DAPPCON 2018: The Road to Futarchy - Martin Köppelmann (Gnosis)

Transcript

yeah first of all just one more remark to mittens talk so we are in fact we are was notice we're working in different capacities together with mr sztyc and one thing I want to encourage everyone to try out so we just a few days ago releases Dutch exchange in kind of this autonomous exchange and that one will be governed by a doll stack now so well there will be very soon very concrete opportunity to try out all that stuff anyways my talk now will be kind of still about in yeah the the one alternative mechanism that tries to

reach the same goal scalable governance and yeah it is as matang already else mentions the word foo turkey so using prediction markets just prediction markets to to make decisions and kind of the radical or the the the promise in my view is very high level promise undeniable go into details is that with voting systems you still have a closed group and it's not necessarily clear that those who have the voting rights or the reputation or the tokens that those are kind of that there isn't knowledge outside of that groups that shouldn't be important and

that or that should be incorporated in the decision process and in contrast with sue Jackie it's totally open so everyone who things can contribute to this decision well can actually contribute and well the the big idea is that there are measures in place that it's still resilient but I will go into it at least if I can go to the next slide it's not working can I do anything oh here we go so I assume people know what it predicts negatives but just kind of to to rise our perspective so you need an unambiguous publicly known event the

outcome or upfront you need to cover all possible outcomes eventually you need an Oracle that that kind of describes what outcome happened and then you can make a payout and you finally need currency or what we call in our wording collateral token so again you have this collateral token you define yeah event with with all the possible outcomes in this case three and you define an Oracle and then it's just a very very simple tool and we have an EAP for that and in the making that you put in this collateral token or anyone could at any time put in one

collateral chokan and create three of those or depending on how many outcomes there are potential outcomes three of those outcome tokens and eventually and at anytime you can create those outcome tokens a full set of outcome tokens for collateral token and at any time you can go back you can take a full set of outcome tokens and convert it back to the collateral token but when eventually the the outcome is resolved and the Oracle says which outcome happened well then you can redeem just the winning outcome for the collateral token so again that means if if you create those outcome tokens and you trades them on a

market and you assume the outcome a happens with the likelihood of let's say 30 percent well the price the price you should be willing to pay for that is 30 percent of the collateral token and if it's trading at if it's trading at 0.25 you haven't clear incentive to buy it all the way up to the price of 0.3 because there is an expected profit and that's just a yeah very very simple concept of the prediction market so again you have a market creation phase where you define those stuff you have the trading phase and you can all

invite it to try that out if your your cards preloaded with only way you can try them out in in in the prediction markets odds that all works today and eventually the market is resolved and paid out yeah I also skip this slide so now that's kind of the general concept of a prediction market and now I want to yeah hint you terms to some observation for the general concept of a prediction market it is important that the collateral tokens of the currency you're using is in a way independent of the event you are predicting so it doesn't

make any sense to make a prediction with easer will the price of isa drop close to zero because if I would well I mean it's obvious I would never place easier in this prediction and and bet on yes because well even if I win I lost and the same thing is i if I would use die to predict whether the die mechanism will fail well I would it wouldn't make sense to me to and then you could even go Safari if said what it was a prediction will they are hit by a met Metroid and kind of everything is destroyed well yeah when I would put all my money on no because in the case of yes well what is it worth to me so so what what we see is again for the normal

use case the the value of the collateral should be totally independent of of the outcome or the outcome or the event should not influence the value of the collateral if your goal is to get just right prediction for the likelihood of that event however in future key and in conditional tokens we are exactly using that that relationship to to kind of discover how events have an effect on on the collateral token and the interesting thing that is that the collateral token can not just be a currency but can be something that measures how close we are

getting to a metric or how close we are getting to go but we'll go into that so the simplest use case of that would be how we want to make use of that is we would take a token yeah the gnosis token or some whatever token and we could yeah we could take the nozzle token and we would ask the question will noses the nose platform have a trading volume of more than whatever 10 million in a year so obviously that outcome well might have an effect on the token price or the collateral price so what we get is we if we just apply the same principle we can

have a token under that condition or kind of a future in the store well actually it is still the promise for the for the real token so it's not just the future its but just under the condition that something else happens and then you can just come it's just on the market and you get an idea of how that event would impact the price so one example where where that could have all been already quite useful was brexit of course you could have have a market in pounds so you could be for breaks it happens and independent whether or not breaks it happens you can already see up front before breaks it

what is the value of the pound or what will be the value of the pound or how do the market value eights value of the power under the condition breaks it and under the condition no breaks it and again people can make those trades independent whether they don't need to have any opinion about whether or not the event will happen whether or not breaks it will happen they just have an opinion about what is the price of it if it happens so they they can yeah have a position that if the event doesn't happen they well they just get refunded or they they have no exposure so so that that mechanism alone you can use to make those those conditional tokens you can

already use to to to reduce your risk basically so today if you buy token or I would use the same thing the for for for stocks now if you buy one stock you're effectively doing a lot of predictions in a way so you're assuming this will happen the day if you a bunch of things need to come together to make to make that a good investment but maybe you don't want to kind of buy into this full set of assumptions or predictions you just want to say well I know this okay I kind of I know this relationship I know that if the team achieves to does two to kind of actually build it the project

will be very successful but I have no idea about whether the team is actually capable so I I think it's a good idea but I have no opinion about the team and and then someone else could could have well different knowledge they yeah it's a good team but I don't know if the idea is good so you can make widows conditional tokens very specific bet so now we we go even a step further and ask how can we how can we uses this concept of conditional tokens to well actually make decision so far we were just like observing if something happens what will be the value and now we are using this

mechanism to say well to actually make the decision of what will happen so we we are looking at the prices of how will the decision or the event affect the future and depending on how it will affect the future depending on that we do make that decision so the the first the the most critical part is to actually define a metric so what you need always for food turkey is a metric what do you want to optimize for and in a case like publicly traded companies that might be simple you say well it

should just be the the the the share price but I would argue or I would argue it's it's quite a heart problem so if you would say for for for for etherium we want to have kind of a we might want to have Unchained governance process it would be very far from obvious to me what is the objective metric for aetherium governance so what is the goal of aetherium governance and I think that is that is very hard question and futa key forces you to answer that question just just to give a few examples if you would just say well the goal of aetherium is to make to increase the value of ether well you could make a

very simple proposal that does that you would just propose to half kind of orchha to to burn half of all ether on kind of force it in all accounts well then maybe you would double the value per user whether that was I were certainly not useful useful proposal and then you could say well then just make semis a market cap but if that is that really the goal and kind of if you would say then you make a proposal that you create new easier and kind of take some some people out then you might increase the market cap but is that release goal are there not a lot of other goals kind

of in there so a lot of other goals that involve some sense of fairness some sense of whatever so I'm just saying the metric is really hard but anyways if we have a metric and and the slogan from Robin Hanson who came up was that was vote on your value so the suggestion was here to vote on that metric but then use the markets to figure out how to achieve this metric so we we need this metric and then we create those conditional markets will be basically asking what is the forecast for this metric under that condition or what is the forecast or

under the assumption that we reduce this decision or under the assumption that we do that decision and we kind of allow trading on those both markets and eventually the system would automatically implement the decision which was a higher forecast on the metric so yeah one standard example is the metric in this case could be just the the value of the share price of the company and the question could be should be fire to CEO or actually gives a quarterly revenue is here the the metric that you want to optimize for and yeah

you could again create those two markets prediction markets what's on the quarterly revenue under that condition under that condition you allow trading eventually market shows yeah the when decision would be better and you implement it so in the context of of blockchain governance one of the promising parts about ya futa key is again what I said earlier this complete openness anyone can participate in the second part it forces discussion on explicit values whether that is a good

thing or a bad thing I do not even have necessarily an opinion about that so a little bit technical again you have those it's kind of how we would do it on smart contract level we have those on the very top you have the collateral token that might be just something a currency easer die whatever and then you in the first step you create those those tokens that represent the currency under the condition that that a or B happens and then you're using that currency in a

second market or in two markets that predict that predict or are prediction markets on on the on the metric and you can go long or short on on the metric yeah a few a few ideas or a few things how we do it completely so the first step in my opinion would be is independent of hudecki just experiment with those conditional tokens just have markets on easter under the condition that this happens or gino another condition that that happens or whatever and the next part is of course the very

heart or finding metrics and a few suggestions or a few things how you can experiment and different doubts that you could set up an experiment with shiitake so one Dow could be just a single event Dow so in this case the metric would just be whether a single event is happening or not how do you do that well you can for example create a prediction market will experience which to prove of stake by the end of whatever 2019 yes or no and now the yes token here in this market is kind of a metric of how likely is that

event to happen and now you could do a footer key that optimizes or the goal of that food a key is just to optimize the value of that yes token effectively now this footer key is the only goal of that food a key is to make it more likely that that that proof of steak will happen until the end of 2019 so you can allocate budget to this thing and the thing will just make proposals and they will only be accepted if the market things that proposal has a positive impact on increasing the value of the token which means making it more likely that this event will happen proof steak will be implemented and the second one would be simple enough hedge fund au so

that would be the same mechanism as the Darrow back in the day so a lot of our people would put money in everyone proportionately gets a token and then you just do the the prediction of the metric you want to optimize it's just the value of the token so the doubt doesn't care about anything it just cares about making money and the third suggestion would be support Dow so we as gnosis could say well we have some resources and part of it we want to allocate just to add ow that supports gnosis for example and you know prize or or other other objectives we might have

and get if we just give it to the style and just supports another objective and the the fourth the fourth proposal is this idea of that the that the objective is a sentiment so again I said earlier it's quite hard to define and to define an objective or a metric but what you could do is you could give any participant in your in your system some voting rights and at some point you could just ask them how happy are they or kind of how happy are they was with

with the decision process or with the state of of the network or the Dow and and that is the metric the metric is kind of the happiness or the sentiment of of some population and again then the metric force Dow would be the prediction on on that so you do this vote every year or something like that and and then in the intervention so you have this prediction market on that vote kind of how happy or kind of what metric something but for example between zero and one will will be the outcome of the vote and again during the time you your decisions are made to optimize that

metric concept we also have implemented and there's currently workshop going on the bit the built room is using future key for TC ours so TC RS usually would use voting but of course you can use future key and here's the metric would just be optimizing the value of the TC our token and that's something we have implemented and are also very close to trying it out so yeah timelines I have to say I put it maybe behind it I have been too often wrong with with timelines but yeah we we have or basically this framework of creating those tokens that is ready and

we will polish it a bit more and make it very accessible for for anyone to just try it out well then we kind of really need to explain this concept and get people used to to trade those conditional tokens and then we can start experimenting with all the different possibilities we have okay thank you [Applause] you