Designing Conditional Markets and Futarchy | Devcon SEA

[Music] uh so I'm Calin as I know as cast. uh I'm the co-founder and CEO of uh futari do5 it's a project uh we just announced uh two days ago uh and I'm going to be talking about uh conditional markets and uh a new way of making decisions called uh futar actually uh it's uh was invented by Professor Robin Hanson who I have have an honor to have with me and as our chief scientist uh and uh but it's um only recently have been implemented uh in uh concrete U applications so let's go so first of all what are conditional markets uh first question everybody asks is it the same as prediction markets uh not really so let's go over it uh so first let's start with an example of a let's go with a prediction Market um about um let's say the gas fees of etherum so U let's imagine uh here uh will the gas fees in ethereum be higher than 100 way right so uh in the next year so that's a prediction Market you all are familiar with um so imagine here's 40% so uh how do we build a conditional Market on top of this so um basically a conditional Market is a trade that place you can make trades on um the different branches of the Multiverse I like to think of different branches of the m so there's one branch where the gas prices are high this is the yes Branch there's one branch where the gas prices are low this is a no branch and uh you can trade e on both sides uh separately and they have different prices because uh uh so how does it work like if you buy uh the if yes the if only if the gas are high uh what happens if the gas end up not being high so uh the way conditional markets work is basically you any trade that you make on on on a branch that doesn't end up existing uh the trades get reverted and um yeah so and get your money back basically so but how do we implement this on a blockchain that's a question so uh here is um bit of a technical explanation so I don't know how many are you are familiar with uh the conditional token framework so it's basically um you can get a token and you can split it into yes and no versions or more versions so uh the yes version is the ver is worth something if the condition satisfied the no version is not is worth something if it's false so here what we can do is we can get a both die and if as assets split them both you have the die yes the if yes the die no the if no and we can actually go around and and buy uh you can actually go around and make pools they DS and ifs the if know and I know so this pools going to basically have the prices of if if the guys are high and the prices of if if the gas are low right so you can actually get conditional prices uh which basically tell you uh whether it's good or bad for if uh if the if the gas are for price if the gas is high of course it doesn't tell the causation uh it cannot tell you if uh it's gas prices being high it's good for ethereum or if it's the other way around if the eum is being doing well if being doing well uh means the gas pric are going to be high so it doesn't tell you this but it tells you the correlation so let's go move forward uh so it we can also do conditional features um so basically if you trade using uh the DI for instance as a Catal you automatically have conditional features uh so here here's an example like we can try to estimate I have features for inflation features for unemployment and we can actually predict this based on uh we can estimate the what's going to happen with inflation and unemployment depending on whether the interest rates are cut are cut or not for instance uh so he we actually implemented so uh or uh new project it Al also owns this exchange called quiver. trade has been working on this on the last few years so we actually implemented conditional features so here's the conditional features on uh the price of Bitcoin if Trump wins or loses the election uh so you see had two different prices right so uh the market was estimating that the the price of Bitcoin would be higher if Trump points and that's exactly what happened the final price was 75700 uh and eventually the market actually settled for 75 75900 right so you can actually bet on how how big of the impact it's going to be so let's move forward so what is futarchy futarchy is uh basically using these markets to estimate the impact based on the difference between the gas and no prices and then using this estimates to make decisions to actually decide things so let's go move forward so just some uh history so there are uh there's been some places where you can make uh already um conditional forecast so there uh manifold that to trade with Mana which is a play money now they have sweep sticks um tackles is a forecasting tool that gives you points based on your forecast it tracks how accurate it is also let you make conditional uh estimates um so here's an example from the um kind unintentional as example you have prediction markets for who's going to win the nomination uh here in the case like the Democratic domination and then you also have the probabilities of they win they they winning the election so if you divide one by another you can get the prices estimate like what's the chance of their winning the election if they get nominated so in fury like the Democratic part could use this to decide that hey nominating new Gavin newon would actually be better for them to win but they didn't do that and they and they lost uh so yeah uh so here's a a pioneering project uh called metad it was the first actual implementation of futar uh governance uh for Rio uh so this is a a project on Solana uh it um uh it's a pretty interesting pioneering project um that we ow a lot to uh and so yeah so BAS basically uh they have their token price and if uh if you create a proposal uh uh they can uh see the prices uh if the proposal passes and the price the proposal doesn't pass and uh so basically the proposals pass if uh the market estimate this is going to make the price go up if the proposal passes so it's a new way you can use that instead of using token voting right uh so F key going forward uh so first uh well uh we just uh announc our new project fy. uh so we're going to bring the very first futar on ethereum uh so and but we going to be focusing very much on uh this question which is well how can we actually get people making great proposals creating great proposals right now mostly only insiders create proposals um or you have people from the outside of Dallas making proposals usually they want to like a make a proposal for them to get some money or something right how to incentivize people to make great proposals regardless of their self-interest so here's uh um um the framework we have in mind it's basically a futarchy cycle uh so first uh don't have people bidding to to create proposals so we cannot have many proposals at once not easily at least uh so let's have an option for the right to make a proposal um step two uh you they present the winner presents the their proposal a draft there's discussion people discuss the edits and then there a final proposal that goes to step three the market evalu ation where uh you can actually estimate the token price if the proposal gets passed and the token price if the proposal fails uh and you have the maret estimate which of the two uh is actually best so basically instead of voting yes or no in a proposal uh voting yes you have to vote for yes uh using voting because you don't actually make a trade you don't make a vote uh to vote Yes is basically tell I want to buy this coin I want to invest on this project if they do this uh to vote no is actually I want to sell my my my coins uh my tokens if they do this so basically it's it's a pretty interesting thing because um not only because you don't have to think that the market is perfect and always does make the right decision the cool thing is uh if you are a token holder and you see the project doing something stupid you can just say hey uh you know what uh I'm going to sell all my tokens conditional on this proposal passing uh I'm going to sell them all uh so basically for a proposal to pass they have to buy all the people in favor have to buy out everybody that is against right so it's a good protection for for minority token holders uh all right so and then after the proposal passes automatic exe execution or uh you can go to also to normal governance if there if it's an advisory FY and then first the last part it's very important that's a unique thing we're bringing uh paying rewards to actually incentivize people to make these proposals uh so here um how we think of a reward distribution we want to give a reward based on the impact the bigger the impact the proposal has on on on on the price uh the better uh uh and the be the reward needs to be so uh who needs to get rewards uh the bidder the one who want the auction to make a proposal uh liquidity providers who make propos liquidity on the conditional markets uh people who and also maybe people who gave intellectual contribution and other kind of contribution other kinds of work that needed for the proposal to actually happen and when when do we give the rewards maybe when the proposal is evaluated and passes by the market when it's actually approved by the governance mechanism and when it's delivered uh actually executed in practice and it actually was done the proposal was actually done so uh so what is being built right now there's some few things being built uh right now uh so there are several projects uh working on using this for Grant evaluation so they're going to have an accurate estimate objective estimate of how good the grant was or how how well the grant performed uh and you have an estimate of this uh conditional on the grant being received so you can actually use that to decide which grants to to give uh so so uh you have the madow itself is using doing that on you have the battery project doing this uh onum uh they got a grant from Unis swap uh so futar B that's a new thing uh so we are uh um going to make a partnership with metalx it's a a a legal cybernetic thing so it's basically how how to create a legal entity that is controlled by a fitar okay so so um U so um another example from um science uh so I don't know how many you know about CICS so um but you can have scientific research you can have a prediction market for what's a chance of a breakthrough in science uh and you can actually make it conditional on like getting $10 million moreal funding or not so maybe the chance of a breakthrough if there's no funding is only 20% but the chance of breakthrough if they get $10 million is um uh 70% right so you can actually estimate H how how much your more money is going to actually help and you can actually use this say to give out um give out uh credit certificates right uh certificates uh impact certificates for the funders uh in a very in a in a in an objective way in a very but Prett uh so um and the last uh Market advice for D so we can use the same mechanism uh for fitar as an advisory mechanism uh so we hope to uh so we have our project has uh been funded by uh lad or by osis D so we hope to uh uh serve as a advisory mechanism to create proposals for for diagnosis Dow and right and uh here are some new ideas so uh many people uh we encourage all of you to think of ways you can use this these ideas of uh of futar and conditional markets uh so fire the CEO markets stock price is lower if the COO is fired means you should fire the CEO uh evaluation of future employee um and uh chance of making deadlines even romantic relationship right there's this project that did this expected GPA so you can use this to decide which students to to to to uh to approve uh which girlfriends to which people to go on a date with uh Capital location for companies well many things many thanks uh yeah and finally uh governance we could use this eventually as a way to to evaluate the impact of Road decisions right so now uh what everybody has been waiting for uh Professor Robin Hanson uh thank you very much so you may have seen him on the during the first day first talk some of his papers were mentioned including one about a market manipulation which is a concern everybody has about vitar so thank Robin thank you very much I'm sorry I waste some of your time thank you hello so people have been concerned a long time about manipulation in prediction markets which include fkey and other markets so I just wanted to address that first before we go on and talk about futar uh so the idea our standard model of financial markets has two kinds of players I call them wolves and sheep the Wolves know something and they go to a the market to trade on what they know the Sheep don't know anything and they're there to trade for some other reason that's pretty much it and the Wolves want to trade against the Sheep when a wolf trads against a wolf on average neither of them can make money but when a wolf trades against a sheep they can make money so the Wolves want to trade against sheep so wolves are attracted where the Sheep go so this means that um the more sheep there are the more wolves are attracted the more wolves who know more things and who try harder to know more things and therefore uh the price gets more accurate with more wolves and therefore with more sheep so a standard we know about financial markets the more trading there is in a market the more accurate is we see this in stocks say Google is bigger than a small stock company markets in the presidential election more accurate than senates Etc so uh this is a basic fact about the markets that the more sheep you can get the more people you can trade in a market for any reason the more wolves show up the more accurate the prices get and the key thing to know that a manipulator somebody who wants to push the price somewhere is basically a sheep that is they don't particularly know something they just have a goal of what they want to do by pushing the price and that means the she the Wolves would love to trade against them and expecting there to be manipulators attracts the wolves and makes the prices more accurate so that's the headline on average when Traders expect manipulators in a market the price is more accurate all right so now going back to futar so there's this key mechanism that I'm been excited about for 25 years now but the key point is you have a metric of what you want what you want out of something so say this is the stock price and we want to know should we fire the CEO and so what we do is we set up a break of two markets one is the stock price if we fire the CEO and the other is the stock price if we don't we track those two prices for period and we try to tell if there's a difference and if we can tell what the difference is then we follow the market advice about the difference so if the price of of the stock giving that you fire the CEO is than given that you don't you fire the CEO and then from that point on the price should track whichever decision you make so you make decision conditional prices you make a decision and then price continues on with the one you had so this is the dream this is the vision of futarchy but I know you guys are designers out there so I thought I would include you in our process of thinking about all the things that can go wrong here or might go wrong and trying to fix them so now that 25 years later years later we're finally implementing this in a number of ways uh there are number of concerns and issues about things that could go wrong so for example uh if you just say okay everybody make proposals you'll get way too many of them what do you do as uh Kelvin mentioned might can hold an auction so that only the best proposals might get uh approved and we'll evaluate those uh if you're looking at these prices that go up and down and it's hard to tell the difference well uh you might want a thicker market and one way to get that is to have subsidies on automated Market maker to subsidize the trading and then you'll get sharper prices and you can see clearer differences but if you make people pay to put uh their proposal up and then you use some of that money to fund a market maker there's the problem that the only people who will pay to do that are people are going to get some sort of rents out of their proposal The Proposal favors them somehow maybe we'd like proposals that just favor everybody how do we do that well what we want to do is reward people if they make a proposal that's passed we say good job you should get a cut of that and if we can we'd like to even see the difference in those two prices because that's a measure of how good a decision it is how much value is produced and give them a cut of that and that's a way to do better on getting good proposals to be made uh and not having too many others distracting all right now there is a potential problem we're not sure how much of a problem it is but it's been reported as a problem that when you have these conditional markets one of the condition is more likely than the other and eventually as the market gets pretty confident what the decision's going to be then the outcome that's unlikely to happen people might feel free to mess with that price uh by pushing it up or down because say if there's only 1% chance it happens there's only a 1% chance they'll actually have to pay anything on that and therefore those prices get might get messed up so I'd like to do some lab experiments to measure this but the key idea is Let's Make a statistical model that includes this effect that the accuracy of prices might get worse and you move into this regime of a very low probability event and then we can take that into account in our formula for telling whether there's a difference in the prices and how much it is and then we can not worry so much about this for that we'll need some sort probably also to have a market in the probability of which outcome is going to happen but that's straightforward to do now we have this middle period when we're measuring the prices and we'll put a market maker in there to make sure that we get sharp prices but now if the market was thinner before that period this is an incentive for people to wait and reveal information during this period because that's when they get paid the most for it but now if information is being revealed during this period that'll complicate the diff the ab our task of trying to distinguish the two prices the difference might vary with the information that comes out so what we really want to do is have the information come out just before this period so then we want to have a especially High subsidy just before the measurement period tell everybody this is the time to get your information out and then low information during the period easier to measure um others problem potentially so say we have a proposal that everybody thinks is great so if we don't approve it right now we're probably just going to propose it again and approve it next period but now the difference is really the difference between what happens if we approve it now versus a month from now and so that might be a very pretty small difference and therefore we'll reject the proposal so to deal with that what we might want to do is first have a proposal we say we're about to consider the following proposal if we don't approve it now we don't get to approve it again for a year and then when we consider the proposal we'll know if you don't approve it now we won't have it for a year now you'll see a bigger difference between yes and um now it's sometimes convenient for organizations to lock assets of some investors and some U managers but if there's a market proposal they really hate it seems tempting to let them sell those assets in the conditional markets as if to sayl if you're going to do this I'm out I don't want to be part of this organization I want to sell my assets that seems to make sense as a structure but we worry that people could sort of use the excuse of selling something in order to sell the assets that were supposed to be locked and so there's a there's a trade-off there to work out how do we set it up so that people who are big players in you know in an organization with assets say I really hate this proposal I want to be out if this happens they can use their money to sell on the no you know sell no on the yes side push the price down on the yes side uh and we want that to work for that now uh there's also an issue that there might be a good proposal and somehow people conspire to knock it down so they can put it up again with their name on it and they should get the reward so we need some sort of intellectual property with proposals some way in which you invent a proposal you put it up somebody else can't just cross your name off and put their name on get paid instead uh and so we need some way to judge you know who contributed how much to proposals um finally well oddly uh prior people might be um this often happens in organizations uh if we use these markets to uh make decisions that previously important people were making they might be unhappy that the Market's making the decision instead of them and they might try to resist these markets and so that's an issue with trying to maybe change Norms or so initially in the prediction Market world people made relatively safe decisions even too safe in order to avoid offending people or decision makers um and now for Q&A thank you very much those were not slides thanks milon guys that was really interesting uh the guess uh how long before we can start putting bets on uh futari um hopefully uh before the end of the year or very early next year that's the Hope wow wow wow that's uh that's really really exciting uh so a few questions here uh we don't have don't have too much time but first one is uh maybe for Robin specifically who did you bet on the presidential election the whole point of these markets is that we want you to shut up if you don't know we want an environment where people don't all just talk because they're interested and want to bla and instead you're supposed to shut up if you don't know so when I don't know I shut up I didn't know nice so in fact I don't actually trade in very many of these because I've fully internalized the norm that unless I think I know as well as the best people who will participate in these markets I should shut up and not participate well okay that's uh that put me in my place uh so next one uh why haven't we seen uh or again this is again I'm not sure if this this holds up but what haven't we seen fetar governance yet and how long until we see it um so the way I see it uh is U some sense some kind of uh civilizational teory uh that uh uh I think uh the timing uh to get uh people excited about FY was is is now that uh uh prediction markets are getting mainstream uh so that's my vision of this look everything happens first at some point this is the point I like it uh okay so a bit more detail one what additional or to prediction markets in general what additional costs are intrinsic to using prediction markets to aggregate knowledge as compared to just pay uh paying a set of experts for their opinions how can you trust these experts well so if you want to pay a bunch of experts first of all you have to figure out who they are and a lot of people volunteer to be experts where they're not remember the we just talked about okay secondly you'll have to get them to be honest with you so often experts know enough to tell you what you want to hear pandering to your preconception so it's hard to get them to tell you what they'd really think if they weren't pandering to you so you need a way to make them be honest and that's why tying their incentives to the truth uh helps a lot got it okay so maybe one more question so you give a few examples one was related to sort of a decision tree for for presidential elections another one was phoc government uh so who do you who do you see as the counterparties or liquidity providers in these mechanisms uh so for the governance mechanisms uh the company itself or the the organization itself is a pretty strong candidate uh because the organization is going to be uh benefit be the one to benefit the most for a good decisions so they have all the incentive to subsidize it uh yeah so the the General Vision here is that we have lots of topics on which we'd like accurate estimates there's a lot of topics on which we could make prediction markets but if we just turn them on people might not participate because why should they bother to give you free information what we want is the people who want the information to pay to create and subsidize the markets so that other people will come and trade to give that person the information they wanted so this isn't a way to get free information this is a way to buy the information you want yeah that that that that's a a perfect note to wrap up on again thanks thanks very much for the talk guys and another round of applause please [Applause]