DAOFest – Weilei Yu, Streamr

test test okay hello everyone this is why from streamer in few minutes I would be presenting the data union framework and in combination how that might work with fatale rollout so in two minutes one two minutes thank you okay let's get started data unions and data Tao hello everyone my name is Wei Lai I'm from the project or streamer we are building a decentralized here to peer Natalie pub/sub method to allow for no large data aggregation in the real time in a decentralized fashion so today I want to walk you through the data union framework they were planning to roll out in the next quarter or coming quarter and also how that might look like in combination to a dhow or implementation on top of it so this is the quick agenda I want to go give a quick overview of our current state and then how they stream of data union framework looks like or what are the layers they are composed of and also how like in our survey lights tap or app deployment on top of it is like we have a few projects from the community that had been you know in progress the quite interesting that can give you examples and then in the end I will show you our proposal of a data down of a community develop up a mobile app that were trying to launch as a initial test maybe one the first data Dao being rolled out so far and how the key part is how I can achieve you know self sustainable self sustainability on the long term so the fallen angels before going to the data union so like you know now part the first question is like why do we need it right do we even need a different way of letting users know sure they did and more ethical ways to aggregate and you know with as alternative we can all decide to shut down or our data sharing you know just like you know put up walls but we all know that's recipe for disaster in terms like innovation for the future so I'm pretty optimistic about technology in general so I guess like I always start with the mentality that projects usually start with the best intentions right you have an issue you have a problem that you want to solve or like you know overarching going on a ridge and then you start tinkering so Google in the early days we can say it wasn't you know a pretty good like you know source of innovation inspiration in many ways it gave us you know as a community know as more contributions right ranging from the top left corner like you have the chromium which the brave browsers actually build on top of it and then you have you know going right you have like you know on the top deep mind they know demonstrate the actual advancement of AI and you have a Hadoop with Google you know started with a research for large scale you know distributed computing and then you have Android which you know you can say was the only one they actually could break the monopoly of the iOS you know I quote you know what are you like or not and then Google Maps that's my favorite because I realized actually you know with Google Maps incentivize people to exploit way more connected globally than you might expect so there was a great contribution but I was trying to summarize how the trend of those you know tech giants evolved over time and I came out with this graph and you know in the graph you see like what is called a Fang the top of five you know tech companies in the US you know Facebook Amazon Apple flakes and Google and then I was trying to summarize like how can you summarize in one graph that the development or you know like of these companies with all that that happened in the latest years so on the x-axis you have the you know use like the product of user space times the market cap and then when the y-axis you have the article you know balance or like you know measure and I think like is not too far it's not like a scientific research but it's not too far to say the larger plasma our platform grows and the more money they make the last day care about uses you know focused ethical or points whenever they make the decision so you know the trade-offs there and we know what is happening with some of those companies so with that in mind how the trend is going the question is can we do something differently you know with all the tools that well develop although you know a lot of you know peer to peer solution technologies are not new so in addition to order you know blockchain Tyrael payment solutions identity can we do something better and what of course we are streaming everybody here's hope so so let's dig into it but no one does simply build our data Union you know you don't just wake up one morning so like let's build a data Union I mean he's a little more complex and you know a little more nuanced I guess then the I guess initial Tom might be in order to achieve the I guess the first release or MVP of a data union that we have in mind we actually needed a different layers of components to you know which reasonable decentralized or like user centric focused approach and the first one is a network I think after I guess like a few decades which we can at least to agreement that if internet was a simple China internet a lot of the innovation will have star and we wouldn't see what it would be today no matter how it is like a still you know pretty decent relies in that sense and say my permission was so we needed something like that we need a you know neutral decentralized infrastructure for the data transfer in the exchange and they cannot come for sure from one of those big tech companies we also need like once you have this transportation layer of the data and the you know aggregation possibly you need a discover discoverability mechanism and just how Google was able to you know help people discover information content easily across the globe we need a mechanism to achieve the same thing with the data sets and then therefore we need something like a marketplace and last is we need a borderless payment solution meaning that from day one we should not restrict our user base or like you know the friction of cross-border payment because it needs to be global by nature by default so going on the network side stream and network is you know pub/sub the peer-to-peer messaging protocol that allows anyone to deploy applications - you know stream like a portion aggregate very large amount of data sets you could think about like you know if you want to aggregate from two thousand two hundred thousand to a million data points right from all users you cannot achieve it easily with your decentralized you know consensus protocol so we have been a streamer network which is all changed solution pops up that takes care of both for the data transfer in exchange and aggregation and also we are trying to roll out the storage the idea of you know having a peer-to-peer decentralized like no one can control the data being pushed away so no one can take advantage of user in this sense and then you know let's say you have a builded application top of the stream a marketplace like a stream a network and now you need someone to find that product right so we have a release what is called a stream your marketplace which is just one of many ways Prada owners the developers could publish their products to make a discover discover for the kree implementation the stream marketplace is a maintenance ntral entity needs to be in check or at least like you don't need to integrate any of the existing centralized platforms for your payments so it connects to the team for the payment solution settlement and all the you know from data aggregation to data styling and to permission grunting everything is a peer-to-peer everything can be decentralized so a developer or you know application doesn't need to deploy any central server through this whole transition nothing stops of course because no the marketplace and the network that decoupled so nothing stops developers or someone interested in this field to deploy their own version of the marketplace what is for like a something more niche something they know meets hi you know the centralized decentralized decentralized criteria you know like extent or if they needed to have like you know some kyc User Agreement in terms like the type of pioneers that wanna set that data to etc the key point of data you know at least like one strong value proposition that we realized when we are doing some focus group study or you know user journey was that the revenue sharing model was a very interesting future for users mean like you as a you know application developer or you know a platformer you don't only aggregate no data in a more user consent fashion by also allow users to join and participate in the revenue creation therefore the marketplace actually allows you to push the revenue directly to us my country or any doubt right address only team in a way that you know it allows some certain logic to be set for how you want to split your revenue of course we have implemented that our solution to handle what is the common micro payment issue because you know from the marketplace point you have me might have like a buy that comes in and say I'm buying this a subscription for this aggregate stream for like $1,000 a month a very common traditional way of the data subscription but on the other side you might have like hundred thousand users contributing to that stream now you questions like every time they pay how you're going to distribute you can say I'm going to wait until you know it was just an amount but that's you know extra friction you want to add to the whole you know decentralized autom automated you know frictionless hopefully flow of data in the payment inconsistent as I mentioned is imagine like a hundred thousand 1 million users right that gets very complicated a stream and we roll out a solution and micro payment solution you can say or you know pay me a side channel or you know mana plasma you know whatever you wanna call it called a mana plasma we actually demoed this solution last year on stage if Denver and what we did with mana plasma is we took a general generic plasma deployment on design and removed every type of future that we don't need in general for our specific use case so we made a hyper specific to this one to many payment distribution system so it's very purpose you know specific in this case for our you know data union revenue sharing we also made it in simplify the transverse was the unidirectional members of the mana plasma cannot send a transaction to each other because the whole goal is that we receive revenue share from the application and withdraw you know for the only use so it's not like you know a payment system for peer-to-peer payment and then like withdraw is the usual we you know allow people to withdraw via proof of like mccottry route - the usual way and we have the operator validators but the interesting I guess simplification is in the exit strategy which is instead of for having the extra game X andr we just impose a freeze period o every time a new fund gets you know sending like a distributed so let's say our buyer comes in today's sends like $1,000 gets distributed to use that $1,000 this field among or user let's say $1 per user that dollar will not be available for withdraw any like two days after let's say 48 hours but after those two days we users can withdraw at any given time so they don't need to wait like say from the request that they would draw seven days for the exit game you know etc for other things and we have you know some operator validators to keep a balance on the scan of course design and this is like you know the simplified flow where you have the operator that you know keeps track of all the distribution on the ledger accounting of like how many you know tokens or revenue is assigned to each user on off chain why the on chain as my contra actually received the payment and whenever someone wants to withdraw - sure that we just send up proof onto their own chain when a program contract to withdraw so now that we have walked through the all the components of let's say of what is needed to create a data union this is the quick recap of what is needed in a traditional sense of someone that wants to deploy a data union like a data you know common aggregation and revenue-sharing generation right you can see from the traditional you use like you know traditional pub sub system payment you will need credit card bank api what is stripe you know paypal or like you know back like banks backhand integration and then you have to build of course you customer api to handle the logic member payment distribution and withdrawal excited and then you have to host your own centralized server to take care of all the data aggregation you know there are you know permission and granting and you know how you do all the usual Japan Y on the you know components that will show in a decentralized fashion for the you know networks I you just know fire-and-forget meaning like you know whatever you publish on a stream and a network is peer-to-peer and decent right so you don't need to care about the infrastructure you as a publisher you fire and then the system the brokers of the network we take care of a Hutu or you know send the data to other to the subscribers for the payment as I mention you know we have mono plasma on TV on your header or like or at the type of flavor payment depending on the preference and the API of course of streaming network IP is my contra the usual and then the buyer would be like the marketplace that I just showed so one app developer or your platform of crack should deploy the whole know did the Union interns like use case without ever having to touch the server side that Dean don't know they won't need to own any you know infrastructure they don't need to write any custom API to be maintained etc so jump into more concrete days because you know we love what is tangible here's example of actual app that was deployed on top of you know our stacks most specifically like the no mana plasma data union and also when I talk these applications called the swash he allows people to actually share anonymously their web browsing data and they get rewarded in exchange with no tokens in this case is that data talking from streamers so as you can see in this UI swatches a browser plugin and that's available for both Firefox and the chromium basis also chrome brave the latest version of HX advance we just this product has been actually a developed by external community team and they launched a beta testing like you know a few months ago to test out the holder functionality and user experience in this UI can see for example users have the ability to enable the modules of the day that I would like to share for example search that are casino search engine like you know Google guitar tag go you can enable like surfing Twitter Amazon etc and what we mean by user centric data you know ecosystem is when you give users the ability to set a preference you know turns like a privacy but also like you know to what extent of you know the data they are willing to share and thus wash allows them to do that you know by simply selecting in the level of the preference and you know if they select medium then you know a lot of information might be hashed or middle if they select high like you know the ideal would be randomly rotated and in a higher frequency and this is example of advanced future where you know it allows like users to set on your text filter mask and also like a you know logic so for example the data filter layer is actually even below the data privacy or whatever theta you put to exclude all those content in regardless if you put like you know low privacy because you know whatever you excluding the filter is supersedes their privacy level and the interesting thing is that this whole deployment is a client side that this means like it's wash does not need to run a service from this side and therefore users don't need to worry they are like you know having a side channel way of abusing the data or collecting the data or like doing like some sketchy things and of course they have an integrated wallet to allow users to withdraw whatever data they have earned own chain from the mano plasma so you might wonder what is the value proposition of enabling I use the century approach like a swash you know why go through the trouble of setting up your application you know through a decentralized network you know set up a payment distribution with like you know blockchain system and maintaining like you know high privacy control this is the example article that shows like not sure if you guys heard about jump shot is a kind of data collection arm of the Avast Antivirus is security firm so what I did was like through the advanced web browser plugin or like you know whenever users installed antivirus in their mobile phone they would collect the browsing data for this user and then package I could gain anonymize eventually and then sell it to you know enterprises the interesting part is most of the Avast users didn't know about this you know side let's say economic branch they Abbas had to sell a day data so once they became public actually of us had to shut down the company because was you know a huge scandal so now the question is that company was generating anywhere between 20 to 30 million dollars a year in terms of like in your data revenue can we invent something or you know deploy something that part of the same value proposition of aggregating browsing data from users but in a user centric and opt-in manner right and this graph outlines laid aside the whole flow of what we mean by data Union and the or their logic underlying logic so users who actually you know leverage as watch to aggregate the data in a you know user centric and kind of abused resistant way because wash simply pushes those aggregate data through the streamer network so they don't control on those data themselves and then you know they will publish this data aggregated s3 into this unlike the streaming marketplace where you can only be unlocked whenever someone like I did a buyer comes in and paste in this case the crypto payment right so whenever the critter payment is a confirm that payment actually gets sent back to the like it's my contract that could be like a mono plasmus my counter for revenue distribution or could be like you know another towel or entity so the whole flow could be like end to end you know here to be ot centralized without this wash entity as app developer itself having like you know abusive power winters like a data ownership or data inside now as I mentioned earlier my goal was to give you a quick overview of what our data Union framework and example app server let's a poor you know like adapt deployment look like with the S wash but also we want to see how we can push this whole notion of you know data common data Union as a community driven or committee manage the another app the our community developers actually develop and this app is called my DM the concept of this app is actually quite straightforward and simple yeah it's a mobile app that allows users to permit or grant the permission to share the app's usage data and just like not the app content self by just like when you use it and how often you use the app and the so is you know relatively privacy free in terms like risk but at the same time this type of data is quite powerful because you have a whole suite of companies of developers that would like to know what are the app trends use it you know not just like you know what is the most downloaded about what is the frequency what are the hours today there's someone use the app what is the sequence right for example if you are I don't know app developer for like a payment do you want to know for example like you know Europe was used before after that another competitor and if someone uses a competitor after you doesn't mean like your app actually has some friction so they're moving away right so this whole studio for insights are quite valuable and the reason why we wanted to focus on a valuable dataset is we need some interested in buying that I said so what has this one to do with like you know data Union passed out deployment we wanted to deploy you know like application with a very simple value proposition that doesn't have need to have like you know Dow and like you know big thing or the whole story about governance etc we just wanna wrap the users understand what is the scope and you know the buyer side understand like you know what is the value of the asset being generated in this case uses you know shared data about apps usage and the buyers get inside in an aggregated fashion on the backend at the same time nothing stops us from distributing distributing like a reputation talkin right you can do that without even having the Dow or the reputation talking as they so value proposition therefore like you once you can verify that your app has a value by itself without the Dow then you can make very interesting like you know experiment for example in our case we want to try with running out the token repetition talks distribution the background as they're using the app and earning the revenues and only after we reach a certain size interest like adoption we may be interested in running our smoke you know small bite size like experiments with okay what would you like you know to vote for this you to a or this future being what do you think about you know this roadmap you know what do you think about if we donate you know certain data assets to like a charity start small and then you know you can roll out educational you know on boarding process about government's ones like everybody has a certain point of involvement in this whole optical system so the whole deal is users you can with because they are the revenue sharing and then the more ethical way but you stay for the government's later on so the doll should be actually hidden at beginning for our testing purpose so this is an example of how you know this whole flow with the my DM Dow will look like users actually will you know manage the Dow but they don't know actually they have power to manage data at beginning on the long run that would be the sole manager because you know they have the voting power that's earned through like usage of the app in time and involvement in the Dow you know through the doubt that will elect and also or you know a grant the funds for all the different segments so you have design department business marketing policy in the operation so then you know the users actually be boarding like who who they want to have as the admins for each of those trucks and through the you know repetition token they earn by using the app that they didn't know they were earning on the early stage and then each of those like you know let's say departments can have the subgroups where they actually break down this specific no more like detailed tasks so one example why we were talking to down stock was for the you know high-level at being election and the fun proposal we actually were thinking about using the Atome but for the sub task when you know each department once they have found approved order you know my stones or like you know project proposed approved they can actually create the old sub sub group of that's a contributors via a simpler UI like a colony where just like a test-driven like you know a contribution and you might wonder like how does this one give power to the user like what if the developer actually screws are poor they wanna you know divert the resource of the data being aggregated I wanna cheat the users etc so another power of this approach is like they agree that I did a greatly aggravated data stream actually generates revenue that gets your back into the data Dao and that's why they you know they fund the mekinese the proposal action would come and that's why the the core value of having the data acid as valuable is so important in this experiment now how can users prevent the application developer for example from controlling the environment instead of them because a streamer you know is that decentralized network appear to be actually the stream to aggregate streaming push is actually owned by the you know entity that could be very well a down which means like users can control Dow that owns the asset of those aggregate data so as revenue come means there's like you know revenue sharing server but imagine for point be in the future that all wants to cheat and sense the revenue somewhere else extra then the user can actually through the Dow cut off the permission to the specific app then say okay the app developer would not have actually permission to push data to the aggregate stream anymore and I don't have a permission to actually fetch data from the stream in a way that actually the app becomes obsolete and they can just move on to another app by electing a different admin they can you know a manner to the development cycle so they can you know just go through the whole flow again so in this scenario the developers not the mind control but they're just like the enabler of the upstream data stream so that you know users can retain much of a control in the long run in that sense and this is what we want to start with the main TM app in the whole data union plus down flow so that's saying this is the end of my presentation feel free to you know a little more about data union framework and you know some the use case of for example swarm idea there's going to be rola or soon in the next quarter as a public beta feel free to follow us on Twitter or if you want to reach out with any specific question might elegant with the handle is that thank you very much