Launching Futarchy: Conditional Spot Markets for GnosisDAO and Beyond

I'll be talking jointly with Kelvin here, he'll talk in a few minutes. In 1981, a Jeffrey Warnick ran a whole company using this concept of future, I'm going to explain in a minute. For three years, he used it to decide who to hire and what features to add to the software he wrote. And after three years, he sold it for a lot of money, he's been an investor ever since. But apparently nobody else has ever heard of this thing, I heard about it last year, so I have to give him credit.

But I came to this field long ago when there was a time before the World Wide Web, I don't know if you know that, and that people hadn't even imagined the web, and I was part of a group of people who were imagining the web, and then I had some doubts about that, and then I started, how else could we do the things we were hoping to do with the web, and I imagined, oh, maybe betting markets could work for that. And so I started to write about betting markets, prediction markets, as a way to aggregate consensus and deal with social problems in the world, 35 years ago. And then, roughly 10 years later, I had the insight that, well, the main value of information is to make decisions, and so the main way to make prediction markets valuable is to put their information right next to a decision, and that originated the idea of decision markets, again, which I'll explain. But I wrote about that 25 years ago, 26 years ago now, and not a lot happened for a really long time, and then three years ago, a secretive government agency in India started to actually use it for decisions, and two years ago, this organization called the MetaDAO started to make its major decisions using this governance mechanism, and they've made 69 decisions since then, they have 13 customer firms, and then in the last year or so, a number of new firms who I'm advising have been starting to explore this concept of what I'm calling futarki, decision markets, and they include Buttery and futarki.fi and some MetaDAO in addition, and in the last few days and the last few weeks, I've been hearing from a lot of other people doing other projects that I'm not directly involved with, so this is now a thing which makes me really happy after only 26 years waiting for it to get tried.

In general, I'm missing a slide. Well, we can use it on this slide. The key concept here, just focus on the top of this slide, is we have some organization with an outcome we care about, so for a for-profit company, it might be the stock of the company, for crypto, it might be the coin, for an election campaign, it might be the chance of winning the election. We have an organization with some goal, it could be an individual too, and we make an asset that tracks this goal, so for the stock, for the company, the price of the coin, and then we can have a market in this asset, as you well know, and that's the unconditional market for this asset. And then at some point, a decision will be made, a possibility of making a decision will come up, well, there'll be a proposal and we'll want to know, do we approve this proposal?

Do we like this proposal? Is it a good proposal? So then the key idea is we now make conditional markets, so an unconditional market, you just trade the stock, say, for cash, but a conditional market, you make the trade, but the trade will be called off, canceled, as if it never happened, if a condition isn't met. So here we're going to trade the stock, say, for cash, in one market, the trades will all be called off if we don't say yes, if we don't approve the decision, but then in another market we will say the trades will all be called off if we say we actually make the decision. So we have a price if yes and a price if no for the decision in terms of the asset that we're tracking the value we care about.

So this might be, for example, firing the CEO of a company, and we could say, what's the stock worth if we keep the CEO or if we fire the CEO? And then at some point we will end these conditional markets and actually make the decision, and the idea is to look at the difference in prices between these two markets as the information about what decision to make. That is, we keep the CEO if the company is worth more, estimated by speculators, if we keep them, and we dump them if it's worth more if we dump them. That's the key idea of pure Tarky, just basically speculators making decisions about what to do. And so the decision will be based on the difference in these prices in this period here, and as I said before, actually, all innovation is a combination of some simple, elegant ideas and a lot of messy details.

And what I've worked on for many years and want to focus on in talking about is that we've been spending a lot of time thinking through the messy details to make this elegant idea work. And so I want to spend some time walking through some of these messy details with you to show you that we have thought through a fair number of them. Okay, so first let's think about the details of subsidizing these markets. So if we just set up these markets and we do nothing else, there's a strong risk that nobody will trade. Nobody has any, you know, there's a vast number of possible markets in the world that just don't exist.

Nobody trades in them, so there has no prices. We don't want that to happen to us, so we need a way to induce people to trade so that they believe they all have a profit from trading here. And a straightforward way to do that is to make a subsidized market maker because the profit from any one trade that changes a price from the current price to what you think the right price is is basically proportional to liquidity of the market times that price difference squared. So when you offer liquidity in a market, for example, by directly subsidizing it, you offer a price for information. You're saying we're willing to pay for information about this topic, so the higher liquidity the higher the price you're offering is.

So the first point is you need to offer something, you should be offering something because the information is valuable to you, the organization, and so pay for what you want. But we have several other complexities here. One is that we're going to want to measure this price difference in some period, and in that period we actually don't want much information coming out. There's going to be people we call manipulators who are trying to influence the decision, and they want to pretend to be informed traders. They want to look like they know something, but they really just want to push the price.

So we'd like to have a period of time where informed traders are not showing up, so anybody who's trading them, they look suspiciously like a manipulator, and people who want to counter that for a profit can identify them and squash them. So how do we make a period of time where not much information comes out? Well, the key point is to have other periods before and after where we're offering a much higher price for information. We say, hey, if you've got information, please come to these other times and get paid for it then. So here we have a high price before and after this key measurement period, and in this key measurement period in the middle, we have the dotted purple line.

Manipulators would actually add liquidity to the market, and so we're trying to make sure the total liquidity is low in this measurement period, and so higher before and after. We also want people to sort of pay, we want to pay the cheapest price we can for the information, and if there are two people who know the same piece of information and they know that, then if we flow the order of the price, they're going to be in a race with each other and hopefully sell it to us at the cheapest possible price. Also the very first price is pretty ignorant, so we don't actually know what that is, so we are, you know, going to be not saying so much about that, so, you know, not paying very much for that. So anyway, there's a number of details about how you set up liquidity to make these markets work, and now I want to tell you about another sort of technical problem, which is that if the market is giving advice to decision makers who know more than the traders, you can have what's called a decision selection bias. So this is a modestly complicated diagram, which no longer has my step-by-step pieces of it.

You have time, Robin, you can go away. Right, no, but I'm apologizing for the messy diagram that was supposed to show you these pieces one at a time, but it didn't do that. Anyway, so the key idea is, imagine we have the fire of the CEO markets, and imagine therefore the true state of the world is the value of the company if we keep the CEO and the value of the company if we don't, and this is the two-dimensional space I'm showing you, and a point up in the upper left there are points where it's best to keep the CEO, because the value of the company is higher there, and in the lower right area is the points where it's best to dump the CEO, because the company is worth more in that case. Now, we're going to imagine that the decision makers who are actually going to decide this, what if they know the exact point, but we the market don't know the exact point, we are uncertain, so we might have a uniform distribution over this yellow area. That's what we know.

Now, according to our information, the average of that area is below the line, saying on average it's best to dump the CEO, and that's what our average information says, but if we know that the actual decision will be based on more information than we have, then we're going to have to make some conditional expectations, so what is the shape of the information with this region? It still says, on average, fire the CEO, but if we're going to think, yeah, but if the point is above the line, then we're going to condition on being in the red area, so we take the average of the red area, and that's the price we're going to get the stock conditional on keeping the CEO, and if it's in the yellow area, the value of the stock conditional on dumping the CEO is the average of the yellow area, so if we take those two lines in the intersection, they're actually up in the keep area, which says the markets are saying keep the CEO, when our best information is to dump him, which is the decision selection bias. So the key problem here is that we have this bias. Now, it only shows up in some situations, clearly. It's not a thing that always shows up, but we still don't like it, but the key problem here was that the market speculators were wondering, what do the decision makers know, because they will make the decision based on more information than we the speculators have, so a direct fix for that, for example, if the market just makes the decision, then the market can know that the market won't know anything more than they know if they know they're making the decision right now, so you need to make the time of the decision really clear.

Right now is when we're making the decision, what do you say right now, and then we don't have this problem. Now, if the market's advisory, we can also avoid this problem by, again, making the time really clear, and then making sure all the insiders who are decision makers, who have special information, they're either allowed to trade, or they're allowed to have friends who are allowed to trade that they can tell, and then we make sure the market has all the information that they have. Now there's other kinds of messy details, which are actually typically more the obstacle for implementing these things, so imagine this is the C-suite of some organization, and there's an autist at this table. What does that mean? Somebody who knows the company really well, but they have no idea what political agendas people have, what people want to hear or don't want to hear, they just blurt out whatever pops into their head regarding any topic that shows up at the table.

They won't last long at the table. Why? They might be a trusted advisor to somebody else at the table, but they won't last very long there, because they will just be very politically disruptive. People can't count on them to say the things that fit in their agenda, they will just be disrupting things. But that's what prediction markets do, unfortunately.

They are just the autist at the table, you turn them on and they just tell you whatever they think, whenever they change their mind, they tell you that, and it doesn't fit very well in organizational politics, and that's one of the key obstacles we often have. I could go through other examples here, but I think I should hand it over to Calvin. You can go, Robin. You can go. We have less time there.

Anyway, so imagine a project with a deadline. Often the person in charge of the project says to themselves, if I fail to make the deadline, what will my excuse be? And they'd rather have a good excuse if they fail than actually learn the answer better. And so the usual favorite excuse is, look, we were going along just fine at every project meeting, everybody said we're going okay, sure, we're going to make the deadline, and then at the last minute something weird happened, came out of left field, so strange, it will never happen again, that's why we failed, but hey, nobody should be held accountable here, we should just forget about it because it was so strange. That's everyone's favorite excuse.

But if you had a prediction market on will we make the deadline going for the last six months, and it kept saying all the time you're not going to make the deadline, you lose your favorite excuse. And people often would rather have their favorite excuse if they fail than actually learn more about how not to fail. So the promise here, the high level promise is, the world has a lot of bad governance. As an economist, I can tell you a lot of huge areas of life, law, medicine, education, all sorts of things are screwed up really badly because we have bad governance. And the main problem we have in bad governance, I'd say, is it aggregates information badly.

And we have this wonderful institution, specular markets, that's spectacular at aggregating information. So futarki is a way to take advantage of the power of specular markets to make governance more informed, which then makes governance more competent. And the promise here that crypto is now at the forefront of pushing is that the world could pioneer, test, and then field much more competent governance, which would then let the world solve many of its key problems by handing the problems to governance systems and saying, do solve. That's the promise that is here if we can achieve developing this technology and making it more widely accessible. Some other applications, do you want to skip over?

Right, so Robby explained the idea behind futarki, and I'm going to go very briefly on a very simple application for GnosisDAO. I know many of you are GNO holders, and I want you to be able to join and contribute to making futarki real, futarki on GnosisDAO, and also very soon on the whole Ethereum ecosystem. So let's go. So we're launching this project called futarki.fi.

So I'm cas.if, you can follow me, you can follow us on X. So first a bit about our story. So we had this previous project called Quiver, it was a futures exchange, leverage trading, and we did this experiment. We always wanted to, we always wanted futarki for ourselves, we wanted to use it for ourselves.

And we did this experiment on the election, the US election last year, we ran this futures market for Bitcoin, and we ran this market for Bitcoin if Trump was going to win the election, Bitcoin if Trump was not going to win the election. And we had these speculative markets estimating in these two different worlds, right, so what would be the difference in expected price for these assets, Bitcoin. And the markets were consistently telling us that in this world where Trump was going to win the election, the Bitcoin price was 12% higher. This was the estimate on our conditional markets. So of course that's not a good enough reason for anybody to vote in a candidate or not, but this is what the market estimated, and it was actually a pretty good estimate.

The estimate for the markets was 75, 700, and when the election actually happened one day later, the average price was actually 75, 900, so it was pretty close, like less than half percent error. So this is like one experiment we ran before. And now we are running futuresmarket.fi, so we're just announcing this private launch for you guys, I want you guys to try it out, we just have it ready today. And we just created our Discord today, and I'm going to share the links with all of you, I want all of you to join.

And so here's how it's going to work, this is a test for you guys to learn how to use this. On GnosisDAO, we are starting out with milestones. So Robin mentioned about proposals, and of course we want to use Futurarchy to make decisions, that's how we get value from conditional markets, we want the markets to tell us what decisions we want to do. So in predictional markets, usually the information you're getting is what kind of thing is going to happen, where in Futurarchy we want to use the markets to get more like what should happen, rather than what is going to happen. So we want to know what thing is best, and according to some metrics, and there are many ways of doing these different kinds of metrics, but the one that is more logical, more natural, it doesn't have to be the only one, but for a for-profit company with a token, the token price is often a very simple and core metric.

So we are going to be, so we have these milestones that you can trade on GNO, on the GNOS token, conditional on the milestone being reached, and the milestone not being reached, so we just have these two milestones now, but we're going to have many more, so the first milestone we have now is, will Circle deploy native USTC on GNOS exchange? So if you think this is important, you can come here, and you can put your money, and you can buy some GNO tokens conditional on this happening, and then if it doesn't happen, you get your money back, right? So this is how our website looks, so we have GNOS now, and we want to have many more DAOs very soon. So yeah, so let's go on this one and see, so this one is the one that is going to end in a couple weeks, will GNOS pay achieve more than 2 million euros in weekly volume? So we're getting very close to that, and we think that's potentially an impactful milestone for GNOS, so let's see what we can do here, so here's a description, put some snapshots here, you see there's a spot price that is a reference, so first of all we can go here and see, and click on Dune, and see what the track record is for volume for GNOS pay, you see it's getting very close to 2 million, so let's see if the next two weeks we can reach this 2 million euro milestone, so we're getting very close, but not sure if we're going to reach it, so let's go, so here's a thing you can do, so it's important to separate this from a prediction market, even though we do have a prediction market that you can go and you can trade on the prediction market outside, and right now it's trading at 38%, the whole idea here is not for you to bet on whether GNOS is going to reach its milestone or not at the given time, the most important thing is for you to bet on what impact it's going to have, so this is the spot price of GNO, and the probability that the milestone is going to be reached is just a reference, what matters is the yes price, what's going to be the price if the milestone is reached, and what's going to be the price if the milestone is not reached, and therefore the impact is calculated based on that, so just to show you how it works, you have these charts, you can see the yes price and no price varying over time, the spot price is the normal price of GNO, it's usually in the middle between the yes and no prices, I'm going to finish up in a few minutes, so usually the spot price is in between the yes and no prices, if it isn't, it means you can make a profit by doing an arbitrage, you can buy the yes and no prices, you can buy both the yes and no and merge them together and get the spot price for free, and so let's imagine here that you think it's getting very close and GNOS is not actually going to be able to make this milestone in the next couple of weeks, but you still think GNOS is doing great, so you don't actually think that the GNO price is going to fall in a couple of weeks if they get very close to the milestone, but you don't think it's going to fall, so what you can do, you can go here and say okay, the no price is undervalued, I'm going to buy this no token, so you're not actually betting on whether GNOS is going to achieve or not the milestone, you're betting on if they don't achieve it, I'm still going to buy GNOS, I don't think the GNOS price is going to fall, you can go here, if no, buy, and here you're going to buy a hundred using a hundred SDAI, and then there you go, you make a trade, you're going to have a position like this, and if the event doesn't happen, you're going to get 1.

09 GNO, you just purchased GNO at a discount, and if the event happens, you're going to get your money back, so here's how it works to make a trade on a conditional market, so this is the same kind of conditional market that you need to make a trade on a proposal, instead of a milestone, so price did not fall, you made a profit, that's how it works, you can potentially buy for a discount, or you can buy for a small premium, but then you're getting sure that you're only buying this world where the milestone is going to succeed, so I want to thank our partners, so we are using FlatOS as our resolution oracle, we are using the Sphere constructs as routers, and we are also using integration with CalSwap, and something really important, so right now it's a testing, it's still in beta mode, so please use with caution, the audits on some of the proxy constructs are still ongoing, so right now there's about $2,000 on liquidity on this market, it's good for small trades, for tryouts, eventually we're going to have like $200,000 liquidity on these milestones, and several million dollars on the proposals, and some people ask, well how are we going to have this much liquidity on proposals, and the answer, the key to that is CalSwap, so you can actually get a significant, we are going to be splitting up into the yes and the no pools, and you can do that because if you want to buy some normal VNO, we can just use CalSwap, and the CalSwap is going to buy them for you, and they're going to buy the yes and the no versions automatically and merge them together, so in practice you can provide liquidity to conditional markets to inform your decisions, and not reduce liquidity on your regular trades, so this is a core thing on how we're going to be able to get like a million dollars of liquidity on these markets. Yeah, so thank you very much, we really want you, so this is just the beginning, we're going to do much, much more, there's so much that we want to do, and this is just the beginning of the journey, so please join us, I want you guys to actually get your phones right now if you can, and I'm going to leave this open while we have questions, I don't know how much time we have for questions, I went a little bit over time, so this is just our Discord, we just created this this morning, so you're going to be the very first to join, and the people who are joining here, I'm going to give them, everybody's going to get a special role for OG, Gapcon, so please join, and we want you to contribute with milestones, what kinds of things would make you want to buy more GNO, what kinds of proposals you think would be good that would make you want to invest more in the GNOS ecosystem, and of course if you have your own DAOs, we would love to talk to you about how we can have this on your DAOs as well. So thank you very much, and thank you.