DAPPCON 2024: Homage to Augur - Edmund Edgar

Transcript

I'm going to be talking today about a dead prediction Market uh which I didn't build um so so I run reality. e which is a crowdsourced smart contract Oracle um I'm building a thing called backstop Layer Two which I will shill a little bit but I'm mainly going to talk about Orga so um the fundamental problem that augre is trying to solve which has some very important design implications is what we used to call at that time the Oracle problem right so the problem is how do you settle a market so for a it was um something like who won the 2020 US presidential election um but this is you

know so so for for them it's very easy to make kind of a smart contract system to do all the trading and all the stuff that you do in a prediction Market um but the fundamental of how you settle is this kind of kind of serious cryto cryptoeconomic problem um so for for them it was a prediction Market problem but it has applications Way Way Beyond prediction markets um so anything where you need data from outside of the chain right um anytime you need truth on the on the Chain um but also it's you have almost the same problem where you need kind of governance type of information so so I call it kind of Truth and Justice as the problem so if you're trying to upgrade a contract and you need to know is this

upgrade malicious then you've got just the same problem um so Igan L now are calling this problem inter subjective work um by which they mean that there's information that everybody in the world knows and people generally have a a common idea of what the answer is but the blockchain needs to know it and how does the blockchain know it okay so I want to go through a few things that don't work and these are things that are very widely PR practiced some sometimes for Oracle stuff particularly for governance um and I'm going to go through a couple of things that don't work and then kind of build up to the Oro system that kind of does work uh and then talk about maybe how we can we can improve on the on the limitations of it um so one thing you

can do is token voting right so if you had the problem of we've got a bunch of token holders uh they have tokens in a prediction Market obviously they have an incentive to make that prediction Market um settle directly right so when it comes to say who won the presidential election they have the uh incentive to say it's Biden um the problem with this is first of all you can do a very simple crude attack on it you buy 51% of the tokens um then you put a bunch of bets on on Trump right at the end of the market when everybody realized that that he'd lost um then you use that to miss settle and you can steal all the money in the bets right um but now that requires 51% which obviously is quite a high hurdle but you can also do much

more um um kind of subtle things um like briber attacks and bribery attacks in theory can be very very cheap um and they actually get worse in a sense the more decentralized you make them um because if you're doing token voting as an individual the chances that your vote is actually going to flip the the answer are not very high right so if you take the bribe and you vote to miss settle the market then it's likely that your vote will not actually swing either it won't the the thing won't go through that it won't get Miss settled and that will um not cost you anything so they'll get the bribe and nothing

will go wrong or the other voters are going to vote to do it anyway they're going to take the bribe and then that will mean that um even if you didn't take the bribe you still lose the value of the system right the system still breaks um so so so a decentralized system kind of is kind of even worse for from the point of view of token voting than a um than a centralized system if it's like three of us it's much easier to coordinate um so so then people might say okay let's add some incentives let's add token voting with slashing um the problem there is okay who are you going to slash right who are you going to reward who are you going to punish um if you slash the minority then you've kind of made the the attacks even worse because if the attacker can now just

make a commitment that they are going to break the system then the honest people are now incentivized to actually go with the attacker's dishonest um solution so for example you could you can make a a contract that commits that shows that you could pay the bribe um and say that if it goes through a b by sufficient margin you're not going to pay the bribe um so if you do that then you know you get to to to reward the dishonest and punish the honest so now we're going to get to to the solution which comes from ultimately from um Paul Schwartz's um paper truth coin which I I very recommend reading which is we back the the the slashing with a fork so what's going to happen is that if there is a

very if there is something that gets kind of escalated I'll talk about escalation in a second we're going to clone the entire prediction market and split the whole thing into two so there's going to be one prediction Market which thinks that Donald Trump won the election and there's be going to be another prediction Market which thinks that Joe Biden won the election and the great think about thing about this is that we as we we can probably assume that a prediction Market that settles incorrectly is worthless so the tokens on the version that says that that uh that Trump won aren't worth anything and the tokens on the version that says that Biden won are worth probably as much as the prediction Market was worth before the fork happened so the great thing about this is that the um the token holders skin is

fully in the game right because if if they say that Trump won they're going to end up on the Trump Fork um and there's no need for anything kind of objective to um for the blockchain to know what happened right people know what happened and people value the the better prediction Market um so vitalic calls this subjective ocy this mechanism um and it's kind of also what secures the base layers right so if if for example we have a censorship ATT density theum and 51% of stakers decide that they're going to sensor blocks they're going to reor out too cash trans transactions then what we'll do is we'll make another version of ethereum where they get slashed right socially slashed um and if

we do that then the people are just going to decide which version they want to use they can use the um the sensored version where their transactions aren't guaranteed to get through or they can use the onest version the uncensored version where their transactions won't get through so it's just this same mechanism that secures the base layers um gives us this very extreme level of security um for securing this application on on top of the of the base layer um a couple of problems with this firstly Forks are very socially expensive right so we can't just Fork the the system for every single market and we'll have like 10,000 different orgers um so what we do here is just used the forking as a back stop so we use an escalation game um reality e by the way is also an escalation game it's not the one that Ori used um but

it's the same mechanism so um people who uh pick one side are going to start by putting a little money bit of money in uh somebody might say I think Biden one and here's a little bit money I'm going to put in the system uh to to back up my claim somebody else can then change that answer by putting out more money in in reality if you have to double it in in in augre it was this kind of custom series of dispute rounds um but anyway so you get these ever larger rounds with ever larger commitments and only when you get to a very high amount of money are you going to actually Fork the system um so this worked really well in orot as a back stop um there was never a need to actually Fork the system it never actually forked um and it was very

good at correctly um settling these prediction markets um having said that okay okay but there's there's another problem which is that um the org users probably don't want to bet in Orga tokens right when we Fork the the prediction Market we're able to to Fork the Orga tokens but they don't want to bet in those they want to bet in ether and die and Orga can't Fork ether and die right so you need some kind of objective way for the blockchain to tell where should we put the ether and die so generally the forking if the Bas layer Forks that's a subjective decision people use whatever they want to use but we have to get some objective choice about which one is better so what the a guys did was they

said the forkable token is is the rep you have to move your forkable token into one fork or the other um the majority will probably enter the most valuable Fork so then we can look at which has the most tokens going into it and we can assume that's the truth um the problem is that now we've got an economic security bound it's it's no longer a purely subjective system um let's imagine that you've got um $100 million worth of bets on on on Trump and Biden um and the Orga token is worth $10 million obviously in that situation it's still worth somebody's while to buy up half the or tokens um or to do a bribery attack um so that they can make the ether and die go into the for that says that Trump won even though Biden won so

what they do in Aug is they say we're going to make sure we never hit that Economic Security bound so they're going to measure what they call the open interest so how much money is at stake that's continually measured you also measure the um governance token value and if the the open interest threatens to get half as high as the um the value of the governance token then we're going to do something about it and what we're going to do about it is either we're going is is we're going to raise the fees and by raising the fees we're going to either um reduce the amount of open interest because it will dissuade people from taking from betting in on the system um or it will maybe increase the value of the token because the token holders are getting this Revenue that

comes from the fees um so a couple of problems with this and and I should say these are basically theoretical problems because Orga didn't fail for these reasons um or Orga you know ultimately um kind of didn't succeed in the marketplace but um it wasn't because of these but if we're trying to use this mechanism more broadly which I think we should be because it's kind of the only me sound mechanism we have for governance uh then there will be problems so one is that these fees may actually be kind of Impractical right so if you if you've got to make sure there's enough Revenue coming in um people are incentivized to hold these tokens and that's got to be uh and the token value has got to stay half the the whole market cap um that's quite a lot of money you may have to be

paying in fees um and it may be that people just don't want to pay that many fees um like I mean I mean if you look at the probably a normal betting market like a traditional bookmaker will have a very small market cap compared to the amount of money they got in bets that's generally how Finance Works um the other problem is parasitism so somebody else can make a different Market that avoids the fees and you they do the trading on their own market and then at the end when it comes to time to settle the market they just have a little market on or that no nobody's betting on and they read that off the or contract so that way you can avoid the fees um this makes it even worse if you've had to raise raise the orer fees because potentially

you then kind of push orer into a a death spiral where the token becomes increasingly worthless because it's going to get parasited off and and like I said this is theoretical this did not happen the system works um but but we need to to still worry about it so that's basically what a did um I'm going to talk a little bit about Alternatives and shill my thing a little bit um one approach might want to take is what if we could make ethereum forkable right that would solve the problem so um we could you know we couldn't Fork ether and die uh and that gave us the this Economic Security bounds but if we could have forked e ether and die then it would have would have worked and we wouldn't have to have had to worry so Martin copelan suggested this a long time ago um when nois was

was still a prediction Market um platform and um uh but but then what happened after that was that we had like the eth ETC fork and that made people kind of ly of M of like Bas layer Forks um Justin Drake kind of suggested having like price feeds enshrined in ethereum consensus that would have had the same result that if there was a dispute about price feed etherum of the for um I can layer version one not the new thing I just brought out but I can layer version one if it had become too big to fail would plausibly have had this effect right so if people had attacked Igan layer with with all these staked eth um and igen L was very much used then it's quite possible that we've ended up

having having a a fork of ethereum to to figure it out and and this is why so many people got mad at IG lay when they tried to do this um Igan L version 2 is quite cool and interesting they have kind of a solution to the parasite problem in that what they say is they change kind of change the goal of the system so instead of trying to get to the right answer what they're going to do is they're going to allow people to ensure against being on a for with the wrong answer um and this means that if you haven't paid the insurance premium then you may still get screwed so this kind of may help with parasites um but it's potentially complex i' like to see how it works in practice and it doesn't solve the basic cost issue that um you need to be paying

enough in fees to the um to to kind of the Oracle to pay them not to steal your money um when when you're being governed by them so just just to say a little bit my my thing what we want to do is kind of more radical so if we could Fork ethereum that would solve the problem um but we can't Fork ethereum so we want to make a new chain that we can Fork have that work as a layer two on top of ethereum so we have the maximum compatibility and the maximum composability with ethereum and any application on this chain will be able after a sufficient escalation to cause our chain to Fork um and and just like Orga the forking is just a back stop we

don't expect it to happen all the time um it may well never happen but the but having this back stop allows you to make crypto economically secure um governance and crypto economically secure um oracles with respect to assets that are native to that chain um if you want to bridge stuff from the layer one that can't be forked by the system then you still have kind of an economic security bound so you still have that limitation anyhow so so that's what we're um we're doing about it um I've put a bunch of links on here there's also a new design for Orga that mikar just came up with that I've hardly had time to process um and I I've linked the Orga white paper there I really um would encourage everybody to read the Orga white paper

even if you're not interested in these mechanisms even if you're not interested in this problem just because it's such a good piece of work it's very clear in laying out you know what what are the problems that this solves what are the limitations what are the assumptions um okay and there's also some stuff about back stop and uh and reality eth and also links to um Paul schwarz's paper which I think is absolutely fundamental to to a lot of what we do on ethereum that he doesn't get enough credit for okay does anyone have any [Applause] questions yeah so first of all pardon me for not

getting all the technical details from your talk but uh I'm curious what happens if the final decision is not clearly obvious uh in some cases and there will remain people who believe in one decision and then others will believe in another decision and then maybe these forks will just grow and increase always yeah so that that's right so you get an economic Fork it's it's the same thing that happened when um for example when uh e ethereum and ethereum classic forked um Al also you know the uh some some of the Bitcoin cash Forks uh the same thing happened um I think generally what's going to happen is that it's going to end up fairly clearly on one side or the other um and

the reason that usually happens is because there's there is a network effect that pushes things right so the same reason if you flip a coin it hardly ever lands on its side or like standing up it normally Falls one one side one side or the other um so I think that's what's most likely to happen um but there is always the possibility that you'll have a have an economic Fork um and that has some consequences for like sort of defi stuff and things like that that it um the kind of weird things happen if your um if your token split exactly into um that that can do kind kind of weird things um this is also something that may happen to ethereum right um like um I I I know people often this hasn't happened for a while so people

are kind of Designing systems kind of with the assumption that ethereum won't Fork um but you you don't necessarily go into these things on purpose right and the only governance we have that governs ethereum ultimately is the ability for users to decide whether to accept an upgrade or upgrade or whether to reject an upgrade right we don't have democracy we don't have a governance process in ethereum Beyond you know some some uh researchers with various potential conflict of interests deciding among themselves what to do which may or may not represent the users um so these systems I think are only really secure if they can Fork um our system will have a greater likelihood of of forking and also a great greater likelihood of forking in a bad way in a

sense of getting one of those 50-50 Forks than the base layer and that's why people don't want to do this to the base layer um but but I but I think it's a price worth paying to get like you know solid governance uh solid oracles all right I think that's all with the questions thank you thank you so much for an amazing presentation