Where Cartesi VM Fits in The Modular Thesis

Transcript

welcome back to the rollup High Roller Community as always nothing on the rollup is to be ConEd as Financial advice we are just an Educators in the defi space toss us a quick like give us a share a repost send it to a friend who may want to watch this and finally give us a sub or a follow it means the world to help support credibly neutral educational content in def5 we're here to educate Empower and enrich you and let's jump in what's up guys welcome back to the modular rollup I mean the rollup we're here with Carston from kesy the technical guy behind your stops your stop losses because you're not exposed long enough to the modul blockchain

revolution Ming around Caron what's going on man welcome to the rollup we have some fun we talk technical Concepts um and yeah we're here to learn more about CTE DM and your view on the modular blockchain experiment that is ongoing so GM sir yeah nice to be yeah I definitely think that full very it's a interesting future when it comes to what we constitute as blockchains or rollups or R machines and other things because I think that in even know one year this space is going to be completely unrecognizable I mean we have some kind of like understanding that uh let's say

that rollups would end up like the sh look like but as things like icon layer and good data availability comes in and let's say very let's say fast finalizing the devel and big they can push through those things I mean it's the craziest thing that he can do and it's almost having your own little let's say cloud infrastructure and I think that it's going to be a little crazy what we being to do I think what what when people think of blockchains they still tend to think of Monolithic

architectures um so could you explain uh kind of like why you know why that's maybe no longer relevant and and why it might look unrecognizable to that model um you know in the in the near future yeah so so so one of the things we have in Europe is this kind of like like let's say no cost aine called Ryan and they kind famous on unbundling pretty much everything including kind of like random randomly seing people and so and uh this is essentially what's happening to blockchains nowadays right because it used to be such that blockchains were the only ones who were

getting able to do this completely it's immutable history of things of events of our timeline uh it used to to be only blockchain environments that could do things like sequencing it used to be only blockchains that could do data avability but now these things are signed to be like let's say many different products in practice like things like uh sequencing and history is something that a data M buil layer could or sequencing layer could take care of like for example with espresso uh they uh provide let's say data availability up to some point but not permanently uh while let's say the etherian blockchain is still

talking about let's say things like State friend CCT maintaining this whole history of things and data availability becomes not necessarily something that that you need to carry around all the time on your not and this whole splitting up including alternative VM of course what cess is doing uh completely changes the way that we build applications for these things because we know no longer tied to the insane things that people came up with in the evm OR restrictions and so on we can sit down and use regular instruction set regular compilers and not pretend that and actually like pretend or ignore

the fact that the last 60 years of computing has am existed sh the VM pieces um what's the uh what's the uh what's the thesis as to why evm wins versus alternative VMS I think that that it's got to be blurry for qu quite a long time to be honest because like uh I think that we're seeing with things like even with arum stylers or the massive waiting explosion of ion layer based co-processors it will startop blending in with each other right I mean back in the day uh people were looking at like let they projects like

trit and so on to kind of like allow us to kind like offload in a nice evm out sourceable way of computation and so on right but uh that never that pattern kind of like went away again which is kind of like a little bit annoying because like off competition is what everybody pretty much needs right uh but I think that everything basically starts blending into each other also with projects like with Z like because things like and I think this is happening at the moment where things like the stock to knock as and you can take the big 200 KOB proof converted down into a small uh SAR proof instead and the recursion so is that that it's good enough that we

don't need to care about all the the C the insane design choices TVM has anymore right like we can actually start people developing applications that actually do things on chain without touching the evm except for proving as that cable and that completely changes around kind of like the ability for us to connect things in ways that doesn't look like we are trying to jump to five different uh hopes of fire at the same time yep yep so so you think this all just kind of comes together from the from the experience of like I guess the user but also just Builders as well that's one of the key things that uh caresia allows us like this onboarding of of deps of Builders uh

which which which I think he further emphasizes the alternative VM bull case which we're starting to really build here at the Lop it's like well the evm bull case was from 2015 2016 until 2020 2021 2021 defi uh realm followed by the 2022 hacks 2023 re-cy attacks this that and the other that is just kind of form the alternative VM bull case um you know which I'm excited about so let's dive into kind of C cz's designed with some of the recent uh Inspirations and kind of announcements um let's start with espresso um how can cartei rollups uh utilize espresso share sequencing layer how is that going to build a better

ecosystem for cartei rollups uh and further you know establish the alternative VM or cartez bull case yeah so the first thing you need to come Rec ask yourself is okay if we are going to have a billion app chains or as specific rollups or uh when somebody finds a proper name for these things let me know about like I the question basically becomes like like how how do we actually do this right and the first thing that kind of like comes to mind is that that well R ux sucks say right that's the first thing part of it is because basically that uh there's very let's say

very uh slow confirmation of actions and so on while it's of like expresso you can literally get a confirmation that your particular transaction has been included with within one or two seconds I mean it's basically the AL experience right that we kind of get stuff quite quickly second uh if we had to scale to let's say a little bit like with cable TV right like we need almost as big bandwidth as a cable TV network which is like something where we need this extreme perform that's something personal uh provides right it basically assigns let's say makes 30 megabyte size blocks and actually doesn't with let's say very large amount of uh large large amount of

notes even and it's the ideal place to come like start rolling things out easily also because they have a focus on this whole let's say block Builders and and let's say proposal Builder separation where we can start thinking about what goes into the blocks and the confirmations of them and all the mechanics around it which is something I don't see in all the data no necessarily so what we've been doing in cessi is experimenting with okay so how quickly can I roll out a functioning app chain and the answer is somewhere up below let's say three minutes right I mean if I want to start put like I declare by

populating a little bit like just block that uh that this app chain exists which sequencer is it what which sequencer block height is it and then basically from that point on like theoretically anybody compute the same application right uh and get the same state so technically speaking exists in a particular State and they kind of like you can just submit in this particular case uh transactions into the sequence or and they kind of like Get included then other different kind computer the bries etc etc etc and that I mean that's experience that has to exist it has to be just as easy as even easier than creating uh and let's say a Twitter account

something like that for rollups and on top of that it's something where these app chains and I'm probably ad my turn but that change should be something where you're in control about what inventor it's it's let's say working with right I should be able to kind of go out and get on the market and decide which provider is let's say providing services for we might even be a network like with things like witness chain was com with for example things like I uh and how do I kind of get the best quality of service for my app and the services that's going on with here right and this is kind of like why we've been playing out with espresso

and the nice thing about Express and kind of like this these uh sequencing in particular is they have a quite nice API right is basically listen to the incoming blocks is there anything distant for me or I lit my virtual machine this particular case uh which is my FJ uh what is in the transactions i p p and I move on to the next BL until there's something for me again right so integrating these different things because of the fact that this is substraction uh sequence really helps in designing or with things like you need to all things like Ys and so on that you can

kind like end up with this nice little let's say nice little architecture to some extent like our note for doing this stuff which is let's say quite fully featured in this particular Point like let's say less than 3,000 lines of rust so it's something that that really changes around and make architecture cleaner for building these things of course where changes let's say a bit up difficulties is uh convincing the poor age in evm uh what the computation state of AAL app chain roll up is because now now suddenly you remve from the problem having one app chain a roll up that you want to settle into ethereum to they you want to settle a billion app chains into

etherum and it becomes like a much more complex problem yeah I mean espresso can um kind of bring those together right and um bring Atomic posability I've been trying to hit home on this point of like difference between atomic inclusion and atomic execution when it comes to using share sequencing something to get into but um I think generally uh when you are able to deploy an achain three minutes then connect it to the shared security and sequencing layer of the rest socialize the costs and also potentially share the liquidity not exactly sure how how you see that playing out would be curious on how that work um yeah

so robust I think that that of course that's kind of like let's say the ability to kind of build that block in a compossible way right like like how like you make promises between them but also very interestingly that you can also logic people to say that that I mean on the same shared sequence I can theoretically read let's say the state of the the previous block of an appj right as well like let's say that uh let's say for example I have a let's say it's a uniswap app chain right I can read the state of uniswap in the clock just before because technically speaking anybody can evaluate that the moment actually settles into some flag interium something that's when you actually need to kind of like Compu then like in a organized manner but totically speaking

the the state only exist at that point I I I think that that when it comes to block building and automic inclusion and those kind of things I mean we haven't seen completely uh uh uh the end result here uh of uh let's say what can actually be possible because it hasn't become easy for people to experiment with it right but if you start combining this into let's say things where uh where uh that people can trivially write something up in let's say the existing

programming languages they can easily let's say get working with you undeploy it and actually start playing well with let's say the architecture of building blog worrying about collecting weird things about that's going in on and 3M then I think that then that basically the amount of things that are possible to do with this stuff is uh going to explore I mean it's going to be completely unimaginable yeah we're talking about the expressibility of let's say being able to do more things with the AIC sequencing and inclusion of whatever those we have terms are exactly because right now it's

really difficult for me for for people to kind of Reason about how does this work like onm but if you say that well this is basically a bunch of L LX virtual machines that are working in let's say on the same clock which is the blocks coming in essentially then suddenly you can start thinking more about like let's say promises that I made uh and the conversation between them and it start looking a lot more like let's say optimizing a workload and let's say a cloud installation essentially yeah and so is is that like heres's mission is is is kind of making

this developer friendly accessible uh tool to launch chains to provide that infrastructure for rollups based on cesi rollups is is part of this is part of this to kind of also figure out like how the best way to get around these certain tradeoffs are with code processors with shared sequencing there's like there's these tradeoffs set we're starting to notice that modular blockchains are trying to kind of break this original Tri trilemma blockchains of Monolithic chains I'm curious how like IG layer in Inspiration espresso um Celestia how how you're kind of putting these pieces to together for the ctez stack to kind of build the best stack yeah so of of

course it helps we have the K Machine which is let's say the virtual machine we ask it is let's say it's not something new it's not like we are inventing some kind of do VM or anything it's not like we are inventing a new instruction set uh it's something that actually works it works today you can run a on it you can R doom on it and um honestly speaking if you have Al M and you can't from do on your lvm I don't consider you on Al VM because it's not general purpose enough um but how all this connects together basically comes down to in the end of the day like aola is essentially a compute primitive being executed multiple times over a series of

inputs right I mean that's what calculates the state uh and how we understand and let's say get to the understanding or confidence that a app chain is in a particular state is something that with very high probability I mean you get quite close I mean some people believe that whatever let's say inua is telling them today right by metamask is the truth right or is a scan and um what you can do beyond that is like things like for example you can stick it into let's say an 8 light claiming trust aw right or you can go out and get a sampling of co-processor computations

over the state basically saying that okay so I know the truth is that that let's say the expresso chain is in this state what let's say uh the calcul state on my app chain which is then I go out and get a sampling using let's say cop ID and then I get even more kind of like higher confidence what is actually true and then if you need like the strong strong strong confirmation the when you can settle to things like theum for example but I think there been many use cases where you you don't need the strongest let's say confirmation and I think also when it layer two layer two let's say to of communication you need more like highly

probability kind of like things that this is good enough right in terms of kind of like breaching data information and messages and so on so we we I think at the same time we haven't seen the end of all these different sequences or datab B builing methods and so on so getting something where you kind quickly can switch between and play with them is quite important so how does ctei position in this data availability Wars are you are you Pro non ethereum da pre EIP 4844 post EIP is there going to be a change in your mind like what how does ctez give the that the developers and the users of this experience how do you

position so my belief is that that that uh it's not my belief to make and it's not my choice to make for people right as if they want if they want to pick a different religion they one welcome to and this is how you do it when a software um and that's why we kind of like we we are basically trying to make it work on everything because in other the day it really comes down to what is this little I mean gospel I think that people call these kind of like APS like Community computer right and that's I mean even arbitrum today is theoretically let's say Gant Community computer to some extent that that they choose what becomes let's say the uation of their

chain um and I think that that as people are kind deploying Things become more and more autonomous there will be different trade offs there will be different benefits uh but I also think that that when we start looking at things like let's sayic conclusion and those kind of sequencing etc etc the lines begin to blur because some things will need to be more safe versus and yeah I thought as use case dependent it's like if if you I leave financialized large tvl for example or large stake you may as well just do everything without extra trust assumptions like if you if you're running some like consumer application or like you know insert your game here

right it's like well why not optimize for the the best user experience and and the best cost and the cheapest fees and all that but that's down to the beauty of this no like with you know you can insert C uh the cartez virtual machine amongst a different stack of other kind of pieces and you can you can almost customize it like a like a like a menu at a restaurant or something right like like a buffet exactly yeah that's that's Al that's also the thing like like and like newest to science like like the the current state of the appj defines where does it get the next block from where they going get the next transactions from I mean why shouldn't we be able to

car like let's say WF between Celestia and espresso in terms of kind of like where we get our input from right it's a little bit like switching cable over from one provider to another right um and I think that that as we are getting let's say further and further ahead we start seeing these things let's say in between let's say roll over actually VMS as more like network interfaces versus let's say a message bses versus let's say in let's say Ting the state of blockchain right I mean that there's no reason today why could a virtual machine we couldn't literally send ping packages between the different

uh machines or may hhtp if you really want it to right and that's like the thing that that I think that we sometimes forget a little bit with thean virtual machine like I actually like the world computer thing except that that well this was welome computer 1.0 and I think we can make it well computer 2.0 I would expect nothing less from the tech mind behind an alternative VM sir um I appreciate the uh the idea of this composability and the different type of framework thought rather than like oh this is one thing this is separate we we need to derive consensus or data from one part no we can kind of bring this all together similarly Am I Wrong to think that basically cartez VM can

integrate with any Ras provider any the availabil layer any inup solution it's very much a one-size fit all uh in terms of being able to to to interrupt with with everything that's currently part of this like modular blockchain kind of thesis narrative like there's no reason why you guys have any technological uh obst to integrate with everybody I think do the technological like let's say yes L machine could run them different way but of course you need to be able to kind of like reliably be able to prove that to other machines right so for example obviously let's say some uh let's say icon layer data

availability thing that goes out us three Stakes operators to make uh data little bit like espresso it's much easier to let's say understand the data availability uh CH like change let Network State because it is tied and anchored into iterior right and uh that's also my preference a little bit that there are certain kind of models out there that that doesn't work very well we're just like let's say this kind of like almost fisherman like style data availability where let say you have to go out and get that proof that some data is poish somewhere or something like that it doesn't work in an ni

structured way or let's say that things are are hosted on a completely separate or not say security divide network from iterior I mean I I I like the iterum for let's say security and so on but some people might let's say want to have kind like their particular app chain let's say with data M from some from things you like learn it might actually make sense for them but in terms of compossibility across uh networks I mean in starts more looking more like instant Insurance agreements and so on very interesting okay so so what is the what is

the this might be not exactly the right question for you but I'm curious like how do how do we get to a thousand ctei rollups but what does that look like I I think I I think that basically it it it has it has to become just as easy as it is right now to upload so from like toel or to H or something like that um like component view of let's say e used to be just like deploying uh Docker containers somewhere um and I think that that's sometimes a little bit what people are misunderstanding with think like repres and so on like like they read that let's say that that me is how you if me had existed we wouldn't need Dock and so on

but I think people tend to forget that the difference between something like the Intel processors and Har processors and rist five processors and web assembly it's just as big distance as between those processors and the evm like it's a complely alien technology uh of course there's things that kind like General computation and so on but like there's a lot of things that basically doesn't compile their to it and I think that I mean it took us with v took it like almost five to seven years from from my python direct running a BB assembly so it's something where like let's say e doesn't come from providing let's say a% based for things

alone I think Simplicity comes from the fact that we can't can have these deistic RIS five lenux running machines that actually act like we expect computers to act to it and they need to be complet like sized in all sorts of different way wor like like let's say uh almost of the setups like like you can't go above two or four GB of RAM while with ours you can theoretically have 64 bits worth of ram right like it just of gigabytes and so on right so that's the level of ease we need to get to and it needs to be not alien to people because like like in one presentation that's this uh snbc comic thing where basically

like like like people talk about this new eara computer and on and and difference between a computer scientist like we have much to learn from this computer uh and then they GS up say boom I got do it right and I don't want a completely alien computer for people it needs to be something where people can use it they can reason about it they can treat it just like something that they expected to actually work yep interesting so from the de side it's like it's about it's it's like a combination of educ tion as well as ease of use just in in the in in the Innovation side um as you're going to come the wrap up here I'm curious about

just your broader Vision outside of cartei um what is 2020 what does 2024 have in have in store for us for the growth of modj blockchain what are you seeing what are some of your predictions or thoughts bull case bear cases and where do you see this this heading with the Advent of Ras providers new ga lers new VMS launching tokens launching um you know how does this kind what's going to what are some second and third order consequences or thought or at least effects that maybe we're not seeing right now but that are that are going to come as a result of of these kind of multiple different rups launching with different uh you know infrast Stacks whether it be interrup or different kind

of things that you're seeing happen I'm just curious like what are some predictions that you may have so one one thing that's for sure clear is that with things like I layer we don't longer need to limit ourselves to the let's say The Innovation on eum becomes permissionless in terms of let say getting just a strong Economic Security and I think that that will both let's say in the bare case this particular Cas I mean we're definitely going to see if uh failed rocket launchers uh but at the same time it might be there will be a couple these rockets that actually make it Ross um I think

that people will be like like let's see there's a more feeling of more pragmatism in terms of adoptability because like I mean I've been around since 2015 2014 something and I mean I remember when we're going to change the world right and I think we still need to let's say actually get there and do something where everybody can shouldn't start using in a in a decent let's say cerent way and and let's say maybe kind of like as the last kind of like prediction is uh I think we're going to start seeing actual verifiable Computing of Bitcoin and I think that's going to let's say change around also let's say how we think about

verifiable Computing like things like bit VM and so on and I think it's not compl put lot realistic in the next few months then we're going to see things like Doom one Bitcoin prediction very cool you know Andy says something uh few videos AG a while back on our Channel and and uh it was about like uh you know that there's all these alternative VMS coming for uh the evm but it's like is there any settlement layer that could potentially kind of Dethrone the ethereum uh consensus layer the settlement layer and it seems like the only one that out there is

potentially Bitcoin because of the the cost of corruption is so high so you thinking about that yeah so so okay so so the main problem with with is that yes we'll be able to see offchain competation because I think like be PRM on bitcoin and so on but I think that let's say the evolution of that takes time and at that particular point I think that and the evolution of things based on things like I can ler and ethereum security we're coming to the Quality point

where you need to be a little extremist to go for Bitcoin in this particular case uh but uh I think that ethereum has in the seeds of itself and the switch to for stake uh it has the nature in it just like phone lines did right that we started running things on phone lines to kind of create this infrastructure inversion that what we're actually using or what actually is kind of like what we consider the etherum blockchain is basically likely say inversed set up on top of uh something like that I can kind of like set up so

I'm not sure how it's ex going to end up but I am thinking little bit that that we'll get to a point where basically we don't care because we actually have use cases running on things and this is like the stuff that none that some of us are experts in and well it's probably going to be be the very lucrative to be experts in kind of like those slower lower layers but in a day the whole thing becomes commoditized where that's something we've been thinking about as well as the race to the bottom at the da layer amongst the other ones it'll be interesting especially at at the RAS l or two it's going to just be like like the era of abundant block space is here so uh it's going it completely becomes a commodity which is uh quite interesting

uh cars I just want to say thank you for coming on just kind of some closing thoughts I would gonna wrap up here and um yeah man thank you I'm aquarious thanks Carson