer as a consumer chain to the Celestia chain cool that I was actually going to ask you a little bit about IBC and then interchange security because in a way The Interchange security model which I don't think we've talked about that much on the show but it's the idea of like the Hub validator set being almost like loaned out to these sort of child chains I guess these are really kind of very much connected what did you just call them you had a name for them

uh consumer chains consumer chains so you could sort of see those consumer chains then using Celestia but would they still be using the Hub in that case as the settlement I actually don't know if the settlement happens through the Hub I'm assuming so but yeah so I'm not saying that consumer chains should use this yeah I'm just saying like in the future like if the community of Celestia wanted to enshrine a Solomon's latest last year that settlement layer could be implemented as a consumer chain this last year I say it would look like that it would sort of Follow that model okay yeah but but in terms of like whether consumer chains should you suggestion I would say that's not something that we want to do because the whole point of Roll-Ups is to provide an alternative to

a more scalable shared security mechanism then interchange security so like in this case The Hub is actually it's a consensus layer anyway so you don't really need this yeah exactly I have a little bit of a question on this is actually I should have probably asked this earlier but like clients right now I'm only hearing it going one way but is there like a celestia-like client on each of the chains on the chains that are hooked up to it yeah so if you're a roll-up light node you also need your bonus Celestial light node okay and that's what that's a big reason why we care a lot about lightners because we don't want everyone to have to run a Celestial full note and follow every single other roll up even though they only care about one roll up or a few Roll-Ups ah but in terms of celestial

light nodes we actually uh recently just announced roll kit which is a modular roll-up framework and it used to be formally known as optiment enrollment but what role kit it does it's basically a modular roll-up framework that can be used as a replacement for tenement to launch new Cosmos chains so let's say you wanted to launch a new Cosmos chain but you don't want to bootstrap your own validate Network you want to launch your Cosmos chain as a roll up you can swap out tenement for wall kit and roll kit provides you a modular sequencing software for your rollup and they also naturally integrates

with Celestia as a DA and the celestial light client software but the kind of like bigger vision for bulkit is that it's supposed to be a kind of public good that's neutral from Celestia and that's why even though we started working on rocket from 2021 because at the time there was no General pepper scroll up software so we had to build it like if you wanted to realize the Celestia Vision we had to build something like rocket because at the time the only roll up so far that existed was you know software that was specific that was extremely like specific to that roll-up team like if it's like I don't know like zkc for example or optimism's old old stack and

so we and the whole the whole point of Celestia is that you can create your own chain the whole point of lecture is that you don't use someone else's roll up but you create your own roll up so that's why we created rocket but now more recently now there's other people creating general purpose roll-up software like like optimism's new Bedrock infrastructure for example so for that reason we decided to kind of like split Out roll kit into its own uh project that is neutral and independent from Celestia because because Celestia as it should be a neutral da layer to all Wallop software and on the flip side that also means that roll kit is neutral to what da layer it supports so you can actually add other the areas to bulkit not just

Celestia cool because we have like a like a modular da interface was it built together with sort of evm centric Roll-Ups as well so like is it easy to work with an evm as well as other systems it was kind of like built very specific to the ABCI interface okay that's tenement exposes so it's like very cosmocentric but that being said that's that's not the end Vision The Envision is that it should be able to support any kind of execution environment so like for example you we have um ether mint running on Market so you there is some EV you can get an evm chain running and see I see but that's also like that's not ideal either like ideally we wanted even more abstract than that like we want it possible to run like execution layer like Geth for

example without the cosmos SDK directly on Rocket hmm and I guess in that role kit going back to that earlier question do you actually have the celestia-like client or like a light client and when you say da I think you're saying data availability level layer right yeah so do you have that like client already kind of built into this role kit or to this like is it an SDK like what it's framework yeah we describe it as a framework okay uh like we were debating actually what should we what is it really is it SDK is a framework is it like no one knows really okay which is cool we just decided to settle on a framework template uh yeah template yeah I mean optimism calls bedrocker template or blueprint okay but roll yeah rocket doesn't embed this list

like client directly but it can interface with the API endpoint of a local selection that client so the idea is like you run a selected that client locally and then you run a roll kit like client locally and bulkit can communicate with that solicit like client via its API or RPC endpoint um and you can also actually add other the LA areas to block it but as I said it's not like it's it's supposed to be neutral from Celestia that you can add other the areas to to bulkit Via its interface like like you could add Bitcoin for example cool I want to kind of just explore even a bit deeper that like that two-way like client situation in the context of the namespaces so like if you have the Sovereign chain the like

line on Celestia The Sovereign chain itself is also running a like client of Celestia is it only pulling into that like client things that match the namespace that they've chosen like does that mean it's even like more sort of streamlined and you have none of the other junk yeah exactly so when you launch your own world kit roll up you define a namespace identifier for your roll up okay and then the role could that client will query uh what's called like the API endpoint on the celestial light node where the API what the API endpoint does is you record you can request some light node to get what's called get data by namespace for a specific block height so you can query block Heights for data for

this specific namespace and behind the scenes this less light node will connect to the peer-to-peer Network and we'll query query the nodes around it for only the data for your roller nice and it doesn't have to connect inferior or any kind of centralized service or anything like that this all happens in a decentralized way peer-to-peer and I know I keep asking this but does this mean that it that it does it faster or is it just like is the no is the actual no like client itself even smaller oh it depends it depends what you mean by Foster I wouldn't say it's necessarily faster than a centralized service it's not so much about the speed it's more about the fact that it's sporty-sexualized and more trust trust

minimized like you don't have to rely on their trusted third party to tell you what your data is you can just ask the network directly oh cool and that's actually kind of like one of my biggest like pet peeves about the blockchain ecosystem post ethereum like actually like people think black clients are really hard but Bitcoin has extremely good Lifeline support and I think that's something that people Miss like if you download a Bitcoin like client on Android the most popular Bitcoin wallet on Android is called Bitcoin wallet it has like five million installs and it doesn't connect to a centralized endpoint it actually connects directly to the Bitcoin Network yeah and it has a black client embedded and it works

beautifully that you can actually see the block headers on the on your phone and stuff and so on and so forth and that's better than like if you look at what's what's the most part what are the most popular ethereum wallets I'm guessing something like metamask which is I'm guessing always querying from inferior yeah yeah and to me you know that's that's like one of the one of my biggest pet peeves about the current blockchain ecosystem and there's also something that Moxie wild and Spike mentioned in his blog posts critiquing web3 um if you don't know Moxie wild and Spike is the creator of signal messaging app and that's something that we want to solve because like one of our I think one of the key values of blockchain network should be that users should be first class citizens of the network because if you're just querying a

centralized database then uh kind of like a road erodes at that value a little bit um yeah we'll try to dig that up and add that article or that post to the the show notes so I want to hear where Celestia is at today how close are you to launch what kind of projects maybe have already emerged in your community like I'm guessing most of them are sovereign chains or existing chains that want to use the data availability yeah so yeah it's been kind of like a wild past year in the past year like before a year ago there wasn't really a roll-up ecosystem outside of ethereum and we were really trying to bootstrap a kind of like a

a modular blockchaining system that was just that wasn't just ethereum Centric but I was also in like for example the cosmic ecosystem and over the past year there's been many many projects that have emerged in the kind of celestial ecosystem and also the broader modular ecosystem just to name a few as I mentioned earlier one of them is called dimension which I'm building Cosmos based Roll-Ups that can connect to their Cosmo cell meth layer another project is Sovereign Labs which are building Sovereign GK Roll-Ups so like you can using risk zero architecture cool they also support Celestia as a dealer and another project is called Eclipse which are building customizable Roll-Ups starting with supporting the Solana

virtual machine and allowing people to deploy Salama virtual machine Roll-Ups interesting of course this year as a database layer would these be in any way connected to the original Solana or is this like Solana mirrored but like using Celestia for da so it's not connected mainnet chain itself it's just using this llama vegetable machine so it's not it's not like a Solana Layer Two okay it's just like a salon and vegetable machine but can it interact like say you made a bridge would it make like any sort of application running over both of them easier to build uh I mean possibly I guess I don't know if I haven't really thought about that but I guess yeah I mean it could be possible to verify Islam of four proofs more easily yeah inside of if if you're using flower

as a virtual machine that could potentially read the case for sure and yeah so there's definitely a really widely expanding modular ecosystem one of the most interesting Trends is that we're seeing a massive emergence of sequencer as a service providers and more specifically like roll up as a service providers and what that means is that in the future and this was like kind of like my original like end vision of Celestia what I want to see in the modular stack is that in the future you'll just be able to as a developer Define the code for your application and go to a role of the service provider and deploy your application as a roll-up in one click and that said this provider will provide the first sequencer for you

and it'll be completely trust minimized because the whole point of Roll-Ups is that you don't have to trust a sequencer you remember like Amazon AWS managed blockchains this is like AWS managed blockchains but it actually makes sense and it's actually trust minimized why does the sequencer become decentralized though in what you just described like why would it be more why would it be more decentralized than an existing roll up because if the roll-up is constructed in this in the most secure way possible then the roll up should inherit liveness and sensitive resistance from from the data it's like if the sequencer goes down you should be able to force transactions to be included if the user submit them directly to the area or should be able

to elect a new sequencer for example ah interesting the most important point is that you don't have to trust the sequencer just to not steal your money because of the fraud proofs or ZK proofs or just running a full node if you what's called a pessimistic roll up yeah like people think pessimistic pull-ups are a joke like like I know like optimism did a jerk the April fools like about it but actually it's a thing uh yeah that's the basic roll up for a thing before optimistic and ZK Roll-Ups like tether is a pessimistic follow-up on bitcoin it's plasma a pessimistic roller no because yeah it's just a roll-up that doesn't have neither fraud or ZK proofs which means

that you have to run a full node that's what we mean by it's pessimistic because you have to run the transactions yourself basically to verify that roll up I see I say but um in terms of like launch uh yeah we plan to launch Celestia as a main net sometime around Q2 or Q3 of this year and we are launching um an incentivized test net next month and it's gonna be like the biggest and I guess the first and only pretty much tests of data availability sampling like clients on a real Network uh we're gonna have around 900 like clients that are doing data availability

sampling so it's going to be like the larger scale test of a data availability layer with trust minimize like clients very exciting I know that the ZK validator is going to try to be in that we don't know over the future but I know it's going to be a real competitive space but we're definitely going to at least try it out we didn't actually get a chance to talk about the data availability sampling maybe we should actually for a minute can you share a little bit about what that is sure so data availability sampling is basically in a nutshell it's a technique that allows a light client to have a very high guarantee like almost 100 guarantee that the entire block was made available and was

published by only sampling a very small portion of that block and the probability is that if the samples come back as accurate then your assumption is the rest of the node is also accurate but is it the validator the validators are doing that right well it's the end clients like the light nodes are doing that okay so it's the light nodes that are doing that are they all sampling differently like is it sort of like every single like client's gonna sample a small subset all at the same time but always different therefore like a lot of ground is actually covered yeah exactly so the idea is that each light node samples a random the different random set of that block and the idea is that if you have

enough light in those in the network and the any data is withheld or then the light nodes can collectively reconstruct that block by sharing examples with each other cool so it's kind of like a self-gielding network in a sense is it the whole thing then like so all of those so by the way you keep saying light nodes I say like clients is it okay to use those interchangeably yes okay okay I'm actually meant to be using light notes okay because like they might not just be clients they might actually be like sequences so yeah okay okay all these that from like from now on but so if you were to add up every single sample that all of the light notes have done it fully covers the whole thing is there overlap though like would you also have a little bit of

overlap between them are they like always sampling different things from one another yeah you can't have some overlap because the clients or as I say the nodes they don't communicate with each other about what they're gonna sample so if there will be some overlap but the general idea is like probabilistically speaking once you hit a minimum threshold of light nodes then you know that for sure do you have enough light notes that they've they collectively have the whole block but why would you not overlap though well you do have an app oh you do you do have them sampling the same pieces for example like you'd have okay okay yeah you do have that because they don't share with each other they think they don't coordinate with each

other uh yeah about what they sampled so they just they locally choose a random number or a random set of chunks in that block interesting so so you might end up with overlap cool so I think we've covered like the community as it is right now but looking forward do you see sort of like in your ecosystem would there be applications specifically or do you see those always as existing only on The Sovereign chains or on the yeah the Roll-Ups that live on top of Celestia yeah so I mean Celestia is kind of a weird R1 in the sense that what kind of like more B2B than b2c okay and by btb I mean blockchain to blockchain yeah yeah so kind of like in many in many segment

stances like where one step removed from the end developer because last year as a d a is usually used in conjunction with the execution environment so it's like the execution environment is where applications live got it and so that's like where roll kits what role kit does so Volcan is more b2c because rocket directly interacts with the actual application Developers and that's why we needed we needed to build wall kit because it was necessary to achieve the full vision but so this year as a DA I think in many cases well like ideally it should be abstracted away because like it should be like magic and they like if you're application developer as there was a quote about this on Twitter or someone said it but like Jimmy Hendrix

does not need to understand how to build a guitar to play the guitar right so it's kind of the same thing like Roll Up application developers should not have to be protocol Developers to to create the application the end goal is to be like magic like you should just develop your own application and have it deployed without understanding how the lower ends of the stack work and like part of the goal is like we want it to be possible to make it that deploying a new blockchain is as easy as deploying a new smart contracts very cool well on that note uh I want to say a big thank you Mustafa for coming on the show and giving us an update on Celestia kind of revisiting it and also letting me ask a lot of these open questions that you

can see I understood parts of this stack but not exactly so yeah thank you so much for walking through it with me thanks for having me so I want to say a big thank you to the podcast team Henrik Rachel and Tanya and to our listeners thanks for listening [Music]