DeFi By Design EP#117: Why Data Availability Is The Missing Piece To The Modular Scaling Puzzle

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 welcome to D5 by Design where we talk all things block chain cryptocurrency while striving to educate EMP power and then R what's going on guys welcome back to episode 117 of Dei by Design podcast today we're g to 100x lower the cost of

ruup transactions and I couldn't be more bullish boys look at that smile on pbal I met pbal out there in Istanbul so did Rob had a had a good time got a sick a veil shirt I'm not wearing it unfortunately but I still got it in my suitcase haven't made it at home yet uh than pretty fresh Rob how you feeling I'm doing good man I'm I'm feeling good it's uh yeah it's good to settle in after the conferences I mean the conferences are a whirlwind uh Kudos and even riding the momentum just got shot out of a cannon at istan bu and just hasn't been the same since so uh we're we're pushing on all fronts learning so much about modularity um and we're definitely going to be going to be

going through the modular stack to day and I think just as as we kind of like like start to get into it a little bit we can all start to frame this mental model where like the stack concept originates from like the layer one layer two layer three and so on and so forth and it it is very clear like the mental model is very very very clearly a stack of these different Technologies on top of one another and then they all they all kind of flow data down to the based level where it settles on chain modularity blows this all up it's like totally unpacking all of these components and just sifting them all throughout space all over the universe

and then there's so many ways where we can connect we can plug in plug out different things choose where we want to combine certain Technologies for different flavors of an application or or a blockchain or or you know SE several use cases from there so purb I'd love to start with um what your mental model is where you you view the general like how you view this General space what is your mental yeah I think call the mental model stems from what we have seen till now like the status quo where we have seen that a lot of different uh a lot of activities are happening on the base layer on a single layer what that means

is we are you know try very hard and for for very long to kind of SK this this one single layer to make it are compatible for doing a lot of these uh computation a lot of these uh you know boot Staffing of security and so on and so forth into a single montic layer and then we realize that you know the montic world doesn't St because you cannot you cannot optimize all the parts of it without breaking uh the part the the main layer itself because it's monolithic any part you want to change you have to make a protable layer change onto the Bas near which is completely invisible um for example if there are many many du

different ways that may can for the users right so there has been a lot of lot of Noel work which has been done for let's say a me pool which is which is uh secured by MPC or there is a comet R mechanism on the me pool and so on and so all of this is great A lot of research have been done but most of them cannot be implemented on design because you at that point have to change the entire R tool for the B layer and that affects all the applications running audit right and you do not have a choice at that point because it's it's a single day um what model and the entire thought of rollup Centric road map that we have

seen actually says is that you can m in bits and pieces of it put it off scene and then push something very succinct on on chain right and then the onchain demands remain very minimal the on the the base M Act only as to so-call give the security right and a minimal things like data availability and supplement um whereas you know everything else the complete execution the complete me pools the ordering um can be done completely of change so that's what the this entire shift in mental model kind of feels like

what that also enables is that individually each one of these pieces can be optimized can be made to be custom um for example simple things like do you really need uh the entire evm architecture for your application specific rollup if the answer is yes then that's great you can just inherit that transaction structure the state and storage structure the you know Trans State trans function and so on and so forth but if not you can now do custom States custom transaction sh to customs you know storage layouts and so on and so forth so what what this essentially feels like and sorry about the long

minded answer but the entire mental model feels like you know how you would how you would start with a single L try to consolidate like but I write a code base I start with something that that just works something might not be broken into functions things are all you know into a single main function which contains all of that code base and then that that lets lets me get the work done right in the in the first iteration but then as you want to make things more much better more optimized and so on you take this of the code base make into functions keep Abra you know you you slowly slowly try to optimize BDS and

parts of it because then your code becomes overall more obtains and that's what at least my mental mod is like yeah and I appreciate the longwinded answer because frankly up until few four months ago I I was quite confused on what this even meant I I remember being at DCC there was modular Summit versus there was like some other L2 events and I was like ah that was back in July I was like ah modular Summit who wants to go to that whatever what does they mean and then now it's like the whole world is the whole world the whole crypto world's talking about it and um and I think that I'm I'm still getting a lot of uh you know questions with regards to what is even the basic stack so I'm glad that we've established your foundational View

and I think it's again in alignment with a lot of what the the the general consensus and thought processes are around the advantages so now building off that um you know and I'd like to Deep dive into each layer into da but just continuing to build off this foundational uh you know kind of Overlook we now have this this this idea in a lot of really smart Builders heads who raise money with B teams that are building this Tech so now what is this modular future based on this Foundation look like six 12 18 months from now and then how is that different from what it looks like today and you're on mute sir sorry and uh how it looks like today

especially let's talk about the ethereum ecosystem right right how it looks like is that whenever there is a lot of activity um the price activ becomes extremely extensive right it's it gets very hard to use the chain because all the activity hits that St there um and the fact of the matter is that as and when we get more and more use cases more and more application this is unsustainable to be able to give uh you know one transaction fee as tens of dollars to even spiking up to hundreds of dollars is is not sustainable for for

a use case specific blockchain if you want to make blockchains main Stree in the in the mod world what will happen is that you know this is this is going to become rollup centry what this means is all that execution will be completely taken off steam and be done with users will get sub confirmations of that acceptance they would be happy with it they will be formed of batches and then the batches will only hit some the DNA and the proofs will hit the settlement layers of the world right so uh in that kind of world is the is is where we will be able to see amortization of cost uh not only it's about just the costs about

the user experience right now uh different lers have different block times block finalities and so on and so forth so even for example um today as a user you have to wait for like say 12 second for ethereum blocks and in one prity you have to wait for 12 15 um minutes of finality right so uh in in in this kind of word depending on what kind of guarantees you need you have to wait for different amounts of time uh for example if you want self confirmation you will get it in millisecond for example I think arbitrum today gives somewhere around 500 millisecond agent lesser times up you know transaction

confirmation which is great to have right for a user you just give it and it feels like a web to gu of an experience where you just submit a transaction you get back a confirmation that it has been accepted um but if you if you are maybe you know transferring a lot of funds let's say I'm sending millions of dollars then of course I would only only only do that on the transaction confirmation the confirmation right you would want to have subjective finality or objective finality in terms of subjective finality you have to wait for this batch to get the DNA which then you know ENT like encapsulates The Ordering of the transactions and after that you

you might even want to wait for objective finality which can be you know a few hours to a few days depending on whether it's a uh whether it's ZK verified or optimistic construction and so on and so so uh you see the options from the users increases a lot right so the experience changes a lot the the apps and the optimizations and the customizations they all change a lot and that is what this entire road map is about to be able to give different kind of experiences depending on the use cases uh yeah I've wrote a lot about and I've got it here as da was a 100x cheaper

transaction cost which is something that I'm really excited about I was at Epic e INF for day in Istanbul and the founder of Neil was there of course they working on a d layer and um you know he I I I asked him like is is this going to allow rollup St transactions that basically are unfathomable in costs to what is current and he was like yes like this is something that people AR really understanding as far as how much lower this is actually going to cost and then in in addition to that you spoke about this this idea of confirmations and this idea of pre-confirmation I is it a is it correct that using a DA layer in will actually increase the time to get a pre confirmation of a transaction or is that

not true like it'll slow sorry it'll make a pre-confirmation quicker rather than going through a theem as a DA layer or n um the the main thing about mod is that there is no straight boundary which uh which is sometimes not uh kind of taken up nicely right but the fact of the matter is it depends uh it depends because if you are using a centralized sequencer the transtion confirmation Tes will not change at all because your confirmation is going to come from the sequence or the entire experience going is going to remain the same if you think thinking about you know the confirmation coming from the DNA a which is the

subjective family that I was talking about the subjective parity will might take longer might take shorter it might take shorter because you know you might you might not be able to fit a big batch into one block of etherium right so when you're unable to F bigger batches you break Dig Down shorter batches and you push it onto ethereum uh which we keep on increasingly seeing we we see some of it today but as as demand increases we will keep on seeing this kind of trend whereas these specific layers they have the capacity to take in a lot a lot of bigger blocks and then it might be longer because there are other constructions

called base rups in base rup constructions you do not have a sequence right you directly send it to the DNA and then depending on the DNA reptiles and so on it might take a bit longer right so all of that is based on orientation is based on what kind of conations are you looking for the then there are very very normal Constructors which are coming up which we will keep on seeing in the in the near future things like shared sequences so if you're thinking of a cross ruup interaction the kind of confirmations which a shared sequencer can give you is probably we like orders of magnitude faster than what you could have done in any of the other configurations right uh

but again is that needed is there supplementary construction which can allow it maybe yes but that's that's that to for us to see like for us to check it out when those kind of infra things come together and then on you will be able to see the benefits but largely to answer your question in one word the answer will be no because today the status schol is centralized sequencers and they just give soft power presses on them the da and the seml is completely abstracted out from the user yeah so for the sake of the conversation of abstracting away edge cases um and also focusing on what is today the the big upside for using da layers is to

find this happy medium between security and decentralization whilst also lowering costs significantly for rollups and then when it comes to the the the the the speed um you know this is something that is kind of a a case by case specific basis now Rob's got something but I just want to add the last thing on to this one is so if my confirmation times on arbiture and optimism in various other modular rollups I think you met it say like 600 milliseconds let's call it 500 to 700 soan is 200 milliseconds the ux is great when I on board friends to saana they love it why why would I even think about using a modular rup if I can use saana yeah I think uh again like every construction does tradeoffs right so if

someone is saying that there is uh like there's no freelance right so every construction makes trade ups and that's what we should appreciate as well uh and that's why I don't want to say that you know don't use S or something like that because if if a user is happy with that experience and and people are building uh like very good use cases there then of course it's a it's a thing thing to kind of study to understand what the tradeoffs are to to kind of keep on using it because we all want more and more use casing in the blockchain ecosystem we don't have enough of it and if if uh someone like Solana can help us bring those news cases then we shouldn't

criticize it just for the just because we cannot you so-called compete with the block right now the the answer to that is verifiability right so it's about the end user verif ability which is which is something which is very close to the ethereum system and that's what we also want to do as far as a and you will see all of the other efforts in this in this community is towards the endu of verifi now if uh if you think about the uh trobs there uh we want the end user to be able to verify whether the transaction where was done corm one in a way what we are trying to do in that

regard is that we are aning data sampling which is a powerful construct which in which users themselves verify the dity rather than relying on any blockchain that is available on BL and we are also kind of trying to do things like onef level ZK verification what that essentially means is if you think of the end user experience and how we look into the world right now um or at least how we envision that the world will be is that the users will actually get proofs of execution and will do sampling to get proof of data availability with these two F structs they can be extremely sure

and self verified that the transaction was done correctly and that it is included and it canot be that changed so this is the kind of end user verifi that we want to G as part of what we are building in the DNA but also what this will in in terms of you know if someone is using let's say a um a ZK stack for for for generating or for processing the transaction so that's the kind of tradeoffs that you are taking uh it's about Verity versus you know faster block times and things like that but again you will you will keep on seeing much much smaller block confirmation

times like it will it will almost feel like a like a web to API call at some point in time time uh because the sequencers can can do you know very very fast confirmations um it is down to what what the trade-offs are right now is that is it sustainable right is it sustainable to keep on growing a chain at the same time keep on getting faster confirmation times right so for example when I seeing a transaction do I just give you a confirmation that it is accepted or do I cross process it and then give you a confirmation of it accepted when I say I process it how long do I take to process

that transaction that if you are talking about an evm Centric roll up which has a lot of Steed bloops then that can be a bit costly right because it has a lot of uh lot of things to pull from the state a lot of IO patches which are harder to optimize on and then if you talk about custom State transmissions and that's why I was mentioning that a bit earlier we will keep on seeing that with custom some State transitions with very very lightweight app rops and you almost feel like a rep API call because all of that already happens in web to API cons as when as soon as you make a con it does some verification some internal checks that it is typed and then give you a past confirmation my body is

ready and uh the these are all trade-offs that Builders are making like as they build these rollups um and you mentioned that you're you're building a veil for user verifiability the reason that that you prioritize user verifiability sounds like something that I heard all the way back from satoshi's white paper about people running a full note um so I I think that's a very noble cause and uh it is important because we've seen like we we've seen why uh like not your keys not your coins why that holds up is it and I I I take it that the data that is going through each user's transaction is

very very similar and core to the to the like native experience about uh of them running this node um so they're they're not running a full node in these cases but they are processing the data themselves could you could you kind of like describe why data availability is important like at a first principles level like the user verifiability that you're enabling what does that enable for the end user yeah yeah on that I think you you made a very good reference to the to the Bitcoin paper and let me just try to you know try to talk a little bit about what's that enabled and what we are

enabling because it's not a app stens comparison and that's why I want to clearly State what are what the value propop is so on a Bitcoin kind of known or even in ethereum uh the CHS are actually doing execution right so you know the entire world computer vision of of ethereum that was there so essentially you can think of it as as like a as like an execution machine which you can trust so it right because it is by the cpto economics of ethereum's validat set um in that kind of a world you have to run a full node to know whether the execution was done

correctly even if you're an end user what you do is you do not rely on the 2x3 plus of ethereum or the 51% 51% of Bitcoin you basically run a full node to know whether they did their job correctly that's the kind of verifiability I'm talking about right so that's why you have to run a full node to know whether it was processed solidly if if and then there are Concepts like social consensus and social talks where even if they commute even if 2x3 pluses and actually processes a wrong transaction you you your full node will actually P off it will not accept that block it will not

accept that blog just because it has enough attestations from the validators That's The Power of of these constructions in uh data availability Focus chain what we want to get is that we want to give the ability to the light clients to verify data availability uh the mind you that the core focus of a execution based sh like etherum is doing the execution for on it's only about data availability and we want the light clients to be able to verify data availability without relying on the 2x3 plus of our chain right and that's the value value that we want to add to this system now what that

collectively means is that as a user you as I mentioned you need two different things you need institutional correctness and you need data availability on the data availability front we are allowing light clients to to do extremely cheap uh data with the sampling only perform like let's say 15 to 30 data samples check it for correctness and then be very sure that the data is actually available we have a web page which can do it on a mobile browser right so that's how easy it is to be to be able to actually do data the same and on top of that all you have to do is verify a ZK proof a ZK proof depending on what kind of proof systems

and so on might take about 60 seconds 60 Minutes second sorry to kind of very right and you can understand the power of this construction like in order of milliseconds you not only know that the execution was done correctly but also the data was available this is a big shift from I think the AI is also understanding that I'm saying it's right but um there is a big shift from crypto economically driven chains to cryptographically driven chains right so in in essence you had chains which were which were secured by crypto economics because you had to rely on the 2x3 Plus in order to do it correctly or you have

to run a full road to verify now you're just relying on the 2x3 Plus on ordering and the rest anything you can verify on your own you can verify the data availability but you can also verify the execution yeah and and the our our audience probably already knows this but the reason that it's so so quick to check the ZK proofs and why this this cryptographic verification is possible is because ZK proofs are very there's no there it's almost impossible to come up with the solution but once you have the solution it's it's very easy to check and make sure that the solution is correct so that means that it's almost impossible right to make that that uh

proof but once you have it uh you can do it in 60 milliseconds you can check that you can verify that proof is correct um I would okay yeah sorry sorry I would like to frame it a bit differently and say that once you do an execution you basically want to create a proof of correct execution making the proof is somewhat very hard right so what you do is you do a lot of this I'm not going to the ZK cryptography side of it but basically depending on start SAR you create those circuits or the a representation and so on and so forth

you basically allow the Prov to do a lot of work after doing the lot of work you you then give a succinct proof to the verifile right and then I'm saying zkp and basically I'm abusing the terminology it's a suin uh proof and so on and so it's a SN things like that but I'm not going into all of that the user can use that succinct proof to verify the execution exactly the complete execution can be verified yeah got okay again yeah go ahead what one step further and then and then yeah I'm sure Annie's jumped at bite at it so so then

we get into Data availability sampling and could you talk us and I heard one of vetal talks at is stamp buul and he was all over this data availability and data availability sampling he described like like onedimensional sampling and then two-dimensional sampling and how there these were a couple different methods and I'm sure there's trade-offs for using different methods of sampling could you describe like once once we understand you know this this uh user verification how how is data availability sampling related and then why might someone choose to go with this one-dimensional approach versus this two-dimensional approach yep I think it's a it's an

interesting topic to kind of discuss and thanks for asking this kind of questions um so essentially what are the powers of data which compute doesn't have in is the ability to be able to efficiency sample it now what I mean by that is when I take up take some data I can be I can eras encode it I can essentially create redundance in that data in order to make it extremely hard for someone to hide parts of it and I remember I like to give which might not be a perfect analoging is that if you had compact discs and sh that you guys have access to it I don't know how many of your

listeners have used it but if you have compa this there with a scratch on top you could still play the video or the audio file which it contained right and the ability for for that to happen is because that this was ual coded so there was a lot of redundancy in place so that even if bits of bits and parts of papw missing the rest of like the computer can actually take in the missing parts and recalls right we are using this in the flip side in the flip side because we are now making it harder for someone to now mind you that if someone was able to suppress data then all of this roll up that we

are talking about their security with like with visibility or actual security depending on how you define it actually completely falls apart if you're able to hide parts of it then in an optimistic design you can't even construct sh a fraud proof so it completely breaks Security even in a zkp kind of world like when are creating relative proofs and so on if you have the data kind of messing then your liess gets compromised so users accounts might get Frozen and so on and so and that's why we have the data availability the data avity constructions where we add redundancy to this data we add redundancy so that it

becomes very hard for someone to actually hide parts of it now comes the onedimensional nature versus a two dimensional nature like how you actually construct this data availability um block right the block of data now in one dimensional again depending on what kind of construction that you use whether you use a commitment based like we use or a house based like some some others use there are different ways in which the data remains secure um in a poal commitment based scheme we are essentially making tradeoffs the tradeoffs are how long does it take to create a commitment how

long does it take to to generate an opening and how long does it take to verify an opening what I mean by opening are the sample proofs so essentially when you when you ask for a sample you essentially choose a small section of the entire data and say that give me a cryptographic proof that this belongs to the commitment that you gave previously and then the network has to generate that proof and then give you the proof you then have to verify the proof so this is this is one one Ang it the second is how much you don't know see how easy is it becoming for someone to verify whether the original data was

available or not right how many samples do I need to do for us for me to be able to get a very high confidence that the actual data is avilable these are the kind of tradeoffs that you are making there are some other trade-offs in some other constructions like for example if you are a fa proof secured system then essentially if you create like how big is your fraud proof is your fraud proof efficiently verifiable By the Light client you cannot expect a fraud proof to be so hard to check that it essentially needs a full note to run that it completely defeats the purpose of creating a life C right so in Ence what we I'm trying to say is all the construction categories kind of

parameters and the construction the the design is there a is a big tradeoff on how much verifiability you want to give how much of it is being offloaded to the generator of the block the block the block production engine that is how much of that is being pushed on to the end user to be able to verify and the proof sizes and so on and so for so I don't know whether I was able to Cap all of it but um hopefully that give some ideas yeah it yeah essentially um based on the on the size um and the level of I would assume the level of security needed um by this specific use case you know one

can assume for something that's not super Financial related you could use you know a a potentially a smaller or less sophisticated system but something with it that has a lot of had stake something a bit more um you know rigorous that that kind of dictates the the design choice of the type of data Del simplying which you would execute but but it also this the amount of data also has to be uh basically it has to fit like a Lego piece in a sense to be able to actually be oper by like clients so that they could actually you know just do their job which which is what they best which is instead of spending a ton and ton of of basically economic value on verification we're able to

cryptographically prove that based on these randomized uh proofs that this that this entire blob of data for La a better term is is accurate and I'm starting to realize a lot of this modular blockchain stuff and especially with da it's about Not Giving into the typical trade-offs that would come with just using ethereum as your um as your GA layer which comes with a lot of costs but you get that you know really high level of security but it's finding where you can kind of mitigate the trade-off of security using something like a like client cryptographically proven um proven Globs up dat for that for for da layer but then also then by doing that you get that exponentially lower cost on

the on the flip side so it's almost like we're like reshaping tradeoffs here in a way where where the we we're trying to find the exact right formula to make it uh the best of both wordss yeah that is true and I think uh it has to be understood that ethereum is by themselves in a DA Centric road map right so they also want in its final shape and form want to be a DNA and that is that is the trend that you will keep on seeing for many base leers you have started seeing all many many l1s or at least one or two of these l1s trying to be a DNA because you will keep on seeing this trend l1s will either start

becoming n2s you have already started seeing that and l1's start trying to becoming data because in order for for a Bas to be able to scale they have to be a DNA because the da is the only thing that scales compute doesn't scale in the similar matter right so that's the kind of PIR off that you will keep on seeing on the other and like it's very important to kind of understand where the cost reduction comes from it should be very clear that this is not a old n one kind of a thesis right that you know I had ethereum I create a clone of ethereum of course the clone of eum is going to be cheaper so just use me

because I'm cheaper but you have not you don't have activity once you have activity you become the exactly the same amount of cost that ATM gives you right so there's a lot of thesis like that I want to I want to make it that clear because often people mistake it like that the the actual cost optimizations comes from the actual code base in if you make a system optimized for a specific job of course it will be more performant and hence it will be cheaper ethereum started off with a very computation Centric road map and only now it is trying to shift towards a dcentric road map because it is inefficient for handling

da you are seeing today like 1,300 to $1600 or even beyond for a single NB of qu it that is because you are you are that does not optimized for the and aail is cheaper not because it doesn't have activity of course it's not in production so I cannot talk about act numbers and and transaction call that will be tunish for me to do but it will probably be orders of magnitude cheaper because it's orders of which are more efficient at handling G right so that is a clear demarcation that I want to make so that it becomes very clear to the end user that what this thesis looks like that we are a scal DNA and the other thing which people fail to understand is

that the super power of Da is that it is able to scale to a bigger block if the demand increases that's that's almost never possible in a computation centry board so for example if I see that my in a typical blockchain if I see that my Glocks are getting filled out I cannot double my space because to double the amount of computation I need double the amount of resources for the validators of course we have seen those kind of things where validators are extremely GI up in order to give High TPS High numbers and that's okay that's also okay because who am I to make that trade off if that's that's a UTI utilizable chain which has a r prob by itself but what

I'm trying to say here is that DNA lers can actually make their blogs bigger without incurring a double cost on the on the validators because the verifiability depends on the amount of light clients and the amount of utility if the utility increases that that's more people are using it then more light clients will run and when more light clients run then then you can have bigger block and that's extremely novel in this use case which not the other kind of you know monolithic chains can give you yeah I appreciate that um yeah that that was well over our head with regards

to the differences between computational Focus chains versus data chains but that actually makes it much more understandable as to like why like something like a you know a just simple like uh chain like a like an app chain on Cosmos or a or even some of these other you know potential da layers are able to then scale effectively what what you're saying is say you can't just spin up another chain that as no activity be like look at me and this has been something that I've been focused on when in in my writings recently which has been the value pral thesis for this world so so so we've been talking about a lot and I'm I I'm very bullish On Da because of all of this uh this last 10 minutes of conversation but one thing

that's interesting is this this transition from ethereum the computational heavy road map to now a data focused uh road map and I feel that it comes out of fear why do I feel that it comes out of fear because I feel that fear is somewhat threatened by the modular thesis because of the the valuble of its importance at the da and at the execution layer I find it hard to refute theorum as the main settlement layer I think that ethereum based robs at settlement level is hard to argue that there's XY Z L1 that can be better and perhaps you have an answer for that but is is there is there a sense or a smell in the air that the modular thesis you know replacing with ebm or using

alternative VMS as well as using alternative Das is a is could be a problem for the speculative or perhaps just for the the value approval of ethereum in the future and perhaps some of that value which may approve uh could approve to these modulators I'm kind of setting you up here but this is something that I've been kind of thinking about um I think there are two ways to answer this right and I will try to be partically neutral here by mentioning both the case both the both the ways and I and I only believe that there are both sides to the story right on one side is the dominance of evm the dominance of evm is is pretty clear right now but my

bet is that it will be broken not not because evm is not good but because there there will be used stases which can move away from evm and we'll be able to tiddes evm doesn't allow all kinds of operations that uh a rollup would want or an application would want and we are already seeing that as part of our exploration team we are trying out different use cases which were not possible in an evm Centric world and they are very bullish on the app specific rollup kind of pieces right where they believe that people are going to you know run a lot of these applications many of them will be custom many of them will inherit the best of

many many worlds which we have seen you have seen things like atos and so you have seen Solana as we we have seen evm but we have also seen things like Risk zero based VMS which are like just ZK verified of any stfe transion function and so on and so forth and there are many good players which are com in which are making really good headways in terms of like this P right which is a very novel structure in terms of the virtual machine and so on and so forth so that is something that we keep on seeing so a domination of ethereum in terms of the ethereum virtual machine is probably at risk and yes there might be a fear that

other things will take over and and that doesn't mean anything for the ethereum because ethereum wants to move away from from the execution layer and evm is just the St of execution there the second thing is the Ethereal user or the liquidity or the value base right so ethereum is already looking to be ultrasound money and and I don't think that that is at risk because ethereum still AVS to be the sement layer which means that it will be the value layer right so in that kind of positioning I don't think it's risk at this point in time at least I don't know what how things will P out in crypto it's very

hard to determine kind of or at least predict futurism but that part I'm un R very sure so if someone is worried that ethereum is losing its position in terms of Da yes but he has the da Centric road map I 4844 is coming after that the dang shotting proposal is coming so we have a p Road up towards a data kind of solution um they are still going to be the dominant settlement layer at least for the for future as as I see it at least and then there are EDM Centric like the execution engine which has been a lot debated a lot right till now and

those are the kind of things where there is a risk but that's a good thing because Innovations are going to come in new things are going to come in and that's how we will will keep on um going ahead but e has already talked about not execution sking so all of this just adds to its own te I think the uh it's a spicy take and it's probably 30 years too early I think the only L1 that could overtake or Dethrone ethereum settlement is Bitcoin because the cost of corruption is higher but we'll say well I mean go ahead if you have thoughts no I think I think think it's it's uh like again as I keep on saying

that we have to understand like what you said is absolutely correct like that the cost of corruption right so at the end of the day it will come down to cost of corruption but what is it that you're corrupting also the definitions will change so Bitcoin ethereum they have actually shown us the way right the Bitcoin has shown us the way how to even use blockchain right so I mean still kind of sometimes open up the Bitcoin paper just to kind of you know refesh my Nary that we shouldn't forget how how distinctive and how novel that entire paper was and then e theum showed us that you can do so much more from this

entire technology so again that is a huge kind of you know Headway that we have made from the D um but when it comes to technology the only thing which we have to go on and do is that keep on understanding the emotional changes that we are seeing as things change around us and for the cost of corruption right now it's huge right so on bitcoin it's huge on etherum that is huge and that's great to that's great to know but there are the other constructions where the where the act of corruption is not has limited impact I say so for example like if you if you today

completely reorder there's a lot of user value at risk so even in DN the corruption can can be extremely extremely dangerous right but at the same time just because you can corrupt a chain doesn't mean that all the exchanges of the world will also accept it and so on and I don't I'm not a very big fan of you know whatever I cannot handle the protocol let's throw it down reveral consense of that kind of a world but as long as like protocol guarantees are incentive compatible it's very hard to corrupt aching and as long as we continue to do that then we would have a very very secure system because most of

the security will come from cryptographic and the Le less Reliance will go on to crypto economy yeah that was what you were saying earlier it's a shift modular blockchains they bring a shift from Economic Security to Crypt graphic security which is which is which is very interesting um but I can't help myself because I'm really curious about the value of crel in the system so let's forget ethereum now we're looking at the stack and the the typical stack is You Know da BMS and settlement so forget the settlement layer for now we've got da and we got BMS I think there's I think that stack is expanded in first half of 2024 to Da BMS interrupt and sequencing and then perhaps more you're shaking your head so let's start with the first

two perhaps how are we thinking about the value of FR here for each player in in in the stat I think you hit a very interesting point is that the value approval needs to be extremely well understood in order for each of these modules to be self sustaining right um there are different trade-offs there and I keep using the word tradeoffs as if to evate questions but really I don't want to use it that way but let's think of specific parts let's say that you have a sequencer right

sequencer is extremely valuable business because it can capture M now that kind of thing probably no one talks about today but the centralized sequencers can capture a lot of energy on the market and they don't need to be bootstrapped even today uh typical production R sequencers captur you know T to hundreds of mil dollar per year so that's a very profitable business to run a sequence even without any May is something that we are not able to see on the balance sheets right so that is that is a very profitable business now for the Prov the Prov is

the part which keeps the objective fin align you need only one correct Pro in order to you know secure the ch right because once the sequencer comes in and publishes a particular batch is set in stone right so either someone comes in and you know completely completely accepts it or you just you just run it to accept it that you know that it is correct so for a pro word there needs to be incentive for someone to do it just someone to do it so as long as there is at least one person who does it it can be altruistic it can be incentivized it can be coming directly from the users

even there are better constructions right now for example that are not in production today but I think in 2025 we will start seeing things like clients side proving right where the clients themselves will proove transactions and then the operator really just kind of you know aggregate proof and so those kind of constructions we will keep on seeing and those kind of uh where the value prop kind of changes the value actual kind of changes because users then are are more self sustained in that manner the operator has a very seatless kind of a role uh in those kind of circumstances but that's something that we will keep on seeing um in terms of again show sequences and such not not trying to

tell about them separately because again they they capture a lot of every and they and give out a big big part of it back to the protocol and things like that um in the end the DNR is where the transaction costs come in right so the the DN is where the batches get submitted and that's why it has a it has a solid Revenue stream that way as well um so overall right now what you see is extremely well deated in terms of value user plays a transaction fee to Operator Operator either extracts more to the m or not typically they don't today at least that's how I like to

believe it then take that and put it onto the DNA for that you have to pay a fee to the DNA uh the DNA is where you D part of your security from so that's the kind of you know subsid like that's a commission that you have to give to inherit the security so to say there's a pro which can be incentivized directly from the user parts of it or through the operator or can the alistic or can be a competition based system where there is a definite reward based on what you are looking at proof of efficiency there there is alteristic behaviors other kind of you know typically a lot of research has gone on to that kind of system and then they generate the rewards it can be

like whoever does it Forest gets most of it or it can be all the people who did it well can get pie of it got are problems with either one of them but there are solutions as well so you see that the all the pieces stay now that I've talked about have a solid you know way to approve value there are other things which are yet to be kind of you know seen for example uh there are separate Services which run let's say you are running an RPC RPC are incentivized from outside the protocol right so the users pay the rpcs for API requests and so on

sometimes rpcs the subsidized costs for others and so on so even when you use metamask you are using an inferior RPC your cost is being subsidized so those kind of models are also being seen um there are there are other models which will be seen in the in the coming future which is that a service will actually subsidize the cost of user transactions uh so as a seing Ser they will not take user fees they will actually subsidize it themselves because you would want the first class web to experience of rep three you wouldn't want a wallet you don't need a wallet for for you to actually um you know sequence a set of transactions so a lot

of different types of uh use cases are going to come up and we will keep on seeing how each one of them you know have Val that even I am I'm also open and looking forward to sometimes an ad based uh web three fun and maybe that's also that to see yeah yeah couple of things first say goodbye to those of those OP retracting grants from sequencer fee boys because those are being subsidized back to the users so enjoy those while they last um the other thing is just to confirm when you say where a transaction goes from the user pays the fee to the operator you're referring to the operator as like the execution layer right the uh the ex environment like that so whatever DM

that is so that's where things start and it kind of trickles down all the way into the whole stack so realistically in the modular Chains It's it's every part uh approves value it just depends on like you know the sequencer theoretically approv more than the rest they do take a ton of me but again if they if they're subsidizing transactions and then there goes a lot of that kind of uh revenue or profit uh margin so basically what what you're saying is as long as we have users clicking buttons each part of the uh each part of the stack is is then at least generating Revenue to which then uh the expenses are obviously laid out on the balance sheet and at the end of the day we're likely hopefully in times of a lot of activity running highly profitable

businesses on chain which is super exciting um my thought on the rpcs is this might this might be a tough answer but you're our da layer you want to give the best experience for all users who use da uh Avail bit packed da uh layers are on Roll so you want to if you're working with uh gelato alt layer n op whatever it is you want to get those users the best experience but there is now 50 rollups there's s 500 modules who's solving the RPC problem for all of these who's paying for that and who in their right mind is indexing 500 yeah you're you're absolutely

correct and that is a problem that we think hard about and that is why you will keep seeing our efforts in the coming days in order to solve that challenge um now to kind of give some some of the insights on how we are thinking about it again these are not solved problems but let me just try to give a few ideas uh that we have been thinking and then you can also think about it the first first is that rpcs right so the one of the first and foremost things that we are trying to do right now is make our light clients the first class citizens of our Network without relying on

rpcs now we want to have minimum Reliance rpcs our light clients and that's the kind of work the the bulk of work we have done in the last two years is on the peer-to-peer L because you know creating a system or creating a blockchain is not hard part the hard part is creating a peer-to-peer n which is stable which is which is you know dependable and which which is performed as SP so what we are trying to do on the light client side is that we are creating a network of light clients all of which contribute to the security and the availability of the network so every time a light CLI samples it not only samples and just check it themselves but

also puts the samples out there on DHD so that the next guy coming in doesn't have to pull from it RPC it can pull only from the peer to-peer layer even going one for one one step up we are trying to populate the DHD with all the proofs needed so that RPC R goes to a bare minimum only the header features can be done from the RPC in the future we also have plans to mitigate that to in order to the life client to only have the RPC as a backup like only when the entire peer-to-peer phase is when they go on to the RTC so that's the kind of things that we are trying to do on the da layer on the execution layer

you are absolutely correct that who in the right m is like interacting with thousands of chains and still being able to you know keep track of all PCS what is down which change is up what is happening have a soft confirmation hard information what is what is there and in those regards what we are trying to create in in aware is we we are not only looking to build just a DNA we are we are we are looking very hard at how we can create a very you know composable kind of ecosystem around it D become the centerpiece because that is the that is

the where where the security comes from so the is going to be our our main focus as we have seen in now but in this world we want to make sure that the experience that the user has is the the the chain remains on a smaller fraction of it and how we are trying to do that is for example suppose you you interact with Amazon right and you it's not that Amazon has Payment Systems in what Amazon does is that when you click the Buy button it sends a call to a payment aggregator the payment aggregator doesn't have um any of your credit card information actually the payment

alligator sends that to someone like it's a Visa Visa doesn't know your account balance so Visa sends that call to someone like your bank your bank is then fulfill the request it gives back a call to the to Visa Visa back to your payment agregator payment agor right and then they they give it back to Amazon as a user you didn't interact with all of these ch right you didn't interact with all of these applications to be honest right of course the URLs changed but you didn't put those URLs and just like that we are trying to scan this scan this entire thing up using asynchronous

message passing between rollups and that is something that we already have a p we are not just talking about a thin air we are we are working every every day to make this a reality so that as a user you would only want to give what you want to do and the rest of it will be chain to chain level interactions which the user don't have to do go and do so for example there might be a payment chain there might be a separate nft chain and all of these are application specific in in our world view what that means is when it's application specific indexing becomes less of a challenge because what is indexing right we have a

shared you have a shared team and you need applications specific information and that is when you have to index it because you don't want this entire thing to come to you you want to filter it and take only parts of it which are relevant think of a world where the applications are only serving one right one the change are serving only one application then that word in that word indexing is the less of I'm not saying it will obliterate the need for indexers it will change the world how we see it not not like that it will happen Delta by Delta right but we are already working on on on features like that where you would want that okay if I want an NFP I check on this chain if I want uh to know my

account balance my accounts remain on this chain and then I can interact freely with one and there can be three or four other CHS which serve only one purpose but you don't need to know it you just need to know that I have this bank card and I want this product from Amazon everything else will be taken care of and that's the kind of power for this rollup Centric ecosystem that we are trying to build asynchronous message passing enter interop into the conversation like that that's where this comes in and that's like the piece that we haven't talked about so much but we had a a great podcast yesterday with hyeran uh who's building permissionless interop in

the rollup space um and kind of open our eyes to this RPC problem but didn't have a a particularly great solution to it I think the asyncronous message passing that you described uh is a very clear tangible visualization of how we can picture how these messages pass through and through and just we we were on a space like not that long ago and there was some conversation between uh Hart lamber from Uma and Susanna Evans from IBC talking about how right now value transfer is the only use case for message passing across chains but what you're talking about isn't value transfer across Chains

It's generic message pass it am I could you kind of like just illuminate a little bit on whether or not the messages that are getting passed across rollups include any any value or is it simply the message that is getting passed it's simply messages like it's simply message just and you can you will be able to see more and I just don't want to paint picture in the air because I I hate when people just uh talk hypotheticals but uh but what it will look like is even you will only talk about intent I intend to pay up to this much in order to get an asset right and then there will be a chain which will

run a solver which will do the solving then then will be another chain which will find the AC chain where the asset is in the minimum price or something like that and then they can go to that chain ask for that asset and then they realize that oh wait I had uh the the user had the funds in usdc but the asset can be only bought me so let's go to a swapchain where there is enough liquidity that I can get the best price and the lowest lick rate let's go and convert it let's convert it take it back and then go to the asset buy it get the rest of it because that can be value can be changed so change the rest of it back to usdc because the user had that and then propag that back

so it's all about messages at this point and promises and Promises getting resolved because we have seen that in the web to right and of course we can sit here and criticize a lot of web to design decisions but there's a lot to learn from it and one of the ways in which web to was able to skip was through this ASAS message request right and the promises and resolutions that you see today on all AP Calles the the entire Amazon to your your band story none of that was sychronous none of that was done in a in a r where there was a s assumption right every one of them were promises and then call backs so or timeouts right so those kind of

constructs we are trying to create but the main refence being that this is going to be trust minimums in web New World all of these were trusted like any one of these calls were becoming a call to a trusted entity in in our sorry in our case the main Delta the main difference is that even if there is a transaction inclusion you can give a proof of transaction inclusion if there was no transaction inclusion you can give a proof of non-inclusion so all of these the relationships are not about trust it's our verifiability and that's why verifiability is so close to our hearts but none of this actually work when you do not have a single ner

because as soon as you cross da boundaries your security tradeoffs actually change once you cross da your security Zone you have to cross it and that's where you have to rely on someone else's cryptoeconomic security for you to cryptographic inates and that's what the kind of world you want to avoid so we don't want to build we don't want to call this Bridges right but they are essentially there they are some form of bridges but with much lower security uh asum like plus asumptions yep yep Sovereign interoperability something that you'll hear over the coming months I A little birdie told me about

it and um we are uh we are coming up on time here and I think yet the most important question for you my good friend the starts with a w sounds like then and the question is when main that my friend when so and you as Andy you of course know that we are running our incentivized test net this is the final test net that we are going to run before mainnet our mainnet plan is q1 next year which means it's just around the corner uh we are doing final rounds of stress testing different components different challenges um any of your listeners are

fre to join in that is challenges for every kind of users we are trying to accommodate as many validators but there is also simple tasks like you know balance transfer and there will be other tasks about light client running of light clients and so on but essentially the target is just to understand if there are any vulnerability anything that we have missed out any last minute Buck fixes and so on that the the feedback that we have got until now has been awesome uh and this is the last mind and the last SL is always the toughest has been so we are towards we very close towards menet qy1 is when we

would launch our menet baring anything significant that we come across under in tet of course but that's the that's the overall plan Bey thank you braal for the time and for the uh di fency the weeds the modular weeds it's always fun cheers F thanks a lot for having me uh it was great chatting there were some topics which uh were breath of fres a because no one really talks about them so thanks a lot thank you for having me