# Interview with Augur -  Ethereum Toronto Meetup July 6, 2015

- Channel: [Ethereum Developers](https://streameth.org/ethereum-developers)
- Date: 2015-10-25
- Duration: 54:00
- Watch: https://streameth.org/watch/yt-qNGnB2D7rEY
- YouTube: https://www.youtube.com/watch?v=qNGnB2D7rEY

## Transcript

so we are live my name is Paul dashos from the and this is the if you're trying to meet up we're joined by Joey Krug from geography thanks again Julie for for joining us today um we're really excited about this project uh as a dance with theorem it's the project is being built on it through it's one of the uh first major uh apps or apps if you're a platform uh so so thanks again for taking the time to join us we might be we are expecting or might be going later on by by Jack uh Denver score Deb Jack Peterson report uh for geography and uh and uh but uh Joey can also answer all our questions and demo the uh the the update offline that's currently running on the your investment so uh uh joint why don't you uh sure thanks for having us um so I guess I'll just screen share here yes and um see here can you see it yeah I'm going to maximize the screen so the crowd into our audience can see it a bit better okay okay here we go so um I'm actually just using our our web client demo version right now because if I ran Keith it would probably make the video laggy um so basically let's go to the overview page or the markets page so in August prediction market and so anybody can create a market on anything on our system and so you see a bunch of questions here and people essentially wager on these in uh the current price is equivalent to the current probability so if you look at the Rand Paul question uh the market says he has a 10.8 chance of winning the election um and so if I buy no shares um I can change the odds so if I buy three no shares so by three no shares it basically makes it so he has a 91.2 percent chance of no and it's going to cost me um two dollars and 71 cents and so if I buy those basically what it does is it sends a transaction to the ethereum network and it basically buys it from this equation called a market scoring rule and so now that I own those shares um if Rand Paul doesn't end up winning I get a dollar for each share I bought so I'd get three dollars back and then um it's basically it updates the market um it should update like automatically but for some reason the uh the odds updating is broken right now so you actually have to refresh the page to do it um let's see here okay and the odds updated on the Rand Paul event and so anybody can essentially you know wage on these markets and anybody can also create the market on anything um so you basically just propose a question and then it's going to ask you a couple things it's going to ask you what trading fee do you want to set and half the fee goes to you and um the other half the fee goes to people who basically report on the event and decide the payouts and another thing you can choose is liquidity um so if you look at this right now if I give it five dollars of liquidity that's not very much so the market odds are going to be very shaky when it first starts off um so let's give it like 5 000 initial liquidity and then I can choose an expiration date so after this date uh trading would end and then the market would be reported on and the event would be resolved and then outcomes will be paid out depending on who won and who lost and then so to submit a market it basically submits two transactions to the ethereum network it submits an event transaction which is basically like um the actual question we're proposing and the outcomes it has in this case it's just yes or no and then it combine takes that event puts it in a market uh the reason they're separate is because with events um you can actually combine multiple events into one market so you could have a a market on like two things and you can figure out the correlations between them uh so like a you know common example that people like to use is like let's have a market on what GDP will be next year and whether we pass a certain bill or not um like say a stimulus planner and then of course there's a page where you can basically you know send cash um in the real system it'll be ether and basically stable coins um and then there's also a list of your Holdings in the market seesaw you've authored and then um the final aspect is basically you know reporting on what actually happened and you don't have to do this but you can if you want right now there's there's no no nothing up for reporting on thank you let's see I'm trying to turn off screen share here and so that's that's our um our latest version of auger and we're basically working on implementing some new features like multi-dimensional markets and scalar markets so right now it's just binary but you can actually have markets that have like a range and you can essentially you know long or short that range now anybody can actually connect to demo.over.net and try try out the demo uh yeah if your investment however that what no the depth the the demo at demodogger.net that is not connected that is not connected oh it actually is now um it wasn't yeah it wasn't connected to the test net like two days ago but we got it connected oh excellent work okay actually uh they should be best on there and it updated it took a while I mentioned that's a block time but it's pretty amazing it's actually working live yeah well [Music] the actual uh uh definitely demographically working with the live met once you're doing those live um so the way I see it is we'll basically have like there's going to be two versions one is people who actually run the ethereum node on their on their desktop and they'll basically just go to a website and they'll download the Javascript file from that um the other group of people is basically people who aren't that familiar with crypto like people who would use something like you know a hosted wallet or something but in our case what we plan on doing is basically making a sort of blockchain out of info style while it's for ethereum where all of your you know cryptography stuff is done client-side um you know the keys would never go to us or anything and you'd be able to basically create an ethereum account there and use auger um on that site and the interesting thing is since our ui's localhost it would basically just download the UI you could check the hash make sure it matches these signed transactions locally and then it would submit them to the ethereum network [Music] um right now you don't have any plans software browser extension um you know maybe after we got after we get the uh the web uio up up so right now the web UI should basically demo.iver.net it's just a shared account but we want people to be able to make their own accounts you how do you get into Bitcoin um I got into 2011 when I saw so let's post on a form called overclock.net and it was saying how you could earn money with your graphics card and so I thought I was you know kind of BS and so I ignored it and then six weeks later it was still the top post on the entire Forum so I decided to check it out and started mine in a bit and playing around with it how many Bitcoins would you mind um not very many like probably around like like 50. I quit once I stopped getting to Bitcoin today because because once I got below Bitcoin today it was like five bucks I was like okay this isn't enough to be worth the the extra heat in my room so my room was like it was like during the summer my room was really hot it was probably like 80 degrees in my room that's like bitcoin's cool so I'll uh do this for like a month or two but after that yeah I think everybody has some similar story yeah um it was basically just so I was building this like this kind of a Bitcoin point of sale thing and um back then so NFC NFC is kind of clunky especially with Android I don't know how Apple pay is it with Android NFC is like the chips located in a specific spot they don't have the antenna like spread across the phone or anything it's just like it's it's in a very specific location and so it doesn't work very well so what I did was I basically created a way to send transactions over sound and play a little noise it's kind of annoying to make your dog bark uh and it lasts like you know probably 0.3 seconds and it would it would transmit to bitcoin payment requests using the sound yeah but uh now it's been like you've now it's just been replaced by you know Bluetooth low energy um and you can actually do ble transactions across IOS and Android so tell us about side chain one motivated you to uh to develop that project oh Cyclone so what Cyclone was was um when we first started working on auger uh we weren't on ethereum we were basically forking Bitcoin and this is like before side chains and block stream are announced and so our plan was to basically make sort of an altcoin um but just basically copy the state of Bitcoin so you take all the Bitcoin addresses and save their balance and then you distribute your new blockchain and people could claim um to clean their Bitcoin on it provided they had more stupidly on the private community and um yeah we shelf that obviously because you know we don't need it anymore but it was kind of a fun project it also um made me not want to build on it before um because it's really tightly coupled on bitcoin core a lot of people that's a side chains whether or not why are you building on bitcoin on that yeah so side chains um so they're you know there is no reason why you couldn't have an ethereum the Bitcoin sidechain except for the fact that um the blockchain guys don't want to support shot three uh SPV proofs um so that that makes it pretty hard so the only way you can really do a side chain with with ethereum to bitcoin is um you essentially make a side chain that's a standard size and you have it except shot three proofs and you can basically transfer to it as an intermediary um you know but at the end of the day people are going to use decentralize exchange for most of their transactions anyway uh even with even with the regular side chains um it's basically just doesn't require to send larger large amounts of money um while being volatility but most transacting to take place with decentralized Exchange but um another reason I'm not like too excited about side chances I think they're quite a while um and so I'm mainly focused on getting something out there on ethereum supporting some stable coins and uh doing that yeah that actually yeah so but initially the yes with Bitcoin as a primary currency out of here um because two Apex side chains are basically a few years off from now and we don't really want to use Federated pigs side chains because those ones um the problem with them is especially with prediction markets it has a huge benefit of the person with just like steal all the money especially if you're doing transactions in about like millions of dollars which in your successful player would have on it so um I imagine we'll first support ether and stable cryptocurrencies and then eventually take turnout will allow you to use um I found out about ethereum let's see I found out about it like probably like February 2014 but I kind of ignored it um and then I started paying serious attention to it again when the counterparty announcement that they like you know forked the evm and youth surfing or whatever I started paying attention to ethereum again and right around there and then basically played around with it and implemented like since I just like started when we've been trying to implement some reward systems and I was like you know I wanted to see how fast is it actually to implement something on this and there's like multiple multiple times faster than where we're getting using Bitcoin core uh so we just made the switch um how did you get to know metallic I understand he's involved in projects how is um so italics is really close advisor he's helped his tons of stuff um where we first talked to him I think I just I just read like a Skype conversation about that and then um didn't really talked to him for a while then eventually started talking to him again when we started building on ethereum because serpent had didn't have features that I think that we wanted and uh besides like you know are there any plans to build these I remember the first in my access it was about uh long strings so it was a string of text and you could only like originally you can only support like a few characters and I was like well we need something longer for our markets and so he came back like an hour and 45 minutes later he's like yeah I implemented this and so he's been extremely helpful it's a really cool guy seriously um so he's he's helped a bit with our math functions um a while back and then he's going to look at our um our math functions again you know later once we have them more more finalized because we're still changing not pretty regularly but what details can you give us about the crowd said first but is there anything you can share yeah so it's basically um so what it's for is um so in our system you have to have a way to determine outcomes and do the payouts um traditional production markets like uh in trade before it got shut down basically just had a third party and then we basically provide a financial incentive for them to report honestly like with what actually happened and if they don't they lose something of value and so the way that works in our system is we have this thing called reputation there's a finite amount of it and it varies according to uh according to how they report and if you report you get a portion of the trading fees which is half of them multiplied by whatever fraction or reputation you have the other half goes to the person who made the actual Market and so we thought about like the um just like we thought about a few ways to distribute reputation one is to just you know raise VC money and give a portion away of it but that means you're giving away networking senses um you can also get away for free um that doesn't really stay in the center compatible and the people aren't incentivized to report honestly so we decided the easiest way to do is just auction It Off and so we plan to basically oxidization up a couple weeks after Frontiers launched well I understand I'll bring some questions to the audience about let them ask as well if you have questions yes reputation uh and owns some of that would report on that event information um so you you have um you can't have an outside yeah I'm not really sure how who would get paid uh the reputation currency um but it's important by uh Lucy right so so the reason we have people reporting things instead of just plugging into an API from like CNN is because you actually can't do that with smart contracts they can't at least ethereums can't access the internet um and so basically we just have real people do it and so it they're actually going to be things that are very easy to determine the outcome of like you're just going to be able to Google it and find the answer um it'd be great if you're just going into an external light guide we've chosen yes so if I don't buy recently you shouldn't start do I can I gain reputation by correctly predicting or correctly reporting suffering does reputation to you know Build It Up is it it opens like civil attack factors which is where you make a bunch of accounts instead of actually interesting let me think about what type of questions and events questions and markets um I couldn't hear the question exactly I think it was you know like what types of questions will people be asking okay so the reason why I'm basically asking this question is because strikes me that um reputation and people are reporting back to work the best when we answering some videos rather than deterministic because it's important from the outside right because everybody produce stock market um I know what the value will be because I think when the answer is ambitious if I ask will I be better off if I use something else right that would be actually more interesting um the main category of markets are the kind that are you know deterministic like it has an outcome you can find that outcome after it happens and you can do payout supporting them um the type of markets where you say you know should I do this or should I do that um those are called at least what I call them is um and it's basically along those lines you have a few options to choose from and you want to see which one will people think will have a better result um the way you do these is typically you have the person who's asking the question or who wants to see the result um he had that first investment Advisory Board that's the vehicles another way you can do it without any subsidy is you basically just say um you can have like conditional markets like if this happens or if that happens um then what and it's one of those things gets chosen and those people's money is at stake for the people who who bet on something that like never happened just get their money returned to them we haven't implemented that yet yes you mentioned uh um um so the initial activity um you basically put it up there um the liquidity actually doesn't go to the reporters at all um except like in a few weird edge cases where like the person asks that question or something um but typically you get their clear turns or whatever is left of it um and the reason there could be you know some of it gone is because just do the nature of the market um if you trade the market and its odds are like 50 50. and then um someone buys much of shares to you know swing them to 0.99 and then they end up being right you're going to lose some International liquidity but the the liquidity is separate from like trading fees so you provide the liquidity and that's just used to kick the market off and then trading videos are what you you know actually make the money that you cut out a non-binary contract yeah so um there's a couple times so the simplest one is a categorical market and that's like um will it the XYZ or will be a b or c um another type is scalar markets um so that's like let's say Apple's currently 150 a share um the minimum price is obviously zero maximum go with like 500 or something you can basically long or short it as a scalar market and then you can also find any of these Market types in any way you want um we only support three dimensions right now but you can basically combine um a binary Market with a categorical binary with scalar all the different combinations you can combine them and what that allows you to do is stick you out the correlations between things so like if you take like one of those future tracking markets which is about like what's the best decision um a really interesting one is like you have multiple decisions to make about a company um whether you should hire this person the CEO or fire the current CEO whether you should spend r d on X or Y and um where you should send your marketing money or something and if you do it in three dimensions like that you can figure out a better combination than you may have gotten if I'm just picking them separately it allows you to see formulations um it may have been unclear when you examine the events separately ready so how do you see I can't hear anything how do you see um so so it's presumed that people will uh people that hold Bitcoin will want to want to um Place predictions with fogger but how are you going to get further adoption to people who are holding Fiat for example all right okay um so just a few projects in the works to basically make uh stable coins on ethereum that allow you to essentially Buy in with like an ACH transfer so you volume you know send the money instantly um that's that's an interesting approach I think there's a lot of mainstream adoption will come out um you know we'll initially will launch and it'll just be like crypto people using it uh but once those sort of stable currency ways to get in like gateways get in um I need to see more mainstream adoption those people I imagine them just using like the web client they go to a website it asks them to sign up for an account they sign up for an account we don't store anything it's all sort of client side they have to remember their password and um and I'll be able to even like you know send money to the network using an ACH transfer um Gavin was also done some work with this he's like made a I've been our iban um transactions such as like International Bank uh something for some transfer network of things use he actually added a transaction type to ethereum specifically for that actually how much volume first year um I have no idea um I guess that's the best way to estimate that would probably just be to um ask our markets uh to be honest um once we have like accounts that are uh where each person can make their own account uh that'd be the sort of question I'd want to ask right now everybody just uses a shared liquidity pool and the shared cash in the demo so it's not very useful um yeah I don't know foreign let's see here yeah that's that's what I do I like it the current the current centralized ones don't have enough volume uh I mean I know personally ones well why don't you have a question I want to ask you because according to realize markets use open order books um what are the advantages of using LSR at logarithmic Market scoring role uh versus the other options okay so um I'll log into the Market Square knows it's basically the idea that so a traditional Market of course has an order book you buy and sell and you meet in the middle somewhere um with a market scoring rule you basically buy and sell from an equation um it's one of the hardest things in the system that wrap your head around so it doesn't really make intuitive sense uh like for each trade it doesn't necessarily mean there's a direct counterpart um then maybe one a few hours later but it it's not like a traditional order but and the benefits of this are where you have a 66 chance of no in a 66 in a 60 percent chance yes um which is pretty useless it makes no sense um yeah it's more than one um and Market scoring rules don't have that problem uh the other thing Market foreign depth so what this means is if I'm a huge whale and I want to you know bet a ton of money on something I can keep buying out as long as I want uh essentially the price is going to go up to like point nine nine nine nine nine repeatedly um but that order depth is still there uh so it provides a lot of nice uh economic incentives the uh I was just gonna say the another interesting thing about Market Square enrolls is there's been there's been like uh some recent research it's shown that there's these algorithms in machine learning that are called No algorithms and basically it's like if you have like you know a few thousand experts and you're asking them questions and if you basically after like a few rounds you reward uh experts who you know said something that actually became true and you D weight people who were wrong um if you look at the math the math is actually practically the same path as Market scoring is um nobody really discovered that on the phone recently another interesting thing is that if you look at um like total entropy like uh information entropy um is uh also the same the same equation as they use in physics for entropy and assistant yeah yeah my question is actually a little bit more about that um sorts of course is the the man who's designed a lot of the out of the mechanism design that you guys are using for the reputation aspect of otter uh you guys have some kind of interest in public interactions with him in particular he seems to take issue with the way that you're influencing and or modifying his algorithm but what do you what do you think about that what do you have to say about that so um at least from what he said is that we're using ethereum um the modifying is algorithm thing is something I kind of actually brought up um and and the reason we're modifying it is so there's when you when you report on the outcomes events uh there's a couple ways to do this basically all you're trying to do is figure out um rather than most values to the system because the idea is that buys are more variable than truth it's why when police interrogate how they do it in separate rooms and so both algorithms try to do the same thing uh the one call originally proposed is a very interface and basically you find this vector or basically a list of numbers let's call the principal component and it basically allows you to figure out how much each reporter contributed to the variance um that's one way of doing it What was decided to do instead is is using a cluster number so in machine learning clustering is basically you have a bunch of different groups and it's basically what it sounds like you need to Cluster them and basically segregate them in a different categories and the benefits of clustering are if you uh one is it's a lot cheaper than PCA PCA is very expensive uh clustering is very simple um and the other benefit is PCA sometimes gets the score is wrong it's like your variance score um you know most of the time it's right most of the time it does a pretty good job um but occasionally it gets a score wrong and so what this means is you're rewarding people who reported dishonestly and uh sometimes it means you take money from people who are reported honestly or I guess take rap and so with clustering that doesn't really happen because it's so much something um what do you find how different someone else or how variant their report is is you basically just use the euclidean distance function which is the function they teach you know everybody in Middle School um you'll find a few points like um X2 minus X1 squared Y2 minus five one squared and just worry about that so you figure out how far each report is from the other reports and then what happens is a cluster of report surfaces which are very similar to each other um and that's basically the truth cluster and then you just penalize people how far they are from the church cluster and uh it works really well and it's it's really cheap um so our implementation of PCA on ethereum it had about 12 contracts and um our cost frame one is at three so at the end of the day it's mostly about uh trying to get that performance more appropriate to the blockchain yeah it's um it's mainly a performance thing um and then it's also a better um clustering has a nice benefit that it's it's um it doesn't get the scores wrong but uh I think that's the part that all sorts might take a little bit of an issue with I think you might say that from the mathematics perspective you don't actually know which is wrong yet that's why right so so the thing I would say to that is basically um so basically he he looked at the way we ran our Monte Carlo simulations and she thought we were basically signing um assign any specific truth values and then like uh manipulating things afterwards according to that and figuring out how truthfully will report it um but the way we actually we actually made these Monte Carlo's is exactly um in its own sample test cases which is you created a report and then you figure it out um what the outcome is and it's just a weighted average um and you can penalize the reporters a different way you can see what the weighted average outcome would be um the first like sort of eyeball group of tests I used were I just took the ones that are on Paul's open source repo uh here's about 30 or 40 tests there and I ran them by hand for my question um they got all of my except for one which is one where people reported basically a bunch of people who didn't report something um it's where you have like say 15 events people only um the basic PCA algorithm gets eight or nine of those wrong um you know that's a small test case it's a small sample size uh so what we did was we ran Monte Carlo simulations um and those shows that clustering is consistently better than PCA um now if someone has you know like if someone came to me is like hell hey um I think clustering might be vulnerable to x y and z um if they tell me about parameters in the past I gladly test it because I want to know just as much as everybody else if there's a problem with it um metallic actually commented our blog post and said hey there there might be a vulnerability if you do this um that's actually really fast because he found that out almost immediately it took me a few hours when I initially wrote the algorithm and um but I already accounted for it so it's not an issue but that doesn't mean there's other issues um you know right now according to the best evidence we have it performs better than PCA um because it doesn't mean there's not a flaw that's why 15 minutes you know Paramount thanks uh Joe thank you for talking about this I have a few questions number one can you tell me a little bit about you know say you have seven people walking probably full time how how do you guys send people get fun like how do you do people to continue working with the revenue or what's the interests Financial benefit for youth so many people um so for that question basically we've gotten some um we've gotten some funding to basically get us to the crowd sale um it's mostly Bridge loans um metallic units um and a few other people um but uh that's that's you know the main source of funding we all think really low salaries there's not like a huge financial incentive for us to work on this like I think around two thousand dollars a month um if I wanted a big Financial incentive like you know go workers right as an engineer but um I'm also focus on getting this out there because I think it's an awesome project how much money do you need to get it to 1.0 release um so we basically need to get to 1.0 release we we pretty much need to get through like December um which means we need around 125 000 to get there um you know after that it's it's mainly like scalability things um you know I was looking at the roadmap that I have for our back end code the other day and it's actually pretty close to being completed uh almost all that components sometime in September um but the main thing we want to hold off on launching for a while is we want to test everything a lot of protect them in other words we think about a sports team football purpose to what extent can your system protect these people them um but I guess to clarify when you say gaming are you referring to gaming on the market side or gaming on the payout side yeah yeah okay um so basically there's a few things here so the the core idea is that um truth is essentially a shelling Point uh this is the same idea that both Paul stork and vitalik stumble upon um metallic uh proposal for shelling point and it's a single idea which is basically you ask a bunch of people something um the truth is accepting that it is that they collude um if they collude they conclude you know to try to make a profit by betting on something that's very unlikely and then reporting that that happened to win money and so we disincentivized that a few ways um the way reputation works is um basically each report is way to buy how much reputation you have and you get trading fees in proportion to how much reputation you have um so your reputation is worth um whatever the future value trading views is uh it's determined by the market and that's how much each rep is worth um if you line up and you don't have a 51 uh Collision collusion um then you actually lose reputation and the main way we disincentivize collusion is basically uh um and of course there's ways to get around this if uh um but right now what you do is you basically get your report you got a random number to it you hash it you submit to the network and then it gets locked down and then you reveal your report to the network and so what this means is um if I want to clue with you to try to cheat on something uh the only way you can do it you know trustly is if I had your hash and I could have your report and I can prove that it hashes to whatever your value is otherwise I have no idea if she colluded with me or not um well the reputation and payout side are like inherently intertwined so reporting decides to pay out football games results I think to give a concrete example of what he's thinking of here I think he's just thinking of I did the simple case of um uh someone in the football team um is their team is projected to win they go on the prediction Market which is working properly and production is right it is likely that they're going to win but but because of this there's a strong economic incentive for them to bet against their team and then take their team's performance so that they actually lose that's that's more the scenarios okay okay um so you know I I haven't really thought of you know a good way around that um when you have huge volume um that becomes a problem uh it's never became a problem in practice before um but it could in theory uh if you look at it's like one of the reasons NJ got shut down was because the cftc said well um wage in a political elections could influence the results elections um you know that never actually happened but it doesn't mean it couldn't um I don't know if there's really any way to prevent that from happening though um because prediction markets they do add Financial incentives to things that otherwise might not have that big of it instead of me yeah I do so I'm trying to understand a little bit more about the crowd sale I know you thought about it a couple of times so what are your expectations in terms of how much you want to raise and this is like a uh it's not getting exactly where's the money going to and that dives into your financial almost like are you 18 percentage every time there is a future transaction otherwise the crowd sale related to the financial model right so we ought to raise like two or three million um in the crowd sale and where that goes is it goes to the forecast foundations basically just a non-profit and it it's basically who we work for um and it'll basically continue development of the software as for the for the profit model um so the organization itself you know has no profit model um it's basically just designed to build the software and get it out there um us you know we don't take ideas will have some reputation uh each of us on the team and so basically um we have an incentive there because we're informed the foundation I think isn't really have any incentive I actually actually three million dollars pardon did you say you're like between two and three million dollars yeah like an Easter okay yeah reputation doesn't work yeah you can you can resell yeah so you can give yourself your reputation yeah that's a question um so uh so I guess the first question is no he asked when you get do you get reputation and you basically as long as you report on the events every few weeks um like you don't did this happen or not uh you get half the trading fees or the events multiplied by whatever fractional reputation you have so if you own five percent reputation and um the average trading fee is one percent uh you'll get you know one percent of all volume multiplied by five percent and that's how much money you'll get um per time period and so that's basically the financial incentive for reporters in the financial incentive to buy into the crowd sale um as for what you can start reputation so I mean in our system you can um it's not it's not traditional reputation it's not like um it's basically like since all it's based off of is financial incentives um it doesn't really matter if you buy it because if you buy it uh you've basically spent this cost and if you report this honestly you'll lose money um the risk is always that if some one person buys more than half of it uh then they can just decide the options of events um but that'd be hard because as soon as you start buying large quantities the prices shoot up Okay so there's a crowd sale that you sort of finite amount of reputation is it still 11 million by the way yeah my reputations he still thinks that 11 million yeah yeah okay so you sell a fixed amount of reputation so let's say that it was only sold to the people in this room with all that some people share and then I one made a wrong report it would lose their reputation and it would revert back to the ownership of the foundation no um so if you report this honestly um basically the reputations distributed amongst people who reported honestly huh right so if you lie you lose your money you can tell the truth you get more information it's it's yeah can you repeat this question the reputation is not it's just basically yeah it's like a verification what's the unit base um it's just like a score so uh each person has an application score uh you can trade it and you know sell everybody or whatever um it's like a token and when you report on an event uh your report is weighted by how interpretation you have so if you have 10 and I have one we each submitter to Port uh your report is essentially submitted 10 times it's just multiplied to attend um and then you would get 10 times as many trading pieces as me because you um that's not just similar from eating ten dollars after every dollar I can't hear you I'm going to try again so it's not too dissimilar for if you made a bed I put ten dollars for every dollar right exactly yeah it's just proportional we just have these units of currency that people are reputation um if there is if there's a lot of people say for example there's events on a sporting event and is there some sort of concern that some people will maybe hook up to some API for the results of the sporting event and then when the results come in they'll just have maybe hundreds of accounts all reporting on what happened and then someone one person with a bunch of accounting will sort of suck up all the reputation um so the interesting thing is if you make a bunch of accounts um so let's say we have a thousand rep that's all we have um you own it all um if you make a thousand accounts or if you have one account um it's no different functionally speaking um it's it's the exact same thing so um but you did touch on an interesting risk which is the risk people just plug into apis um right um which it could be a problem if you know the API essentially lied on as an outcome um yeah yeah I think for I think for things like sporting matches it it kind of makes sense to just use an API um to do your reports good one is there a risk to be perceived as a gambling service and you shut down um so what our plans are with like the foundation is um we're basically building software and we don't plan to make any markets on it and we none of us like in our employment agreements none of us to make markets uh none of us can actually repeat that on Market um and so there is some risk there uh it's but it's kind of similar to the risk bit torrent face when when they launched um they basically have this platform and you could do legal stuff on it there's legal markets you can make and you can do illegal stuff on it um the people who tend to speak on that are the people who you know make big aggregations of legal content so with BitTorrent it's a torrent sites um it'd be large market makers I don't understand that after the event is closed and in 30 days you have a first report in which is uh uh uh not disclosed and then another security day but you have another important do you combine these two recording together or are there are they separate uh only based on the open book um so they're they're essentially combined so if you if you submit um if you don't submit you know your your sealed report in the first half uh you can't reveal anything in a second right um so you have to participate in both halves um second half is pretty easy you just have a client up and you just turn your client on the same time in that month and it will automatically submit it so what's what's the benefit of uh two separating one is a is um so after you submit the closed part um in the in the open part if you submit something that doesn't match up to the closed part um I mean you're basically trying to change your report um and so what this does is it forces people to essentially it provides a huge incentive to not cheat with other people because let's say we're in the close part and you and I want to cheat together and we want to report on something that didn't happen um ancestors close to the network uh the only way we can see is if you provide me some secret data and I verify that that it matches your closed report now if I do that I have two choices one I can just cheat with you and that's the end of it or two I can submit to the network and essentially steal your reputation because you were I can basically prove that you were trying to engage me um the idea is that people who are you know wouldn't mind still another person's reputation they're trying to achieve but it's it's like you know um if you're a criminal you necessarily probably yeah and also that the crowd sales repetition you think that the 11 million um open unit um is that sufficient uh I imagine that let's say The Parkers become very successful and in the future you could have so many questions that requires so many uh reporting otherwise that each reporter uh replication token Holder will be overwhelmed by among the questions so um each application is divisible um just like Bitcoin so you can you know you have Miller Bitcoins satoshi's all that stuff um and then that's for sale building um there's there's two ways one is you basically make separate subject categories and so you choose to report on a certain subject um whether it's politics Sports and you could segregate those farther as like you know political elections and events um the other way to scale is basically do randomized supporting so if you're a reporter you have a certain percent chance of getting selected to report on an event uh so basically subsets report on events um the subject-based uh scaling is a bit more secure in my opinion but if it got like a huge system with you know hundreds of millions in volume as in subset base where you basically have a random chance to report on a fence that would be pretty scary as well I think one more question okay Jordan let's give my hand has been really great [Applause] thanks thanks for having me okay take care bye bye