How MPC Technology is Powering Private & Secure Data Storage in Crypto with Nillion Network

Transcript

welcome back to the rollup High Roller Community as always nothing on the rollup is to be ConEd as Financial advice we are just an Educators in the defi space toss us a quick like give us a share a repost send it to a friend who may want to watch this and finally give us a sub or a follow it means the world to help support credibly neutral educational content in def5 we're here to educate Empower and enrich you and let's jump in what's going on guys welcome back to the rollup here to today with Miguel from million we're going to talk about a interesting Dynamic when it comes to the differences in scaling using sharding um and some MPC which is a separate uh type of Technology from

you know the typical type of scaling mechanisms that we see uh in crypto today Miguel we're not NPCs we are MPCs what's going on man what's going on Andy great to have you uh to be here with you today Yeah man so how uh how did you go from um being a surfer to being an MPC cryptographer oh it all happened you know when when I left the Canary Islands where where I grew up um I moved to the Madrid and then Munich and then you know places without uh waves so it had to happen but I did have um like my revenge with snowboarding um so it you

know there is a still a a surface a surfer in me somewhere and which is your favorite boarding or surfing I think actually boarding boarding is yeah it's pretty cool I think I actually my cousins live in Munich a little bit off topic here but this is like this is what the Ro's all about it's infotainment we're going to dive deep but I'm gonna go skiing in the Alps in about a week and a half so I'm stoked on that maybe I'll be thinking about the MPC design of them drawing my lines down the uh slopes yeah yeah so when did you join the team um were you there from the beginning yeah I was there from the beginning uh I think the story started like back in 2017 when I met Andrew masanto he's um

like a Serial entrepreneur and he's founded a couple of crypto unicorns at era hash craft and reserve and initially we were talking about you know playing with the idea of creating a privacy coin like zcash or Monero but he's got a real good sense for market and back then he said no this is not the the right moment for something like this and then in 2021 we met again in brussles I had been working on NPC multiparty computation uh so by the way maybe I can explain what what that is it's it's just a way of allowing multiple parties or nodes or computers to compute something together whereby they contribute private inputs to that computation that are not

seen by the rest and they all get the result from that computation somehow and so it's a cryptographic protocol and it is like um it has many applications to to crypto and we were and I had been working on ideas on how to accelerate MPC and make multiplications be more efficient and so we met in Brussels in 2021 with also with Alex paig the CEO of nilon and Andrew yo the chief marketing officer and we may brainstormed about you know how to use MPC in the in the context of this centralized networks and we got really excited I mean with looking at it from different angles uh we saw the potential of um creating a new technology based on these ideas that

would be like a new type of decentralized infrastructure different than blockchain um we can talk a little about the difference differences if you want um but nevertheless that provides something something else right something um something new uh to the space and in essence you could summarize it as as being like a secure computation and storage Network that uh decentralizes trust for high value data just in the same way that blockchain uh decentralizes uh transactions it's more or less kind of the summary so this breakthrough is more more focused on uh secure decentralized uh kind of uh data storage um and message passing and this kind of

a transfer of information um and you guys kind of came together in one of these like uh brainstorm Mastermind uh how to get rich conversations and was like Wow guys we're gonna actually make an impact in this space um and be able to kind of put this together and you guys actually decided to opt away from the typical blockchain space which was probably kind of you know still early when you guys had that conversation so this breakthrough that you guys have made with MPC see um you know I'd love to kind of learn more how that idea has came from an idea to now kind of what you guys are putting in practice with

alignment um with the overall Vision yeah so I I think the story can begin with the notion of of um high value data right in in a sense transactions are high value data because they're they're this the movement of assets but they kind of prescribe a certain way of architecting dat data and compute for example data it has to be immutable because you have this Global Ledger of transactions so you don't want to delete transactions and also you need um kind of a common consensus on the state of that ledger so there's a specific type of compute which is a consensus mechanism that kind of provides um like supports the the This Global notion of

um of an agreement on the state of that ledger and also The Ledger is primarily public so so but that's not the only type of high value data right there's other types that also call for different models for example the data model could allow for storage and retrieve and also deletion and updates of of the data that's kind of more similar to what we're used to um experiencing in a normal computer and it also doesn't need to be like um you know like a monolith um agreeing on the state of the of the blockchain but rather you could have like clusters operating almost independently just like threats in a in a CPU so that's kind of the starting point for for this network to

acknowledge the differences compared to to a blockchain infrastructure and once you acknowledge that then you start designing um like like the diff the different types of networks like Dev net test net and so on so that entirely mimics what you would expect from a blockchain but each one of those are are very different um we we make use of blockchains because you know it is very useful for certain cases when you get Global when you need Global consensus on certain things and you need to orchestrate the different elements in the network then you need a blockchain but the rest is it looks like a clustered network of you know of different groups of nodes they operate rather independently and each one of

them kind of serve um they they perform computations on behalf of the users on data that is kind of stored um privately so it's not in the clear it's not like in a blockchain and um and provid like the answers for those for those computations back to the to the users or whoever whoever needs them and so this is this is something that you guys came up with and have drawn up with in different technical papers over the last couple of years to kind of come up with these um answers right and and I know that you guys are working on one now um as well which I have this feeling that you guys are just never going to stop working on papers it's just going to keep on going until

until you guys are like all right we're done uh which who knows but yeah so what's the so so what is the current paper that you guys are working on now that you know we've established this Foundation of what uh you know kind of the vision and what the operational capacity of million is you know where are you guys looking to go now with your upcoming paper and kind of research yeah so I agree with you that this is a Never Ending Story of of papers is never going to end up and so far we've been you know focusing on development on on developing the the core infrastructure so we've been a little bit shy on on you know showing the the work to the rest of of the community so now we're really excited because there's a time where we

um we've hit a time where we are going to change that and we're going to start publishing a number of of papers beginning with the first one which is essentially about that core technology I was talking about is is one of them by the way we have also other other different uh Technologies but this was um a foundational one and it's about essentially how to do MPC with um multiplications in an efficient way so any computation that you can think of can be boiled down to additions and multiplications additions are are typically performed by NPC protocols very efficiently but multiplications typically require the notes to exchange information and so this technology allows you to do certain number of multiplications and additions for free

like with without communication and in doing so it provides a really efficient and fast way of of of doing things and you can you can use it for things like um authentication signatures private database um retrieval of information you know all those things that are not too big in terms of computation but that require that that are very time sensitive so they they need to run very fast because there's a user waiting for that and users are are are you know kind of used to certain standards in terms of for example AU indication they they cannot wait there for 10 seconds it has to be fast so it is it is a technology that enables all that all all those use cases and following that we've already

published um all all these two paper these two papers have been talking about so one is the authentication that paper is important because if you think about decentralizing you know you have high value data you don't want to give it to one server to one company because they have full control of it so that natural thing to do is decentralize it right so you break it in into bits every bit is private so they cannot gain any information from that single bit so that's good but next comes the problem of accessing that data so you need a data access structure that is also decentralized because if you rely on centralized authentication against a single server then that single server has your key that can access all the data so you need to decentralize

authentication and that's a bit tricky so we published this paper that looks at the problem from different angles the legal angle the engineering challenges and also the cryptographic protocols that are based on the on the previous one I mentioned uh so that's this other paper and then there's another paper on signatures which is again a very interesting problem like okay you're signing transactions sending them to the network that private key controlling those signatures is really important so what do you do with them that do you do you store it like in your computer probably not a good idea do you keep it off of line on paper okay that's better but still it has problems attached to that so MPC um solves that because it

can if if allow it allows you to decentralize that key so you can break it into pieces give each piece a different note and now no one has the all the information to reconstruct it but you can still instruct the network to sign on your behalf so there's a number of companies already in this space I think it's kind of the main use case for MPC technology right now and we thought about okay why don't we use the thise technology I was talking about and see how it performs in in when we apply it to signatures so that's this paper and it again it's a technical report it's not a full-blown mathematical paper but it does present uh protocols based on on this core

technology and the main feature is that it reduces the number of rounds of communication to the bare minimum which is I need one to send you the the hash of the message and to get the signature back that's it two messages and the rest of the signature computation is local and it also has um more like technical advantages um that are maybe not for this audience yeah yeah we're getting into the technicals but I'm I'm curious about it because with these um with with MPC um you know what's coming to to my mind with what you're saying which um it all sounds extremely efficient and effective for decentralization of data for a storage of data for transfers for secure uh passing of of U information

and data but a question that I have is about the incentive structures um and perhaps you could give us a practical example of how you see this technology with what your building being either implemented today or planned implementation where and I'll give our audience a simple example where in proof of stake models on blockchains the actors put up their stake you know their coins their their assets you know and they can get slash if they act outside of the network you know or um if they try to do something uh you know kind of in in bad intent they can you know they get penalized so there's this kind of like incentive design super basic right there's tons more incentive designs and blockchains and computers but with million where is this where is

with MPC where is the incentive structures aligned for these different nodes and um people or entities or servers who are acting within the network to kind of in these protocols that you're explaining and like kind of abide by the bounds of of what the ideal kind of structure is in this decentralized fashion is there like is there a monetary incentive is there is is there like a a a penalty is there some sort of idea that we can understand from a practical level as to how this works um incentive design that's that's a great question by the way so MPC is not pro proof of stake but there is staking so the the way we're implementing is it is

uh we require the nodes to to put up stake and also we build control mechanisms that allow to identify situations where the nodes are acting in in a malicious way and that is slightly different than what happens in a blockchain so in in MPC you can act primarily in two two main bad ways so the first one is you silent you follow the protocol so there's there's no way to really detect you but you're silently leaking those pieces of information you're getting from the from the you know from the users um so you're sharing that information you know with the other

nodes with the hopes of gathering enough information to reconstruct the the data that's those are called passive adversaries um and actually they're they're hard to to detect uh because they follow the protocol so there's there's no not um an easy way to detect them but you can you can place both like incentives to for those nodes to report uh if there's another note approaching them with those intentions and also of course the staking that once that is reported then you get slashed and then you you are penalized by by doing that so you can address that from the token economic perspective there's also ways in which you can address that from a

more cryptographical sense which is to use encryption so that those Shares are never in the in PL text so that's for example what we use for the signature paper so the shares or those bits of information are actually encrypted so even if you gather them all you still you will still have to find the private key to decrypt them so that's a kind of full you know another layer of protection but then you have the other type of adversary which are active adversaries which is the Byzantine adversary you're used to from blockchains those are on the ones who deviate from the protocol in any way or form and they just want to in our context what they can do is is change the result from a computation not in a predictable way but they can make it like produce random you know gibber um

but those are interestingly enough there those are easier to detect because you can build in mechanisms into your protocol that have by the way the same ratios of like up to three onethird of of the notes can can be bad actors and you get Byzantine full F tolerance and stuff like that because the underlying mathematics for both um Byzantine agreement and MPC um computations are the same so you get the same um kind of trade-off um and then you can you can identify those nodes so you can detect there's three levels of of um protection here for number one is you detect that something went wrong and then you abort number two you get even better and you

can also identify y the no who were um Bad actors and you can slash them so that's the minimum level we're interested in because that's the one that allows you to slash and then the third level is where you can even correct the bad actions introduced by those noes so they can send wrong messages and stuff like that but as long as those are within a certain ratio you can actually correct them because there's enough redundance in the network to um to do that gotcha okay and then in million's case is the in I guess also in other NPC networks is the stake simply software and tokens or is there also Hardware

involved it's primarily software is in tokens um and also we see like a path to decentralization like we are not building from day one the full-blown security model that I just described we're going to go from scenario where the initial noes are more trusted to progressively adding those mechanisms to detect um bad actions and to progressively um you know add more permissionless into the network so that the onboarding process of the notes becomes more and more relaxed um that's kind of the the goal yeah okay gotcha yeah that kind of is is the reason why I put sharding Maxi in my uh stub name here because in the

previous conversation that we had with um one of the lads from from million it was it was directed around shardan now was that a previous paper and kind of focus and that it's it's it's it's shifted or is that still tying into kind of what how you see MPC and million playing out like is this distributed technology of data also a form of of like charting in the sense where there's different uh nodes and different pieces of data separated amongst all of the nodes that makes the the the whole is that how we think about charting in this sense or is that just something that we're not really focused on any any you know anymore in a sense what how the way um data is protected in NPC resembles

charting like you are breaking down information into into pieces and those pieces do not you need a a minimum number of pieces collected to be able to reconstruct the the information and you distribute those pieces among the notes in the network so it kind of remains resembles a little bit what you're describing but um well formally speaking it's called a threshold um threshold um linear secret sharing scheme so it's a way of breaking down information into pieces and if you cross a threshold meaning you get more than certain number of pieces then you can reconstruct but if you don't then you can't learn anything from from the data right it's

like going to the bank and it's like giving half your private key to your mom the other half to your dad and one word for yourself Ex two examples that come wait go ahead yeah if they collude and and they don't get this to this critical number of of pieces then there's nothing they can do it's it's just as good as if they hadn't colluded at all right and then obviously the incentive structure will be aligned so that doesn't happen that's that's more of like the uh in in that example that's more of the passive adversary word exactly yes gotta yeah yeah makes sense and again that doesn't make any sense for people to do necessarily unless the validator account is very is very low where where in in the beginning you'll

have you know trusted validators likely people you know perhaps investors you know trusted people that you know kyc and things but eventually there's a world where permissionless anybody can be a validator however they want they just need the required stake um so I assume that'll kind of open up that open up the uh the strength of the network more and more to where it scales somewhat exponentially as you add validators I'm curious if you know uh and or if you've heard much about like EIP 4844 where PLO dank sharting on ethereum where they're separating the Blobs of data from the transactions of the on on ethereum this is just got got me thinking kind of similar to where where you guys are where it's like um

you know separating obviously from from from blockchains but just kind of focus on on on the data aspect I just bring it up because it's coming soon but it's something that that's interesting yeah there are again similarities to what to to the the structure of our architecture which is based on clusters we call them clusters and each cluster is operating on a on a like a segmented way of you know segmented subset of the data so the the whole data on on on the network is actually divided in in segments that are assigned to the different clusters and within a cluster you can operate you can do whatever you want with the data with in that cluster uh and and in a sense this achieved certain degree of

decentralization along the logical axis that vitalic describes like he described the three AIS of logical decentralization architectural and and political blockchains typically just achieve the two of them like architectural and political but this one is going a little bit on the other direction as well because to a certain extent the operation of those clusters is fairly independent so you can slice the network into two and okay now you have the cluster was divided in two but that still can function to a certain extent because of that division of data that you were mentioning um precisely yeah another some another kind of U you know because a lot of the stuff that that that you know I've been learning about on the modular and the rollups and

ethereum side there's I'm drawing similarities here so bear with me as I draw another one a big concept right now uh distributed doat technology in etherum basically it's been expensive to run noes you have to 32 eth rocket pool Li came you you can just do you know liquid staking now they're taking that even further where you can run like a solo Staker no which is one three e 4 e and basically what happens is these validator Keys get split into like uh into you know however many uh parties there are in that 32 e um and and so similarly I'm curious about the the barrier to entry to become a node in million uh if it's going to be if it's a set value or if it is dependent on the

token price or kind of what that even looks like um and then is there is there a thought about like how you can potentially democratize that as well and like offer you know let's say smaller players to be able to uh participate in the network without necessarily needing to buy the whole node or to to own a whole node yeah good question um so the the detail of token economics have not been figured out yet so we haven't defined them yet um in my head there there will be like a minimum staking requirement uh for notes to join but it's an interesting thought what you're mentioning that actually what is a note behind the scenes could be like a network of smaller parties clusters yeah yeah clusters uh kind of

yeah sitting behind it's like a logical abstraction of of that level of detail and you for the network is it just behaves as a Noe but and and there advantages too I mean the advantages become that node could then be shared among you know half a dozen different GE geographical regions right further en enhancing the decentralization of that of that node thus the network um and then you you can allow more people to get in I just think lowering the barrier to entry for any uh distributed like uh stake proof of stake or just any distributed NPC network of information data value is like it's it's it's huge for yeah the only thing we have to be careful about is is civil attacks

essentially so those little notes if they get all the information from from their logical abstraction the chances because their staking is lower the chances that they collude with others without you know being economically sufficiently penalized are increasing so but but I agree with you in that it should be easy it should be that dead easy to join a network and it should also be kind of relatively affordable so that almost anyone can run notes another thing you can do is um trust like the notes don't are not like Anonymous but they are meaningful to you so maybe is you mentioned like half jokingly your your parents and you know your your

brother your sister your the people you would trust your sensitive information with those could be be could form a network or a cluster that are that is the cluster actually taking care of your data um so there's not just like random Anonymous notes they mean something to you there's C certain social bound um to to them and Trust lion and and that offers additional Security in that sense so that's another way in which things could be made cheaper and still without compromising security yeah I mean the more and more I I kind of understand the uh idea here in the vision the more and more I just think about like this private this private internet this decentralized web this kind of trying to disrupt these

adversaries or you could just I mean I view them as adversaries because I'm so deep down this rabbit hole but the uh common folk you know just views as like tech companies like the Facebooks and the ex and the uh social medias and all just like these the the internet clusters if you will that exist I mean the vision for you guys probably has shifted since when you started but you know I if it has could you maybe shed some light on where you see the best and most valuable product Market fit kind of implementation of this technology that you're building and perhaps like if you have anything like that's under wraps or kind of like um you know planned maybe or you can maybe at least give us a

direction as to where you see this kind of going like how like where do you guys make a change or impact or implementations in in the internet that we know it I think there's like two main avenues that U we we are exploring one is taking the current state of web 3 and trying to remove like um um or or unleash its potential for growth because if if you think about like there's been already like more than 10 years since it it was created and it hasn't replaced web 2 entirely one of the reasons is because there's limitations as I as I was explaining before as to what you can do in terms of

data access control and in terms of computational model behind um the the web 3 technology so in expanding and and making making them more flexible those those two models we we hope to make it you know feasible to kind of broaden the use cases that we're seeing with with web 3 to the point where they start p permeating through actually Web Two use cases and the other way to look at this is from the other angle like okay we have web two we have barriers into you know entering a web 3 ecosystem because essentially web 3 is missing a key component which is the ability to keep information confidential and web web 2 without confidentiality is almost

nothing so from that other perspective there's companies and and use cases in web 2 awaiting to be decentralized but they're hitting this Blocker of of confidentiality and MPC technology could be a an enabling factor that would enable you know would remove that blocker essentially and would enable Web Two use cases to to be decentralized as well yeah yeah it's interesting so so there's possibilities to somehow integrate with like you know blockchain based applications regardless of where they are and Implement these this NPC technology at like the foundational data level yes and in fact every every um

neon node is also blockchain node so it's also Layer Two blockchain node so there is a very close interaction between the two concepts to the extent that you know nodes are bilingual so to speak um and you can open up all those you know Rel you know use cases that involve the the exchange of assets on a blockchain but as well some sort of computation that requires decentralization but also handling uh other high value data that is not just transactions yeah authentication would be one one use case like okay you have this information which are the authentication factors like something you are your biometric information your password you know all all that

information that is not exactly a transaction but you can do an NPC computation that allows you to authenticate to a blockchain network which now allows you to then send transactions uh sign sign transactions gotcha yeah makes sense it's like the other part of not the other part but the it's the aspect that we honestly don't even see when interacting with blockchains it's kind of just like you sign in with your wallet or um with you know whatever uh data you have and you're just so used to it you just start transacting you're you don't really think about that but like that foundational level has a mass uh Market to to be able to to kind of apply too or

take for instance interoperability right so instead of just moving assets from one blockchain to another one you could move messages or information but also with processing in between so you could add some you know additional information that is high value from the outside world and and mix all together in one computation that then now moves one event from one blockchain to another one um the you know the the use cases are are really vast in that sense yeah so are there other MPC blockchains or NPC networks that you know of uh that that have tokens today and are they running on like is the token native to that NPC Network or is it like like different

type of token standard or yeah how does that how does tokens work when it comes to NPC yeah so tokens are not NPC are not an NPC native thing so you you don't run the token in NPC because as I said before like you should we should use blockchains for whatever their best equipped for which is token management so typically they they sit along like along the dmpc network and in in our case we will probably start with an ethereum based token erc20 or or the likes of it um but there is also value in in you know deploying your your own token because you have more control over governance um you you get you know fewer

fees and and stuff like that so there's other projects projects out there um they present different flavors of multi-party computation like based on fully homomorphic encryption or based on more standard linear secret sharing and they some of them don't have tokens so it's not a must have in in this space but some of them do and and they do what I just described so the they they have like a normal regular blockchain token usually on on ethereum and that's kind of tied up to the value of of offered by by the network very cool yeah that was a bit of a blank spot for me there I think I have a pretty pretty uh you know wide and

also pretty uh um deep understanding of uh of MPC from at least the abstract and high level obviously the cryptographer aspect is quite different um and kind of as we as we come to a a close in the coming minutes um you've got a cryptography background also founder uh your team members are some serial entrepreneurs I'm an entrepreneur uh there's builders in our audience a lot of developers Builders tons of defi users as well um what's kind of some thoughts that you have for early stage Founders whether they've raised or not um and whether they you know are engaged in cryptography or are you know in just doing other types of programming and building cool stuff you know what are

some words of wisdom that you can share with our with our audience as far as maybe um getting off the ground finding product Market fit uh you know having grit and kind of perseverance to see your ideas come through um you know team these types of things okay um I I think um one of one of them would be um that you you you need to deepen your knowledge on cryptography because it's becoming more and more prevalent in every web 3 project so understanding what not just hashing which is the basics but also what Ser knowledge groups are um and also other things like fully homomorphic encryption and NPC um you have to stay updated because this is

a really changing environment so you know following some leaders in the space would would be a good idea participating in conferences and all that I I would also recommend to focus on security from day one as in it's not an optional thing in in web 3 it's really important it's foundational even though it's a non functional requirement but is a very important one I would also consider engaging with Community from the very early stage because web 3 you know the ethos is about building with the community it's not just like a private company and the rest are customers it's like they have a piece of your company if they hold tokens so they are like

part of of you uh so embracing that from the beginning I think is very important and open source is is part of of that um I would also Focus like you said on product Market feed on use cases that's probably not the case for every crypto project um but it should start being like the norm I think focus on on Real uh product Market fit focus on on user experience as well um it's something that is not always that good in in in crypto project so still user experience is not like um you know um is is is is only geared towards certain people with certain skills so it's not

like embracing a wider audience of of users and I I think I would also care about regulations the haven't been that important yet but I think they're coming and it's a muddy and tricky um space because sometimes they're not clear about what they mean to to to to crypto right they they just regulate web to centralized environments but it's not really clear how to apply them to crypto so having someone in your team with that knowledge I think would be a great idea and um yeah maybe embrace the embrace the ethos of decentralization I think you really have have to believe in what it means and

represents otherwise otherwise if you don't do that um it feels like you're you're making a hard time for yourself because implementing something as simple as for example authentication in in in a centralized setup become becomes suddenly something that is really complex in a decentralized setup so if you don't really believe that you're doing that for a very good reason um then you know it doesn't it doesn't make any sense to complicate life for for yourself yeah yeah I appreciate that um and um some some valuable insights for the broader community at whole developers Founders so thank you Miguel for your time and um excited to see the

vision of Nilan come to play um and stay up to date thank you Andy it was great talking to you and U for everyone listening uh remember that if if you want to check out the papers uh you can find them pinned in our Twitter and yeah happy to to join the conversation awesome thank you thank you