So hi, everyone. I'm gonna give you a presentation of Claros and a few of its application which are currently live. First, let let's start out to why we need something like Claros. So on a scenario or pretty much any other blockchain, you can have contracts which are complete at the smart contract level. So, for example, if you have Uniswap and you want to swap some ETH for some die, there is absolutely no subjectivity. It's just about executing some pieces of code, which only refer to what is already on the chain. So you can exchange this die for the ATH. You can make a contract which does that and does it without any human input. But you also have incomplete contracts. So for example, if you have something like an insurance, you cannot know on chain if the event the insurance is protecting you against has occurred or not. If you want to buy some good or service, the chain cannot know that this good or service was delivered or not. So for Urus contract, you will need a way to solve disputes about what happened outside the chain because the chain cannot access what is happening outside the chain, or can also be used when what is happening on this chain is too subjective to be written and expressed in code, and we need some human intervention to interpret what is happening and what should happen. We can start with a really simple contract. You can have Alice, which is gonna employ Bob to make a website. And for that, she's gonna put some money in a smart contract. Well, she ends if imagine if you imagine that she doesn't put someone in a smart contract, she paid directly Bob, and Bob doesn't make the website, well, she will not be able to get her money back. Or or otherwise, you get the other problem of Bob making the job, and then Alice is not satisfied, and he cannot get his money back. So for that, we can ask Alice to put his money in a smart contract, which would be an escrow smart contract. So now the t h, which will be paid to Bob, I'm not in control of Alice. I'm not in control of Bob. They are in a smart contract, which is autonomous. But, obviously, if there is a problem, you will need, to get an arbitrator. So in the scale of dispute between Alice and Bob about whether his website was made according to specification or not, it can create a dispute, and an arbitrator can rule to reimburse Alice or to pay Bob with the money in the contract. What is pretty interesting is that you can execute automatically the decision of this arbitrator. So you don't get, like, a non forcible ruling in the classical sense where you can have it enforced by your state by state authorities. But here, it's automatically enforced by some smart contracts. The limitation of this is, obviously, that you can only enforce ruling an asset which has been explicitly given in control to a smart contract. So you cannot, for example, make Alice pay more than what she puts in a smart contract in the first place. So now let's look how we could make this arbitrator and how we made it at Keros. So Keros is a staking system with a random drawing of Jiras. So you will have a bunch of Jirals which will stake p and k, and one p and k stake is one chance of being drawn in a case. So we have this case with Alice and Bob. A few p and k are drawn. And if it's one of your p and k which is drawn, you will be a juror in this case. That's made to be scalable because if you ask everyone to put on everything, that will cost way too much in term of arbitration fees. Or people will not carry enough if you put arbitration fee too low, and they will not spend time to review the disputes. In addition, we have to give incentive for jurors to to rule correctly. So that's inspired by a shilling coin systems, where in this case, if Bob loses a dispute, his arbitration fee, so we we have depoted arbitration fee, They will go for jurors who voted for Alice. So in this case, you have Daniel, Ezekiel, Gabriela, and Adas who voted for Alice. So they will split the ETH arbitration fee paid by Bob, and they will also get part of the P and K loss by Ignacy. So if you are not coherent with the result of a dispute as a juror, you will lose a bit of P and K. If you are coherent, you will win a bit of P and K, and you will get arbitration fees. So it's a system which is made to transfer in the long term P and K from Juror which are not really good or which are outright malicious to Juror which are honest and doing their work thoroughly and in a in a good way, which are making the system behave as intended, which are giving good good reading. Obviously, even if you are honest, sometimes you will not be current with the majority. It's perfectly fine, because amount of p n k that you will now lose, is pretty small compared to what what you stake. So you don't just lose everything when when you are not coherent once. It's a way more nuanced, redistribution mechanism. But To give you an idea, let's let's play this z shading game. You need to pick the same square as what would the majority pick. So you can you're gonna be able to do that in the chat. Like, you can sing for thirty seconds on which square you want to pick, and you want to pick the same square as a majority of people in this call will pick. So I'm gonna open the chat. And okay. Now you can give me the the square you will pick. And I get one I get I get 4 a 6 and one d 4 5 A6 and one d 4. Well, now we add on the result. Obviously, you know which one is the one tick by the majority. So maybe someone will be maybe we are the first one who picked a six. So could you give us a small explanation of why you picked a six? It's red. Yeah. So it's red, and the other square are not red. So if you want to find something which stands out compared to the other solution, you're gonna pick this red red square. And you see the majority of people pick the red square. We have someone who did not pick the red square. So maybe, Nadine, if you can tell us why you pick the d four. I knew everybody was gonna go basics, so I just wanted to be contrary and throw a wrench into the system. Okay. Okay. So you you would just want to lose in purpose. In this case, you will lose some token, but in most of the time, people don't want to lose and people want to win. So we we can see here that we have this, shading shading point, which is a six. And in case of Claro's ruling, the shading point is gonna be the honest ruling. So if you see, for example, that Bob, like, really spend half an hour to make a to make a website by just using already available components, and it's absolutely not what Alice had asked and absolutely not what they had contracted. What is gonna stand out is to reload for Alice because Alice is the right answer of who should get the money back.

So we have a parameter of support which show to make some private votes. We serve what we call a commit and reveal where you first commit to a vote and after, you can only review this particular vote. But it's not activated in most of the courts because of the user experience as it will require it requires to come back a second time, pay us a second time. But we expect in the in the future to be activated in more courts to make voting private, at least, like, private until the votes are revealed. And that will be one of the one of the improvement, for example, what we can do if we are on the second layer where the gas fee quite low or where some, like, Jurok will install some some small program which are gonna which will reveal in instead of them so that you don't have to come back to reveal. So for now, it's just about user experience because user experience would be worse if they had to reveal and use private voting. But in the in the other world, we would like to go to private voting. Also in the structural heroes, you have appeals. So you can see what people voted before, but you cannot see what they will vote in case of API. But we're gonna go in the future side to the API structure. So the basic idea is that people who try to gain the system on average lose money. So people who are like want to be contrarian and want to lose. Well, it's again, I lose money. And people who try to rule correctly and be honest, they will make money on average because it's a positive sum game. PNTR distribution is a zero sum game because what the PNPA, which is won by someone is lost by someone else. But the arbitration fees are some external source of income, so it makes it a positive sum game for euros. Now let's look at a few potential, like, problem you could have on such an arbitration system, and all the they are solved by Cleros. The first one, I think it's maybe, like, the first thing you should ask every time it comes up with with a blockchain or a governance system or or a dApp, which is how do you protect against the civil attack? So the spill attack is what if you make tons of fake accounts? Well, not really fake, like tons of accounts for one person. So in this case, if you have, let's say, like, Lawrence will make another account. Would it allow her to get more chance of being drawn? If we were to do just one person, one chance of being drawn, that will be the case. You could just make tons of accounts, and you could make one year account so that you get one inch of being drawn, and you can attack the system quite easily. So that's why in Keros, it's a staking system where one token is one chance of being drawn. So it's not one person, one chance. It's one token, one chance. So this way, for example, let's say, like, if as a kid is splitting all this P and K to 10 different accounts, each with one P and K, you would have the same chance of being drawn as if he puts all the P and K in the same account. So So it's a system which is which is a CB resident. Now let's look at how we we can solve the issue of, like, the sort of trade off between having a secure system and a costly system. So if you only take a small amount of Jira, well, your system is gonna be pretty cheap because you don't have that much people to pay. But it's gonna be pretty insecure because you also don't have much people to bribe. So we we would like to get a good compromise or like in the best of both worlds, if it was possible. So that is made with the Aptiv system where at first, you draw a small amount of URLs, But if the result you get is you believe the result is wrong, you can appeal, and it will every time it will double plus one the number of Jirals. So if you try to to bribe people or if you stake a lot of PAK in this particular call to get chance of being grilled and try to have the majority of the of the seat of the jurors, the defense against that is to make more appeals, to get more and more and more jurors. And, ultimately, if you want to win at this game, you will end up with a with a juror population, which will be statistically similar to the general population, of of Claro stakers. And that basically means that you will need to make a few 10% attack. You will need to get most of the p and k. So as you appeal, you get higher amount of zero, so you need to put higher deposit. And if you lose ultimately, you will lose your deposit. So you can can, like, store the system, store the process if you want to keep appealing, but it's gonna cost you more and more every time you appeal. And if you lose, you will ultimately lose this cost. Lose the deposit. Sorry. But if you win, you will actually get your own deposit back and also some part of the deposit of the of the losing party. Let's now look at why Keros needs a specific token. So for that, we could ask, okay, how would it work if we were to use ETH instead of a specific token, t and k? So how will this work if we are to use ETH instead? So if we use ETH and someone would like to take over Claros, they will need for that to get 51% of the ETH which is taking Keros. So that will be easier than getting 5010% of the PNK in Keros because ETH has a way larger market cap and liquidity. So if you want to buy, maybe, like, Kleros is $15,000,000 market cap, something like that. Let's say you want to you will have also 50,000,000 of market cap oh, sorry, of worth of ETH stake. You will need to buy 51% of you want to have 51,000,000 worth of ETH that will cost you 51,000,000. You will have a budget of 15,000,000 to get 15% attack with Clerox working with ETH. And buying 51,000,000 is probably gonna move the price of ETH a bit, but only a few percent. So not that much. Now if we have a specific token and you want to be 10% attack it, as you try to get more and more PNK, you're gonna impact the price. And, actually, you would probably not even find 310% of the token you need available on the market. So you should you would need to keep offering higher and higher price to try to get this 15% amount of the tokens, and you will not even be sure because 15% of the token are probably not between the hand of people who would like to to sell them. So that's a higher using a specific token leads to a higher budget required to do a 50% attack. Now we can look at what will happen if this attack is successful. So in this case, it's the cost of the attack. So the budget is the amount of capital you need, and the cost is the amount of capital you will lose by doing this attack. Now you by doing this attack. Now you get to attempt to attack a Clerox with ETH. And, okay, you control the system. And, okay, you you destroy Clerox, but it's fine because the price of ETH is not that much correlated with the working of of Keros. So you will still sit on a on a pile of valuable ETH even after breaking Keros. Now if you have a specific token, if you have some k and k and you 50% attack it while you broke the system, when people will stop using it, so they will stop paying arbitration fees. So there will not be any interest in to holding the P and K because the ultimate interest of P and K is to stake it to get drawn as a and to get paid arbitration fees. But now if you break it, people are gonna stop using it, so you will not have any value in PnK anymore. So in this case, if you have use some ETH, you have a cost, which is zero. Why if you use PMK, you will have a cost which is approximately 59% of the price of the PMK market cap. So using a specific token is also leading to a lower to, sorry, to higher cost of 50% attacks. And finally, what can we do in case of an attack? If you were to use ETH, well, the attackers would still be in control of Cleroses and there is not any way to recover from that. You can say, oh, we make a new Cleroses, but then the attacker could move his ETH into the new Cleroses. I know if you have a specific token, you can fork it, and you can say, okay. We make a copy of Cleroses, which will remove all the token of the attacker, and that's a new Clerox, which is only owned by honest people. So that's one of the other advantage of having a specific token, which is all to recover from some attack. Let's continue a bit on what is how can you see Keros in an ecosystem. So Keros is basically a brick. It's a dispute resolution brick. By itself, it doesn't really do anything. It's it's giving ruling. But what to do with a ruling is something to be decided by the contract which are arbitrable by Keros. So in the example, you got an escrow, and the escrow will send some money to Bob depending on the on the ruling. That's a really simple arbitrable contract, but you can make slightly more complex arbitrable contracts. And we tend to see that we have three main type of arbitrable contract, which are connected to Cleros. You have the escrows that we already mentioned, which even if it was a first MDJ and use case, tend to be the type which is user list. You have curated registries. When in a curated registry, you want to verify that some information is correct. So for that, you will put a bond when you certify the information, and someone can challenge you if they think the information you gave is incorrect. And if you lose, the information is removed from the registry, and you will lose your deposit. So it's not only transferring some value, some tokens. It can also be used to create information. And the other use case is Oracles. So for Oracles, you want to get a result about something outside of the chain. For example, who won the US election, which was an emblematic case recently. You make a prediction market, which is gonna pay out people who are who have chosen the right candidate, the candidate who won the election. But how can this prediction market know about who won the election? For that, you need an Oracle system. So in this case, you can get a dispute about who owns the election. And depending on the result of this oracle, you can perform arbitrary actions. So for example, adding some token reading for some ETH or anything or pretty much anything you want. You can that you can program in a acceptable amount of gas. So here you can see how how Claros is currently used in the current ecosystem. So you can see, for example, that we have some SQL application. We have some token registry, which is a list of token, which is one of the list that you can use for Uniswap. You have some other application looking at this list. You have, Real Issue, which is, a general purpose oracle, that can, select, Claro as its arbitrator, which also has here, Reality cards and home end, which are production market. You also have a SafeSnap, with Ignosi's tape Safe, which is also connected to our LCO. So it's really the brick. You you you can connect stuff to Claros. You can even connect stuff with to stuff which are connected to Claros. That's really how we envision the ecosystem. We provide a really specialized brick of dispute resolution that can be used in pretty much any kind of application which require either out of chain or subjective data to to function. Oh, sorry. So to give you an idea of what it it looks, you can see here escrow where we would make a marketing service agreement with John. We put some some TH in the escrow. And if the agreement you have people who are disagreeing, like the buyer and the seller disagree on whether this agreement has been completed, you can create some disputes. So you will say to the party did not comply with the agreement, and you will get a dispute and Carole Girard will decide if Alice or Bob should be paid in this express transaction. The Trusted Registry, which is a registry of tokens. As you can see, you get fake tokens, which was like quite a common scam, mainly on making a fake token that they put on Uniswap and giving some link to that to kinda like trick people into buying token which are fake. Now you're associated with NFTs. We recently had someone making some fake NFTs pretending to be made by Deepals, where it was obviously not DIPELS, but it was, like, kinda like showing how the system was broken. So it was not made in malicious manner. He he called them NFT theft. So it was obviously fast fake, but it was to show the vulnerability of some NFT platform, which basically list everything and do not curate token which are listed. And for that, we have this curated list of token. So the idea is you want to add a token to this list. Let's say you want to add PnK. You put in for this token. You put a deposit. And if no one complains, the token is gonna be added to the list, and you're gonna get your deposit back. Now let's say you put a token which does not comply to the list with the list criterion. Anyone can see it and can challenge you by putting some some auto deposit, and you will have Claro's process randomly during some p and k. Some Durer are gonna be able to vote, and depending of the Durer vote, so it again can either be added to the list or rejected. And in case if it's rejected, Bob will get Bob the challenger will get the deposit of Alice. So it's a way to to create information. You can see a few screenshot of the token registry. And we have also another version of this, which is great, where it allows to create pretty much anything, not just token. You can create your list of whatever you you wish. Here, what will happen if we get oh, sorry. Yeah. So it's it was used by some exchange, and now it's mainly used as one of the Uniswap list. So you can select this this Claro's list on and that will allow you to see all the token which are there that you can trade on Uniswap, and you get some pretty strong guarantees that they are the regular token and not not scams. Let's show you some example of cases we got with with Keros. So at this time, we had an exchange, which was HNex, which now, like, kinda like became diversified, but at the time, it was HNex, which was released to linked to Bitfinex, where they had made some list of, criterion which would be fulfilled by some token project, to be admissible on on the exchange. So you get Bioshin, which, apply to this list. But what some Jira found oh, sorry. What some people found is that they had fake profile. So you had a Scott Bingley, which was supposed to be their CTO, which presented to be working at the Blockchain Research Laboratory in Oxford in Oxford, which apparently did not really exist. So you had someone finding that, calling or sending some mail to ox Oxford. And Oxford administrative people told them that, yeah, they never heard about Scott Bingley and that the laboratory, they were referring to, does not exist. So that allowed to reject the token. And, obviously, people of Biogen had to put a deposit. So, basically, it's like if they got some sort of fine for making false in for disclosing false information about the African project. And it's a fine which is made in a self you know, sorry, in an obscene manner because they had to put the deposit first. So, like, Claro cannot find people which do not accept it, but you can have application where people are required to put deposit, and they can lose a deposit if they misbehave in this case by giving fake information about the team of their project. So you can see some other elements which were pointed out when someone found out that the batch and project was probably not a legit project. Let's see some example of Oracle use cases that we got. I think the most emblematic is Omen. So this is a good example of using Keras as a brick because you have reality.eth, which is an Oracle system. And on this reality.eth, you can choose any arbitrator, and, you know, Claros is one of the arbitrator that you can choose. And Oman is connected through relity.eth. Oman is a prediction market. So the basic idea is that you can make a market. For example, let's say, let's take the presidential election. You put one die, and you bet on whether Trump will win or lose election. So that gives you one token, which is, yes, Trump, which you're waiting for one die if Trump wins, and the one token, which is no Trump, which we're winning for one die if Trump loses. And if you think that Trump is gonna win, well, you sell the token no Trump. And if you think that Trump is gonna lose, you can sell the token yes Trump. So we had this market about the presidential election. And we got a pretty large dispute because the time of the resolution of the market was slightly before the new president will take office, but still after the vote and after the delegate vote. And so you had some people arguing that, well, Donald Trump did not lose yet the election, and some of the people arguing that, okay, like, people have voted, delegates have voted, and all the result and almost all the media are declaring Biden as winning the election. So it should resolve as Biden won the election. So Trump loses the election. So we had a dispute on that. The consensus ultimately was that Trump had already lost the election. So it means that people who had bought some token of no Trump will get some reward or people who bought some token of Yes Trump didn't get any reward for this token. And so they lost some money in this prediction market. And recently, the last lunch related to Keras is proof of humanity, which is also a created registry. But in this case, it's a created registry of human. If you remember at the beginning of the presentation, we spoke about the spill attack. So here is to make a system with one person, one vote or one person, one token, for example, where people can register, but can only register once. So they can register. They need to get someone to vouch for them to register, which will be removed from the registry if their registration is malicious. So in this case, you have not only losing a deposit, but you can also remove people from the registry if they behave in a malicious manner. And if some submission are incorrect or malicious, that can lead to a clear or dispute, which will determine whether or not the submission is correct, whether or not this person already registered, or whether or not this person, is basically non existing, like a defect, something like that. We we saw someone trying to make a defect, but you not even get a voucher. I did not even trade a dispute. And to give you an idea of why you need Claros if you want to get a list of human because otherwise, people can just register multiple times. And it's currently used by a project with Democracy Earth, which is a UBI, which which is a token which is given to everyone registered at the rate of one per hour. And that's why you need Clerals because otherwise it went. You would just get someone registering 1,000,000 profile to get this 1,000,000 per hour. To swear at which I would say state, we got 361 Jira, which will at least one case. We actually got more Jira, but some some Jira have really low stakes. So they never get run-in the in the case yet. We got 574 cases yet, and there have been more than $4,000,000 worth of value, which has transited by some contract, which were ultimately secured by Claro. So it's not a formula of dispute, but it's formula of value. And even if there is no dispute, Claroos is obviously very useful, because it's impact straight of a dispute, would make people, behave, in the first place. So thank you everyone for for listening. And now I'm gonna go to questions.

Thanks for for in in asking or for for inviting me to. So, yeah, first comment is just that in the in the case of stuff like Claros, there's tons of great examples where this isn't really an issue, but there there are there are the cases where the underlying ground truth is reasonably fundamental, like the examples you've shown. But the Kinsey and beauty contest is the this sort of thought experiment or analysis that that demonstrates that the economic incentives and the social incentives in a consensus based outcome are are out of alignment when there's private information or when people believe that other people believe something. I guess, for simplicity, the case that I outlined was if I have a minority opinion or at least I believe that I hold a minority opinion, economically, I still vote with the what I believe is the majority opinion in the presence of economic incentives based on the consensus of the vote versus the sort of yeah. The the it's the the breakage of the economic and the social equilibrium that match when the information is relatively public verifiable or available like in the Oxford case you gave. And so the question is really about, like, how you guys see this and how you deal with it and, like, you know, maybe even just, like, the characterizations of when to use a consensus based court versus not and what would you do in the cases when maybe you don't want a consensus based core?

Yeah. So exactly. Like, obviously, if I had to use Keras for something where it's not beta, it's not gonna work. Like, you probably don't want to use Keras to to make some good governance decision about should we do x or should we do y. Cleras is better suited for issue which are slightly subjective. If it's completely not subjective, well, you don't need care what you can do, you can just code it. If it's completely subjective, you are a governance system. In this case, you don't want to encourage group syncing. But for cases with low subjectivity, where you want to have a high productivity, you want to be able to predict what will be the outcome. So for example, miss Alice and Bob, what you want to predict is that if the website is not done, Alice will get reimbursed. If the website is done, Bob will be paid. So in this case, if you think most people are seeing that the website is done according to specification, you should vote for yes. Even if you may have some other reason, you can say, okay, maybe Bob has been unlucky, and I think you will be more fair. I I really will be fair. I I have this money. No. Because we want to have a predictable system. So we have some rule and that we want to follow the rule, to be some predictable outcomes. So Keras is really about solving disputes. It's really about, cases with, low subjectivity. So it doesn't mean zero subjectivity. It means low subjectivity, but it's not like. You're not gonna decide, for example, on, like, the tax rate, for example, with girls.

The jurors are fulfilling a a a functional information processing task that's not well suited to custom like, you wouldn't build a computer to evaluate the thing you want. Like, it's fast and easy for a handful of jurors to check the thing, determine the answer, and you render the answer. But the the what they're doing as humans is not really about the subjectivity. It's about the flexibility of their processing. They just see the thing and they go, okay. Yeah. Maybe they look up some stuff, but the mapping is more or less deterministic. It's just not appropriate for coding an oracle.

Sometimes you have some slight activity. Sometimes you have, dispute which are not really about the fact, but more about the interpretation of, or the question was asked, for example. So you sometimes have dispute which are more subjective, but it's all way about, what should your rule lead to, and never about what is the best thing to do. It's what will this rule lead you to have. And then for this kind of question, you actually want to be coherent with what you think or the people are gonna think because the point of of contracting is to give you some flexibility. So if you put a world in your contract and for you, you you this world means something different for you, but you know that for most of the people, it means x or it means y, but, you know, for most of the people, it means x. But you should work for x because you want a contract which has maximum predictability where the most amount of people are gonna be able to know what it means and what will happen if this or this event happen. So that you that want this image is I predictive, but you don't want the system which can change, like, you will get in two task programs. So it's not like a system to go down the dial. It's it's ready to to execute what should happen depending to a particular header code.

Right. So I guess then I'll close on saying that I think the critical thing is to help people understand what they're doing when they construct their incomplete contract effectively. Right? You're you're filling in you're supporting the in supporting incomplete contracts, which is necessary, but there's a spectrum. Because what you already started to describe in this slightly, you know, more or less subjective is a sort of continuum of subjectivity that eventually gets to the point where this consensus problem binds where you might really get answers that you don't want because people are saying what they think other people think instead of what they really think. So just being aware that that's the spectra. It's like knowing when to stop making it more subjective and then making the subjective choice about how much subjectivity to incorporate in your contract knowing full well that as you push it too far, you're you're exposing yourself to this issue.

Yeah. Yeah. Exactly. Like, I don't I I am meeting in those representation where we have this nice, I think I got it here, a nice graph of where Cleroes will will fit in the subjectivity scale. Yeah. This is this. Let me

people on the outside And that's it. Insult the Yeah.

Maybe just answering to to Thomas first. We haven't seen any, like, indication that you will get some sort of, like, advantage of being popular. You often see, well, like, popular project or individual, which gets into some case and loses. And, actually, you you got, like, some for example, you had the people of Great Views which were already pissed off or losing their case, but That's a simple application of the rule. If you apply the rule, they were not eligible to the registry where they were applying to. So we haven't seen we haven't seen that you describe, like, the sort of What we saw may maybe the other way is that you tend to have Jira, we tend to have a pretty much literal literal interpretation. Not Not completely literal, so they are they are worse. Like, they are not like literal. They are less literal than, but tend to have interpretation which are more literal than what you would see, like, maybe, like, in the state court systems. So I think the way that, like, they can organize, not something that was planned by us. It's really, like, some self emerging property is that they give a really high importance to the rule. So that's it doesn't they, like, leads to this kind of popularity contest. Sometimes you see people which are kinda trying to play on that. So, for example, getting, like, Internet celebrities to to to to to make their cases, but it doesn't it doesn't specifically work. Like, we got two people doing that, and one time they won, one time they lost. So there is not really any kind of, like, correlation on whether it's it's working or not. So I think getting two literate interpretation probably more problematic currently than getting, like, a popularity contest and if you're famous, you will win because that's not what we we've observed in practice.

Oh, yeah. On precedent. So that's actually, also something which was not specifically designed, but also the notion property of the system. So we did not design any precedent rules. If you look at Claro's policies, there is no rule which are giving any like, saying anything about precedence. But in practice, we saw that quite often people in their cases are gonna refer to similar cases and how they were resolved to argue that the case should be resolved in a similar manner. So you definitely see some sort of precedence emerging, but it's like an emerging property. It's not something which was designed. Like myself, I I'm I'm from France, so we we are in a more like a civil law system, not not command law, which where command law is more related to to precedent. But I guess it it did to some system a bit in between where you still have people using precedent even if the system was not explicitly made for that.

So anyone can create some contracts, which are gonna connect with Claros. So we have some arbitration standard to do so. So we have a standard standard way sorry. Standard set of function to set up in your smart contract if you want it to be arbitrable by Keros, and it's a completely open process. If If you want to get a specialized quote for it, you will need to ask Keras governance to add this quote. Frankly, they're all accepted to to add the new quotes. So it's also pretty pretty open. So, yeah, Keras is really a big that pretty much everyone can use with the exception of some specific set of cases. So you have, like, general policy, which for example, would prohibit stuff like that assassination markets and stuff like that. Like, stuff that we believe are highly immoral. Claro has the option to rule, like, rule invalid. They say they do not want to rule this sort of cases. But beside that, there are 90% chance that we can connect your application to Perot.

Yeah. So you have multiple courts. You have a, like, a court harbor essence, where you can get into more specialized court. So for example, you will get, like, a blockchain technical court, which would have which will rule about some translation of plain English text into some smart contract code or smart contract code. So this one, you don't want random people to be there. So this is a specialized court. Everyone can always stay in any court, but, obviously, if you stay in a court you are not knowledgeable at, well, you will end up either spending a lot of time to give your rulings or being incoherent. So you will either lose time or money by being in a court you are not qualified to. So you have this incentive to only stake in courts that you are qualified even if there is no technical limitation in on which court you can stay.

Yeah. So the cooperative, it's a structure which is different than Keros. So so Keros itself is a protocol in a DAO, and you have the Keros cooperative, which is an entity which had developed most of the Keros protocol, which is a SCIC, which associate collective well, it's like a company of a, like, a French law, which include those producer and, users in the governance. So this means that if you are using Polaris, as a juror, as someone making application connected to Polaris, or, as a end user, so someone using an application connected to Claros, you can register in in this kick and get some voting right on this particular entity. But this entity is like a classic quality. It's like a I will say, like, it's a like, it's a which is linked to to state, in this case, registered in France. And this entity is developing the Keras protocol. And at the beginning, at some sort of, like, ending the governance of the Clarias protocol before the governance was ready. I know that the governance is ready. The governance is completely decentralized, so the cooperative doesn't have any specific right in the Clarias protocol and also has to ask the Claroos DAO to get some t and k to to fund its activities. And in theory, you could have any other entity which could also ask the Claroos DAO to get some grant to fund its activities related to Claro's. And why why cooperative? Because, obviously, you want to get different stakeholder represented. You want people working on Claro's itself because they are the one who knows best, like, the system. And you have users, which are very important because they are the key stakeholder in the system. So it's made to have a bit of everyone represented.

So I'll just jump in really quickly. So by the calculation, by the numbers you gave me, the average payout per drawer is about $450. So that seems relatively low. Can you speak more to what are the incentives for jurors to participate in Claros? And are there incentives other than the financial?

Well, you can also have people who, like, find it interesting to to judge on cases. You actually saw some other project of crowdsourcing dispute resolution, which we are not paying the Jira and just relying on on people liking to to be Jira's. But this project tend to be, like, not not enforcing. So they tend to just give some sort of advice, just project in, like, before blockchain. The real difference here is that, well, all all the all the result of dispute are enforced, so it's quite important to to pay people. You can see that most of the cases are probably not that complicated, so it doesn't require to to pay people a really high amount. So you have the arbitration fees, and you also have token redistribution. So if you are working currently, you can do some token. If you are working currently, you will win some token. But those token redistribution are zero sum game. So on average, you will win arbitration fees. But because most of the cases would tend to be, like, way smaller than what you will see, like, for example, in classic arbitration courts, like, I mean, in the in the in the legacy space, that's why the costs are not that high. But for some really particular cases, in particular cases which require people to to be able to encode in Solidity, you will see arbitration fees which are higher. Yeah.

much, sir. Cheers. Thanks, Heidi.

Thank you.

Thanks. Thank you.

Bye, everyone.

Bye.