ZK Proofs and Data Markets in DePIN: A Technical Panel

[00:00:00]uh we have a really special fireside chat next with Kevin guo co-founder and CTO of iotech Mark little CMO and BD lead of streamer and yev uh CTO and co-founder of demo and they're going to be talking about zero knowledge proofs which have applications in both privacy and scalability for deepin so let's give them a hand guys no there's exra check check testing hello cool on hey guys um welcome everyone to this session uh this fireside chat is uh ZK proof and data marketing Deep In following a recent uh proof of concept that was built um with streamer demo iotex webstream um so very excited to be having this conversation today uh some quick introductions uh my name is Mark little CM streamer um if you're not familiar with streamer we do uh PTP decentralized real time data streaming uh we're deep in infrastructure we've got a booth at the back if you're building in deep in come and talk to us and we can tell you more um great pleasure to be sitting here uh with uh yev and Kevin um mostly because stream has been working with demo since day one almost I'd say and uh recently we're working more and more with iotex and the webstream guys and building a loads of cool things one of which we're going to talk about today um we're going to run the session for about 20 minutes got some questions obviously uh but then we're going to open up to the floor so uh we appreciate some questions so if you can have a think about questions that' be great and I'll preface this session just by saying I think everyone's really excited about the deepin narrative in this bull run um and long may it continue you know deep in's here to stay for the next 10 years we all hope all of us building in the space and uh you know if we think about the amount of data and and how sensitive and personal that is there is a need for a privacy solution privacy lay on top of deepin and uh today we're going to talk about pref concept that's been kind of built in that space for the first time so it's really really interesting one cool um enough from me uh let's get into introductions Uh Kevin do you want to introduce yourself hello yeah sure uh I'm Kevin uh co-founder CTO of aex um yeah I I I think just now uh I think J has done a lot of introduction already and uh um primarily I work on technology products and aex uh and uh our goal is try to build the best modare infer for all deing projects um I think that's a brief introduction awesome check check hi everybody my name is e geny I'm the CTO and co-founder of demo we've been building it for about 2 years and it's a decentralized transportation Network awesome thanks guys so yeah we're going to talk about this um proof concept with um ZK proof on top of deep in and uh going to kick off with a sort of just a background question so yeah guys could you tell us a bit about where this idea came from how it came into the making how how long you've been working on it um yeah and how you guys going to work together with streamer on this so you know we've hit 50,000 cars on demo just very recently but as the number grows we have more and more Mobility companies reaching out to us to see how we can work together one of those companies uh is in the category of insurance and that's a big one over the last two years Insurance in the US has roughly doubled right so there's a lot of incentive for the average C average person really to try and save money but the problem with insurance and the way it's usually built today you usually get it you know with no data attached from your actual driving or if you do get get one of those plug-in devices from your insurance company you don't know what they're doing with it so we want to you know obviously fit this use case and one thing that came up is how do we let somebody share their driving data without really divulging where they're going really how fast how fast they're driving and talking with streamer and iotex teams we thought hey why don't we put something together and see if we can build a more privacy preserving insurance policy for people awesome thanks Y and um okay let's let's let's build on that kind of area of data and car data in particular metalic actually gave us an absolute gift towards the end of January he tweeted um I'm going to read it out actually he said as a reminder modern cars are highly networked computers and are a privacy nightmare very underere explored sector for privacy improvements some fan some fancy ZK please um I like that so yeah guys could you tell us a bit about you know how much data has been generated via deepin devices via demo um maybe specific specifically within the kind of car industry um and what are the kind of privacy concerns around that today and over the long term so the average car nowadays that's coming off the line has over 100 computers in it if you record the entire message bus of the car you will end up with 5 gigabytes after just one hour and one thing that makes it interesting is it's not just the performance of the engine or what gear it's in it's the weather outside it's the cell towers that it sees if your car is equipped with cameras it's record ing all of that video around the car right so there is almost an infinite amount of everything you see when you're in a car the car sees as well and that data has a lot of different use cases but it's gigabytes upon terabytes from just a little bit of driving right it's not it's not just the car data per se is it because you're now mining other data types as well is that right we're so the device the demo device will collect as much as it can right depending on the vehicle support we can get more we can get less but the idea is to put the user in charge of as much data as possible and let them make the final choice of what they want to do with it some of it they want to share some of it they might not yeah okay you especially right now the the cars are electric and they can be their office on and they can collect all all the data instead like could just was a par down in the past yeah yeah so really the whole thing is about putting power back into the the user right um giving them full control okay cool um okay so I guess there's going to be a few devs in the audience um we should talk about the the technical side of things which is what everyone's here for um yeah guys can one of you just given an overview of the kind of architecture and how this was built and and what's involved I can start with the demo side right so a demo user purchases a device they plug it into their car that device is a secure element and every piece of data coming off that device is signed and we're then allowing the user to permission that it to someone else else in this case we permissioned it to the iotech web 3B stream node yeah yeah so uh um we build the INF uh webstream um yeah I think there are actually two part of the making the de project decentralized one part is a uh we call it a DP verification basically uh whenever uh the devices connect to to S out like a uh data availability layer and uh the DP verification can actually pull the data and do the verification and put on chain so that uh when whenever uh the d project wants to either issue token or like nfts they can directly talk to the smart contractor to verify whatever data option and uh in this demo specifically there's another I think also very important uh use case which is the Privacy like uh um a lot of data um when the user want to preserve at their own uh space they only want to enve certain type of data to whatever providers either centralized or decentralized and Z is a perfect technology here and uh it it uh it can actually prove whatever the user has done but uh uh not actually sending any raw data uh to the system in this case in the smart contract I think in the in this demo uh project uh we we particularly do to uh have the raw data of the uh cars driving uh Ser I think certain data and come together to prove they have driven how many miles and as a result they can issue empty online and without actually uning where they have gone and uh and so that potentially uh insurance insurance companies can utilize the nft to uh issue whatever kind of benefits to those those guys to the users yeah awesome um we can get into more detail on the technical side because we'll we'll open up to questions at the moment yeah sorry for the tech part no I think the guys want to know more about the tech I'm sure um so yeah so have we seen any like uh insurance companies come to us with kind of an interest in this kind of data or are they kind of more on the back foot with this and it's kind of okay ZK is a new technology I don't really understand it like what's the kind of appetite from the the people you've spoken to with this oh so we showed the demo that we built to a few insurance companies and it's definitely early right um the their first initial take was what is this machine gargo here that you're showing me right it's all hex um so there's definitely a human element to it right um we went through this I think with TLS encryption back in the day no one really understands why GPS works but everyone just trusts that it does and the math behind it is real there's the same kind of element when it comes to ZK where I think will probably get it working before the average public and you know the Enterprise companies of the world actually believe in it and that's something that's something too you know for us to solve I think as a space it's not the only problem right ZK is just a small piece of it web 3 in general has a big bar to entry for for them at the moment yeah yeah that's a whole separate discussion so yeah okay um all right great and then on the kind of education side of things as well obviously there's a job to be done there not only with kind of Enterprise customers but maybe with users as well so um I know you guys build a demo uh can you talk a bit about the demo um yeah be good to hear about that yeah I was just going to say yes so you know the data is sent from the device um it's republished on a streamer in the streamer Network the iotex team is subscribed to that topic that the user has permissioned and then you guys are from what I understand running a proof on it per request of the user right is that you you will have to tell me more because I don't know how EX exactly your system works inally sure sure yeah I I think the the temp part is is always online so people can actually try it uh we do not have here but yeah maybe uh later in the in the in the Boost people can go there and uh give a try I've got it it's uh yeah but but in general I think the the way that we we try to demo there was uh how we can make uh uh all the business logic decentralized way yeah like streamer uh yeah people can directly uh stream their data from from the car devices to the uh to uh to webstream webstream can do the decentralized proof and put it on chain and uh and then there is some business party who can actually leverage on the uh the UN proof to me empty whatever yeah yeah yeah so so the the demo if you want to check out is EOD driving. verel doapp and uh uh yeah essentially if someone submits their mileage they can get a discount based on how many miles they have or haven't driven so that's kind of the concept there and I think there's an insurance contract as well that's been built so yeah check it out yeah I forgot yeah I forgot a discount part yeah that's very L people's yeah okay um all right great so guys just just one final question for myself and then we'll open it up so um what are your kind of views on terms of like the longer term of the Privacy layer on deepin like do you see CK proof as the solution for that kind of longterm um what are the future use cases for this you know maybe within Automotive or maybe further field within the deepin sector um yeah so uh I think there are many uh interesting topics on this one but but I think the today's Hot Topic is AI definitely U when we have the DP we have a lot of data and also also just now we talk about how to uh leverage the data using Z technology I think they come together is a natural uh uh kind of like a feel for for for AI um I think in this particular I I still want to uh mostly uh focus on the technology part I think in this case uh ZK actually can can do uh yeah many different ways one is kind of federal learning like uh people do not have to uh and VI their data uh when they are doing this kind of like learning systems and the second one is uh kind of pre uh privacy preserving kind of like learning so that people do not have to uh actually uh yeah expose their personal data and but they can still uh uh prove whatever they have there and the third one is verifiable uh kind proof which is uh what Z is you do but in this case it's kind of like a uh ai verifiable ai piece yeah so yeah uh I think that's that's a very natural expansion in the future for uh dping plus DK yeah so just to build on that line of thinking because the conversation that I've been involved in recently is about um autonomous driving an AI inference at the edge so when you start thinking about that then you know that data and those streams that's coming from the edge it needs to be verified it needs to be private uh and all those things so um yeah that's that's probably where the immediate future looks looks like okay um thanks guys uh I'm going to hand over to the the audience and see if we've got any questions um anyone uh want to ask the guys a question about ZK on on deepen and privacy layers I got a question on the left do you want to do you want to say your name and affiliation or build insurance without and arees today it would be pretty difficult to build because I'm not I'm not sure who else is in the defense face uh off the top of my head that could provide that functionality and well you know a big part of it is interest in the other side to actually accomplish it but I would say your other question was the use cases around besides that right I mean insurance is one that impacts most people so it's an easy want to focus on but you know with where Dio wants to go uh I believe on the last panel we were talking a bit about web 3 Uber um I don't think anyone loves what what Uber and lift have become taking you know basically 50% profit margin on everything and putting all the other taxi cabs out of business but I'd love to see somebody come along and build a ride hail Network or a moving Goods Network NW work using ZK proofs right where anywhere in the world you could submit to a contract that I want to go from here to here and be able to not necessarily say exactly where I'm trying to go and then be able to verify that this trip occurred without divulging where and where I went from to so I that's one that's a shiny object but it really is uh you know I the way I think about it every single data point could be shared either way it just depends one how much are you willing to spend on the infrastructure because EK proof are expensive and not all data needs to be that way and the second part is you know as we as zek gets better maybe we can move into having that be the default all the time awesome thanks for the question uh any other questions from the audience yeah one at the back hey this is a little bit self- serving but I think we're doing something interesting with that right um so one of the things that we do is verify that the vehicle is actually moving and one of the ways we do that is by using the helium network if it's the macaron right so it may not be the Unchained interaction that everybody see you know wants to see but at the same time we are using the trust layer of helium as the base to make demo better right and I think as we dive deeper into deep in you know we get into smart charging and energy grid all of these pieces will start will start coming together more and more into what it really is uh an experience for the user that's better than what we have today yeah to add on that um streamer for the data streaming as well yeah and potentially I think the webstream network will uh whenever it's become like all public run network and uh it will be another deeping part uh deeping uh Network as well I think the infrastructure wise is very easy to reuse for different different de projects but ENT interaction still uh very early at this moment but I I do believe like uh for example like if we in our uh demo projects uh when we issue the Mt to prove people who are driving every day or whatever uh insurance company maybe they can just uh building some like a def5 project to give the dis discount directly there uh maybe other project can actually directly use the evidence online on train to do to prove something else yeah great thanks guys um time for one more question any takers okay I think we're going to wrap up there um big thanks to Kevin andev uh for coming on the panel having discussion about this proof of concept I think it's an interesting one um so yeah we'll see how that develops and watch this space thanks [Applause] guys all right thank you for a great fireside chat on all things ZK looking forward to what's next in that regard