Building Modular Move Networks on Ethereum | Nick White, Rushi Manche & Avery Ching

Transcript

[Music] [Music] okay welcome everyone uh we have a fascinating uh and one of our first live stream um pods here uh and we're joined by Avery rousi and Nick from Aptos

movement and Celestia um and I think if you if you've been still in crypto haven't pivoted to AI or haven't left I think you've heard the word modular before by now hope you have and so this is going to be a fascinating discussion because there's a lot of intersection happening between between these ecosystems um and so welcome everyone uh why don't we get started with quick intros Avery hi folks I'm Avery Chen uh I was at meta for 10 years before starting up co-founding opt house Labs with moake and before that I have a PhD and high performance Computing U I'm the CTO currently at opt Labs great to meet you all rushi I'm rushi co-founder movement Labs background is engineering um system

based and databased security at United Health Group um and then got into early the engineering and Aptos engineering and here are my movement I'm Nick I'm the COO at Celestia labs and we're building the first modular data availability layer and essentially you can think of Celestia as this um new type of layer one that is built specifically to scale l2s and rollups and and enable more flexibility for developers uh um that's great well welcome everyone again um I'm curious uh if any one of you guys can just broadly simplistically kind of explain what modular blockchains do what they uniquely enable kind of what the problem

is that you guys are solving yeah I can take that um so modular blockchains are you know contrasted with monolithic blockchains and most uh existing sort of l1s since the birth of blockchains with Bitcoin and ethereum have been designed in a monolithic uh sort of architecture where essentially the same set of nodes do all the tasks that a blockchain needs to do to uh provide the infrastructure to run applications and so those tasks are consensus um data availability and and execution and when you combine all those things into one monolithic stack it kind of limits it kind of imposes some limitations on scalability as well as

flexibility so the modular blockchain thesis is that you can actually decompose these functions into separate protocols that can then be stacked on top of each other and be combined and when you do that you can get better scalability you can get better flexibility because each one of those layers can be you know specialized to perform that function really really effectively and so you know there's different layers of the stack you know Celestia being sort of like a consensus and data availability layer but then you have things like movement who are building execution layers that run on top of things like Celestia yeah excellent so rushi maybe uh that's a good segue to talk about like what movement actually is um and how you're working with like Aptos and Celestial yeah so movement is building

the first ethereum layer 2 powered by the move virt machine it leverages Celestia for data availability um I think in the past we've been using evm for past few years and most developers and ecosystem can be like this is horrible um I'm getting hacked for $4 billion dollars every year um and now what we're seeing with the modu thesis and what modularity means to me is the ability to truly build whatever choose whatever execution layer you want to use choose whatever da you want to use um it's use case specific blockchains so whether you're building a blockchain for gaming and you paralyze throughput low gas fees or you're building a def5 specific one and you need smart contract security so kyber swap attack doesn't happen again the curve attack doesn't happen again um and I think we're focused more on Smart contract security

scalability coming to ethereum with low gas fees powered by Celestia got it and and I guess you're using the move language which obviously you guys at Aptos um have taken what was I guess the initial team of Facebook then spun out and then you you guys took that and can you just explain the novelty of Aptos and and movement as a programming language relative to solidity um because you know for context I would say the line share of developers still are working with solidity and exist in the ethereum ecosystem um but you guys obviously came out and uh with move and and um so manyways if you could explain that but

yeah thanks for thanks for that segue um back in 2018 when when meta started looking at what it would mean to build out a scalable payments Network that could support uh for for you know billions of people around the world we had to kind of you know we were definitely inspired by what ethereum had done what it had done in the space um in terms of what was possible but we realized that technology would not scale to the use cases that we wanted to design for and so that meant a new blockchain architecture which was DM a new program language which was move and the idea behind move was how do we kind of get from ideation to production as quickly and safely as possible and being very inspired by rust around pushing a lot of the bugs from the runtime into

compile time earlier on the programing stack and then thinking about where the Primitive is necessary to make sure money movement is as safe as possible so can be dropped on the floor can be accidentally M out thin air all those things uh as built in you know ways that program can't hurt themselves so just a safer way to get smart contract language development into the hands of programmers now I think it's worth noting that many very few smart contract work at that time we were writing for very specific use cases so things like travel rule um governance were something early or Primitives that we worked on it uh in the early days now fast forwarding a little bit you know apas we wanted to definitely take this forward to massive degree and so a whole bunch of the team has has come with us to make that happen

and mo has taken on adap has taken on a massive uh pivot kind of from its original Vision in terms of how do we take on especially parallelism uh from from uh programmers uh to to extract it automatically so uh with BLM technology I think which is being used by many other you know protocols in the space and being inspired by that we kind of call that implicit parallelism so when a programmer starts to design their contract whenever there's par is impossible it'll be extracted automatically uh from from the program without having to impose any necessary boundaries on them them also when you think about programs they naturally touch different pieces of of shared data and being allowed that shared data to be kind of in the same program as opposed to being split across programs to extract parallelism which is what

happens with another a lot of other program par paradigms uh produces a lot of challenges and so I think you know that's some of the core things that we have really been taking forward in the language also formal specification and verification is very important to us so the move prover code has been an amazing asset uh for making sure programs are safe uh running in the in in the move space um the entire framework at up TOS is actually formally verified which is really really incredible uh as a as a particular tool and so you know there are other move variants out there just again being very transparent uh there's the original kind of DM move that was there there's the S move uh and I think epas is kind of the most fully featured superet of these these different moves out there and the most supported uh and

the easiest for integration into different platforms and I think that's kind of why we're excited about movement being one of the many different kind of providers that are taking move into different ecosystems going forward into the future yeah and so maybe if you can rousi or Avery if you want to take this but unpack like from a developer experience um how how is it um if if you're trying to convince someone that perhaps is coming new to crypto wants to build an application how would you frame it to them it's here's why you should build on on move here's why you should uh utilize movement um and how would you kind of frame it and Pitch it to them because the natural perhaps inclination can still maybe I want to build where where the users are and the liquidity is

today which in my opinion is sort of like we're all beta testers and it's not really like we need to think about at scale like where are most users going to exist and build towards that but that's just my bias so if you guys could just comment on that I think analogy I use is move is to rust as react is to JavaScript in that it's a very specialized version of programming Russ has been around for years it's a standard now in multiple Industries I believe Microsoft is Now using for AI biotech is huge it's clear that Russ is like a very hot programming language that's very powerful um and what Avery and a few other great guys really did and girls um is made move the program like rust made move the blockchain programming language powered by rust

bace provs rust bace mechanisms um and the way we kind of approach it is like Aver and app doing a phenomenal job innovating the move language some of the best engineering work I've ever witnessed we're approaching it as like a very web Focus angle we acknowledge that in the short to midterm solity is still going have mind share it's going be difficult to get eum adoption away um especially when liquidity lives on the evm all the developers know solidity in a bare Market if you're going to developer and say hey learn this new language it's probably going to be a very tough battle to win um which is why we kind of explore this idea of a transpower which allows any evm code to launch saying solidity code curve contracts a contracts Unis swap contracts can seemlessly deploy on rdk it'll transpile down m b code and

launches the VM itself so we kind of hedge our bets in the short term where we're saying okay we believe in move as thesis we believe the future of programming is move but in the short term we acknowledge that city is a line share of developers so we allow seamless deployments to ethereum Json RPC metam Mass compatibility hard hat compatibility that all comes out of the box and any city code can now run the great apas dmvm um or move VM um and live on ethereum so you get the liquidity of ethereum you get the benefits of evm tooling and devel um while still harnessing the benefits of appas and move so this is really um this bridge is so if you're a these are if you're a project that has already built on Aptos or or perhaps Su some

other you allow it just to seamlessly like deploy and run an evm instance on movement and like attract all the users and the liquidity that may exist is the end goal to ultimately just like bring it all back um to something like Aptos where once people realize that it's just better I'm curious how you think about like because the analogy once you see it like you don't unsee it and so once you start interacting like in a move base like for the user it's like well you don't have a a re-entrancy attack and so def5 becomes a safer environment and so how do you think about like that user flow um you know perhaps in the next you know one three five years of the state

of Aptos movement relative to ethere and the growth of like liquidity users um I I can take this one I mean I think the goal would be that as it's a technology argument right you brought up a lot of advantages of what evm has today definitely users uh developers but as what we've seen and observed from people building on on move with Aptos is that people who start running this code just love it they don't want to go back um and we've seen Flagship projects uh you know from the even world come over to Aptos uh using move and being really really excited about that whether it's sushi swap or it's live live pier and many others actually coming in the you know haven't announced yet but are be moving over know pun intended uh in the

coming months and um you know as they as people start to see like again the the idea from ideation to production as safely as possible and these R attacks are not possible you have the additional like um safeness of formal verification in the place and also you have this very rich way to get parallelism kind of both optimistic as well as pessimistically coming in the future and then seeing the way that we evolve the the language the V compiler and the VM we have a very long road map of like taking to the next level I think that is going to be the reason why people start to use these different networks that are move compatible so whether it's gonna be movement uh chains or you know what What's happen top of Celestia we want move to be the the interface into how

all the smart contract program you know happens in the future Nick I want to go to you and rushi as well commenting on how important was it to like this data availability um like unbundling for you guys where you can now use Celestia from a cost perspective like what does it mean to be able to utilize Celestia versus do da and consensus inhouse and I'm curious if you could put a finer point on on the benefits um of of using something like Celestial yeah I'll I'll start off and then I'd love for rushi to jump in too um so there's a lot of advantages to building on a modular stack and a modular Foundation like like Celestia so

one of the key things right if we go back to modularity is decoupling execution from consensus and data availability so what that means is like you know when you launch a normal monolithic chain it comes with the execution environment already defined in it so you as a developer when you're building on that L1 are stuck with that execution environment so it could be evm it could be Salon MVM or it could be move right and um kind of like Celestia's thesis or the modular thesis it's captured in in the phrase that rousi uh said earlier is is build whatever which is that uh we want to remove constraints from from developers and and enable people to to choose whatever things best suit their use case right and and we firmly believe that

there's there is no one you know stack to rule them all like this like monolithic architectures tend to breed sort of maximalism right because you have to say that like well you have to justify that you're the best you're better than everyone else right and in in the modular world it's like well there isn't actually a best you know stack and instead it's like well what are you trying to build and then from there move work backwards and say like okay well in that case you know evm might be the best for you or maybe it's Cosmos SDK or maybe it's move and so first and foremost I think that's one of the benefits of building on something like like Celestia is that it it removes any like lockin I guess into our particular ecosystem or stack um and so

one of the things that we're really excited about is supporting all these different execution environments and and we're and and we like all of them it's like a pluralistic view right so we recently there's also eclipse is launching a salon of virtual machine roll up on top of Celestia um and now with movement like it's another offering that that can kind of be be built on Celestia and like long term the way we see things playing out is that it's it will be as easy for develop to deploy a new execution layer a new blockchain as it is right now to deploy a smart contract so like in the long term um it will be this uh really easy outof thebox experience to just like spin up a move VM chain and um be off to the racist building whatever applications you want

and that's that's like the future that we want to get to um but rousi why don't you add your perspective yeah I think Avery touched on this earlier but one of the benefits of move is a parel runtime in that you you have local fee markets um I've been using arbitrum for a while now and every time that GMX goes crazy the gas fees go through the roof um so I think one of the benefits of nextg VMS whether it's slant VM move VM um fuel guys are doing a great job with sway um is really allowing gas fees to be controlled and minimized um so when we're looking at architecturally um building on ethereum sure ethereum is great um settlement layer you have a lot of liquidity um it's great for M share and just generally good for defi but if your gas fees are going be the roof it

creates a really poor user experience so when we're looking at different da layers and studying the modularity thesis what struck out to me is look if I'm going to be bringing this next gen VM to ethereum it can't use um ethereum's main da as his da layer until Pro dang sharting which is years out um it needs a different solution so that's why the Celestia research the Celestia um integration made a lot of sense because now we can essentially bring an ethereum scaling solution bring a better nextg VM to ethereum sement and not have to worry about da being a bottom instead it can be a fraction of the cost basically mimic um some of the cheapest if not the cheapest rollup on ethereum so that enables a lot of use cases for defi where we kind of abstract away um any gas spikes but also on the consumer

gaming angle right we work with a lot of gaming clients that can't afford $10 per transaction be minting millions of nfts killing a thousand zombies um that would probably run the player like out of all their money especially if it's like a 14-year-old kid um so having alternative D layers like Celestia make a lot of sense to us yeah um for me the some of the like the big aha moment um and I think a lot of the market is kind of coming there's realization is um like like there are incredible benefits to this nextg kind of movement like like building an optos um like parall execution and like Dynamic fee markets like just make a lot of sense when you start thinking about all the different use cases like um that are now uniquely powered by that and and

I think um you know I want to get into discussion around modular integrated however you want to use it versus um um sorry modular versus integrated monolithic um and and like particularly in the context of like why do you still like what do you think is the benefit to monolithic blockchains today like because one could argue well there's all the incredible benefit with um modular and modularity makes a lot of sense like the software like like you know pre bubble crash 99 and then software to service like I I kind of liken it to that like there's this huge unbundling and in technology movements the unbundling like creates a lot of new

possibilities but I think we're seeing that with blockchains execution environments and so I'm curious like how you guys think about that debate of modular versus monolithic and and if there's merits to still looking at a monolithic integrated blockchain saying yeah I want to build there versus the modular approach yeah I'd be I'd be happy to take this one um you know first of all definitely respect the work that you know those in the modular space like Nick are doing I think it's an interesting design exploration you know for me uh you know a couple things like you know I think at two examples I'll pick on one is I think Linux colel is a great example of a monolithic Kel where Linn tvold had a very strong opinion

that that monolithic kernel would actually allow them to move faster and be more performant and I think it's done extremely well in terms of you know achiev those goals and being deployed to you know pretty much you know all major clusters and Cloud centers today compared to um modular modular kernels um or mic kernels and uh I think another example was you know even from our early days within within meta when we started to build out the Libra blockchain you know we actually tried to build it as a microsurfaces architecture to some degree all connected through grpc and we quickly realized that like you know this wasn't a good approach due to like it was super inefficient um you know in theory these things all sound great like I can deploy this component here this

component here you know and uh you know we end up rewriting the whole thing as instead as you know just more like internal apis and the whole system got much more efficient much faster to develop I think it comes down to like you know a thesis around like do you if you believe strongly you know the you have a strong like the right path forward you have the right interfaces like the language is the way it should be the way that the system interfaces with it is is kind of you know built really well for it or if you feel like there's a lot of like use cases that are not going to be covered by these particular components then sure uh adding all the optionality makes sense and it doesn't mean you can't be more modular in the future when it comes to apis interfaces you can always add those

in as you kind of discover the right points for those but today we haven't had any concerns where people are like well you know we love move for this but we really wish we had evm for that or we love you know the way the block works today but we wish the consensus protocal was slightly different um so far we have you know every use case that we've seen being thrown at optos has worked really really well and so there's just no no need to to kind of explore you know the Alternatives right at this second that could change of course in the future and the other thing I would just kind of point point a point on modularity is upos is a modular stack we're kind of designed as several phases we have data dissemination we have ordering we have par execution parallel rights and then

kind of uh proof verification after that and so like you have the modularity as a design choice and we can actually take a lot of benefit from like upgrading individual components of that and testing individual components but you know one of the downsides you know if you start to also get two two modular uh in certain services for example that testing becomes very complex like suppose you want to test interaction of you know chain a with chain B and chain a is changing this thing and chain B is changing that thing you know if you want to test holistically it gets very challenging like the number of combinations explodes exponentially and that's something we just we don't again that we don't think that right now we haven't seen use cases for that uh our developers are are really happy with what we offer them and uh you know

building that full stack is is just simpler and more performant for us and optimizing for or currently today yeah we don't have someone uh from salana here but perhaps um I I may take that um I'll try to I'll attempt to do that but the argument would be look salana just makes a very clear um trade off that they're just going to accept higher Hardware costs and requirements um and try to optimize the bandwidth like because when I think of like blockchains like you have the hardware costs you have bandwidth and storage and those are in my mind the three biggest categories that ethereum guys say well our Hardware costs are not 3,000 they're 200 and anyone can run a

node and that's great when you actually look at the data they're actually the same when you look at the Nakamoto coefficients salana actually has way higher than salana so in practice like these things also manifest themselves in different ways but someone in my salana say well why don't just come and build in salana like why do I have to go to optos or why do I have to use like a modular design when you know you know like you still need to run like you still depend on full nodes and running a full node has a real meaningful cost and if you just accept that then we can just focus on solving like Hardware cost can come down over time or solving on bandwidth or optimizing these things but I'm just kind of curious like taking the

camp of you could build in salana like what would you say to a builder there's a lot of excitement of course now in salana for variety of reasons but what is that tradeoff between you know Aptos or you know movement yeah yeah um I hav to answer that question about salana I mean I think you know salana is definitely one of the first in the space around high performance uh blockchains you know obviously we we're in the same field we've taken different design trade-offs we think our design is is much more scalable from a thers perspective and there's different goals around latency right so I think they talk a lot about the block times but you know finality time is actually quite different than block time and so you know finality time in apps is is generally about a second or so and that's kind of where uh all

valers agree on on the uh the the execution of the transaction uh from a Byzantine F to perspective two-thirds uh so I think from a throughput you know and and a latency perspective you know you can try app test you can see the numbers we put out we have put out verifiable numbers even in a benchmark in March for anyone can rerun the same experiment we ran and get the same numbers of I think our experiment showed about 20,000 transactions per second in uh a main like setup end to end results for that and we also have new numbers that'll be coming out fairly soon that'll even blow those numbers away uh and so again our goal is to keep pushing on the on on the spectrum of parallelism uh latency and other aspects and yes so if we're trying we're not trying to

support all the use cases out there it's really about High through put low latency highly safe and available use cases and kind of Enterprise grade technology that you can trust to build on for billions of people yeah Nick or rousi I'm curious when when you think about the projects that have become interested in working with um Celestia and also using move as a programming language what kind of use cases and projects um have you g been getting the most amount of interest from uh yeah well there's a lot to talk about in the monolithic verse modular debate which maybe we can revisit but in terms of uh adop keep going you know make the transition I think it's super

it's probably the most important discussion that we can have like in crypto today realistically yeah well I think I mean okay so first of all it's it's clear I think from from everyone that we're entering into like a multi-chain world right like even if uh each of those chains is is monolithic we're ENT a world where there's going to be lots and lots of different experimentation of execution environments and a lot of different chains kind of living side by side with their own ecosystems and I think when you like start to play that out further the modular approach makes more and more sense in the sense that you can uh sort of still share security via having this common consensus and

data availability layer while still experimenting with all these different chains and execution environments like all the optimizations that are happening in the move world or the salon virtual machine world that speed things up and make them faster Etc like are absolutely really crucial but they they're they're only really scaling like the uh execution aspect and not so much specifically the other major bottleneck to scaling blockchains was which is data availability and I think and so I think that's that's one of the the the key differences and like um I would say another thing that's that's really important at least in the modular space that I think it doesn't kind of gets lost in this debate is is the emphasis on verifiability too right so like

increasing the throughput of your chain simply by requiring higher bandwidth and higher um like execution like sort of like CPU or or whatever like computational resources in the node is not really scaling the blockchain because you're just you're increasing the cost to run a node and making it less and less accessible for the average person so the only solutions that I think actually scale blockchains is a combination of data availability sampling and rollups and what they allow you to do is they allow someone to verify the chain very efficiently with low resources so even as the amount of data in the block grows

you don't have to do more work to verify the the chain even as the number of transactions that are processed grows you don't have to do more work to verify that execution and so what that means is we can preserve the core value proposition of blockchains which is end users can verify the chain and audited and and uphold the rules of the chain and not have to delegate trust to validators or other people running full nodes on their behalf and so I think that's a really core thing that like and I think that that we need we need to like keep that very Central because it's it's the core value proposition of of blockchains and I think the modular sack kind of stands for that obviously a monolithic chain can Implement data availability sampling

and can Implement these proof systems that that are like rollups um but then when you enter in that world it kind of it modularity makes more and more sense because data availability makes sense to pool pool resources because you get all these benefits of scale and rollups naturally lend themselves to these like just oh spin up your own chain so like I to me all Ro all roads kind of lead to to modularity one way or another can you go a little bit deeper into this specialization and like by specialization you have all these advantages like what what is uniquely enabled when you kind of Outsource it to Celestia when you Outsource da and consensus to Celestia like how does this actually like scale

and yeah so so Celestia is the first L1 that was designed for people to just only run l2s and so you could for example R run a roll up on Solo or ethereum or all these things but it ends up being very hacky because those l1s were never actually designed specifically for that use case right so like even the way the rollups run on ethereum today is by using call data which is I don't think it was like ever planned that like oh people are G to want to run these rollups and then therefore we'll Design This call data interface as a way that's that's optimal for that versus Celestia was like designed from the ground up for this use case so there's there's a bunch of

things first is that we scale specifically data availability via data availability sampling which means that we can scale the block size with the number of sampling nodes in the network and so no other blockchain has this property where you can increase the throughput without increasing the node requirements it's just you just need more people participating and I think that's that's a huge unlock and that enables us for example to have these extremely low fees for for l2s and why you would launch for example movement is launching using Celestia rather than ethereum da for example um in addition to that there are things just a put it in perspective that's a order magnitude save uh probably multiple orders of magn

MTI compar to EA yeah but like even compared to a like if you're a normal L2 launching on the op stack whatever and you're not using Celestia and if you use Celestia like that cost save translates into one two like one or couple orders of magnitude I would I think we're we're still working on exact numbers and we'll have that more concretely when some of the first rollups launch on Celestia but I think it'll be multiple orders of magnitude and so because the the the main cost driver for rollups especially optimistic Roll-Ups is not the settlement cost but the da costs and so when you can offload the da cost to Celestia It Can it dramatically reduces

the the end user costs but anyway just back on the specialization Point some other things that celesia were designed for is like the um is overhead minimized meaning that so when you're running a rollup on top of let's say ethereum or any other L1 you have to verify what's actually happening on the L1 too so you kind of have to you have to run like a full node of that L1 and if there's a lot of execution and state and things to track on that L1 then that adds overhead to every single roll up that's running on top so Celestia is designed in a way where it has basically no execution has a minimal minimal amount of execution and minimal minimal amount of state and and so like verifying Celestia is as light as possible so all the rollups that are

built on top don't get don't inherit all this like baggage and overhead that you would for example if you're running on on ethereum or other like monolithic l1s um so and then also like there's things about it's designed specifically to post data uh to it so like the the interface for posting data is very clean and the way you pay for that is all like optimized for that use case and and that's just something that a monolithic L1 can't do or could but it has to like completely start from scratch yeah can we comment on the like around this like throughput and you know I'm not a TPS maximalist but yeah like if we just compare TPS as a figure of like all these different use cases that

can be enabled right if you're going to Visa well they care about TPS you know I guess of course other components but one could argue like and I'm referencing some figures here but like salana since with fire dancer has I would say like probably the highest DPS um and then you go down from there um like with Celestia like what are the improvements I guess I'm curious like movement like where where does zto stand in term in that Spectrum relative to ethereum relative to salana and where does like a movement VM with Celestia stand in that spectrum and I'm not saying TPS is the most important thing but I I do think it it's important to just like call it out

I guess yeah I think I think by the way TPS is a metric that has to go away at some point I just want to be very clear like it's a really poor way to like that's Avery's fine way of saying it was a very stupid question but anyway no no no no no I I I think by the way it's a very fair question everyone asks us all the time what is the transaction through putut of your system the issue with TPS is that it really depends on the transaction types that are being issued right and um you know that's why databases have TBC benchmarks so you can actually test out a suite of different types of operations and understand for this type of work this is the performance you're going to get for this type of workload this is the performance you're going to get and in our world

unfortunately there just isn't standards out there you know we've done our best to try to put out some standards again we put out repeatable benchmarks we'd love others to do the same in the space so you know we have a medium post or describes like here's our exact setup for how we ran our experiment here's the exact types of transactions we've run we have coming again very very soon improving upon that where we'll describe different types of workloads that we think are important uh for different kind of crypto use cases and we'd love others to go ahead and participate and share you know what they think is important and also ideally adopt some of those benchmarks that we put out there so we can have and Apples to Apples to Apples comparison instead of apples to oranges to bananas to pears to whatever uh you know jack fruit you

know thing that's out there and so you know I think you know fire dancer you know is probably interesting product it is something though that is kind of still designed within the current Salon of parameters it's just a faster implementation of it and so you know today I don't know we've actually seen any verifiable numbers from what salonic can do in a act controlled maintenance type environment and so I I think you know these are not apples apps numbers that we compare and the only numbers that we know of like you know that kind of show benchmarks on the space are the ones that we put out there so again like as of now we think that we've shown the fastest numbers in transaction throughput they are verifiable they're repeatable you can run them on the same gcp Hardware you'll actually get better numbers today we haven't shared those

yet we will share those in the coming weeks um but you'll get you know the same kind of results that we have at this particular time and I think they're the only ones that are verified for now yeah I guess the better way to reframe the question for for all of you guys is like when you think about scalability like what is exciting now is that there are many different use cases that have been tried in the past didn't work we're in a different environment now because you know with Celestial with different environment because of optos and then with the movement um VM and so I am curious um when you think about like scalability is something that to your point Avery is heavily discussed it can mean a whole variety of things um it may be the most important thing for C like but um it just depends on the use

case and and again there's tradeoffs and whatnot so it's a very Nuance discussion and I don't think we have enough time to do it but um I do think the more exciting thing maybe transitioning a bit into the discussion is like um rushi like as you guys have set out to to do this kind of connectivity between ethereum and those and I'm curious like what type of use cases do you envision what type of projects um do you envision building um and utilizing kind of this um bridge for a lack of a better term or just environment that kind of gets the best of both worlds I think was that what gets lost in like the infrastructure discussion and like the whole modu mon discussion is like what do the builders actually think right um

so me and my co-founder were just D Builders as a DEX Builder appos my co-founder build a first yield agapos um so we're really looking at it from like how can we build better Financial products for our customers um it didn't really matter to us what the TPS was because if we got five TPS we were throwing a party um it didn't really matter much on like what latency was at that point because we were just trying to figure out like how do we get 10 users on our app I think when what kind of led us to the modular thesis is that we could permissionless innovate if that makes sense um for example when we were building yield arator there was only three assets that we had at the time on testnet it was like AP native token and like two live tokens and that was on

testnet so we're trying to get customers to use our app it was like do you guys really want to play around with three tokens that like not that many people know about or do you want to use native eth on your protocol do you want to use btcb um which is powered by a snowm sensus that we leverage do you want to use evm assets which are usually more speculated upon so you're taking like a defi app to meain that um think about DX pers lending It's usually the evm assets that are usually speculated upon that drive demand to your protocol so that was the bottom neck that we face as builders in the space which is often forgotten in these discussions and I really haven't seen it come up like cool latency is cool cool like the fees are like cool but most of that's in theory

what we're looking at infrastructure today is I want to build the decks in the next six months if I'm going to monic chains right now especially ones that are starting up on the come up usually these assets aren't speculated upon heavily and it takes time for development to occur it takes time for stable coins to come it takes time for bridging to happen um whereas when we went to a modu chain on top of ethereum with Celestia we built a hyper bridge in a month right so now we have full connectivity with the rest of mod ecosystem with other evm chains and we didn't have to wait for a third party like later zero to warhole to prioritize us instead we could do it ourselves um that ties in IBC we were able to fully bring IBC to move which is pretty

exciting um and also like assets right so when we have defi apps that build on top of a chain now you have apos andu assets which are move assets we can also have native evm assets so native eth in your protocol BCB in your protocol um Lio your protocol native ethereum assets which can actually drive adoption for these Mo protocols um and make my life as a builder a lot easier um but I'm sure Nick has more thoughts on like modularity that sense well I I think you hit the nail and the head in terms of like permissionless Innovation I that's a big part of the modular thesis is this of like just set set the developers and the builders free you know and and give them as many options as they uh could want

and then let them build build whatever essentially and I think um it's really I'm glad to hear that that's like one of the things that that it does end up mattering to to builders like like rushi and um I I think uh just on the on the throughput uh conversation earlier uh I I totally agree I think transaction per second is is like a a bad metric for all the reasons that Avery mentioned and at least for for us and what I think will eventually become like kind of the standard at least for the the base layer like data availability stack is just simply data availability throughput uh as a metric and that is universal because you don't have these things about like oh well what kind of

transactions are there it's just like how many bytes can you publish and make available on this L1 and I think that's something that um it's it's great it's another benefit of this modular Paradigm is that it really clarifies it's like okay we know what the objective of this protocol is like what are what are the some of the things we need to maximize right and one of them is just data availability throughput now with the constraint of it must be actually easy to verify and all those other things that I mentioned earlier um but I think um that that that to me is a very clear standard I think will emerge is like how much what's the data throughput of your chain I B thing um then a can go is why it drew me to

move is the concept of velocity production I think I was an evm Builder before that um and for me to build a SL dap as like Dex or yield dagger it took me months and have spent another few months auditing the code and then pay 100 Grand to an auditor who a of the code and then it still probably get hacked right um so that was a whole headache for me and when I Su across move and the work that Aver done it's like okay so like 90% of the hacker attacks that was spending like countless hours at 3:00 a.m. like are kind of taking care of for me I don't have to pay this auditor 100 Grand to not that cre of a job and still get hacked um so we focus on velocity production if you're R developer think of Mark Zuckerberg in his college dorm you

should be able to ship maina code as quickly as possible not worrying about security instead worrying about actual functionality build the best social f app don't worry about that 108th line of the code that you glossed over that can lost like lose your protocol 3 million and lose all your credibility um that's the biggest issue in development as a young and aspire ing like being infrastructure application developer it is too difficult to build the killer app or making the killer app easier with move yeah I can agree more with that and I and I don't get me wrong I really love the fact that like movement and and Celestia are partnering together make this happen I know Nick you're putting your Hedges your bets on all these different you know ways of languages whether it's be Salon ofm or evm but you know I think hopefully the right stack

will emerge over time and maybe maybe they'll be less focused on the ones that are not as interesting to your developers um as plays out yeah I'm curious like just looking ahead like the synergies between the teams and also just how do you think about you know all this is fairly new um how do you envision the next I don't want to call it the next cycle but just like the next two three years um in terms of just developer interest activity type of applications um with all this kind of new possibilities out there I quickly I think the evm burns um I think every is sick of the evm like there's literally a major hacker attack every year every month um it's the same reentry bug over

and over and over again it's like getting punched by the same kid over and over again and you're not doing anything about it um so hopefully next two or three years um we see more adoption get the blue chip D5 protocols actually on move some other networks just anything besides the evm and I'm happy I I completely agree with that I mean I think the technology space is going to evolve rapidly uh like I said you know we were doing 20,000 transactions in a verified environment uh that anyone can rerun in in March that number is going to increase significantly in the coming weeks as we put those numbers out and in next year those numbers will increase even further and so as we see the technology start to grow to really rival those of traditional databases or other systems

that are used for money movement uh it's going to be a pretty massive technology advantage to move more into the blockchain space and I you know our we're taking a very opinionate approach that that move is our version of move uh is going to be the right the test version of move is be the one that drives this Innovation forward with new compiler coming out with new VM technology coming out with folks like rousi and others helping us to derive this from a developer standpoint going forward uh as well as the system you know being design co-designed with it like Hardware softare design is something really important we think about full optimization across the stack and you know as Nick pointed out these metrics around like whether it's going to be data movement or like I like to think about resource usage uh I worked

on distribut schuling for a long time across 100,000 machines how do we measure the real like performance usage and benefit from these different infrastructures so become very very clear and uh massive massive adoption in in the coming years because these big Enterprises and other large applications will be very confident to put their their use cases on top of blockchain because it's it's truly scalable it's truly reliable it's easy to build on uh and the user pain points around how the user experience of onboarding is going to be much much simpler when you have things like a com abstraction built directly into things like optos move as well as impure movement what they were doing as well yeah I agree that the evm is not the Beall and endall I think a lot of

people over index on that I think it just had the first mover Advantage but there's just you know part of the modular thesis again is experimentation and especially experimentation at the execution layer has been held back by the fact that to try out like a new execution environment you had to launch a whole new L1 and now when we remove that barrier we're going to uh accelerate the velocity of innovation at the execution layer and um even that even includes like new flavors of the evm like we have teams like Manta Pacific who have built an evm that has like ZK op codes inside of it um and so I'm really excited for that like just unleashing a whole new uh wave of innovation at the execution layer itself and I think in the future what I see is

um just you know right now L2 be has something like 30 rollups or 2's listed on it I think in 2024 into 2025 we're going to have thousands maybe tens of thousands of of of rollups eventually it's just going to be like smart contracts is going to be the same thing as we had kind of in the wave when ethereum first shipped of all these people just like deploying new apps um I think we're going to see something similar and what I'm and and it's going to also have this explosion of all these different combinations of different components modular components that you can mix and match and uh you know achieve different trade-offs and so uh I'm just exited for experimentation because wherever there's experimentation there's Innovation and there's going to be a net new applications and use cases that haven't existed before so the whole

Space can move forward and um I'm just stoked to see what what comes out of it I can't predict what what will be the breakout use cases but I'm sure that we'll find some yeah I don't think anyone envisioned Uber when the smartphone came out but um I am um as a general point of just curiosity for me is in this decoupling unbundling between you know execution settlement and da consensus and kind of like this Rubik's Cube analogy like where are we in the state of like integration when you do that um you know you you talk about IBC like connecting all these different chains but like can one of you guys just simplistically comment on like how these like when you do this un bundling are you losing some sort of communication or performance by

doing this un bundling um or not and I'm just curious I think I'd be the first to say that like it's going to you know integrated chains have an advantage in the sense of you know if you already Define how all the pieces fit together and you you've tested them Etc like what Avery is saying like it it's it's going to work out of the box and I think you know one of the near-term challenges for the modular space is going to be defining those interfaces in a way that makes it actually like more pluggable and then also making it easy to test all these things so I I will be the first to say that I think uh there will be challenges uh to to it but like there's a lot the good thing on the flip side is

that there's all these different teams that are incentivized to make this work and so there's a just a huge amount of uh really talented people tackling these problems and you have you have teams like roll up as a service projects who are sort of like system integrators who are like okay here are all the components and we're going to plug them all together into this stack and like sort of sell that as like an end end product to people who want to deploy a chain and so like you have you have people from across the modular ecosystem that are like actively really working on this I think we're going to still struggle for a while to have better bridging um I think that's just not solved like not as that problem exists like everywhere not just on the modular space but any cross

chain thing is still not quite as smooth as I think it could be and that does diminish the value proposition of everyone launching their own chain because then you you are when until bridging becomes more SE seamless yeah it will be a little bit more fragmented composable but I operate under the uh assumption that or the belief that in the long-term future cross chain composability will be really seamless um and and so like eventually we'll get there but for the for the near term I could see an advantage to sort of like a shared execution layers um like salana or aptose or or even like you know arbitrum for example as a rollup example where um you know it makes sense for people to build there because like you

do get this seamless composability that might still be a challenge for for like cross L2 stuff yeah I go ahead richy I think it's largely use case dependent um at the current St like if you're trying to build a Veno for crypto it's I would argue it's better than salana apos S just makes the most sense it's easiest to do out of the box um I think if you're trying to unlock different use cases that weren't possible at monic chains like I touched on like Innovative defi apps that can use assets from different chains like IBC enablement um stuff like that that's where modular trains have an edge because you can kind of customize it you don't need to wait for a third party to add dependencies add features you can kind of innovate by

yourself if you're just looking to use blockchain in the background um build a payments app build a gaming app and get to Market quickly as possible in the current state I would say that monic Chains would win so I think it's monic and modular in the current environment not or would you Avery Nick would you agree I I think um you know I I think they're just different exploration spaces and I I like I said I respect the fact that people are exploring different ideas and I I think you can probably innovate faster in a monolithic chain as sorry a modular chain as rushi suggested you our goal is always to support the 99% of use cases in opos and you know there might be 1% we don't hit but we haven't seen it yet so we would love to would love to wait until kind of what I've learned as

a developer for a long time though is it's really really hard to build apis before before you know the use cases and you almost always get it wrong and that's the problem and Nick alluded to the earlier right it's tough because you don't know what you want until you have a use case that you're driving it and then you know like exactly what you need in those apis uh to to push forward on um so you know we're always open to adding apis at the right layers and allowing people to plug in and play different things um but it we kind of wait for those use cases to emerge first rather than being a little bit more proactive about it that way allows us to focus on our core stack support those you know pretty much most of the use if not all the use cases out there we've

seen and and then buil in a simpler and easier test and more performant from the Geto yeah if you were to predict like what is the killer use case that's going to emerge like obviously in the last cycle a lot of it was just Innovations in Defi and defi summer and yield farming which was fun until it wasn't um what do you think is an X Catalyst and like the first type of application that just gets Mass adoption this is a hard one you want to go ahead Nick you can go first I you know I don't I don't have I feel like you know it's one of those things where uh it's very hard to predict because and if you could maybe you you would go out and like build that right um like uh so

I don't I don't have a good answer I think that uh some things that I I think are kind of exciting are things like uh unchain gaming I feel like gaming is something that um naturally has this like virality and mass Market appeal and if you bake in the right kind of like economics or incentives it could really like take off and I think Al things like like social fi applications similarly have this baked in kind of like Mass market and like if you have the if you crack the incentive code could like spread virally um so I think about those things I think uh stable coins are something that already have like really strong product Market fit and so there could be something around payments or just leveraging like stable coins that

that could also really really take off and those are just some some ideas off the top of my head but I think universally I'll just say that all those re in order to reach mass mass adoption or like these breakout use cases you need to have the underlying infrastructure ready to absorb all that demand and so that's why it's so important that like teams like Aptos or or like Celestia Etc are like trying to build this like underlying capacity to to like so when if we do hit that hyper growth as an industry we don't just scale fail and like yeah we're not in dialog we're Broadband we need to get Broadband yes yeah that's right um complet I think those are all great use cases uh that Nick Nick laid out the one I think that is hopefully

coming a little bit sooner than or later and maybe it's more of a hope than a reality is really about how blockchain can can provide much more accountability and responsibility to AI Innovation going forward and some we've been working very closely with Microsoft on a couple of different initiatives in this space um to understand like you know how does how does identity play to the picture how do we um get responsible AI where you know if if you're using uh uh if if if people are using your data for training uh you're either like giving your explicit permission for it you're getting compensated for it or uh you know other other you know as or you're just not giving that permission at all right and so those kind of things um are very interesting also like verification of content of AI right so if someone

posts a video of of of rushi dancing on a tree like rushi can say like you know this is this was really me or this was like someone faking that content this is not me uh your private keys are proof of humanity I guess yeah yeah proof of humanity is a really important thing that blockchain can do as a you know auditability with private keys and and user signing to to to show that kind of you know attestation of of whatever content is being put out there in market so that's an area that we we're we were pushing very hard with a couple of Partners on and we we really want to see that change happen in the world sooner rather than later Rich what's going to be the killer app in the move VM I think payments outside of infrastructure is just like where the

obvious use case is sending money from like Thailand to like Ethiopia and under like a second without using like Western Union without having pay PayPal like 30% um and having like centralized authorities um we're still waiting for the venmo of crypto to come out but I think that that's what gets me personally excited whether it's on Aptos it's a rollup it doesn't really matter um getting the average user to use uscc in back end not mention blockchain once um is what kind of keeps me going how it's accomplished I don't know I don't know what chain it's on um don't know what wallet on U but the team that does that and gets people actually using blockchains for payments and usdc everywhere in the back end um wins that's fair maybe as a parting

question um I'd like to get um two components but the first question really is whenever there's an unbundling decoupling they over time there there tends to go back to bundling like you see these unbundling and bundling waves and Technology Cycles um as you think about we're I think now entering a phase of mass experimentation right because you all of a sudden can decouple them and we're going to try all the different flavors svm movm and all the different components do you see a time where we aify into one standard of the stack that is the most optimal um or do we or do we never get there like will we Define like one two

or three very nicely like Stacks um or it will just be like there's going to be so many different verticals and use cases that we never kind of like Define a standard in this modular uh movement and you just continue to iterate and innovate that's a good question I the way that I see it is uh not so much in like the rebundling that's not the way I would frame it but like consolidation definitely I think like you know you have that this exploration Divergence and then like there's like kind ofle clear winners that kind of like emerge and and they become like sort of network effects around certain execution environments and and what have you so I do expect

that to happen and there might be sort of like a power law distribution of like here's like the most common execution environment that everyone uses and then it kind of goes down and there's like a long tail but I think there will always be a long tail we should enable that long tail for sure and um one way like I I often use a uh sort of analogy the way I think about modular blockchains is very similar to like Cloud infrastructure and that like a rollup is is like a like sort of like a virtual machine that you Deploy on like a data center but instead of a data center it's a Data Network like Celestia and um the cool thing is if you think about like how that like the the virtual machine world has evolved like you have all these like Linux is is like

you know obviously the most common you know operating system but there's all these different distributions that are like you know have different like trade-offs Etc and and so there's like this really long tale of different Linux distributions that like people use and so I think there you could see a similar thing kind of emerging with with execution layers in in um in in modular watchings where they they follow one kind of common standard but then there's like customizations uh and different sort of like they Branch out into like this family of of different things the best analogy is like there's a billion websites right but there's one Google there's one Amazon it's the website that have the best underlying infrastructure the best Partnerships the

best distribution that win similarly with like blockchains whether it's a layer one layer two layer three layer 500 um is going to be the blockchain that can get the most Partnerships get the most users um and then whatever underlying infrastructure that allows them to accomplish that job the best wins so whether it's ethereum whether it's any other settlement layer with other da layers execution layers um it's going to be it's going to be a billion blockchain sure but it's going to be three ones that actually get users yeah I think we use the same analogy hear it out to us like thinking about blockchains is very similar to Cloud there's several major Cloud providers it drops off after number four or five pretty pretty rapidly and that's the expectation we have and you know we

think that we're building out the right execution environment the right consensus protocol the right fully integrated stack that is going to support a huge chunk if not all the use cases out there but again very respectful of others that are trying different approaches in the space and ultimately though I think we all agree that we'll they we see consolidation to a small number of technology stocks in the future that ultimately serve the most user use cases as rushi mentioned absolutely well I'm all for experimentation and I think maximalism doesn't serve you well and I think we're in this incredible Renaissance of let's try all the different variations and see what sticks and ultimately let's we're so early we have like we're all you know it's a it's a spec of dust and we we haven't even reached any sort of

mainstream adoption so I think this is a uniquely uh in the like like it's a requirement to get there so great work from from all you guys really exciting I guess any parting thoughts um things you're excited about your building challenges or you know stuff that you want to call out before I wrap it up I like what you said that we we're so early I I agree I think we're we're so early on so many different dimensions um and and one being just the actual infrastructure like we uh you know when I first got into crypto in 2017 I thought wow I'm so late like it feels like everything is already done and like you know here we are like six years later and I actually now feel like we're earlier we're still extremely early and

there's so much more Innovation and experimentation to be done and we're also early from an adoption standpoint like very few people actually use blockchains today there's very little value that that is that are like sort of transacted on blockchains today compared to like you know the actual like scale of the the global economy and so I think uh yeah we're just a spec of dust so I'm excited for hopefully in the next six to 10 years like we'll actually get there yeah I think for me it's like on Twitter I see like ethereum already won or evm already won and the question I ask is like what did you win like you got the 10 users you got like like four people knowing about you like is that what your definition of win is so I think people are finally recognizing

like Network effects especially in crypto very small and all it takes is like one Kickstart team one killer use case one app that gets more than 10 users um to really take away the network effects um and then evm and ethereum is not W anymore uh completely with agree with those points I love the fact that we're early I love the fact that also the adoption is is not not there yet uh I mean I I wanted that to change obviously over time but I think the one thing I I want us to leave us with is that the infrastructure is gonna get ready really fast it's not going to take a long time it's actually pretty much here that people can build reliably people have the ux experience that is you know if it's not here yet exactly it's coming really really soon and you know our team

is hyper focused on providing at least one clear path of making sure that there's a great fully integrated stack a great user experience uh you know L fees and something that hopefully folks like Celestia and movement you know leveraging other networks we love that and we're going to push that future as quickly as we can that's great well guys amazing work well obviously link to your twit handles and projects um so really exciting uh I I guess we we to probably do a checkup in 6 months because to your point Avery like if you weren't paying attention and you left the bar Mark you nowk back like you would be shocked at the state of infastructure and how that is so quickly uh evolving so great work guys I really appreciate you coming on and sharing your thoughts uh I really enjoyed it

thank you San nice chatting with you Avery and rushi bye