ZKPodcast: Validium with DeversiFi & Starkware

Transcript

[Music] welcome to zero knowledge a podcast where we talk about the latest in zero knowledge research and the decentralized web the show is hosted by me Anna and me Fredrik so today I want to welcome Tom who's a product manager from Starke we're and we'll who's the co-founder of diversified to the show hey guys hi Anna so the the majority of this episode is going to be us exploring this

collaboration between stark ware and diversify but Before we jump into this I actually want to kind of welcome back will because you you actually were on the show last I think it was like about a year ago I guess a lots changed exactly a year ago I think we were in Berlin and we talked mainly about Dex's and of course yeah the dead space has changed a lot in the last year but also I started a new company diversify which was specifically to create a really scalable you know high throughput decks and previously it was known as FNX trustless so in particular we were fun out of bit the Nets

and had focused on building a prototype proof-of-concept for doing self custody trading and diversify is really kind of the sort of child of that project and but in a much more sort of serious and scalable way for time now being able to really handle high speeds many many traders and you've actually incorporated some ZK into it which is cool through this through this collaboration now I should say our listeners may have heard your name mentioned by me because well and I have also we founded a company together a separate company not diversified and that is the Zeki validator that I've mentioned a couple times I wonder actually will like

because like do you feel like that in a weird way influenced your interest in zero knowledge stuff or were you already into it before I think it was definitely you know I was it was definitely a theme that I was interested in but probably even a year ago I saw this is bigger an opportunity to really dive deeper into this into the topic and certainly you know in retrospect it's almost felt like a very lucky that we ended up now doing that together a separate as well as you know diversify ending up working with Stark where to kind of have the crossover and see the full spectrum of ZK so for anyone who isn't familiar with this EQ validator I have actually mentioned it a couple times on the show but it's the purpose of the ZK validator

is the idea here is that we can actually validate on different networks and use any of the rewards or the Commission on their rewards to promote or you know be advocates for privacy on those networks and so this is kind of a very unique validator proposition which is focused more on one particular kind of research than anything else that's kind of the beauty of what the evaluator is doing is that by participating in these proves two big networks and trying to actively encourage research and the right mindset around making sure that actually exactly what we're doing on a theory and we start where in terms of scaling a an application for a much wider number of users can be done on the future previous

date networks so for example cosmos and polka dot as UK validator is now active will hopefully be able in quite a short amount of time to support applications just like diversify and other scalable layer 2 solutions I mean one day actually will I'd love to do a whole episode maybe not a whole episode on us but I'd love to have a chance to talk more about it but this episode is really about the work that you've been doing with dark work so you've just given a quick brief to the listeners about what diversify is and even though we've had Elly on and alessandra on from Starke where Tom maybe you can share a little bit about what star is so star queries I think almost two and a half years old company which was funded by four co-founders professor elements the Sun

professor Alessandro key is that I think it was by now with dr. Michael reactive and our CEO very coldly and a technology company which develop and implement a full stack of proofing systems in our main goal is to solve the two most critical problems the blockchain which is privacy and scaling using proof systems now I'm not sure if your audience is familiar with zero knowledge proof but the gladly zero notes proofs in my opinion can solve those two problems by privacy by using the zero knowledge part and scaling by using the fact that the proofs are logarithmic in in the size of

the computation and our first partnership in our first product is diversify and the star X which is a self custodial layer to start based exchange so as a follow-up to that though I want to hear a little bit more about you and how you got to this you know position and the work that you're doing is dark we're sure so my name is Tom I'm working a star go for almost two years now at the product group I'm in charge of the collaboration with diversify and also I do research on protocols which I research the connection or the integration of proof systems into blockchains did you study this kind of stuff before like how did you get to this point it's

a good question so I first studied physics and mathematics and I worked in this field for almost six years and then I somehow I guess a lot of people are came into this topic the same way but I stumbled into Bitcoin and it like it captured me or fascinated me so I decided to do a master in computer science and specifically research this topic somehow I stumbled across any business on office it's a poor complicated story but anyway he agreed to two to be my advisor on materials and I did a research and those topics specifically on proof of stake for a cost and then I landed it on Stark where

after that Wow but you like you got interested in Bitcoin and blockchain and then jump to like the forefront of all this stuff right kind of skipped a few steps or something yeah actually my education on computer science was mainly through blockchain so I mean a lot of the things which for example for my colleagues in stark war which have a very deep background in computer science finds very strange in blockchains for me it's like this is the way I was introduced to computers so in many sense that those those stuff make a lot of sense to me so let's talk about this collaboration and what exactly you've built because in the past like we've

heard of ZK roll-ups we've heard of you know optimistic roll-ups and and Dex's but like what is the stark word diversify collaboration exactly diversify is aiming to offer a high speed exchange that it looks and feels if you're if you're happy with that like a centralized exchange in terms of speed performance privacy but where ultimately you have the knowledge and ability to verify that every trade every action that you're taking is cryptographically proven and at the kind of end of the process essentially proven directly to maintain etherium where we at least because all the assets it also issued on

that have and trust the economic guarantees that it gives us what that sort of means is that from a customer point of view they're interacting with diversified but once they're sort of signing orders for example so every single action that they take they sign with a fee that they own we sort of on our back end sort of once a let's say a trade happens send these cost and we're who generates a start proof and submit that directly to the blockchain so essentially what we've kind of done is is move the exchange one layer up off aetherium so that everything's happening off the blockchain but in between the two we have this proof system which is what was thought by stark ware and really gives the kind of ability to

suddenly allow us to have tens of thousands of trades hundreds of thousands of trades potentially and every minute rather than you're sort of a much smaller throughput you can get through to the blockchain with other X's and long term that means sort of more efficient trading and the ability to get much better price discovery and bring in a much bigger and sort of more mainstream set of traders Ben's currently using us to do exchanges I wanna I want to understand a little bit what this actually looks like under the hood like is I mean when you when you talk about this I I automatically go to like the Z K rool up model of like a batching system where you have like all these transactions and then a single proof

being written on chain I'm really curious to hear though like how do you describe this system how is it actually working under the hood sure what cost today on it cerium are three things one is computation the other one is storage and the last one is transmission or call data the way it's called material the basic notion of layer 2 solution is is to take off the chain some of those parts and but still maintain the security of the maintenance the way we do it is that we batch together bunch of transactions that happens off chain diversify sends them to us and we create from them a batch and then for this batch we create a proof and basically what this proof is saying is that three

things one is that for each transaction we saw a valid signature the second one is that the user had enough balance in their accounts to do this transaction and the last one is that after we applied all the updates to the user balances this is the merkel route and we commit it to the chain like the first two parts the fact that you saw a valid signature and the user had the correct balance or enough tokens in their account to be able to make this transaction that sounds a lot like z cash and the start the snarks that they're using to basically prove every transaction the merkel route proving to the other chain though is potentially

where it diverges I guess it's actually very similar because also in z case you have a miracle tree the main difference with the cash is that in thickish it's a you take so based abduction right so what do you what do you actually prove is that you know a commitment which for this commitment you know a valid you take so and you prove that you know this commitment because the miracle route is part of the of the block itself of the block indication right so in a sense it's very similar right because you don't actually store the state of the commitment tree on the blockchain

but it's the state of the notes but you still can relate to it in a proof because it's part of the block so it's very similar to what we're doing only that we're not using YouTube so but we're using accounts but the ability to relate to a state is possible because we have the commitment on chain but it's it's so the difference here is like you're also sitting on top of an existing blockchain so you're you know writing to the etherion blockchain basically and also you are using Starks and not snarks I guess and also that the proof is not for a single transaction but for a batch of transactions okay and that's I mean the batching part is where

it's very similar to the zk roll-up then yeah what I described up until now is the same for what is being called today phylidia --m and ZK roll-up and if you remember I said there are three things that cost gas on material so in both system we remove both storage and computation of chain but where we defer is what we do with the transmission right so in ZK roll-up you post the transaction the necessary data to reconstruct the state as part of the data of the transaction unchain it it is not part of the storage of the EVM but it is part of the transaction and invalid you don't do it and instead what you do is that you provide along with

the proof and at the station from from a committee and we can talk more about this committee in a few minutes and at the station from from a committee that they have the necessary data to reconstruct the state I want to just point out you're using this term phylidia m-- is a very new term and I think actually later on in this episode we're gonna start to talk about the quadrants and the types of these layer to scaling solutions and hopefully we'll get a chance to dig in a little bit deeper on what that means so to our listeners don't worry we're getting to it but just know that the diversify exchange is considered this validity I'll layer two solution okay so I think we've we've kind of

started to scratch the surface on like how this is working you just mentioned a committee and I want to understand like where the committee actually lives in this in everything you've just described so is the committee off chain yes as homies just saying really the the main method of creating these batches and submitting them on showing to prove a certain state that's the same way however you transmit the data about what the state is and one way you can do it and I think the kind of simplest way potentially is just to put it all on chain publicly as cool data and that doesn't require this committee basically because at any point anyway anybody

who's used this or actually anyone else can reconstruct and from the data that's been put on chain what the state is and therefore see that the state is valid and potentially check any future state transitions but that means that everything all this data is public and actually you know that's the case now with most decentralized exchanges on layer one aetherium but that actually doesn't meet the design requirements of what diversify was aiming for with private trading and that's very important particularly to let's say any larger trader who has a specific trading algorithm that they're making decisions based on where if they're giving away

their trades take you over a long period of time and if they're making very high creep you know higher frequency trades they're basically losing their strategy because it can be been you know seen especially if it's profitable by anyone else and poppy and so we needed to make sure that that data wasn't on chain and that users could you know know that they're able to trade privately and so this second option which is to put the data off chain and held you know somewhere else and in a second like the committee is being asked that with with the way that we you had to go and essentially these committee are we trusted or known organizations with reputations at stake who exist in this community in this ecosystem one of which

is stock where another which is diversified but also a series of they're sort of well known brands like in Fuhrer and never mind consensus who most people at some level actually if you're using aetherium particularly through like you know a browser and where you're using metal mask or whatever else sort of have anyway some sort of implicit trust in basically the solution is that they are always holding a copy of all users data and will sign any any state transition before it can be submitted to the blockchain which means that there's a guarantee that a certain number of these organizations always have a copy of the data and if for any reason sort of start where I or diversified you know both

disappeared at the same time any of these organizations could publish all of that data to ipfs or you know also a school day to Andrei theorem and that could then be used by every user of this system to withdraw their funds from diversify and safety kind of take them back to their wallet to more committee members sorry I clusion and cephalopod which are also in the business of proof of stake validators yeah exactly and cephalopod is Ethan's we can smooth man yeah oh yeah okay like you just talked about kind of a worst-case scenario if you know two of the entities disappear and you needed to potentially reclaim

those tokens but what about like where does it normally live like does it live with you does it live with each of the committee members or so the committee member does not have any role in the standard operation of the system the user does not have any interaction with them and we just send them the copy of the data and receive a signature from them on the new state that they compute but it is not necessary for the validity did not necessary for anything but there's a station that they have the data so when a user normally wants to do with response and this was your question right well actually my question is where does the ledger live that's what I'm actually asking I'm like where where does the record of all of this live

somehow it's on a database that each of these members hold including updated and they then use this to you do not pollute and and sign that it's correct I see and each of the committee members would do that on each bench like would one okay would it be like a committee member will do one batch or each one does each batch all of them all of them each batch okay it can still proceed but if you know if nearly all of them go offline basically the system would have to pause and wait for you know come online to sign the next update okay so I know that like when this was released the committee was a

point of kind of like I was pointed out as something that sort of makes it a little bit less decentralized how do you how do you kind of take that like is that yeah I mean I assume that that was a trade-off that was made so it's like a purposeful decision yes I think that the the criticism was not about the decentralization because in any case other solutions can can say many things but now all the latest solutions have a centralized operator so it's not about decentralization it's it's more about trust or more concretely about security risks so what the attacks in the system described some way to to use this committee in order to to lower the

security of the system so this was the criticism in the and I understand it but I think like any other decision in the system there are trade-offs I mean there are many things we already have a design for and wanted to do looking forward that can lower this risk but since the fundamental requirement was the traders will have creating privacy so this was the first solution we decided to go with got it so it's almost like this is potentially the first shift version on a road map I guess yeah and I understand what you're saying here so it's not about decentralization for decentralization sake it's rather like when people want it they want decentralization because there's no point of failure you've created the committee to sort of

mitigate having a central point of failure but it's not all the way to being like completely mitigated or like I mean I'm sure it's mitigate it enough but yeah I mean the thing is what the solution mitigates is for example neither diversify or Starcraft 2 malicious these take users funds and use them for any other purpose we could never financially benefit from any you know disaster scenario essentially what we initiated is is these sorts of issues but there are still scenarios where for example if we all go up line or you know the later sucker front where funds can get lost you can get stuck basically and by having this committee we've you know mitigated that but not as much as if all the data was fully published on chain

but it's still extraordinarily unlikely that you know these essentially sort of very well-known organizations who anyway you know in the business as Tom stared with occlusion and cephalic part of essentially one increment state systems you know at high availability are all going to go offline at the same time and not be able to you know and so it's a very good mitigation and by the way like I I think the you know the concerns are really interesting and things that will be improved on in the future as you said it's part of the roadmap but none of these concerns come from traders that's something these are from you know real decentralization and with experts who of course you know wanna see an improve and so do we but it's probably does right now I think actually solve exactly what traders need

and anything else is almost a bonus to you know the actual users of the system and this is this is significantly different from the centralized exchanges where there are central points failure or there could be central very centralized malicious actors who could do things to your funds this is already like a pretty big leap away from that it sounds like yeah definitely and I think the I mean something I got really excited about just last week which i think is another sort of quite strong indicator of why defy and other debt systems are going to start to you know continue and continue to grow was diversify launched now five weeks ago and we got insurance

last week from Nexus mutual which is you know this on ethereum like stating system of path essentially where people can put beef at stake on different smart contracts and if there's ever like a hack or breach or security incident with lost funds customers can get refunded and five weeks after launch this system is now insured up to a thousand bees which you know seems maybe trivial but if you have to think about that that's something that no centralized exchanges or very very few centralized exchanges on any scale have ever achieved and this is just you know a starting point where it can continue to grow and that's because you know the way the way that is designed and the way that it's auditable and transparent on chain and despite the

fact that yes there's no some centralized points means that it can actually be the risk can be quantified and if a huge yeah difference cool I want to move on to I want to talk a little bit more about starch themselves because we I mean so we've had Elie and Alessandro on I've actually had Ellie on twice to talk about Starks and Stark where so there's some episodes that I can actually link to in the show notes if people want to dig in there but what I want to find out about now is how Starks inform how this system functions because as we understand as I've always understood snarks and starts differ in terms of proof size and verify our verifications

I or sorry you ready rather they differ in terms of proof time and verification time so how does that actually change this system first of all one more thing that they are I mean they're dafair in their mathematical construction which implies that one did their dafair in in their trusted set up or stark does not have trusted set up and sonarak do have trusted set up both in the universal trusted set up version or in the specific circuit trusted set up version and also they differ in their cryptographic assumptions that they're using so Starks only need collision resistant hash function which is very

very actually state well researched and yeah it's not exotic it has a lot of I mean as early stated it's very nicely in is a lecture and post it has a lot of value on top of it so if someone will break this then I mean there's much more value currently right then the diversify system that they can take if if they could actually break collision-resistant hash functions yeah so and I will get to the proof size because there is difference but I mean it's you need to compare apples to apples right so okay so but you're also right there is difference in the proof

size so for example the proofs that we're now using prove diversified batches are on the order of 40 kilobytes and if for example you will look on the on the proof of snark of sapling of secret of dickish then need 200 bytes right so it looks a big difference but actually so let's say we want to prove 300k transactions in a single proof okay so if you will go to snark proofing system then you are limited by the size of the circuit that you created your trusted setup for so you can't use one truth in order to prove this amount of transactions so suddenly you need more

proof let's add an O 10 and also if for example will we look at the on the proving system that ZK sync is using its name is plunk and their proof size is approximately and I hope I'm not wrong but I think it's approximately one kilobyte so suddenly for example there might be statements that we can prove in one single proof which will take another 40 kilobytes but they will need trend proofs or even more proofs so suddenly it becomes I mean it depends exactly what you're what you're comparing if you think using plunk I didn't know that actually yeah okay cool so in a like a naive comparison of the proof size really yield the the fact

that stark proofs are larger but their ability to prove much bigger statements I think mitigates it and it's not things that I can talk about them now but in the coming month we will have very I don't know in my eyes amazing improvements which will I mean close this gap okay so this is one thing now on the verification time then I even don't think that this is correct for example we released I think it was last week or two weeks ago the East is stark if you're familiar with it it's an open source prover that we were released as part of a project with the interim foundation which proves hash chain of rescue which

is some start friendly proof and I don't want to misquote the numbers but we proved a lot of hashes in a very little time okay I think it's something like 100k a hash chain in a two point three millisecond but again I don't think there's any snark system that can make this proof and the verification time would be lower so again it's it's I mean there are many numbers which are quoted from the original academic papers but when you actually go down to the engineering details and how you construct those systems then the differences can can can can be very situational can you say it yeah what I wanted to say is that we are at Starke

where we have no like a wedding which starts we are a technological company and we want to solve problems so once we will see that there is a better proof system I mean we have the manpower and and the all the things we need in order to move to those proof systems which would be better so that we don't have any stake in stark-raving system and and we're also always keep on on evolving it but once we will find something better we will immediately switch for it and it might be the case that for some things one proof system will be better and for other things other proof seems systems would be better the way we see it now is

that Starks has almost any advantage that you can think of I want to ask sort of a question about the collaboration itself like how did these two organizations actually work together I think the I mean the kind of line between the two is pretty clear and some experiment so essentially from you know from the customer perspective there is no there is no stark well almost so that their interaction is all with the smart contract on chain or with our api's or user interface and essentially all of the kind of sort of main exchange software like you know these the website that you're doing to face the api's food

to the point where trades basically get matched or transactions decorated in let's say diversifies database it'll diversify side actually I think the easiest panelist is talked about the way that the platform used to work so you know when I last came on it you know a year ago actually we were running what was then called s the next trustus it was a similar you know concept exchange but every single transaction when it was ready this is like a trade for example we would send directly to etherium and we would pay you know however much gas let's say 150,000 gas for that transaction I would get accepted into the blockchain and and that's that would be the users trade complete but basically that step where you know a transaction would get

submitted to aetherium instead of doing that we send it to like an API provided by Stark where and from that point on you know everything along the journey is then stark wise side of the system where they would collect you know I don't see enough of these transactions to generate proof and then broadcast it to aetherium so that's kind of yeah where the split is and you know in terms of also kind of linking to you know the efficiency point of view and the proof sizes basically the old system compared to how it is now you would have needed around it's only around 20 transactions for it to become more efficient using the best proof

because beyond that point basically that even the cost is emitted ethereum becomes cheaper does the user I mean the users still hooking up a wallet though right like like they they don't their accounts they own their accounts when they use this this system so they link up their wallet to your to do diversify they make a trade I'm kind of trying to follow that that route like there's basically a smart contract that's been deployed that interacts with their wallet I guess so actually it's only at the point where you deposit and withdraw that you interact with the smart contract I see come along you deposit which is a transaction on chain where you send your

funds to the smart contract and actually from that point on all of your interactions are no longer with the blockchain because your funds are registered to you sort of in this off chain state and when the user with them wants to trade what they're essentially doing is saying I want to place an order in order to buy ten aetherium at this price and they sign it with actually a separate key and it's a key that's been linked to their etherium address which was part of the kind of registration registration and deposits sort of step on chain and the reason for that is that this particular key that they're using is more efficient in the in the start proof that's data used and that piece of

information that they own sign would get sent to diversify we would determine it you know would match with another order basically by adding it to an order book and having a matching engine that then you know connects to traders one wasn't studying one person's buying and then having decided that would send basically that original information that the user signed to stock where which they can use you know this signs information to generate the proof cool and so what is it happening like and you do I know you just described it kind of earlier on Tom but like let's follow that journey so now it's it's with Stark where what like it's is it just sitting like is it the prover itself does it sit on your servers like where is that where does that live in the cloud and that is but

this is like a this is basically creating the proof to validate that yes are correct yeah yes it's true so as will said the only truth that we know is what diversify center so they sent a stream of transactions at some point we we decide either if there are enough transactions or enough time has passed to close the batch this batch actually what it represents is a state transition right you transition from a specific state to a new state which represents the Oh all those transactions taken effect so for example deposit will mean I add tokens to the user balance withdraw mean ice obstruct tokens from dues about us

and a trade is just you know a trade so we closed the batch we generate the proof as I said before the proof attest to the validity of the batch the signatures the balances that we've never used this transaction before and then we send the proof and the state update unchanged so I mean a self custodial system is itself custody up to the point of its weakest link right so if I have a perfect self custody system but the user does not know what what it's signing on so anything else is meaningless because I can I can give it whatever I want to sign right so the wallet is a really big

part right if you truly want the self custodial system you need the wallets to be fully integrated so we both the verse defiant and Stark we're put a lot of effort into this specific topic for example as well said we have a slightly different signature mode so we worked together with ledger to insert all the necessary parameters through their femur and to integrate our all our methods into their nativity remap so we ledger user now can use diversify system and he trip and it's truly see the confirmation

it doesn't see just sign this blob of data but when when you want to do in order the ledgers show a confirmation of an order and exactly what the user is signing on and this is also true hopefully soon for argent and we have a lot more coming so for us it was very important to have a true not just to have in browser wallet but to have a truly integration with the with the users wallets okay so when when it came out there was whole conversation around kind of a like what are the different kinds of layer two solutions is it as a rollup does it fall under that umbrella term does it

have something else is it using optimistic ideas like what basically this like all of the different layer twos came back on the table and people felt that they needed to redefine a little bit where they fell and the axes that they seem to be working on was whether the data was on chain or off chain and whether it's doing validity proofs or fault proofs so maybe we can talk a little bit about this I don't know what to call it like a table or these these axes and where your solution falls so I think the the conversation about it or the discussion about it actually started week before we launched and it was this week we start or demonstrated the like the migration of a full subreddit into etherium and then the

question was asked what you've just demonstrated was it Zika roll-up or worth it some undefined system which doesn't have the data on chain and what we demonstrated was with the data off chain and so the community said and I think it was it was correct to say that this needs to have a proper name so the name of alidium came up so as you've said there is there are two axes and in the very naive descriptions of the systems you have validity proofs and fraud troops and the different on the second axis you have data on chain or data off chain okay so I guess that the first system which were in which was in

research was fraud proofs with data off chain and their name was plasma yeah right you have a single operator and and those systems required very complicated exit games because you had to prove with the data of chain that the operator was somehow malicious so this was the first then I don't know exactly the order but fraud proofs with that on channel called the optimistic robots so that just described that so fault proofs if you're just looking in that category you have plasma which is off chain and optimistic roll-up which is on chain yes exactly and here for example you have arbitral which I think falls in this category and

here again you publish that own chain so in some sense it is it is supposed to be simpler but still you have you need some kind of fraud proof mechanism to to prove an invalid state transition right this is what you what you use the fraud proofs for what did you just turn you just said the arbitral is that a product that's a project right yeah it's a project okay optimistic yes I think I think they were they existed even before it was called up to mr. Corolla but this is basically what they're doing and invalidity proofs then you don't need to you don't have this problem of invalid state received transitions right because

you can't move to an imbalanced invalid state because you want to be able to generate a proof of it originally the the I think it was described by Vitalik in a post in its research so you had Zika roll-up which was using validity proofs and data on chain and what we opted for was was used it was their putting data off chain and using a committee and for this family of solutions the name the lidium came up so back again in this case we're looking at the validity proofs and the two kinds you have the data on one chain which equals the DK roll-up and you have the data off chain which equals this new category ville idiom yeah yeah I think

you've interesting as well to think about also the difference between the chord proofs and the vanity proofs in terms of trading as an application because obviously there's different trade-offs where each of these you know different scaling solutions make sense but particularly for trading systems where the amount of value at stake in quite a short period of time can be very high particularly if you have large traders operating with big price moves and you don't want to have the risk of potential war backs or changes where the game's theory also becomes more complex that you know the higher that's stating the bastards changing so it is a really strong push I think towards validity proofs in this application you know from diversify point of view as

well thinking back to a year and a half ago now when we were first considering different layer two solutions and is a very strong push towards at least you know that that kind of right-hand side of the of the quadrants and with Polizzi proofs as opposed to four please yeah that actually makes sense I'm going back to the fault proof in the case of like say we take that exact example that you just used of like large trades if you were doing it with fault proofs what you basically have is I mean there's sort of you could potentially do one of these large trades that was sort of wrong and it would take some time for that to possibly be rolled back right like it's this after the fact kind of game theory

that would like have to correct something rather than what you're doing where it's like it must be correct before it's written it just can't be invalid okay yeah we can't prove something that either the user didn't sign on on we can't prove a state which was not updated correctly it's just not possible mathematically so wait what I just said is that basically correct no it's true and this is why this is why so for example in our in our system you have finality when when they're proof which is the blockchain right because because then the state transition to a new state in fraud proofs you don't have this kind of finality because you have to allow users to verify the state

understand there is a an invalid fate transition and then start to reconstruct the fraud proof and then there's some kind of exit game on the blockchain which needs to to to occur so for example for some optimistic roll-up designs which were described the the finality time for a state to be actually finalized was two weeks yeah that's what I've heard to like like not just days but potentially like quite a bit of time which is worrisome if you're using if you're like dealing with large that potentially have something wrong exactly and there's a lot of game theory analysis to those kind of systems

because it's very hard to mathematically prove it right it's it's it's game theory and it's basis so it's very hard to make arguments for very large amounted to take ok so actually we're at the point now where I feel like we've really nicely mapped out these kind of different quadrants and stuff and I think maybe this is a good time to talk a little bit about what's coming up because you have a project in the works as I understand it which actually tries to to bridge that on chain off chain I want to hear a little bit about this ok so there are low there are a lot of things going on and which will come to

the future and let's try to take it one by one so we talked about that availability and I think there are two two major improvements that we can offer in the near future in this in this area one is it's called volition and this is a solution which basically is a hybrid between the lidium and is ek roll up in the sense that the user can choose whether its data will be on chain or off chain right so the user can choose whether it wants to pay a premium and does not have the risk of what we in the beginning of the episode described as some worst case scenario with the

committee or it can it can choose to have its data often and have the full scaling capabilities of the system would one of these would one of those options be faster like if you're doing on chain is no just you have to pay for the data that you that you send some chain okay and and and the nice thing about it is that you can choose it pair transaction I mean it's not a choice you make when you register to the system but it can be a dynamic decision for example a big trader can can say generally I mean I don't know at night when I'm not trading I want to have my my fun stored in the Unchained at apart and then I

don't pose any risk but when I'm trading I will transfer them to the off chain data part have my trades in in a very cheap system and once I finish trading I will transfer them back to the to the end unchanged at the part so so this is one improvement and the other improvement that I can I will hint about but I won't describe exactly because we still haven't published it but as we'll have mentioned at the beginning the decision to go with that off chain was was because of trading privacy right so right now there's no solution or at

least there's no widely-known solution to two books have data on chain and maintain privacy so another solution might go down this road you can have that unchain but still maintain privacy of the traders there is something I don't know if you've heard of this there is the zk zk roll-up but I don't think it's implemented at all yet I don't know if you've heard of that but that's the private zk roll-up exactly so we have this this design and what's nice about this design is that it has benefits both if we choose to have that option and on chain on chain it's very clear right you can have the security but lose your trading strategies but often you also

have a lot of benefits because now you can suddenly have much wider committee because one problem with the committee's that they know all the users data so you have to be very careful and have very strict rules on what they do with this data but once you make this data private then you have less problem with expanding the committee I see what you're saying Yeah right now you need the committee members to be somehow trusted because they do have sort of some site into what's actually happening whereas going forward you could potentially expand out the committee because you're actually gonna make it like unviewable by each

member so they wouldn't be able to be malicious yeah in the ideal case every large trader becomes a committee member and holds all the data at which point and the smaller traders don't care as much - I mean really both of these two solutions end up pretty much meaning that as a customer you have the full range of choice over whether you know how much you care about your security in your privacy and right away you want your own trade-off to be and of course you know that they can be default but basically you're saying the customer isn't you whatever you want yes and I think that cool I mean if we think the deuterium and the financial activity and etherium will be what we hope it will be and it will I don't know

expand to a worldwide I mean it's it is worldwide but let's say mainstream I don't know how to say and there if we want to have big adoption yeah big adoption then opting for the data unchanged solution just does not hold right because it is a short-term solution but we must keep researching what what solutions we can afford that allows us to have logarithmic scale meaning that the amount of resources that we consume on chain grows logarithmically with the with the amount of transactions we do often because otherwise we want to be able even even

if we go from I don't know three trades per second which was before layer 2 solutions to I don't know 300 trades per second which what the curve Olaf can can give you today this is really this is not enough yeah I think that's reducing point in that already you know the size of the state on ethereum is already too big and even though yes you down the amount of it you need to put on with Unchained a tip is EK roll-ups it's still actually you know potentially gonna grow way too big if there's ever an option yeah so I think there's no escape from when finding a really good data availability solution and saying we

just put all the data on chain because this is like the safest is won't hold got it okay so it sounds like this is not it's not like you guys did a one-off partnership and then it's over it sounds like there's still a lot of work going on so I'm curious to hear like what else is happening what's what's gonna be coming as you move forward absolutely I mean entirely the premise behind this and I think what makes you such a strong partnership is that diversify specialized on exchange self custody exchange and equally you know starquest specialize in the scaling solution and so the combined effort means that we can move very quickly and part of the plan is to keep you know pushing out changes

and improvements over the next however long and and one of the ones that's found soonest and which is actually the thing I'm personally most excited about is basically an upgrade which allows essentially better and faster communication between layer one being etherium and layer two being essentially you know where these balances are held off chain and so the simplest example of this is the right now one of the you know common problems for someone who wants to use anywhere to exchange including diversify is that because of this batching system actually you may have to let's say you want to withdraw funds out of the system you have to wait for the next batch to be complete before

you can get your funds out and because there's a cost per batch actually right now batch maybe only every hour or even less often because we want to make sure there's a certain number of transactions included in that bath and so if you want to withdraw your funds you have to wait now this upgrade which I think I'll let Tom talk about the kind of real in-depth photography behind how the proofs work basically allows you to have a conditional proof that says for example I want to withdraw funds from diversified and I want to get it right now what some someone can send funds on Shane directly to the to the user and they can sign this transaction that's

valid in the next proof that says ok in return I will - let's say they're withdrawing five five aetherium in return I will transfer five aetherium on diversify to whoever sent that on chain transaction and it but it will only be valid if I received my fire pit by the theory of Montaigne you know within certain amount of time and so that then allows you to have direct interaction between you know this like wanting transactions and the proof of chain and that actually can be extended even further to mean that you can have direct interaction between any d-5 sort of protocol on chain and and trades on diversified so let's say you want to deposit funds into compound interest you could again do a

conditional transaction to say like they say someone else has funds on chain that they deposit the compound you could receive those in diversify and vice versa which allows all sorts of arbitrage opportunities and basically and meaning that this layer two system is no longer sort of isolated and separated from all of this exciting stuff that's going on in the rest of d-5 it can become fully integrated and have all the same sort of interactions that that can happen with other protocol so I mean to me that's basically that's the big missing piece right now in terms of layer two because you kind of made this trade-off to say yes it's scalable but it's no longer connected with all of this sort of you know money lego type stuff that everyone is you know going crazy about finally we kind of connect

the two yeah yeah so I think I mean I share the excitement of will because I mean as I've said at the beginning Texas was really matured last last year and now we have a scaling solution for them and I mean the next step will be to connect this escalate solution with with a whole ecosystem so what we actually constructed is a way to condition and offer chain transaction on an on chain transaction okay so I mean the easiest example is it is actually exist in the

system today right if diversify wants to send to tell us off chain this user deposited I don't know 5/8 then the proof will only be accepted on chain if an actual deposit transaction happened on chain right the first if I can tell us this user deposited and the proof the truth won't be accepted if real deposit didn't happen on chain right okay we can start that's how it is currently in that first sort of element that's that's the way it is yes exaction that's how it is yeah exactly so and and and now we can expand this idea to basically any transaction

on chain so as will said the fast-food role is is you conditioning an off chain transfer on an on chain transfer right I'm I'm saying I will give you an five ether unchain only few will give me five ether on chain off chain it depends so let's say I want to to get my money on chain very fast and you're you know a big liquidity provider with the Zeki validator so I want to get my money get my faith on chain so I conditioned Mike

my transfer to you off chain on the fact that you transferred if you eat to me on chain so so you get this transaction it's signed it's valid you know you can you can include it in the proof all you need to do is transfer me the funds on chain once you did this we can include it in a proof in the proof will be valid because the proof guarantee that it will only be accepted if this transaction really happened on chain and what will said is is very nice because the transaction can be any theory on transaction so I can condition an option operation based on any on chain transaction I mean the potential there

like I what what do you think could be built out of that like what kind of systems so I mean I'm very excited about it because it's it's it's it's an infrastructure that we will include in the next version that will allow basically the developer community to make innovations for the system right because anybody can write those types I mean the way the way we do it is that you deploy a contract and you do this operation through this contract so basically anybody will be able to write a contract I mean you have a new idea and you want to provide it to diversify in exchange to diversify users in exchange for some fees then you will

just be able to implement it the same way d5 works today right this is not I mean it sounds good but it just like the way that it's happening in India by now where people come up with ideas like slash loans and and anything they want and suddenly they can all interact together and I think some of the things that I'm most excited about it definitely going to be to start with you know quite simple just allowing people to interact with borrowing and lending protocols for example to take long and short positions on diversify instead of only long positions you know right now you can only buy a token you couldn't come and short it and but you know if

you can interact potentially on chain with following it borrowing platforms then suddenly you can do that on diversify and but yet as Tom said you know the real the most exciting pieces that anyone can actually start to come up with other ways of doing interactions and I think that will create a whole whole range of opportunities that most people haven't we suddenly haven't thought of yet yes I mean that's cool well I guess it sounds like it's uh it's a project to keep keep our eye on congrats on this launch and everything good grandma cool so this is this sounds amazing I'm really excited to see what comes next me too yeah thanks for coming on the show yeah it was a real honor being here

ok and to our listeners thanks for listening you [Music]