William George, a mathematician working on mechanism design for the Kleros decentralized dispute resolution protocol, explains how the absence of robust identity tools historically forced protocols like Kleros v1 into purely token-weighted, financialized juror-selection mechanisms vulnerable to Sybil attacks via address splitting. He shows how combining imperfect proof-of-personhood systems (like Proof of Humanity) with financial staking creates a 'Swiss cheese' security model where both layers must be broken simultaneously, and extends this to soulbound tokens (from Weyl/Ohlhaver/Buterin) for credential-filtered, diversity-weighted juror panels inspired by quadratic funding. In Q&A, Venkatesh Rao pushes back that such mechanisms impose an esoteric 'homo economicus' framing that clashes with how real communities handle trust and cheater-detection socially.
Shared by the speakers for the symposium. Email addresses have been removed.
Transcript
YouTube’s automatic captions, unedited, so names and technical terms are often misheard. Each timestamp opens the video at that moment.
0:00 Hey everybody. So, my name is William George. My background is a mathematician. Uh so, I will try to put my slides in in screen sharing mode. Okay, so I should have slides. If you don’t see them, let me know.
0:16 Uh so, my day job these days is I work thinking about the Kleros protocol, which is a decentralized blockchain-based dispute resolution protocol. I do like mechanism design for that. And uh I’m going to specifically talk today about like some of our learnings around the role of identity that have come out of years of research for this specific dispute resolution protocol, but I feel have like very general themes that apply to a lot of decentralized web3 protocols, decentralized protocols in general. So, I’ll try to like make my sort of like the the case study running through will be the this dispute resolution use case. But, I hope that the themes will be broader and like, you know, will be applicable to a lot of people, I guess. Okay. So, just to like briefly recall how Kleros the Kleros protocol or dispute resolution protocol works for the so the for the sake of the running use case. Imagine you have some dispute between Imagine you have some kind of commercial relationship between two entities. Alice, a small business owner, hires Bob, a freelancer.
1:20 They create some kind of business contract and an interface that looks like this. Alice puts the money to pay Bob in a smart contract escrow. And if like Bob does his job correctly, then Alice clicks a button, money’s released. And if there’s a dispute over whether Alice did a good job, then the Kleros protocol, so this particular decentralized dispute resolution protocol, uses crowdsourcing. So, a bunch of like people from the crowd are chosen and they vote on your case. And depending on whether they vote for Alice or for Bob, the money in the The goes back to Alice or is used to pay pay the bond. Uh and um so then there’s a question of like who the people are uh in the you know what what what members of my crowd are chosen uh which gets very deeply to identity. Uh so in the Kleros 1.0 version which was launched in early 2019, uh it’s a very financialized mechanism. Uh people have a Kleros token that they stake uh and then their odds of being drawn are directly proportional to the tokens they staked.
2:27 Uh and um the um like there’s some nuances like you can stake into different courts that would handle different kinds of cases. Uh like maybe you have cases on website design and you’re comfortable judging those and you’re not comfortable judging the like smart contract bug cases. So you’ll self-select into one or the other. Uh and um your odds of being drawn for a case that you’ve self-selected into is the the equal to the percentage of tokens you have among the people who have staked into that kind of case. Uh but once you’re you know within the pool of people that want to do a specific kind of case, uh then you have X% of tokens, you have X% of being drawn. Uh and particularly you have an X% chance of being drawn for each juror spot. Uh so imagine you draw a panel of three or five jurors uh and you have a distribution that looks like this uh where you know like this like attacker devil character has 45% of the tokens. Uh the attacker has a 45% chance of being drawn in the first slot and if he gets drawn uh on the first spot, he has a 45% chance of being drawn in the second spot. Uh so maybe somebody else is drawn in the second spot. Uh on the third spot, the attacker’s odds still 45% uh and you can calculate what the total chance of the attacker being drawn more than half of the spots is um using you know this this distribution. This is a calculable thing. Turns out with with this to be 42 and 1/2% chance of getting more than half of the spots out of three.
3:52 Uh and you can ask, okay, like what Okay, one, that seems very financial and I’ll I’ll like get back to why it’s so financial. Uh and two, like, okay, is is it at least secure? Uh and the security model of of a model like that uh is that um no participant or colluding participant should be able to have more than 50% of the tokens staked for a category of use cases. Uh as long as you’re under 50%, as long as the attacker double character is less than 50%, if you pick a big enough panel of people, uh there’s like an appeal process, you pick bigger and bigger panels, uh then by the law of large numbers, uh you know, as the panel gets larger and larger, the attacker’s chances of getting more than half of the spots, if they actually have less than half of the stake, go to zero. Whereas, if the attacker has more than half of the stake, then, you know, like, the bigger the panel you pick, the more likely it is they’ll have more than half of the spots. Uh and um Okay, um why like sort of you one can ask lots of reasonable questions about like why one would choose to have a mechanism like that. Uh if I’ve picked somebody, why should I be willing to pick them again? Uh I I’ve already drawn, you know, the attacker double character.
5:01 Why can’t I just like remove his address from the pool uh and just not draw him again? Like, I draw on a a blockchain address, uh I’m just going to like not draw the same address again. Uh and um this gets to questions about identity uh and the kinds of identity tools that are available in these centralized environments. Uh particularly in 2019, there were very few. Uh and even today, there are there are more, but they’re like imperfect in various ways that I’ll that I’ll talk about. Uh that um like fundamentally, there’s like if you had like it it’s trivial for an attacker to generate a bunch of blockchain addresses. Uh this is like native identity tool that like blockchain’s historically used. It’s like addresses associated with a key uh as your kind of base identity marker, is trivial to generate as many of those as you want. You can just generate thousands of keys. So, if you have a mechanism where you remove somebody once they’ve been drawn, remove the address, anyone who’s participating honestly, you know, maybe they’ll keep all of their tokens on a single address because they’re they’re interacting with the protocol in like an honest way.
6:11 But, you know, the attacker can just split their tokens over you know, all addresses. So, that like if you draw them once, the the attacker is still there. They they can be drawn as many times as their their tokens are allowed to be drawn. So, this limitation of of blockchain identity for, you know, source first like specific kinds of mechanisms on you. Earlier today, Nathan Schneider had sort of interesting talk about like the virtues that protocols and different different systems like push upon their users. And he made the comment that web 3 protocols like often push like a sort of homo economicus, everybody kind of follows their incentives like as an ideology on their users and like users kind of interacting with these protocols start to adopt the ideology. And I would agree with that. And that the you know, that like those kinds of values often come from the designers of protocols. A lot of people who build web 3 applications like believe stuff like that. They like human beings they think of human beings as very homo economicus. However, like there’s like this extra layer to this here in that like already if you have a very homo economicus application that pushes financialization on you.
7:19 But, in order to have protocols where like users are seeing who’s as human beings and not like piles of tokens, you need to have identity mechanisms that can distinguish human beings and piles of tokens.
7:36 So, like what exists today? Like what kind of identity mechanisms do we have now? So, I’ll start with one mechanism that the Kleros Cooperative itself developed as a means partially to like resolve this problem for ourselves. We developed this several years ago called Proof of Humanity. This is a proof of personhood tool. It’s kind of a proto It’s like a very simple kind of social network almost. It’s a civil resistance tool. You can think of it as an alternative to WorldCoin if you know them the thing with like you you scan your iris in an orb and they give you cryptocurrency. Uh, and then like for both of these mechanisms both Proof of Humanity and WorldCoin you’re only supposed to do be able to do the thing once. This is a like a list Proof of Humanity is like a list of users with the property that you should only be able to get like one profile. You submit like this little video and photo of yourself and you shouldn’t be able to submit multiple profiles. Uh, and the way that it works is that there’s this list of existing users and when I want to submit myself to the list I take my my little video and I place a deposit in cryptocurrency when I’m submitting my profile and there’s a wait period where observers can challenge my deposit. They challenge my profile say okay like there’s this list of of criteria that the profile is supposed to conform to.
8:57 Most importantly not supposed to be a duplicate of something that’s already on the list. There are also some other like formatting criteria. And if you fail those criteria you can be challenged. This creates a dispute implicit dispute between the submitter and the challenger about whether the protocol the profile actually satisfies the criteria whether it’s actually duplicate. Uh, Kleros the the cooperative built Kleros cooperative built this so of course we use Kleros to resolve these disputes but you could use other systems to decide whether like the profile was was legitimate or not. Uh, and then ultimately you made a decision either that your profile is acceptable or not and if the profile is rejected then like the challenger gets some reward withdrawn out of the deposit of the submitter. Uh and uh you know, once you have a secure proof of personhood, like either with a proof of humanity or something else, uh it you have like a totally different mechanism design space for like what kinds of mechanisms you can securely have in a decentralized environment. Uh so imagine that I, you know, I have my dist- uh my distribution of tokens from before, uh and I’m going to for reasons that will become more clear in a moment, uh I will say that I like draw people with uh with weight that depends on their token stake. So I’ll, you know, in the first round, uh if the attacker if this like a devil character has 45% of the stake, I’ll draw him with 45% chance. Uh but if he’s once he’s chosen, like say that he needs to have this all this all the stake on a an address that’s registered on proof of humanity, so he shouldn’t be able to get more than one such address, uh and I can have confidence that like all the attacker’s tokens were on that address. Uh so once he’s chosen, I’m just going to remove the address, and the attacker’s gone.
10:37 Uh and uh then like I choose other people at once every time I remove somebody, like they’re just removed from the the pool, and I necessarily get a panel of three different people or however many different people, uh and the attacker’s overall percent chance of getting a half of the stake or half of the spots, half of the like votes on a on a case is zero because he can’t get more than one. Uh so that that’s great. Uh like you got that’s very secure, uh as long as the proof of personhood system is secure. Uh and um I say this with humility, we we built proof of humanity ourselves, um like you know, like it can it’s it’s difficult to get more than one profile on proof of humanity, uh but if you’re like a really sophisticated attacker that has lots of resources, um these reputation system, these identity mechanisms we have, you know, are not like ironclad. There are there is some there there there is some space to be attacked, uh and we can talk talk about like how resistant and like what the attack models are on the proof of personhood models. But for example, with proof of humanity, we’ve noticed cases of people going around in like train stations and offering to pay people. Here, I’ll give you $20 and like if you record this little video of me of yourself and I’ll put it in, you know, put it on this like blockchain thing. And people were happy to take the $20, at least, you know, some people, in exchange for having their, you know, like giving up this video of themselves that sure whatever, you know, you know, it’s it’s not a very sensitive video. It’s basically they took a photo of a stranger.
12:06 So now the attacker controls this address that theoretically belongs to a distinct human being, but in practice the attacker can accumulate a bunch of addresses. But that like costs time and effort and they might get caught. We, you know, some people like that were noticed and removed. So um, with resources like an attacker can attack the identity mechanism, but it adds friction. It adds robustness. And so one can ask, what happens if I have a proof of like a identity mechanism proof of personhood system that is insecure and then I couple it with like a token weighted staking. So I’m going to draw people with the percentage of the percentage of tokens they have, again, and every time I draw someone I’m going to remove them. You know, they have they have to be on proof of humanity and like once their their proof of humanity address is drawn, it gets removed. And the attacker has a bunch of addresses because they’re really sophisticated, but nobody else does. So attacker’s drawn, his 45% stays in the pool because it’s split up over all these other different addresses from people he brought in train stations. And then I draw the second person, like Alice, and she’s removed because she only had her one address that, you know, she was participating legitimately.
13:19 And that means everybody else’s odds go up, including the attacker. So now the attacker has 56% chances of being drawn on the on the third draw. Uh and you can use this distribution you can calculate the attacker’s overall chance of getting a hot more than half of the spots. Uh it turns out to be around 52%. Uh and what I find interesting about this uh is that um like before when I had just had the purely token weighted model, uh the attacker had like a 47% chance of being drawn or something like something in the 40s. Uh and here he has 52, but it’s not 100.
13:53 Even though he completely broke the identity mechanism, he like only got a slight additional edge uh in terms of his chance of breaking the system. Uh and you end up with a security model uh where you kind of have an either-or Swiss cheese security model. Uh you know, like the layers of Swiss cheese, the holes aren’t all in the same place. That as long as the proof of personhood system is secure, you’re good. And even if the proof of personhood system is not secure, you’re still good as long as no one has like almost half of the stake. Uh so you wind up like the the security advantages of having an identity system and like a financialized staking system are complementary. Uh they layer on top of each other. Uh and as long as you can’t break both, uh then your system is secure. Uh so um that raises interesting questions about what other more rich identity we could have. Um but like so far I was just talking about these like kind of civil resistant tools uh where it’s like, okay, yeah, you’re you two human beings are distinct. Uh but human beings are are much much more than that.
14:58 Uh you can ask, okay, like what if I wanted like make claims about like you know, I belong to some kind of community or I have some kind of expertise. Um you know, like specifically in the context of dispute resolution, maybe some community disputes wants their disputes to be resolved among them. Like if I have disputes among Argentinian neighborhood community disputes or something, maybe they want like in order to be like to resolve of the you know, I want some kind of credential saying that I’m Argentinian. Uh Uh or like you have certain categories of disputes require expertise saying I I know smart contract development. Uh and um so like this richer form of identity, there’s been a lot of work, interesting work over the past few years about like how you can make decentralized applications cognizant of of stuff like that. Uh there’s this great paper from like 2022 uh by Glen Weyl, Puja Ohlhaver, and Vitalik Buterin, uh where they introduce kind of what what they call soulbound tokens. Uh so these are these like non-transferable NFTs. Uh and you know, it like they call it a token because it like, you know, you you can apply sort of like a a token standard to it. It will apply in blockchain It will You’ll see it in blockchain explorers and such. Uh but ultimately it’s just like a credential. You know, it’s non-transferable, so it’s intrinsically non-financializable.
16:11 Uh and um it, you know, it’s just like some credential that you as as an individual might have. And you can accumulate them at a given address. Uh you could might get such things in different ways. Uh like, you know, if you want a credential saying that I finished university, you can have your university issue you that. Ideally, you know, in like a perfect world, if your university’s like really technophilic and and uh and agile, maybe they have some kind of private key public key pair, uh and they issue you kind of a digital diploma where it just like appears in your SBT wallet saying, “Yeah, you you finished your program of study.” Uh some universities are maybe more willing to have them than others. Uh but uh you could have just like a community of people that um you know, people already have the credential and they say, “Yeah, yeah, you’re you’re one of us.” And then they vote and they they take you to they they allow you to join. Uh you could have stuff that’s similar to the kind of proof of humanity system, uh where you could self-declare that you’re an expert in something. Uh like say, “Okay, I’m I’m an expert in Solidity Solidity development, so here’s my portfolio of my past work uh that I’m making available to see as for public review.
17:13 Uh I’m going to place a deposit. Uh and then some waiting period goes through and if someone submits incorrectly and they they don’t really satisfy the criteria to be to have this kind of expertise credential, uh, then some in some third party can challenge them. Uh, so then you have like a committee community review where ultimately like conflicts of whether one belongs among the experts or not can be resolved by a jury, uh, like a Kleros jury or some other jury, uh, maybe that of people that already have that credential. Okay. Uh, and there are lots of other people working in this kind of burgeoning like blockchain identity space. So just to give a shout out to some of those, uh, so I think this this recent project Ethos is trying to develop blockchain, uh, this kind of decentralized credit scores based on past user history of a given address. Uh, so that that’s really cool. Like, you know, it’s slightly different model that can have slightly different security properties and different, you know, be more or capable more or less capable of certain things, you know, used for different different requirements. Uh, but that that’s another thing in this space and like related to that, uh, is people are starting to ask a lot of questions about agentic identity.
18:20 Uh, there are these uh, there’s this like ERC-8004, uh, which is a sort of a standardization proposal about like giving reputation and identity to agents. Uh, again sort of based on their past usage. Uh, so that, you know, maybe when an agent is born, I don’t know a lot about it, but as as it develops a bunch of uh, feedback over time, you you get like this like full picture of the agent’s identity. Um, Okay. Uh, so like once you have identity tools like that, uh, I’ll just talk briefly about, um, how you can use this sort of more richer credential-based identity uh, in a system like Kleros and then I hope that that will like spur some, you know, be able to take some like broader lessons from that. Uh, so you know, now that I have the all these credentials saying, “Okay, you belong to this community. You have this expertise.” Uh, you can just filter jurors based on having that credential. Uh, in order to be drawn for a juror to to be you know a Claros juror at all, uh you you need to have the appropriate credentials. Uh and uh then like once you’ve passed the filtering stage, okay, yeah, you you sure you you belong to the community, you have the community credential, you can be drawn, then you use the sort of token weighted staking again. Uh these people can do that.
19:32 Uh and if you do that model, token weighted staking, but only after you pass a series of filters, then you have the same security model as what you had before uh with the proof of personhood thing, uh where um it’s basically an and or uh you are as strong as your weakest SBT uh or the financial stake. Uh as long Even if someone breaks all of the SBTs you need, uh you know, okay, so I needed like the SBT saying I was like I belonged to um you know, Mendoza like the community of like whatever in Argentina, uh and I also need the SBT saying that I like I’m an expert in banking or whatever. So, I have all of those. Uh and uh but I also have some stake. Uh you know, in order for the system to break, someone like some attacker needs to be able to like maliciously generate all of the relevant SBTs. Uh and uh even then they still need a bunch of stake uh in order to be able to break the system. Uh and um what becomes more interesting is if you don’t just say okay, like you need I have n SBTs and and credentials, they’re all relevant, you need all of them. Uh what if I have just like an ecosystem of people that have different credentials, different identities? Uh and I wanted It’s not realistic to expect everybody to have every credential, uh but I want to pick people that have like a diverse set of skills, a diverse set of credentials.
20:48 Uh and uh I think that like based on social science research, uh in some cases diverse panels have better decision they outperform uh in their decision-making ability. Um homogeneous panels, even if the people on the homogeneous panels were really high performing, uh so like maybe diversity is is like positive for the sake of making better decisions in of itself. Uh and I can measure that diversity based on the credentials people have, uh where I’m like, “Okay, I want you know, I want people that would like a variety of credentials.” And what are the best ways to to do that, uh to like pick people like that, uh in in a decentralized application? Uh and just to sort of wrap up, uh I think there’s an interesting parallel between Carol’s juror drawing uh and some other proposals around SBT, filtering SBT weighting, uh particularly in the um original paper that um Glen Weyl, uh Oliver, and and Buterin wrote. Uh they uh like one of their examples, I think it’s like in the appendix at the end, is weighting of quadratic funding contributions. So, if you’re not familiar with quadratic funding, it’s just like crowd-funding mechanism where like some kind of entity, maybe a government, maybe a grant agency, uh gives grants uh based on how many people how much people from the crowd fund. Uh so, if I have some participants that each submit like these contribute like C1 {dot} {dot} {dot} to CK uh to uh the project P, uh you know, like I have you know, I’m doing some kind of crowd-funding for a pro- for some project I really want to see succeed and I give it I give it some subs- I give it some funding, uh then this like grant matcher agency uh will give it a match like a grant based on the formula at the bottom uh with these like square roots, you know, you have a square root of the contributions sum to squared minus the contributions. Um And the the point of that uh is that basically you get um more grants if you have a broad set of support for your project. Uh if you have a whole bunch of people that fund your project, uh you’ll get a good grant. And if you just have like one person, like here I have two scenarios where like there were seven units of total uh crowd-funding, uh but on the right the attacker like just gave seven things seven units himself.
23:06 And the formula can pick up on that and it cancels out and like the attacker doesn’t get any any matching grants. Um Uh and then like like all like Glen Weyl and his collaborators have interesting ideas around waiting funding like contributions like that by SBTs. Uh where basically, you know, you want like projects that have broad community support maybe cutting across communities that not only compose of different people but like look different. Like they have like you know, different backgrounds, different priorities. Uh so basically you like put users into a group space on SBTs. Uh and then you um they kind of like go into the same square root together. There’s this like square root in quadratic funding that tapers off people’s contributions if they give too much if like one person gives too much. And now like the idea is like, “Okay, you know, you have the same SBTs as other person. I’m just going to put you together. You kind of look like the same person now.” Like, “Yeah, you can you can give more contributions to a project uh and that project will get more money but like, you know, it wouldn’t get as much more more money from the grants as if, you know, someone who looks less like the people who are already funding it gave it gave it a contribution.
24:12 Uh so this is the formula that looks like that. I have some examples. Um I can share these slides with people that want to look through them in in detail. I’ll skip over most of this as I want to have time some time for questions. Uh and then the kind of like the thing that I’ve thought about at Claros is doing like a similar thing where you group people based on SBTs uh and then you have some kind of like weighted total stake uh where you um you know, you do this kind of square root thing to put together like functions of people’s stake. Uh and then the chance of being drawn depends on their stake but also on like how similar they look to um via their SBTs via their credentials to other people in the juror pool. So that you try to like give extra weight to people with rare skills. Um And then just finally like there are lots of interesting research questions about like the security of this, how like the taco resistant, like exactly what the best way to do this is. If you think of like, okay, I want these like like groups of people. I want to like cluster basically using a clustering algorithm and like different people in my community. Like if the attacker can break some of the SBTs, can they give like fake SBTs to like other users in which case like make them look more similar than they really are.
25:24 Can the attacker make themselves look less dissimilar than they really are?
25:29 And this gets to like questions around data poisoning of machine learning like clustering algorithms that I think are you know, there’s lots of interesting technical research here if there’s anybody interested in that. So just to recap, historically the lack of identity tools forced decentralized applications to be really financialized. Some of the identity tools are getting better. So there’s a new space of stuff you can do, but there’s a lot of interesting research questions. So I would be happy to take questions if people have them.
26:02 Yeah, people have questions. Go ahead and raise your hand. Yeah, I think this was kind of interesting. Um I I feel like it’s nice as always to see like the safety concepts like the Swiss cheese model make their way to like um security and like there’s kind of an intuitive way for people to think about it even if the the technical stuff is like really complicated. Like math too. Venkat, you got a question?
26:33 Yeah. Thanks William. This this topic keeps coming back around. We had a project a couple of years ago on um plurality [clears throat] voting. We we kind of used a quadratic mechanism to do some follow-up grants for uh one of our summer programs. Uh and I think after 3 4 years of circling this kind of thing in various forms, um I I am beginning to take the blockchain sector’s homo economicus uh uh orientation criticism I I should say more seriously, which is um people don’t take kindly to being put in these I don’t know somewhat esoteric uh you know, mechanism design uh systems. Uh and I think the crypto sector’s starting to recognize that and they think of it as a UX problem, but I I think there’s more to it than that.
27:40 Uh I think yes, you have to make the UX much easier. You have to like make it more intuitive to like, you know, think about what is quadratic, what is a civil attack. So, all those things somehow need to be more made more intuitive, but I think there’s also a missing ingredient of um ultimately, this is a relational communal kind of technology and it feels like there’s a whole missing side. So, this is why I’m calling it part of the homo economicus critique. It’s uh uh Like, how do you how do you put this stuff in the background and make it invisible, where we call like, you know, um making it a Whitehead protocol as in it becomes an invisible system and foreground the social relational aspect um again. And to make it concrete, to use your example, you had very conceptual diagrams of like, you know, the devil character and how you remove that character if uh that person is using um his stake and moving the stake around and stuff. And as an engineer, I love the description of that and I like where that’s going, but when I think about how uh in real human communities cheater detection and like uh, you know, is there an untrustworthy person who’s come into town and what do we do about them?
28:53 It seems to work very differently. So, there’s a gap there. Like, how do you like get to something that resembles kind of the native human way of dealing with things like uh, in our village uh, grifter has come along and has tried to cheat everybody. Like, if you watch The Simpsons monorail episode, that might be a good example, right? You’ve got this grifter coming to town selling everybody on the monorail and like getting them hyped up and it turns into a disaster of a project and communities have immune responses against that. Can you speak to the gap between this homo economicus way economicus ways of thinking about this stuff versus, you know, like how real human communities deal with it and can we bridge that gap? Does it need to be bridged?
29:38 Yeah, I feel a lot of that is downstream from the like core identity protocols and it’s like identity protocols continue to improve like some of this gap will be bridged. Um, like, you know, like come people, you know, the sort of the perspective of the maybe the like sort of perspective of crypto people thinking of this is like UX problem, you know, users like normal people come to these protocols, you know, unless they’re like crypto native people expecting them kind of to work like web two protocols. You know, like you you expect, okay, I have my crypto marketplace or something, you know, and your default, if you’re not familiar with the norms of crypto communities, is it uh, it should sort of work like eBay. Uh, and then when it like doesn’t work like eBay uh, because like the protocol had to make design choices that were based on like security models that were used in the web three environments where there’s no centralized company to intervene if something’s going wrong and be like, ah, yeah, that’s a new attacker. I’m just going to like ad hoc do something about that. Um, you know, they had the protocol had to be designed differently uh, and it winds up being designed in ways that feel unintuitive to people. Uh, so uh, as you get better and better identity mechanisms uh, and also, you know, you just do research to develop other kinds of mechanisms to handle, you know,
30:46 like doing fun things without trusted entities. Uh like ideally you would get back to something that feels more like a web two experience. Uh so I’d say that that’s the the biggest thing. I mean, like ultimately it’s there are limitations that there you know, like people interact differently online than they have to in person, you know, like the way you you look at someone in the eye, you know, and like, you know, your like reactions and stuff. The trust that you have with your like villagers or something is difficult to reproduce. Web three environment is also difficult to reproduce in an internet web two environment. Uh so I would say that as you um like as tools develop, at least the gap between web two and three and web three should should be able to be bridged.
31:27 Yeah, good answer. Um thank you, William. That was a a great presentation. And if people want to talk more about um security identity tools, there are lots of identity nerds um in and around this community. So like pinging each other on Discord, continuing in the Zoom chat. And uh next up we have Krishna Kumar.
Simone Cicero of Boundaryless presents O2A (Open Organisation Alliance), a proposed open standard for making organizations 'composable' by replacing implicit coordination mechanisms like hierarchy and meetings with an explicit, machine-readable contract graph of units, offerings, and relationships. He argues the protocol's compact grammar—three economic primitives (purchase, revenue split, investment) plus composition mechanisms—can represent bespoke institutional arrangements like Haier's Rendanheyi model or joint ventures as 'sentences' in a common organizational language, becoming essential as AI agents enter firms and transaction costs collapse.
Amber Case presents the Calm Technology Institute's 81-point certification framework, arguing that human attention operates like a law of nature—distinguishing prefrontal-cortex-demanding 'center' attention from pass-through 'peripheral' attention—and that products (from car buttons to AI interfaces) should be designed and governed accordingly. She traces the framework's lineage to Mark Weiser's 1995 calm technology principles and discusses the legitimacy problem of building certification authority without statutory power, comparing CTI to Underwriter Laboratories and the Good Housekeeping Seal. The session closes with audience discussion, including Venkatesh Rao connecting the talk to Kairos/Chronos time and AI's 'staircase' temporality.
Daniel Kronovet presents 'Chores as Complex Coordination,' introducing 'regenerative labor' as a concept distinct from maintenance or care work, and detailing Chore Wheel, a Slack-based protocol suite (chores, hearts, things) built on Elinor Ostrom's commons-governance principles plus four digital-era additions: no privileged roles, simple inputs, humans-for-sensing/machines-for-bookkeeping, and continuous asynchronous participation. Drawing on his Los Angeles co-living house Sage House as a real-world case study, he argues that separating task definition from task performance (akin to double-entry bookkeeping) allows a fixed, continuously-replenished point economy to turn dish-washing and trash duty into a system of social computation that generates social capital alongside physical upkeep.