Interview with Stefan George and Martin Koeppelmann of GroupGnosis at DEVCON1 in London

Transcript

Oh now we are joined by the founders of group noses which is a prediction market built on etherium so we're joined by stefan thomas and stefan George and Martin couple month so for us we would like to have your introductions starting with Martin sure yeah I'm Martin quite quite a while into prediction markets and now working on Oh sister yeah we like couple of years ago we started to build a prediction market betting market named fairly calm and then like the beginning of this year we started then

to focus on a theorem and we are built the initial version of group noses which is already life you can try it out so my first question is um we've had prediction markets for the last 10 or 15 years and all of them have been centralized and now with the tedium the possibility of a censorship resistant distribute breaks and market opens up why do you think this is an important idea at all why do you think that going from centralized to a sense of resistant prediction market will open up new markets I think I think the sensor resistant is is obviously that's one big thing but it's it's it's already

important on other levels so for example with whisking also with was fairly we centrally keep money and that means first of all it's unnecessary costs for us it's it's a it's cost to to have secure storage of of bitcoins and with theorem we can use the security bulletin into theorem so that's directly from a from a from our perspective or from a perspective of the prediction market that's superior and on the other end for the user obviously it's better to to

trust or hopefully trust a contract instead of instead of a company your user of course the censorship is also problem for prediction markets as you said there are strongly regulated and so of course in all countries where this strong regulation it potentially opens up a new market because runs in decentralized men and no one can really sends it and now I think those are the two main advantages that the theorem has in the context of prediction markets so in the in a prediction market space especially on it helium arm or girl had a lot of publicity because they raised a

lot of money selling rep can you summarize what the difference between group noses and an auger is from an architectural standpoint yeah absolutely so our goal is kind of to be the what worked what WordPress is four blocks that that's what we want to be for prediction markets so we do not necessarily want to run a specific prediction market ourselves we want to provide tools and the framework for anyone to create a prediction market with with a few clicks so the final goal would be to if you are interested in a specific topic let's say what the what what's happening and then a TV show or

what's what some what what what are the name the next Supreme Court decisions or whatever whatever your niche is it should be as simple as starting a blog it should be a simple to start a prediction market and that's why we or that's what we plan with kenosis uber they a lot of work in this for this decentralized Oracle and I think that that's a good idea but we want to have a flexible model for Oracle's so in our architecture you can just choose one Oracle that could be a decentralized one

like Oracle like order but it could also be a centralized Oracle or multi fixing or yeah all kinds of possible Oracle's or just if the if the if the data comes directly from the blockchain then you don't need oracle or Oracle's very simple it's a simple contract that just pulls data from the book chain I think from an architectural point of view with the main difference is that we really try to be a platform a modular platform where anyone can sign up and become an Oracle and anyone can provide this information and not only Oracle's but also market makers and and also tokens so you can use any kind of times they

don't have to use either you can use any kind of tokens you can create your own toten just recently spoke with someone he asked me i can use colored coins can you use a coin specifically for specific people and i said oh you haven't implemented this but you can use any kind of coin that you like because our platform allows us so i think that the main point that we create software where other people can easily extend our platform and yeah make it like they wanted so so to summarize would it be right to say that auger by selling all of these reps is promoting a specific type of Oracle in which people who hold

the reps are the actual people who put the data in and are responsible for the accuracy of the data group doses on the other side is saying that the design of the Oracle should not be specified by the prediction market and and it should be swappable so if somebody wants a centralized oracle swap it in if somebody wants an Oracle like auger show up that in if there's a third model that comes up like selling point or equal or whatever then swap swap that in so you are really trying to make the oracle modular right yeah it's a good summer summary and also maybe as addition we think that there should be like a two step process and

this process is first as you said anyone can decide what should be the oracle and the easiest way of course the easiest implementation of Oracle's centralized one where one third party gives its information and it's also the cheapest one and like in the two-step oracle we would say then if this oracle provides any malicious information information then there's the possibility for users to challenge this oracle and then maybe swap it for an Oracle which is more sophisticated like the one that Otto provides or other approaches that we have already thought about and

implemented something that we call the ultimate Oracle so so basically you could have a centralized Oracle and then if there's a dispute then you could the augurs Oracle could resolve the dispute for you it that's just another form of an Oracle so so you can I mean in the Oracle you can specify that that's only one step in the centralized or that's only one step in decentralized but you can also specify an Oracle that takes first to centralized Oracle and then has the option to some dispute mechanism and then calling I I think that's just an example what makes sense to make it flexible the the Oracle so what does the

ultimate Oracle then yes the item a torque is now as it Martin came up with so so the dual take a little bit longer but I so the idea is a northern Oracle is as secure as or it's it's equally secured to the amount of money kind of that's at stake if it's in centralized Oracle then for example the reputation of this person is at stake with arbor for example all rep tokens are at stake and the ultimate Oracle has the idea that all Easter is at stake and basically the simple description is if

there's a dispute so let's say was Barack Obama elected in two thousand thousand twelve or eight or whatever is an ounce or whatever yeah right then you you get you a few the two possible outcomes and and voting for one outcome would just be sending ether to this outcome and in the end the outcome is chosen where people send more institute and and as an incentive the the ether that sent to the other outcome that's distributed to those who picked the one

that more people send this or two so now you could ask okay from game theoretic perspective that's a symmetric game from from from the mechanics of this game it doesn't matter what's what the true outcome is but there are two to two reasons that breaks the symmetry first I would argue that it's easier to coordinate on the truth that's kind of the natural shelling point and the second argument is and that's that's really only the credible threat that's nothing that I want to see but that's something that yeah the credible threat

that that exists that hopefully because of its exit is existence it's never used the threat would be that you could fork athyrium so let's let's think through so in the extreme case of someone is betting on or lots of people are putting money on the wrong side and then be told so many as a back side and it completely escalates and then there are after the 80 million ether there are 6 70 million or something involved in this dispute and then finally the wrong side would win then the true side would have could say okay we just fought

either and we just start before this process began and this is option the the the evil coalition does not have because they they cannot really say okay we try to screw someone over it doesn't work we lost a lot of ether now we focus please join our fork so while the other side could say could argue okay we want to honest kind of honesty sir yeah so that's reasonable so so basically what you're saying is if if there are two centralized Oracle's on the same that a pretty information for the same event and one of them says hillary clinton was elected president 2016 and the other

says no she was not then basically you have a betting game where ether holders can bet ether on one outcome or the other and then whichever outcome gets most of the ether bet on it wins and in general people are incentivized to bet on the truth because they expect that to be deepening outcome it's it's slightly different so you only need one centralized oracle in the first place and it publishes written result and then there is and before bets are resolved and there is a dispute period of let's say five hours at twelve hours of 24 hours and if during this dispute period someone challenges the resultant said

okay that's not true then he would need to to to put a bounty of the earth in like something like 500 ether found either to trigger the ultimate oracle and then it would be triggered otherwise just the solution of the just the outcome of the Pinterest Oracle would be taken so in terms of the in terms of the market maker technology are you using automated market makers yes we do reduce the LMS are a market maker and that's the standard that we use but we also use offer the possibility to complete off chain trading so anyone can also become a market maker by offering odds for fault

comes and markets and no so enter this modular as well so you can use different knocked Nick algorithm that's also optional right so I went a little bit describe how its modular so you you have an event that's defined by somehow an event description and it's defined by who's responsible to resolve it and from this event you can generate tokens and those tokens can be traded with an automated market maker and we have the option to just choose an algorithm or just again choose a contract so it's again modular and we have implemented why not to that can already be chosen but you can create any market maker you

want but you have also the option to just for example sends those shares because they they implement the the token interface so you can just send them to someone and then kind of make over-the-counter trading or you could use a decentralized exchange like ETA X or oh there are different options yeah cool so our markets already live on group noses and what kind of markets are you seeing right now right so it's more experimental currently we had we we are I guess we are the most used Syrian conflict right now so we have like 5 600

traits or transactions attraction to be no no no no no no a nine total so so that's a it's right in the beginning wet currently it's stuff like a surprise or he's a bit difficulty or some insights on on the DEF CON here and something like that I i thought the coolest one you implemented was how much how much repeat how much money was all were wholly raised in dollars that was the perfect demonstration of group gnosis wasn't it yeah it was our first market it was pretty accurate forecast

so now in terms of in terms of building on top like let's say let's say somebody wants to so this group gnosis wanting to be the actual user interface to a prediction market or are you trying to be more like the back end let me let me take an example prediction markets could conceivably be used for sports betting right because sports betting is a is a prediction markets and sports betting is a huge market around the world and especially there are many countries that do not allow certain forms of it Andy theorem could considerably skirt through all of these regulations now if there

are markets like these is group gnosis trying to be the backend for other people to build these kinds of applications or are you actually trying to be the front end and make these friendly for users so the the ideas the back end we have to do it in any case so that's what we build right now but we want to like there was a talk today about the average user and that we don't really know whatever to this Oh everyone has a different opinion on what the average users and that's why we want to try out different skin so different front ends for all back end and so we will start with a skin called hunch game and that's the predictions about

anything any gossips are around and celebrities everything in this area and I will try to to test out different skins and see what's possible but of course the back end is is there in any case and we are open for other people to build on top yeah the the ultimate goal would be that we have a bunch of tools of different front ends and that it would be easy to to create your own front end and your own market for your topic but since it's all compatible in principle you could also do the sports

betting on a skin that's more oriented for financial markets or you so so it's all into exchangeable but yeah it's clearly very important or it is important that different topics will have just very different user interfaces if if you want to do something on sports betting the new you most likely should should do the interfaces that people are used to with if they are used to sport betting if you do something like the hunch game that's more oriented to a less very less technical audience then you would have a very very different

interface from I mean there's much more that prediction markets can do like insurance contracts or forecasting under specific condition and obviously that it's a completely different user interface compared to will Kim Kardashian do this or that so so so finally one of the one of the interesting concerns regarding prediction markets is ethir IAM as we know is not a very scalable system and conserving gas and the amount of computations on the network is very important and I've seen posts from group gnosis regarding off chain methods to increase scalability in prediction markets so perhaps you could tell us about some of these methods that you are

thinking about sure so we are implementing of chain trading so it bit has said basically it's similar to to the lightning network approach so the general idea is that to users lock up some money and say or I'll lock up some kind of asset could doesn't need to be either could be any kind of tokens and then they they privately have a their own state so they agree on a state and only if there's a dispute then this

state is settled on the chain that's very powerful you could even do predictions or markets even the market does not need to be on the on the etherium chain yeah so you can run your whole prediction market completely off chain and it's still trust less it's still it is instant you have no transaction costs the only thing the only downside is that if you want to settle it then then you have you need some time to give the other one the chance to to make a different claim so basically you are submitting the latest state but who knows if it is the latest

so you have to give the other part of your trading partner the chance to come up with a new at state if he comes up with a new estate then you got punished because you tried to cheat but yeah the incentives and the economics works in a way that no one is incentivized to cheat because he most likely would get punished so the assumption is that he does not do it in the first place and everything works move yeah so so so basically what what's happening regarding this is that just as in as in Bitcoin you have you have payments that can take place off chain on etherium you

have the possibility of whole computation to take place of chain right and you are using these kind of mechanisms where the computation happens on chambered is finally settled on chain as a way of scaling prediction market and also guaranteeing low latency in Indian traits right it's just the opportunity to settle it on chain doesn't make it necessary to trade on chain so you can always trade off chain and if you think about sport betting markets you have so many events so many bits that will be just not feasible to do it on chain cool and how can I listeners discover more about group gnosis what are the resources they could

go to so we're life group noses come you can try out the DEP the other bigger resources we are quite open in our development so we have a forum forum that group gnosis and we discuss what's going on and without the latest development steps and that's a good way to to get in touch our seats latest progress cool so that's it from the group who knows his team thank you thank you thank you