Building the Data DAO – Weilei Yu

hello everyone my name is Wayne from streamer in a minute or so we'll be starting to talk about how to build a data or data Union welcome to our data dot presentation if you cannot hear me well just let me know and we try to speak as loud as possible so I'm from streamer we are trying to build our peer-to-peer decentralized network to empower the next generation of data economy and the new paradigm on how we can leverage use the data in a more ethical and you know open today agenda is going to be divided into three section I will give a quick overview on the current state of our data economy and then I will walk you through some examples a data union framework are being rolled out so far from our community member side and then also you know how a service gap or you know data app can be deployed and and I will show a proposal that we're trying to roll out that's going to become maybe the first data Dao in the open you know user there's a issue with the presentation I think that someone plug the projector by mistake give a second so let's go to the initial let's say recap the current state well I'm usually up optimistic guys so usually I take with you know good point of view the project usually start with four you know the best intentions for example in the early days of Google we know that actually did contribute quite a bit to the you know open society or like you know back to the community and some of the projects you know they share the release were like you know Cora me on they help with like deepmind Hadoop Android and Google Maps or other suits but all the time we started noticing a trend across all the major tech companies and in the graph you see like you know the usual Fang and the top 5 tech companies in the US and I tried to summarize one graph what was the China was happening in the last decade and in the x-axis you can see the multiplication value of user base times the market cap of each company while on the y-axis the ethics or ethical value that those companies hold whenever they think about decision trade-offs and I guess I try to convey in this message what was happening meaning like it seemed like the larger the platform grew and the more money they made the less ethical the you know decision making process became so you know if this is the real trend you know since the inevitable inevitable for the future is that something that we can do to change your ship the course and the days are all saying like you know large boat steer very slowly this means hoping for those larger enterprise to change is going to be very hard so can we do something on the other side the bottom lap this is what we wanted to introduce the data unions framework the stream is a plan to roll out in the next quarter or so in how they can help you know developers build more user centric maybe open data Commons or even more like you know for decentralized data organizations before going to the details I wanted to give a quick overview of what we mean by data union another perspective is a framework to help developers build user centric data applications and then to empower those users to get more control over their data and in this context framework can be actually interpret in many ways it could be like pure technology wise like you know a full stack of a framework to incentivize developer to developing like you know more like decentralized stacks or more like a conceptual design wise how can you design someday cannot be evil you know from the start for example or you roll out a timeline to reach the goal I think we have heard many data hacks or did abused you know in the news and especially in the last two years or maybe four depending how you can do things like you know 2016 Commission in Ithaca or like even earlier did abuse you know like all the maybe like Dragonfly project from Google if you have a heard about that and many others but one day was very recent and the interesting is actually a vast which is a you know antivirus software company they had a scandal because they had a subsidiary jump shot they actually was using the data they aggregated about users browsing behavior and we packed them as a products to sell and jump shot was actually making revenue around like anywhere between twenty thirty million dollar a year so was interesting because it put a price tag to what the data value meant on very you know straightforward context and of course they had to shut down the jump shot once the scanner came out because was you know for ethical reasons then he prompt to our community members to think about is there like turn ative that can still maintain flowing data economy but reaching like a much more ethical way and that's what this watch team did is actually browser plug-in developed by one our community members the developers team so it's not internal to streamer but they build on top of our you know network like peer to peer network pops up in in conjunction with few other stocks that we were what you learn in next few slides later so what this watch offer was a simple value proposition because every platform is already tracking your data and they're doing that without much consent or like those like kind of force concern that you cannot opt out unless you wanna be go back to the Dark Age swatch it implements a more like opt-in I mean like you can decide what type of data or module you're willing to turn on like you know enable and to what extent you are feeling comfortable anything like another change in the paradigm is you know as when my friend Chelsea was saying last night Swach wants to distribute back value where value is created meaning that we wanna create a more like democratic distribution of the volume not just on the top actually were actually users created so as I mentioned before us watch is not just a new paradigm for data aggregation of data you know sharing revenue sharing must also much more user centric privacy focused they allow you to select whatever you feel comfortable so is everything like user opt-in and user preference but more importantly swash is the only client side this means they don't own any server and they don't send any data across their own server but is leveraging the streamer in a peer-to-peer network to push an aggregate data so they're already eliminating one big segment of the abuse which is like centralized server and centralized silos of data and to making my father you know more user centric and user privacy prone they actually build on the client-side entire data filter layers that users can enable in advanced topics that can you know filled out for example you know you are for like a personal website content etc so to go around you know swatch is what we see as the next phase of user centric in the data privacy prong alternative to you know a lot more let's say ethical driven data sharing and how can we sort of guarantee that swatch cannot abuse users data right and this is a quick you know diagram or graph to show how actually how works on the backend of swash which is users will averages watch to aggregate the data in a you know anonymous way and that's what you will push the data across the student peer-to-peer network which is a you know decentralized with the intention of implementing encryption based permission granted in the next milestone that way they didn't ever touches wash control server and all the permissions are controlled via registry that can easily be on chain so is a transparent and whoever is the other end of that is pub/sub you know aggregate stream which means could be like one or many that can simply purchase the stream and lock the transaction and those are whole the whole flow will be recorded and transparent so there's no way easily for swatch to cheat in the middle and to add I guess the icing on the cake is wash code it's actually open source so they cannot even push to a private stream where they control and whenever the buyers on the other side you know purchase the stream for like certain subscription or the revenue can be forwarded to as my contra or Dao where you can send a specific rules about like value distribution you know governance etc so we can create actually end-to-end you know more transparent and democratic system of a data sharing and I guess user centric approach now I'm not sure if you guys seen like read the article about Facebook running out that study up which has user to install like you know in grande roue permission so the Facebook team could you know truck even more about what they're already tracking of every user or you know the study group so once the this article came out one of our community member a developer actually started to think about is the more is the more you know alternative way to achieve similar results of you know opening up this data set for like you know app developers analytical company you know like market study without having to worry about a central party controlling all those data set and this is what you know brought my DN to life it was developed by another team from our community developers and you know the whole value proposition of this app is you know users will be able to share anonymously they apps uses statistics like you know simply the time you use the app the length you use with double you know sharing the app content itself this data's fairly you know anonymous and fairly innocent but can unlock very powerful insights on the you know data economy sign and there are many buyers interested of course in this data my instead of doing like this wash approach of you know a benign team making sure like you know the ecosystem cannot be cheated or abused we actually trying to allow even a further approach with my DM which is running as a data down what does this mean is that we won actually the users that use the app and you know the addition to receive the revenue sharing from you know other let's say subscriptions from other side from data bias we want them to be able in the long term to control also the direction and policy of this app in the ecosystem but the same time we don't want a rollout as you know the usual traditional down well you know the value of the Dow sometimes intrinsic value is correlated to the Dow involvement and the involvement is in internal correlated to the value of the Dow so you cradle chicken-egg dilemma we actually wanna approach with a simple proposition we want to start with a simple value the users can stand which is you know shaded you didn't animus Lena aggregate and receive revenue for a revenue sharing but also you might receive you know insights about your usage pattern without running the risk of you know being tracked by central party in addition of course there are more ethical values like offer like open data comments or like you know more democratic access to data the other user might follow but we want to start with like you know you received some valuable for what you share and then you'll be very interesting to you know with the app ecosystem users to distribute actually you know something like a token reputation tokens in the background without them even knowing the early stage but let's say once you reach like you know ten thousand hundred thousand users what happens once you unlock this a hidden future or potential to the you know how user base and that's where we want to start like you know once the app has utility for the user how does the government look like and we wanna you know at that point roll out like onboarding educational information about the governance and let them test out like in on stage by stage progressive because governments participation and as they stake in the governance increases because the revenue from the upstream like you know the data stream increases we will increase for example the requirements for like participation said to remove all the worries about like you know how easy to do like support a core you know like fake accounts a test etc but there's no reason why we should worry about that in the early stage because they would just create a like a huge roadblock in a sense and this diagram shows actually a quick flow of what you know data that rollout might look like for my DM which is now users control you know adapt based on the mix of you know usage and also participation based a repetition token and then they'll be able to elect different administrators for different segments that could be you know design development business marketing and also policy and also the same time that would be able to approve a funds for digital and these are like different segments or departments like themselves can run like subgroups of you know dowel task oriented for example colony where you know you have like a participation from the community where the rewards are based on those sub stacks sub tasks as I mentioned earlier how can we make this whole ecosystem leader self-sustaining is we start with the proposition that this Dow should generate revenue before even you know the rolling out the Dow itself and that's what the data aggregation of data you know Union comes from if the Dow itself has a steady revenue stream then everything else can trickle down you know in a more like a smooth way and when you have a specific proposition and then the users have a reason why they want to participate and why they want to maintain the ecosystem because the value that goes back to them and also another point of you know I guess friction who come from like what who develops the app and not what is the extent of power control they have you know over these data down like can they actually cheat or scan and the users by either you know hiding some transaction forward introduction or data to a personal server or even like you know doing some other unethical hacks they could but that would be like short-lived because actually like the Dow managed by the user is the one that controls the permission to the stream in the sense like because the string is a peer-to-peer and the storage is the peer-to-peer it can be controlled easily via own chain registry on who it can grant permission so actually the data app itself where is my deal on my DM version Chu is actually uh in a renter of that permission so the app can aggregated and push data stream to the specific you know aggregated stream and then create a revenue sharing as long as the DAO grants permission to the app this means like let's say you know one day the doubt which is a certain value and then the app developer decides to do something funny well what users can do is is a straightforward once you know you can actually get auditors to auditor apps deployed to make sure they are not having any malicious intent but once they figure out they were malicious then they can simply shut down the access to the a great stream to the app so the app cannot do anything anymore to that point and they're not stealing value from the now because the storage and the stream is where the value is being created the app is just like a means to end at that point there you know another app can come up you know especially for the app is open source or like you know the design is well known to the community they can simply you know nominate another administrator in the app development team etc to take over the new stream against permission and move on from there just like never happened before and there's no value stealing in this sense because the app the new app actually builds on top of the old stream aggregated that was generated with the previous app so as you can see I highlighted here where actually is the value in the value resides in an aggregated stream which is controlled by the Dow and the Dow is you know over time run by the users of course there are many attack vectors that you may think like you know what about Sybil talk you know and other vectors and that's why I put earlier like you know the rollout structure mean like there's no really value in attacking the you know let's add this data down early on because there's not much doubt value itself embedding where as the user base grow the value would be accrued over time but the same time participation requirement my increase over time so you know creating like you have those like data farm for example that can run like in your many users but at the same time does the value of Ronnie like cost of running those you know fake users match with the value that we'll get out of it because you know the Dow alikes administrators that need to propose actual functional I guess milestone and work projects that you know means there's a lot of work actually behind it and to just simply hack Madison time you know all these like looks very promising and as I mentioned before swash is already out in beta testing and we had a thousand users already testing the environment experimenting with the dow did you know aggregation value or exchange etc my DN is going probably roll out in the next quarter so and the goal is actually to expand you know whatever adoption ecosystem to the larger community we are trying to reach those users they know airdrop can reach so you know the traditional token distribution is never going to touch them so you know the goal can be you know of course broadly speaking you know what happens if we onboard hundred thousand mini users from the traditional web atchoo ecosystem and then they suddenly realized they have a data talking or you know talking about a boy and then say you know what what am I going to do with that and you know there's nothing that pushes more corrosive than having something valuable that you're willing to trade off at the same time to reach this goal was in a hard journey like wasn't an easy journey as this is a funny meme says no one does not simply build a data doll especially not other blue and I want to quickly walk over some day layer stacks there or ecosystem components and okay that you know this data union framework is composed of you have the network layer the neon that takes care of the data transfer and the storage is peer-to-peer you know we get more decentralized over time and the next milestone we are rolling out like for example the encryption permission granting scheme and also like once you have this data aggregation then you need a marketplace to discover those data streams and also a payment of course system which will average interior so stream and I took as I mentioned before is a peer-to-peer pub/sub real-time network is the almost perfect for those data aggregations where you can aggregate hundred thousand meaning of data users stream into a single one and from a developer perspective you just fire forget meaning like don't need to care about how many subscribe on the other side because the network will take care of them you just fire you know you did into a single stream and then the dental will distribute via the permission granting and the whole purchase flow result emitted on chain in the stream marketplace is our example deployment where you know developers can actually publish the data for you know three four three or you know other like you know whitelist or blacklist like many flavors you can even deploy your own marketplace on top of streamer if you want to run like a niche market or your you know subgroup and for the micro payment we actually roll out what is called a micro plasma which is a one-to-many micro payment distribution because as I mentioned before behind the data union or like you know the data that could be hundred thousand users or mini users right so whenever our subscriber data buyer pays for that stream you need to have a way efficiently to distribute those revenue across users and this is what we did with mono plasma we tried like we we had to cut down like a similar trade-off which makes this like very useful specific for streaming and this is a quick overview of what our centralized you know data application looks like compared to did a tile framework which is you can roll out everything with the existing infrastructure this means you can have a data app that never needs a server by itself so you can cut down so much maintenance headache but also like security other issues that would be impossible nowadays so if you're interested in any of those spaces we would love for you to reach out because the interesting and intriguing thing about the data union framework is it touches upon many components you know from data application building to data union management to dial governance from like you know federated learning so how can for example create a common machine learning that is a privacy centric because we have a way to aggregate user data right on their own storage for example and that's where you know this centralized storage becomes the big pain of course like God baptizing and zero-knowledge are very big use case so that's it thank you for listening if you're interested please do reach out follow us on Twitter we have read it you know visit our website for more information and my telegram Twitter is the same handle if you want to chat thank you very much [Music] you [Music]