How ZK-Proofs Will Change Crosschain Interop Forever | S7 E5 | Uma

Transcript

there's still going to be many layers of a stack but I just think ZK is going to be like the base glue for everything and like basically yes like having truth communicated everywhere and hopefully that leads to like better security like no more hacks and like leads to like oh there's not rollups that are just multi with like no fra proofs and stuff it's just like everything hope which are you talking about we don't name names we don't name hey everyone this episode is brought to you by say the Blazing fast parallelized blockchain which is unlocking salonik performance for the vast ocean of e devs out there now you're going to be hearing all about say and their new V2 upgrade but if you take away one thing evm is here to stay there are some problems with it which we're going to get into

later in the episode but say and especially their V2 upgrade is helping solve that so thank you very much say for making this episode possible hey everyone this episode is brought to you by Unis swap delivering the best onchain trading experience in crypto period Bar None one thing I want to call out is the Unis swap extension so say goodbye these days are these annoying sort popup wallet extensions you lose your place trading you have to open it back up unisoft now has a nice Sleek sidebar that persists no matter where you are in the web it's much easier to use if you click the link at the bottom of this episode you can join the waiting list and I'll see what I can do to get you moved up that waiting list but definitely go click the link check it out all right everyone welcome back to

another episode of bell curve before we jump in quick disclaimer the views expressed by my co-host today are their personal views and they do not represent the views of any organization with which the co-hosts are associated with uh nothing in the episode is construed or relied upon as Financial technical tax legal or other advice you know the deal now let's jump into the episode hey everyone welcome back to another episode of bell curve today Hart and I are going to be interviewing Umar Roy of succinct and we're going to be doing the long- awaited much talked about already on this season zero knowledge proofs episode now this is a really interesting episode because as many of our guests have already brought

up this season ZK has massive implications for cross chain interop and so we get into all the details about how that's going to work with Uma and unpack the implications and for those of you who pretend like you know what ZK actually is but you don't AKA 95% of you we spent a good amount of time in the beginning of the episode just reviewing what ZK is uh and talking about the ecosystem which is super super interesting so really excited to get into this one with um and Hart thanks for listening guys all right everyone welcome back to another episode of B curve today Hart and I are joined by Uma I'm succinct Uma welcome to the show hello thanks for having me yeah we are we're really excited for this one Hart and I have said I think at the end of every single episode so far this season

all right we need to do a ZK episode and there is no one better to do that with than you so I'm GNA start with uh you know um I'm gonna give you the caveat here that we don't um like treat me basically as you would a five-year-old who knows very little about zero knowledge that'll probably be about the right level um actually as a funny story flying F back from eth Denver I told heart this so I was flying back from eth Denver and I was like I'm G to do some research for this episode with Uma and I started off by reading zky actually put out a five-year-old like picture book description of what ZK Tech is and some guy was reading me saw me reading that and was like hey are you guys you trying to learn about ZK Tech

and he had led the investment into risk zero and I talked with him about ZK for the rest of podcast but or the flight but still pretty embarrassing out myself a very smart ZK guy you know Uma I won't dox him but it was just very funny he's sitting beside Mike watching his picture book and being like hey let me actually tell you about ZK stuff uh that was his flight that's so funny yeah um so with that Uma can you just give us at a high level like just explain to us like what is like what are ZK proofs if you had to describe them yeah so I guess at the highest level um a zero knowledge proof does two things uh uh the first one that I'll start with is like I have some computation or some function and I want

to prove to you that when I run that function on a certain input it results in like a certain output so as an example it could be like when I do this hash 500,000 times on this input it results in this output it could also be like a state transition function where when I take a state route and a bunch of transactions when I run those transactions it results in a new State Route so I want to prove to use some computation and you want to verify it without having to redo the computation because the computation can be expensive a Zer knowledge proof lets you basically prove to someone else this function when run on this input results in this output uh and then the key part is it's succinct which means that like when you verify the proof it takes much less time

than it would to actually do the original computation got it so on that Uma is it like accurate one thing that somebody told me that I I think makes sense but I want you to verify this is to think of it like a compression technique in a sense where I'm compress compressing this giant computation into something I can run in linear time um and that's the succinct this property so first of all do you agree with that yeah I mean except it's like constant time but yes so even better okay yes so I have a constant time proof of this incredible like it could be an incredibly complex computation doesn't really matter how big the computation is and I can compress it into something

that um is easy to verify and um that's the succinct property which is the name of your company cin Labs but also why don't you talk about like the zero knowledge bit too because like I think for our when we're talking about ZK for interop I don't think we care about the zero knowledge part I think we care about the succinct property um but like we might as well mention it because we're on the Pod right yeah that's exactly right so the other thing I was going to mention is you have the a signus property and then the other thing you can do is you can prove that I have some function when applied to some inputs results in an output but the inputs can be hidden to the final person so as an example of that you can say

like I know the password or like the key to some something I'm not going to tell you what the key is so I'm going to hide that as part of the inputs but then it results in like some output that the verifier can understand and so that is like the Zer knowledge property is the verifier doesn't need to know what all the inputs are and Uma could you just briefly explain the concept of circuits in there as well and what some of the computation is honestly not even super relevant for this for the what we're going to talk about here but like I'm just just for the audience I'm curious how it works yeah so people talk about this concept of circuits in Zer knowledge

proofs um basically if I want to like take well okay historically if I wanted to write a computation in a way that I can generate a zero knowledge proof of it I'd have to write the computation as a circuit a circuit is just like a fixed computational graph of your logic um and it's like a concept that exists even outside of CK like there's like you know in E there's like circuits uh and things like that um but basically to generate a Zer knowledge proof you need to like format your computation in this like very structured way and that's kind of why you have like all these dsls and sdks and like all this like specialized stuff for generating Zer knowledge proofs and so a circuit is just like a fixed computational graph of your

computation that is an easy form to generate a zero knowledge proof of if that makes sense yeah I think this is where the circuit stuff is when we start getting into the moon maath but I think at a high level right for this pod mik there's this way that historically um tell me is correct there's this way historically you need to um describe the logic of your function in order to be able to generate the ZK proof um yes and actually maybe that can then lead into kind of the frontier or state-of-the-art of a lot of the ZK stuff is abstracting away that circuit generation or making it so the average developer doesn't need to build circuits anymore right I think I think

you chanted death to circuits I think you did that on a panel with me I did I did yeah so historically basically if I have some function say I could describe the function I could be like Mike I want to prove that when I take some number and I hash it 50,000 times it results in this answer for me to actually generate a zero knowledge proof of that function I can't just like write it in normal code like I can't just use a normal out of thebox hashing Library what I'd have to do is I'd have to write it as this like circuit um in like some sort of zero knowledge proof library in order to be able to generate this zge proof and that'd be really annoying like if you look at circuit libraries today it's really annoying you have to like basically rewrite your computation in

this like fixed graph it it's very bad basically as a developer experience and it takes a lot of time and it's it's very intensive and then with the frontier of today and especially one of the things we're building at succinct is this open source ZK VM called sp1 but what that lets you do is instead of having to like write all this handrolled circuit stuff you can just write your normal computation as you would with like normal rust code and then you can generate a proof just based off the normal rust code so that's kind of like the frontier of where Zer knowledge proofs is heading is like instead of having people do this really specialized moonm maath stuff you just write normal code and any programmer can use it so well I want to start to transition here I think and and a sort of a

powerful um effect of zero knowledge proving as opposed to the way that Computing is worked historically in crypto is where the computation is taking place so historically the way a transaction gets submitted to kind of this public blockchain gets gosp around in this mempool a bunch of nodes within the ethereum network sort of take those up execute them package them into a block but everything happens within this one virtual machine and one of the side effects of what we're describing here what you just mentioned with what sp1 allows you to do is it allows you have many different virtu like machines or sort of computing environments a lot of computation gets done offchain and then the proof is submitted onchain and one of the uh one of something that that allows you to do is to knit together

these very different execution environments which has historically not been super easy to do and that's where I think we want to drill into sort of the interoperability piece but be before we even get there I would actually just be curious to understand you know there's a lot of new infrastructure when it comes to this zero knowledge space they're approving networks U there's like new types of applications um that leverage zkps so can you just kind of give us a sense of like what does the landscape sort of look like what are the different buckets uh that exist in this new ZK world M yeah so there's kind of like the app layer that uses zon knowledge proofs and I actually think there's many different types of applications and infrastructure that can use ekps so I think for example for bridging like all

bridges can use their proofs and maybe we'll like dive more into that later considering this is like focus on inter op I think all rollups I think people are probably really familiar with ZK rollups I think all rollups can use zero knowledge proofs um as like a validity mechanism um then there's like other zero know apps that are more focused on like the Privacy aspect so I think tornado cat is a good example but then there's like other ones um for example like identity applications where you kind of prove like hey I belong to this set of people but I'm not going to tell you exactly who I am and that's just like a blockchain application um so not in the infrastructure category so there's like all these consumers of zero knowledge proofs who use them for like validity security or like privacy or whatever um

then there's like the actual proof system um that basically people use to like Express their function that they like care about proving so there's like various different proof systems and then there's also like ZK VMS where instead of having to like write all your stuff custom in a proof system you can just like write normal code and then have it proven in a particular zkm then there's like the proofing networks um or infrastructure to actually like generate the proofs so it's kind of like you're an application you write your zero like you write your zero knowledge comp like you want you write your computation that you want proven in a zero knowledge proof in a proof system or in a zkm and then you actually have people generating the proofs um and so that's kind of where proving networks come into

play is they like connect the people who are consuming the proofs with the people generating the proofs and the people who are generating the proofs right now it can look like companies like us like we just like run some Cloud provs on AWS but it can also in the future look like Hardware teams who are running who are creating specialized hardware for Zone noge proofs that like can really accelerate the proof generation time and then they can like basically be the suppliers of proofs and then the marketplace layer or network layer is like basically I see like the shelling point where these proof consumers and applications request proofs and then you have like the proof fulfillers who are generating the proofs who are like the hardware teams or people like us with

our Cloud infrastructure actually generating the proofs and fulfilling them so that's like kind of like top to bottom the layers of stock that makes sense so to summarize this so your applications two use cases privacy use cases and um I'm going to call it like compression or succinctness use cases like scaling scaling use cases scaling use cases so and again going back to even how you're explaining ZK for me personally trying to understand how the zero knowledge works for the Privacy part that's what makes my head hurt like I still have it it's still hard for me to understand how you can prove something without tell anybody about it right but I I I kind of but anyway so there's the Privacy use cases which for rpod and for the multi-chain kind of endgame we're not going to talk too much

about and then there's the scaling or proving a complex computation succinct property that we are going to talk about okay so those are the applications and then you're proving system um is it right when you talk about a VM is it right to like if I make an analogy for our audience it's it's like are you using like Linux or Windows or like the operating system you're picking right and here you're specifically talking about the math or the the math framework that's being used for the CK proof yeah yes yes exactly okay and then once you pick that you've got to actually do the work to generate the proof and that you're calling that proof generation and then there's this idea of pro networks where effectively we're gonna have different provs in essence competing

kind of like a Marketplace um to see who will generate that proof H and then you have your proof and then you can go and use it yep yeah that sounds about right cool so I have a question about the pro networks in general and I what I ultimately want to understand is like the cost of generating of like proving today and like where that I I understand that that's trending down but like as specific as we could sort of get about that I mean are who are the types of entities that are running these proving networks obviously succinct um is running one are there going to be eventually like almost chorus one or figment like entities that sort of spin these things up is it actually going to be like chains themselves like who ultimately runs these sorts of things well I think it's not like we are

running the approving Network I think it's like we are designing approving Network and protocol and then there will be entities that participate in it so for example it's kind of like you wouldn't say like Celestia runs Celestia Network you know like Celestia Network's a thing and then there's node operators that participate in that Network and then there's like users that participate in the network and like post blogs so it's very similar there will be like cinct Network that's like a decentralized protocol and then on one side you have like node operators who are like running provs and yeah I think like figment or course one or traditional node operators like maybe they could get into ZK proving and like start running provs on the network and they will get paid for generating proves

and then on the other side you have like applications and users submitting proof requests to that Network and being like hey I want this proof um so those are kind of like the two participants in the marketplace it's like the applications who want proofs and then the people who are actually like generating the proofs got it and they could include like traditional node operators for sure he everyone want to give a big shout out to today's title sponsor SE now I want to talk to you guys a little bit about why I think say is cool from a design standpoint a big problem that it solves for E theevs out there and then some cool stuff that se has coming up the reason I like say from an architecture perspective is again it's a very fast blockchain parallelization all of that stuff but say has essentially been

custom building block space which is for Consumer apps and Dees now they have some very cool features which enable that so TN turbo consensus optimistic parallelization say DB all of this stuff allows you to reduce the time to finality make for very very fast transaction so if you're building a consumer app or a DEX this is basically the blockchain for you if you've been building the evm you love the evm but there are some restrictions about it that don't support your app so maybe you can't do fast enough transactions or it's not parallelized whatever it is you can now take all that stuff that you built you don't have to start from scratch and you can build it on sale now recently they've launched V2 but also public Dev net so the way that you can follow that and keep up to date is go

and follow say Network on Twitter all right thanks guys what are the economics around proving what what does that look like today yeah so I think there some ballpark numbers I think right now for example um we can just take a few different applications so one popular one is bridging uh at the SN we've written a tender A ZK tender like client where basically we verify like that in a tender consensus protocol enough validators have signed off and that lets us do like trust minimized bridging between you know networks who use tenderman um like Celestia or you know dyx or all these other people to

ethereum or something like that and so the proof generation cost for one of those proofs is actually really cheap I think it's like a couple cents if that um and so it's very very inexpensive um today just in terms of raw compute cost to generate a proof um so wait um that's cheap enough that I want to like push and understand that so this is like to be clear this is generating um a ZK proof of a tender bin consensus um and you're saying that the computation cost of that are like couple cents which is awesome um for for context to kind of share with the audience how quickly this is developing like a year ago what do you think the computation cost would have been on this

or yeah like a year ago to be honest was like kind of impossible like it just the circuits and like all that kind of stuff didn't exist um and then yeah we were like one of the first teams to like kind of put all put all of that together and like get it working end to end like I think a lot maybe a year or a year and a half ago people have thought it's like impossible like they would thought like oh the proof generation would take like hours or something like that um I think our proof generation time takes like five minutes um it is on a pretty big machine but like with Cloud cost the machine probably cost like you know a couple bucks an hour and so the costs are very cheap it's like you know a couple cents so and actually we should

go into this too as Mike's pushing on the economics which I really want to hear you describe more as well um it's proving time is directly related to computation costs basically like how long you got to run this beefing machine for and how much that costs so your two cent proving time here is like five minutes on a beefy machine in a us right now yeah yeah that's how we compute cost it's like it takes five minutes to gener your proof The Machine's like pricing is like two bucks an hour you just do like that math other examples people might care about is for a blockchain for example proving uh a state transition function of for example a ZK evm uh recently with sp1 we released some numbers that showed

that the proving cost per transaction and again this is like the same methodology where like you take the proving time on a BP CPU and you just say like the CPU is a couple bucks an hour and so you do that math advertises to like around one to two cents per transaction so like the proving cost to prove the validity of a transaction and like it State transition from like the previous state route to the next state route comes out to maybe for an average block around like 1 to two cents which is already super good because like for normal transactions ethereum the ethereum data availability costs are like at least 10 to 20 cents um and so yeah the proving overhead today already is like pretty cheap uh

for transactions and yeah only going down can we though just to be complete here um once you generate the proof you also have to validate or verify the proof on um at least on a block if we're using it in blockchain context we have to verify the proof what does and I mean this very much depends on what chain you're on but like um what does it roughly cost in dollars or maybe gas units to verify these proofs on different chains yeah so I mean the gas costs are pretty negligible on all chains except for ethereum uh but on ethereum May it's like 250k gas uh which yeah I mean it depends on the eth price and things and the gas

price and things like that but it's like that's like a now it's like more than 100 200 bucks I think something like that that sounds or I don't know I think gas might have been a lot recently that sounds kind of high but I think back when I was looking at it was maybe like $20 um yeah maybe recently it's gone up because eth went up obviously and then the gas price has also gone up yeah um and and in terms of on a layer two is there like a lot of call data or other costs associated with with uh validating one of these proofs um the call data is actually pretty negligible on a layer or two like that probably costs like back when we were verifying proofs it was

maybe like 20 to 50 cents as my best guess don't quote me on those numbers but I think it was like in that range where you just pay for call data obviously computation on a layer 2 is like basically free um so yeah it wasn't too bad um but I think there's this kind of myth I think for ZK that like oh it's so expensive it's going to cost so much money and I think historically that has been true but just like every year things are getting orders of magnitude better and like the cost curves are coming down so much and so I think even today with the numbers we have have like it's one cent or two cents a transaction like that's already awesome like that's already something that's very feasible

right now for tender Al also it's like a couple pennies per tender proof so I think there's this kind of big myth about zek that it's like too expensive and I think that's just like not even true right now and I think there needs to be like education done on that and Uma isn't aggregation a part of this story as well like can't you aggregate some of these proofs together and then that ends up making it cheaper oh yeah 100% yeah so as Hart mentioned like if you want to verify your proof on ethereum today it's like pretty expensive but what's awesome about ZK is you can verify a ZK proof in a ZK proof so we call that recursion and basically all ZK systems use that like rollups use

that a lot even we use that for our bridges and stuff like that and with recursion you can just like compress an arbitrary amount of computation into one proof and so so for what a lot of like Layer Two Roll-Ups do is like they'll have maybe one proof for Block then they aggregate all those proofs across a chain of blocks so maybe like every 3 hours they'll aggregate all the proofs they've generated across those three hours into one proof and then they'll verify that proof on chain so you only pay that amortized like 250k gas every 3 hours or something um and in general like you can actually aggregate arbitrarily many computations into one and so I think in the end the amortised

cost for aggregating a proof on ethereum should basically go to zero because everyone will be aggregating all their proofs together yep that makes sense and I just double checked so 250k gas is like 70 bucks right now oh right now oh wow what is the gas what is the guay right now um it's like 80 guay okay e price has gone up like you know um but like I this old 70 bucks amortized over many transactions is completely reasonable right yeah totally and like this is the way a cross works too right like we batch together all of our intent verification um and amortise it over hundreds or thousands of transactions and makes it cheap so I'm a big big

believer in that concept um I did kind of think it would be worthwhile though going into some of the limitations of ZK here from an interrupt perspective too because I think there I have come to believe that this is like incredibly powerful and transformational Technology but I also think at times people think they don't they don't always push on the limitations uh of what it allows us so I was thinking through some examples that um I was hoping you I could walk you through and you could tell me where I'm wrong or help me make sense of things so proof of work blockchain proof of work blockchain is basically one giant

computation and so it any point in time I could do a ZK proof of that whole computation up to the current state route of that blockchain right but the problem would be I could also then mine one additional block that I mine that I uh control and do bad things in and as long as I can generate one additional block I can make a ZK proof on that one additional block right and then my ZK proof would be valid but it would be for a bad State Route yeah yeah so for example if you wanted to make a Bitcoin to ethereum like client using ZK uh you have to be kind of careful because as

Hart is saying like I can just mine like my own Fork of the chain that's malicious and then what happens is when I verify that on ethereum like sure it is the longest Chain by amount of work um but it's like not really the valid chain because no no one else is like updating that light client so if you have just a like longest chain based light client on ethereum for something like Bitcoin there's like a strong liveness assumption that like that like client is getting pinged Often by many different people with like the latest blocks and the latest amount of work um so that you can't just have this like malicious fork or or you could do

something where you like prove the zcape route for the the current state route but then look back 10 blocks or 15 blocks and be like okay here's where I actually want to say here's the data I'm trying to say is true because the cost to generate 10 or 15 blocks right now would be so astronomically High you could put like an economic guarantee around that or do things like that right yeah yeah yeah you can yeah so I think in general for most Bitcoin like lines they do that where it's like you do longest chain and then you walk back a bunch so you could definitely do that another great thing you could do is you could prove like a basically you prove like the VA State transition validity with the ZK proof so not only do you just prove like hey here's like a header

with the most work attached to it you also prove that like oh all the headers are connected with a valid State transition of like the utxos with valid signatures and stuff like that and then that makes your guarantees like significantly better obviously it still might not be the canonical chain but at least then you can't just be like oh my balance on bitcoin is like 1 million because like there would there would never be a valid transaction that would make your balance like an arbitrary number uh so you can still do the double spend attack but you can't just like set your balance to something arbitrary okay yeah so basically what you're saying is like if we're proving computation which we're doing with proof of work blockchains um because you can theoretically compute the blockchain

you've gotta you got to do some work to put like guarantees around how difficult it would be to do that computation is right um okay now um proof of stake blockchains like ethereum um as I understand it right uh you can do this ZK proof on the current state of ethereum or on the state transition of ethereum but ethereum finality comes from this like sync committee concept actually I don't want to keep talking why don't you talk about like what are the limitations around how I can transport um the current state of ethereum in a ZK proof to other chance yeah so ethereum's like consensus

mechanism finalizes after like two Epoch which are like six minutes each so that's why ethereum has this like 12 minute finality delay so even with zero knowledge proofs you're never going to like increase ethereum's finality so there's always going to be room for things like a cross where like if I want my money on the other side from ethereum faster than 12 minutes I'm going to need someone like across to basically eat the finality risk and then provide me money on the other side and like that actually seems like a great use case for something like a cross that zika doesn't solve CU like there's no computation you could prove that ethereum has finalized before 12 minutes because it just hasn't um so that's like one limitation of zero knowledge proofs um I think the

other thing is ethereum right now has ethereum's consensus protocol has basically like around a million or something BLS signatures um or sorry BLS public kees that kind of like get aggregated into one BLS signature to prove across like the finality of the chain this is because they have this like 32 eth staking thing where you can only deposit 32 eth and each 32 eth is associated with the public key in practice there's actually only maybe 13 10,000 or 13,000 like logically different validator entities and like they're working on this like Max validator balance that will increase the 32 e the lot so it reduces load on the Network um but right now if you want to

like prove that two-thirds of the entire ether validators that have signed off on a particular block which is needed for finality it's like a really really really large computation because like you're having to basically prove that these two-thirds of one million public Keys have like signed off on something so that's very hard to do in a zero knowledge proof because like zero knowledge proofs already have a lot of overhead and then on top of that you're proving this like really really expensive computation so at came up with this protocol called the sync committee where basically they choose 512 random validators that's like a subset of their million validator public key list and then those 512 validators sign off on the headers and basically you can like

verify the sync committee's signature instead of having to verify from the entire validator set now the sync committee statistically speaking because it's randomly chosen is is very likely to have like the same portion of honest validators as ethereum itself so like the probability the S committee is honest is like very very high because like we're already making the assumption that two-thirds of the ethereum validators are honest um but obviously it doesn't have the exact same guarantees as ethereum because it's like this probabilistic sampling thing if that makes sense so right now in a Zer knowledge proof for us for ethereum proofs of consensus we always verify the sync committee not like full ethereum consensus because the overhead of that computation just very high so that's

like kind of another limitation is like the sync committee's properties are just like not the same as like full ethereum consensus um and so yeah some people think that that's kind of bad so basic okay but but basically you're saying that the efficient way to do a ZK proof on the state of ethereum today is to make a proof that these 512 sync Community um validators signed off and finalized the block and then then you can and that's after the 12-minute epoch has passed and then I can make a ZK proof around that that ZK proof costs me a couple cents to make and I can export it anywhere um yeah if I yes okay um and there might be other things that I think we've gone deep enough and this is

actually super fascinating but like how the sick Community gets picked and how you know who the S syn committee is but like let's leave that aside for now so there's like there are some limitations um there around this and then just for the audience too if I want to make a ZK proof on an optimistic roll up so on arbitrum or optimism um technically because they don't have finality for seven days if I'm GNA be like purist about it I can't really make that ZK proof of the state of optimism or arbitrum for that 7-Day period is that right is that like kind of inaccurate or how would you like think through that I mean you can make a z knowledge proof of optimism in arbitrum state

transition function I mean my real R maybe spicy to take is just like optims and marum are going to become ZK rollups before everyone thinks they will be like I just think the tech is ready it's pretty cheap and like why wouldn't you have the best tech you know like why wouldn't you have like the future of tech securing your rollup so I think they will become zero knowledge proofs in a much shorter amount of time than people think but I'm obviously very biased so today you could make a zeron notch proof of optimism or arbitrum state transition function um now like but so that is something you could do if you like coded their state transac function and like took their call data from ethereum and were like hey when do you have all this call data

and you do this stuff it results in this new state transition um but in an optimistic if you don't do that fully in an optimistic system yeah you can't have finality for like seven days just because like you need that like challenge period basically and yeah and then there's like lrange is one team that's working on like a s a thing to get more more finality from optimistic rollups and stuff like that too yeah they're working on this like basically Economic Security solution where like you have the igen layer re or node operators who are economically bonded like sign off with economic security on optimism and arbitrum state routes and then they get slash later um so that's like one

intermediate option but I think the end game of all of that is just like zero knowledge stuff like then you don't even have to worry about operators you don't have to worry about any of that stuff you just have like approv like a proof of simple okay yeah Pro but basically so to to wrap this up right um if if you have a chain with finality unless you're doing some other mechanism that you really your your cleanest way is you make the ZK proof after you've achieved finality right so uh and you you can so so ZK proof after 12 minutes for ethereum or after like some optimistic time seven days for the optimistic rollups that's like the cleanest way to make these proofs and if you have a proof of work blockchain um your ZK rollup probably needs to be walked back

uh from the head of the chain to have some economic guarantees around it being not manipulated is that kind of you kind of agree with that yeah so I think it's important to talk about like for different chains like what is finality so for example in Bitcoin there's never any true finality it's all probabilistic finality so yeah for Bitcoin like you just have to prove like hey here's like the longest chain and like if I'm walking back a certain amount it just gives you more and more like economic finality for ethereum you do have like finality finality with economic security you just know like oh these many people get slashed if this block gets reverted for a ZK rollup on ethereum you have first for a ZK rollup on ethereum to be finalized first ethereum has to

finalize and then on top of that you you have to prove like a ZK proof of the rollup State transition function on top of ethereum so like on top of all the call data post to ethereum and then for an optimistic rollup an optimistic rollups fin out or for any rollup on ethereum it's actually finalized once all the data is on ethereum and finalized but the state route might not be available on ethereum for a while because of either this like optimistic challenge process or like the ZK proof verification doesn't happen every 12 minutes it like happens every 3 hours just because like posting aof to AUM is expensive but the rollup itself is still finalized if that makes sense yep yep I

love this okay keep going sorry or sorry this is like really deep into like but but the point here I want to make and this is where I think we're gonna like lose the listeners and I don't want to go any further right but um uh the utility of like ZK Tech is related at a pretty fundamental level to blockchain architectures I think um from like an interupt perspective and all although it's like it is clearly the future in my opinion and clearly the way these um these networks should be connected um there is some latency concerns um so we basically get like we get maximum security uh but we probably have some latency concerns because many of these

chains don't achieve finality particularly quickly right or I think basically at the highest level zeron knowledge Tech can make you achieve finality faster because if it there like rollups actually do have finality fairly quickly it's like as soon as ethereum's finalized and all the da is available they are finalized it's just like you can't bridge out of them and you can't interoperate them because like it's hard to prove that finality to someone else but then with a zeron notch proof you can prove that finality very quickly because like the like the computation is already true after like the 12 minutes so yep that like zeron knowledge proofs are useful for just like finalizing things faster and like

having a validity proof of that like as fast as possible hey everyone this episode is brought to you by Unis swop delivering the best onchain trading experience period Bar None in all of crypto here's how I would divide the protocol today there's a web part A mobile part and an extension part you can say goodbye to the popup wallet extension was just not great from a ux perspective now what they've delivered is this nice clean sidebar where you can just very easily track swaps sign transactions send to receive crypto anywhere just a huge ux Improvement on the web part of it huge improvements here you can buy and sell 700 plus tokens at your price on your terms and they've got limit orders which are powered by Unis swap X and there's gasless they've also got real-time charts transaction logs pool data project information all this cool stuff

that just really improves your ux trading on the web and again all this is powered by the smartest protocol in terms of Unis swap or Unis swap X click the link at the bottom Unis swap extension is in Alpha right now and we'll see what you can do about getting you moved up that list thank you very much JIS SW I I want to get a sense of like how you see ZK playing role with an interop and you know we've been talking to so far this season actually Chris goes of anoma sort of painted this world right where you know you kind of have within the anoma framework there are these heterogeneous environments sort of call them Computing environments and one of the big light bulb moments for him that he introduced to us very early on in the season is that the thing that

changed and makes um interupt work within the enoma framework and sort of blockchains more uh more generally is ZK proofs and I guess probably just because it's it's been in the Zeitgeist a little bit like this kind of the polygon sort of a layer type thing as well right where you've got the their cdk their you know chain developer kit where you can spin up multiple different chains and the way that they're going to talk to one another is there's kind of a uh you know two chains that want to talk proofs are generated they're aggregated at this layer and then posted down to ethereum and that all sounds very good uh I think there there might be some limitations to that and Hart was getting at it a little bit with the latency discussion but but I mean Uma when you broadly kind of

think about uh ZK proofs how do you think about that within the context of interupt what does that bring out or enable so yeah I think for interop today for example um first of all I think let's talk about like optimistic rollups I think it's very hard to scale interupt with optimistic rollups even for people like across where the rollups actually reach finality very quickly because you know they'll post all their data and once the data is on ethereum it works but then these relays in order to get paid back like the withdrawal window from these rollups to transfer assets is like seven days and so heart's relayer on a cross or whoever is doing the relaying on across has to wait seven

days for their Capital to be like withdrawn and that's just like very inefficient and so I think with zero knowledge proofs you can just have much quicker finality because you don't have to wait for this like challenge period someday thing you can just have a proof posted within like a couple hours and so it just like really improves capital efficiency for interop and I think that's like really exciting because it's just unlocking like so much more Capital I'll add one other thing one other thing there too Uma like if there ever is a like a fault proof or whatever like fault proofs in the optimistic system are just very messy to deal with and like the ZK stuff just is like way way cleaner because like you said it's a proof so I'll further add that like dealing with fullprof sun optimistic

systems is just seems like a whole messy can of worm that nobody really wants to deal with but keep going I interrupted your yeah yeah yeah they're planing for like oh if there's ever a fraud it's like oh well or yeah I mean yeah that it's just like much cleaner to just have like I always say like ZK is like truth and it's just like it's very much cleaner because you're just like I prove this function happened um so yeah I think Capital efficiency is really nice I also think for example for across today all these like we're we're like rapidly in this world where there's like thousands and thousands of rollups right like you can go to rollup go to conduit or like Calder or whatever and like I could like even spin up a rollup in like the in like two minutes and so for all these

different rollups it's really hard for someone like across to like support all of them because like you have to run all their nodes and like that actually costs a lot in AWS um and yeah you have to like run all this infrastructure and so that like fundamentally doesn't scale super super well and with a zero knowledge proof it's like maybe one day a cross doesn't even have to like run a node for a different chain it can just like verify or proof that like this happened and then be like okay like let's do the relaying or stuff like that um and so I think that's also like pretty powerful so I think those are like the exciting applications of ZK and interop also even moving out of rollups we're still in a world where there's like a ton of l1s and for L1 to L1

bridging you basically today the status quo is like you have some multi sick of like five people and then you have like someone deposits money on one side the multi says yep okay the money was deposited and then they unlock tokens on the other side or they mint tokens on the other side um and that's obviously very risky because these multi6 can't get hacked because like you know you just have like five people you just hack into an AWS account or two and suddenly like the keys get compromised and if instead on the other side you're verifying a zero knowledge proof that like the original validators of the source chain actually signed off on a current state of the world is just much more secure cuz that's like very hard to forge so I think yeah cross cell like

replacing multi and making cossell one bridging way more secure is very exciting and then transitioning all Roll-Ups to zero ZK rollups and unlocking like Capital efficiency and like more seamless smooth interop and like better scalability across thousands of rollups is also really exciting okay Uma I'd love to get your opinion on the overall demand for like how useful synchronous versus asyn composability ultimately is because that's one of the limitations here at least my understanding of um like roll up to rollup uh interop that gets facilitated by ZK that's that's a sync right that's not happening at the exact same time and like maybe to challenge a little bit how useful that is is one of the reasons why

people are extremely excited about or at least at one point we're really excited about that interoperability there's a me aspect to this right uh where you could do some sort of like the simplest form like an Arbitrage between two different is and in this um and in this world you could maybe place a trade where there are multiple legs you submit a proof which corresponds to each leg that gets aggregated into this one layer and then posted and then that's very simple the problem uh as we've talked about a little bit is like the latency there um because again like ethereum moves in 12C block times and five minutes there it kind of limits that specific uh use case so I guess you know what are your thoughts on the limitations that the latency in this world uh has on

different um you know sort of cross chain use cases yeah so cross chain m is like an interesting topic because I feel like a lot of people are excited about it but I think so okay at the highest level you could argue like the real cross chain meev is like the rollups and like binance or like centralized exchanges because like those people are always going to have a higher like a faster block time than anything decentralized so one card well is there even going to be like cross roll up a cross roll up B ARB when like the true leg is really roll up a to binance roll up B to binance and then like I mean maybe that's not super ideal but like that does feel like that's going to

probably be like the most important volume at least for a long time so there's a question of like does crushing M even really matter um even so even if it's a sync you know um the second thing is I guess if you have synchronous in across two different chains fundamentally I think the proposer like the person who is deciding what that block is has to be the same entity the chains can be separate but like the proposer at that point has to be the same entity and so that can happen with ZK without ZK whatever if that makes sense um like yeah they're I think the proposer sequencer is like the most

important person not the validity mechanism if that makes sense it does make sense and I think it's a nuanced and interesting and tricky point but I I think where I'm aligning and Mike we still have more episodes in this season to try to Hash this out but if you want synchronous cross-chain things the block proposer is got to be proposing that same block on both those chains and in a sense it almost looks like they're the same chain but not and maybe there's sort of like some weird Nuance here where they're separate chains most of the time but every so often when they need the synchronous thing there's like a block that is proposed by the same entity and it kind of becomes the same

chain for that moment um anyways I think that does make sense Uma but it's definitely a nuanced Point um yeah and yeah regardless what you're saying here is to to just reiterate the stuff that's really like in my brain and it's like ZK becomes this glue to uh connect disperate system with quote unquote truths because you literally just have a math proof of what the truth is um your limiting factor here is that in order to create that proof of Truth you need that chain to achieve its true truth like which is AK finality right which kind of makes total sense if it's at some intermediate state

where it's not fully fully proven on that chain then you can't quite do the proof of Truth um yeah I don't know that just seems really powerful yeah I agree so Uma if if this is uh already really cheap the latency doesn't ultimately end up matter mattering that much why aren't we aggregating all of like why doesn't it feel like all these chains are basically one why isn't everyone doing this today you know I asked myself that question too Mike um I not Visionaries like you Uma that's the problem no no I think okay a couple of things so I think historically to use ZK you'd have to

have this team with like really built out ZK expertise like you'd have to be a ZK wizard you'd have to have a whole cryptography team you'd have to build your own proof system you'd handroll your proof system it sucked we we did that for all our bridges and it's just not very sustainable I think there's been this recent wave of ZK VMS where basically it allows any team to just like Express their computation in like rust or like normal code and then get that proven for a cheap enough cost and that's what like things like sp1 kind of enable is like you can get like all the performance you ever wanted but you can still write it in normal code and that combination of like amazing developer experience plus amazing performance like that combination is I think what's been

kind of missing until this point and now with things like sp1 I feel like it's kind of at like the turning point where it's like the devx is good enough and the cost is cheap enough where now everyone can go adopt it so like and that only came out like a few weeks ago so I'm very optimistic that like hopefully this is kind of like the turning point for ZK where like suddenly even teams that have pursued an optimistic approach for a long time are like oh wow like now the tech is good enough and like now the tech is cheap enough like let's go do it you know yeah like Uma I guess just to push on this this is so recent right like sp1 you guys announced uh but is it meet or what do you call it devnet right now too right now it's de like right now it's

just like still like Alpha mode like is not intended for production use that's like our highest priority as a team and like what we're working on very actively towards so hopefully that comes soon like we hope in like a few months um so yeah that that is also a bottleneck yeah well okay but like it's just we're we're like of the moment like you guys have just put this out uh risk zero sort of in the same space is getting coming out soon like very soon too I think there's another project Gul lot that's like approver Network all this stuff is like not like this is the point why um I think the next six months to a year are going to be super impactful because

these tools are really truly going to be live and usable um with a good experience yes exactly no yeah to answer your question Mike I mean I was kind of joking before like I think a lot of this like stuff is really new so historically ZK has been very hard to use and its performance is bad then a lot of people fix the performance but it's still really hard to use and then now with things like sp1 it's like you get the best of both worlds but sp1 obviously is very recent I think it's like a month old at this point so now it's kind of like okay now people can start like thinking seriously about it um and so but this is exactly we want to talk to you it's the frontier you are the frontier so right that's too nice can you help us think through what are some

of the second order implications of if ZK Tech were widely adopted how would that change blockchain architecture and sort of where value crues like here's one thing I was thinking about so one of the layers that Network effects AC crew to is at the execution layer we see massive returns to the evm now we're seeing it with the svm and maybe this like move ecosystem ends up being valuable as well but one of the at least when I think through how Computing changes with ZK is like suddenly it doesn't really matter what execution environment you're using in like most of that computation happens offchain and then these sort of proofs there's like some new kind of aggregation layer that ultimately ends up happening and then a lot of blockchains as we think about them are

just kind of like bullet and Boards right and that's like where I understand the sort of Celestia Vision going the ethereum vision going so I mean does that make execution environment slightly less valuable um like what are just help us think through like some of these second order implications If This Were to be widely adopted yeah it's interesting I feel like if ZK proofs become widely adopted there's kind of like it adds like new layer to the modular stock so like there's data availability and consensus which is like uh you know one part of the modular stock there's settlement which is kind of like what asset you're using where's all the bridging happening so like for example right now like I think a lot of people use ethereum as a settlement layer so that's another part of the modular stock there will be like proving um which becomes a part of the

modular stock which is like how are you actually getting your validity proofs to like settle to the settlement layer um and then there's actually like the user facing stuff which is like what net networ are they using and things like that um in terms of value acral I mean I obviously think like proving will be very important and it'll be more of like a volume thing where it's like oh you just pay like one cent or maybe even a fraction of a scent for a proof but like there's so many every user interaction with the chain will use a proof and so that will be like proving will be like a volume based thing where it maybe ends up making a lot of money but like it's just because there's so many proofs being

generated um I think obviously the economics of like settlement look pretty different because like it's like oh eth is Money And so there's like monetary premium and then like maybe with da there's like yeah I don't know like what the value of cor story looks like there but that's just like another part of the stack so yeah it's hard for I guess it's hard for me to say but I think the tldr for me is like there will be another really important piece of the maio stack in addition to execution or in addition to Da and settlement and execution there's also like now there will be like proving so that that is another question I have in my mind for you Uma too where it's like how many of these proofs proofs do you think there are um in this and like and I think there's

going to be this Evolution here too so if a proof takes five minutes to generate right now it's not Po in my mind it doesn't make sense for there to be like 10 million proofs being generated a day um from from like at least an interop perspective because like you know you're going to batch your messages and send them every 10 minutes or something like that um but I still think the ZK component here is incredibly valuable because you are sending truth um like every 10 minutes and sidebar this fits in like our across thesis very well because across like batches together uh repayments every hour or whatever um but this sort of runs counter to I think the thesis of a lot of other messaging

Frameworks um Axel Wormhole layer zero I think they're all kind of uh built around the this idea that there's going to be tens of millions of messages getting sent between blockchains and I'm kind of curious but that are not necessarily ZK right and I'm kind of curious where your mind goes where if ZK takes five minutes to generate now and it's going to take 30 seconds later but we're not down to 200 millisecond proof times or 20 millisecond proof times for a while while I think how many ZK messages do you think or ZK proofs are getting generated in you know a year from now and like how does that influence blockchain interop and like architecture um I guess I think for

example for a ZK rollup it's kind of like there's not a proof generated for individual transaction but there's like a proof generated for the entire block and maybe actually there's like proof generated for every block that get aggregated together so so in the end it's kind of like every transaction does trigger a proof it just might not like it might just be everything's like batched together um and so I think with messaging it could be pretty similar where it's like maybe not literally every single individual message triggers a proof but you have a proof that like oh ethereum came to finality on this batch of blocks and then oh I have like 15 messages that were contained in that range um I think obviously there will always be like non ZK solutions for things that where it's like maybe the

finality doesn't matter as much or like the security doesn't matter as much um and so yeah I don't necessarily see like oh everything has to be ZK all at once I just think like more and more things will like transition towards that starting with like the highest value use cases so like for example right now I feel like rollups being zik rollups makes a ton of sense because like they have a ton of TBL on the line and like I'm sure if you're a rollup team it's really stressful to operate one and like adding Z ZK security that'ss like your peace of mind if that makes sense uh and then I think as it gets cheaper and cheaper like more and more people will like just start using it because it's like why not um so that's kind of how I see it evolving okay so this is actually kind

of interesting so if we're if you're doing a ZK proof of each block and then every um I don't know 100 blocks you're doing a ZK proof of those 100 blocks recursively and then posting that to ethereum let's say right um and if you're block time is 2 seconds I just did some math um that means you're making 43,000 plus a little bit overhead making like 45,000 ZK proofs a day um and then in a world of a thousand rollups you're making millions of ZK proofs a day um here okay cool so this is like good business then for people doing ZK things where if we have a a world of a th000 rollups or 10,000

rollups all ZK based they're doing a ton of ZK proofs um on the interop side uh maybe that's true maybe that's not true though because if you aren't able to make those ZK proofs uh immediately anyways because of finality con concerns like you can't you have to wait for chain finality then probably does make sense to batch them together and so maybe you just you like you said you you don't have nearly as many ZK proofs between blockchains but they're just really useful I don't know yeah I can imagine that happening yeah interesting okay I'm learning things I should ask this probably a really basic

question ask right at the end of the podcast but it like in L L1 to L1 um like communication that makes use of like a ZK like line or something like that do you still need a general message passing protocol like or transport layer like layer zero to deliver that or no yeah like layer zero actually is like this and I think a lot of other protocols like hyperlane and stuff are like modular so basically like they just provide you smart contracts and kind of like this framework for sending messages but then the security mechanism of how the messages are delivered can be abstracted away so like you can use layer Zer multisig you can use ZK proofs you can use so sync ZK proofs you can use someone else as ZK proofs you can use like some other system so the you

still need the messaging framework because like you still need like the solidity side of things to like keep track of the messages you still need like the Rel layers like all that kind of stuff and then the security mechanisms like abstracted out and can be either ZK or not ZK so I think yes you still need layer zero like we would for example work with the team like ler zero like we work with Wormhole to like help provide them their ZK security and so we could also work with a team like ler zero to help add a ZK work to like their protocol awesome um Uma in in conclusion like one sort of big question we've been asking guests from this season is you know as crypto grows and there are many different chains and many rollups that exist um obviously the space is getting

more more complicated uh sometimes these Solutions like the best Solutions end up taking a really long time to get to Market other solutions that maybe uh except some less positive trade-offs or cut Corners kind of end up happening and we kind of have to have these very hard conversations with ourselves about trade-offs and something har and I've been questioning is like what ultimately would be a good end state for crypto and this sort of on like a 10 to 15E time Horizon like if we accomplish something here as this industry in this future like what would what would be a successful outcome from your perspective like with regards to ZK or like more broadly more broadly but like theoretically maybe you spend a lot of time with ZK so I'm assuming that's part of the

vision yeah I think at the ZK level like I just think ZK is like simple because it's just such a fundamentally simple like it's not simple to implement like our team knows that but like once you have it it's just such a simple primitive it's like I some function I proved to you the function is like this value with this input okay now you're good to go and it's just like I like to tell people it's like like ZK is like truth and like the truth is simple and so I just see ZK being integrated into everywhere like I think every bridge will be a ZK bridge I think every rollup will be a ZK rollup I think basically all blockchain infrastructure will like run on ZK rails and it'll all be like seamlessly interoperable it'll be like

all verifiable there's none of like complicated optimistic stuff there's not this complicated multisig stuff it's just like verifiable compute being used everywhere that's kind of like the end game Vision I see for all this like blockchain infrastructure stuff is like ZK is like the glue that holds everything together and that's actually why it's like a great fit for an interop episode because interops all about glue and I think ZK is kind of like the ultimate glue that's not to say there aren't won't be more layers like for example I think there will always still be room for a cross because like you know sometimes you do need to eat finality risk um yeah we um I do think there's obviously going to be for wallet um there will always be room for wallets who like provide a really seamless user experience and like

integrate with AC Cross or like stuff like that there's always going to be room for aggregators like there's still going to be many layers of a stack but I just think ZK is going to be like the base glue for everything and like basically yes like having truth communicated everywhere um yeah so that's kind of the end game of how I see the space evolving and how I see things talking to each other and hopefully that leads to like better security like no more hacks and like leads to like oh there's not rollups that are just multi with like no fra proofs and stuff it's just like everything hope which are you talking about we don't name names we don't name I think okay I think Mike wants to is gonna wrap this episode but actually Uma

I want to ask you one other question what's your most exciting use case for ZK outside of blockchains I think I have a guess what you're gonna say but I'm really curious like where this go I I actually have an answer for this too so Mike and I have our bet Mike and I have our bet let's see what you say I think some of the attested image editing stuff is really exciting and interesting like basically you have some attested sensor but none of the images you see on like Facebook and stuff are like the raw sensor data because that's like way too big and basically you can just prove like this data was Tak this image was taken from like a camera and then modified with these Transformations and like jpeg compressed and cropped and whatever and then this is like a real image not from mid Journey like I think

that's already super exciting actually some people at a Stanford hackathon tried out SP attested image editing with sp1 and it was really exciting and interesting so I'm very excited for that kind of stuff especially as we get like more and more crazy like image Generation video generation whatever that was pretty much what I was gonna say something in AI space which is like it's like this idea of verifiable compute so even verifying that an air ran account computation is interesting too but the attest Imaging stuff having all our iPhones with like a chip in them that is going to attest that our our our photos are reals pretty cool Mike what was your what was your guess I had such a lame I mean mine is much too optimistic I don't think it's ever gonna

happen but I was listening to this podcast shout out Demitri cofenas uh about basically like how the modern surveillance State got created and we don't need to go like all the way into it but basically to like drastically dumb down what ended up happening was the US government just created essentially a list of everyone they're like in order to make sure that people aren't doing bad things we just need a whole big list of people you know this I don't know how this would actually translate into something like ZK Tech but I just remember think like it would be really nice if we could actually get the desirable government outcome that the government wants which is like hey are you doing like horrible like kitty porn things online without just like aggregating all of that in one place because that's always been my big problem with the crypto anarch Anarchist

like governments actually do do a lot of you know like stuff that we really want them to do and they do have hard jobs so if we could give them this primitive this tech building block maybe actually that could lead to the outcomes that we all want with the Privacy that's extremely important um I I like Mike you're kind of like this like upstanding American like ta good taxpayer Cipher Punk it's like a great combination I I I kind of love it so I apprec I'm definitely not like a burn it all down kind of person like I'm actually pretty stoked I like living in America for the most part I love to not burn it down um but I don't love what's going on now so I appreciate that heart that's very nice Uma if people want to find out more about you succinct what's the best way

to do that yeah you can follow us on Twitter like cinct laabs um and yeah like we post a bunch of stuff of what we're up to there um yeah it's probably the best way okay cool this was really fun Uma thanks for joining us thanks yeah this was super fun all right good to see you guys all right bye all right har that was a great episode almost made a ton of sense like uh like we should just obviously be leaning pretty hard into ZK proofs what do you think uh yeah makes a ton of sense I mean I think Guma is brilliant um you know full disclosure I've gotten to know her from like basically day one of her starting succinct her company and she's

just like clear thinker great energy um it's hard to think that that same person is like a super nerdy math geek that can do like the crazy moon math in her sleep um and her co-founder John's pretty brilliant too so just amazing people and I think like those are the types of folks and there's a lot of them in the ZK space that are really taking this Tech and making it uh a thing you know three years ago the idea of this existing was an impossibility people just thought that was no way and this industry is wild so yeah the ZK stuff seems like it's a really important part of the stack and what I really enjoyed

about uma's conversation is like she was also very real about the limitations of it like where it works and where it doesn't um and I thought that was super super compelling and one thing that we didn't really get into but you know Uma originally cinct was really focused on the bridging uh sort of application and then they I don't know if pivoted is the right word but sort of expanded into this this new release this uh VM the ZK VM that they just launched S sp1 I believe it's the name and you you can hear her you you can hear her on other podcasts sort of talk about going through the pain of developing a ZK application it was kind of beyond the scope of this podcast but like how difficult it was to just like debug circuits and stuff like that and that was actually part of the reason why she

decided to build out this more general purpose infrastructure so I think that was one of the bigger takeaways for me is that it actually just used to be extremely hard to build these applications out but now with you know succinct and I understand risk zero does something pretty similar like I think it's just going to get a lot easier and there might be this massive camb braan explosion and um yeah absolutely I mean I it's funny I can speak because like across we were working with succinct on their old product their Bridge product to actually Bridge messages from ethereum to other l1s like BSC um and it is pretty interesting because I basically think uma's story here is they were hand rolling and hand tuning all of these circuits for specific use cases

and it was almost like they were becoming like Consultants or like service providers that would just build this like circuit for a specific use case um and they're like H this sucks this isn't scalable this takes forever um and then they figured out how to abstract it away um which is amazing and makes so much sense and yeah they're not the only ones in the space uh doing it so we should give a lot of props to um all the other great teams in the ZK space that are that are building cool stuff but they you know they're an example of a team that was um painstakingly tuning their own circuits or writing their own circuits and they're like let's let's substract this away and I don't know enough to know

exactly how well it works uh but from what I've been told and from what I can observe it works really well um and so that's pretty cool um you know the other stat that that shocked me actually was this like two SEC two cent uh um kind of proof generation cost and this is in a specific context this is like a tendermint consensus proof but like that was way cheaper than I was expecting um I thought that was pretty pretty cool too yeah the cost aspect of it is coming down the latency aspect of it is coming down you know in my mind ultimately went was I was drifting back to that first episode that we did with um that we did with Chris and trying to think about these different environments and how

they ultimately end up getting connected and one thing I've been sort of thinking about is like I'm starting to question this whole modular versus monolithic debate and how much it really makes sense it's kind of just you know like if you you know what I'm saying I I think so I'm the way that I'm sort of starting to look at it is which actions which Computing actions are coupled tightly and I think once we solve a lot of the frictions that exist in these ecosystem it's ultimately going to be like which where there's an economic incentive to like tightly couple certain interactions and I sort of feel like where I'm heading now towards this like ultimate sort of end state is that we're going to basically break all of these blockchains apart into this like super modular uh

sort of framework but then we're going to get the recoupling of these um of certain interactions where it makes sense to actually couple them and then I kind of just think it sounds like ZK is going to connect everything else for you know maybe there's not actually an enormous amount of demand to trade a theory on Jupiter right but for that we'll have zero knowledge proofs and that'll be fine that and then in and then it's like a different system will will uh will support trading salana right on on Jupiter so I that's what that's the where I'm kind of landing on this ultimately yeah Mike I like all your questions around you know um I'll call them business models like value capture like where is value going to be captured here and increasingly to me

it's coming across like a lot of these systems there isn't really a moat um it is actually like there's a service that's going to be provided in a highly competitive market um at what is going to be extremely low cost and um uh the the the value capture is just volume it's like doing a lot of it really cheaply um and I honestly think that that this is first of all not a bad thing for a consumer um and when you think about blockchains being permissionless uh decentralized uh organisms it's almost the way it should be too so like Uma we didn't I didn't we didn't push her on this but when she was talking about like ZK and how you make

money here it's just like gonna generate a ton of ZK proofs gonna charge not much more than the cost of them but a little bit and it's going to be a volume game um just kind of like AWS is mostly a volume game there's like some profit margin but it's not huge right yeah well I think you know Sam said this so well in the second episode it's like all these people are asking about business model and value captur and that's such a luxury for me because I just want it to work and I think that's one of the by the way people ask all the time like why are all these podcasts so technical it's because the smartest people are still figuring out how to make it work and they're going to figure out the business model and value capture stuff later on but I do think it's like one thing I

wish we had poked a little bit more on is just the um you know where where value is going to end up getting created on like in like these proving networks and there is a lot of interesting stuff going on in different layers of aggregations within ZK and I I think right correct me if I'm wrong but I'm pretty sure that the sort of plan from the polygon perspective in their a layer is to kind of have these economics economies of scale right in the aggregation layer when once you get a whole bunch of these things together you kind of uh you know you kind of batch them up and then submit them down to to main chain eth and ultimately that ends up being pretty sticky because if you were submitting this outside of that aggregation layer it's much much more expensive so there are these kind of

interesting games that I can see people starting to play um but maybe it's just too early and we just got to see how it all shakes out yeah I mean I think it's pretty clear to me well if I just gut reaction think about like uma's like cinc Labs uh sp1 business model they're building a Marketplace for provs provs are going to be hyper competitive this is going to talk about a lot of the other things we talk about our space it's like the hft space in in in finance so hyper competitive um maybe there's not a ton of Prov provs um because they're super specialized Uma hinted at like maybe some of these provs build their own Hardware like literally to generate

these EK proofs cheaper maybe that happens but you have really competitive uh Marketplace of provs that are just there to um offer this proving service at the lowest cost um and soy's business model here is they they're the marketplace for it um and that Marketplace can't be usurious or can't charge too high rent because some they'll just another Marketplace will pop up that's cheaper um but they can charge something for being the the central point of where like supply and demand meet and I think that makes sense um I do too yeah we love a good Marketplace model and you know we we started that's a popular model do people like um all right here's here's one

interesting place that we could have gone as well that we like sort of jokingly did did at the end but I do think zero knowledge is one of those areas and I think vitalic has actually said that zero knowledge as a technology is comparable to blockchain as a technology in terms of what it unlocks like that's the scale that we're operating on here and it does feel like you know there's so much that's been said about how much can you bridge what's really happening in a real world context in into a blockchain I mean ZK proofs are really interesting part of that story right like we've called this the Oracle problem right like how do we Rel the stuff that's happening in the real world and one of the things that it seems like ZK proofs do really well is they take these extremely you know

heterogeneous sort of computing environments and it puts it spits out this output which you can put onto a blockchain and that actually like ZK kind of could be the missing link in between getting some of this trafi stuff that we've been talking about forever on crypto rails also um so yeah yeah I think it's hugely powerful I I I agreee um this but like emphasizing what you said like this is a technology is as big as blockchains that's a big statement and we should probably fact check it before we say vitalic said that but I think he did uh or said something like that David told me that vitalic said that last night so that's why it's your all right so we we'll we'll fact

check this at some point but like before we say but I'm willing to go so far to think that vitalic definitely thinks ZK is a big deal um and we all should because I think it is one of those like um transformational texts that there is a lot and again what's so interesting even about the conversation with Uma the whole thing was about the succinctness property of ZK we really didn't even get into the Privacy bit too and like there's a whole other unlock of what the Privacy stuff gives us as well that I think is um frankly that's where I've spent less of my own brain power thinking through what the implications of that are um but there's like it's like one technology that gives you two major unlocks um at the same time and

that's pretty wild I agree and I've actually decided um I'm gonna for the intro to the show when I do it I'm gonna walk you through but getting ready for the show I actually did a there's on Wikipedia there are a couple of like very plain plain World sort of examples of how the ZK and the Privacy model work and the Privacy stuff I mean again going back to our conversation with Chris like that was super important to him he actually view that as a fundamental issue with how blockchains are designed for all these different reasons it's just very suboptimal to have your transactions revealed either because you don't people have a right to privacy and people don't want to be exposed or because you know some bot can come in and grab your transaction then sandwich you right like there are all these great

reasons why privacy should actually be at the core of the design and yeah we didn't really get into that but I agree heart this was a really fun one my friend I will uh yeah I will see you same time next week cheers man see you next week M hey everyone want to give a final shout out to this episode's title sponsor say now there are a whole bunch of really exciting reasons to be building on say V2 outside of just parallelization I want you to head over to say. to looking into building on their public devet again click the link at the bottom of this episode and head over to say. start building something [Music] today