My kind of fall into the crypto space, I think, is a little bit atypical for most folks. I think most folks are brought in for the technology for the technology.

I was like, alright. This tech is insane. What's possible?

real engagement with these protocols. So, like, creating fake GitHub accounts, fake commits, fake pull requests Yeah. Fake tweets. You know, literally, if you've seen the videos of these farms, it's and I don't wanna call any specific chains out, but you'll have these farms where they have just racks and racks and racks of cell phone motherboards that are preprogrammed to just do certain actions and just constantly click farming. Your iris data is uniquely identifying and it's public. Right? Like, my iris data, I'm walking around constantly with my iris data just on display.

Yeah. I I love I love telling the story. Thank you for the intro, by the way. I'm really excited to be here. Yeah. My my kind of fall into the crypto space, I think is a little bit atypical for most folks. I think most folks are brought in for the technology for the technology.

Yeah. That's what I mean. That's definitely, you know, if you did, like, some sort of, like, latent analysis of all the reasons, that's definitely, like, a main ex explainer of the of the variance there. But, or for the, you know, I think for the community, it also is, like, a good a good reason. But for me, it was it was kind of, none of the three. I was trying to solve an existing problem. I was actually at, Georgetown at the time finishing up my PhD working on an inter institutional cons a consortium, data data sharing consortium that was looking at how do we track the developing adolescent brain with, magnetic resonance imaging and, behavioral studies. And so there was this huge amount of data all across, you know, it was like there was a site in Brazil and in in university University Grande do Sul in, Porto Alegre. There's a data set in, University of Melbourne in Australia. We had collaborators all across the East Coast in The United States. And it was, you know, hundreds of terabytes, this this dataset. And we needed a huge dataset. Right? Because we were trying to predict cognitive age in developing adolescence. So if your brain and your behavior looks a certain way, can we predict where along the normative growth curve are you? Are you, like, more advanced than your peers, or are you developing a little bit slowly, and how does your environment play into it? So huge challenge in getting all of this data in one place, even just on a cloud instance so that we could pre process it and train train models, synthesize data together to train models. But it wasn't as easy as you think because all of the different universities, because they were in different countries, had different rules and processes for approving, like, where the data can go because it's really sensitive patient health, records and and patient data. So that was really tough. And I was stuck for quite a bit. COVID hits, and I just had to come up with a solution or else, like PhD, TBD. I wouldn't be able to finish. So I happened to come across you know, at the time, like, there was a ton of excitement and interest in, Web three space. And I think, like, search engine optimization was really being, optimized, I guess. But I was able to come across IPFS, as part of my research and how to figure this stuff out and organ ended up organizing a hackathon where I met my cofounder, Nanak, and we basically hacked together a solution that allowed the universities to share their datasets directly instead of going through a third party server. Mhmm. Now it never ended up making it to, like, this really large scale production, but I was able to kind of get what I needed to get done for my research. And then after that, I was like, alright. This tech is insane. What's possible? I can't even, you know, begin to match what's possible, and I wanna take some time to really absorb it all and and see what other really cool infrastructure upgrades I can build for science, for doing science at scale and distributed and distributed research communities.

Yeah. So I I looked into this actually. So I I have this email. I could probably pull it up. I have this email that I sent to one of my, PhD defense, one of the members on my PhD defense, defense board. And I remember she asked me, she's like, okay, well, you know, what's next for you? Are you, you know, gonna head off to North, you know, University of North Carolina to do your post doc? You know, I've heard you might continue down the same field. And I said, hey, well, you know, actually, I'm not quite sure. I'm really excited by this idea of decentralized science, in DeSci, and really building up infrastructure to do research in the cloud, where I could set up my my lab in some piece of software and be able to capture funding and do research kind of, you know, in parallel, but independently from the traditional academia, world of academia. And I just never got a response back from that. Okay. Whatever that is. Right. Right. Right. Like, we'll write we'll write that one off seven years later. But yeah. So later I looked into this actually, because I had this idea and actually for the hackathon I named, it's in the global brain hack. I guess you could say, like, records on GitHub. Global brain hack is it's like ETH global, but for neuroscience.

And it's styled after some other really global hackathon. I think there is this kind of tradition or culture about hackathons, especially on the Internet, that all go back to this one, this one type. But, yeah, I know that the brain hack was modeled after this. And so, yeah, we we were thinking like, okay, well, how do we take these hackers from all around the world and give them infrastructure to share neuroscience data? So decentralized science seemed to be the right way to talk about it. But funny enough, decentralized science or DeSci also sounds like DeFi. And I I did spend some time looking on Twitter coming up with fancy filters to see when it was first mentioned. And there's just like this random guy from 2019 with like a like a bootleg punk PFP, RIP guy. And, he's like, when when when are people gonna build Desai? All of his other tweets, I think, were, like, about Compound and Aave, the future of France. I mean, finance.

Yeah. No. It's, yeah. I think it's like David or Donald Hebb, I think is his name. And he was this guy who was doing single unit recordings in in in neurons and was researching auto associative learning. So, like, how do two pieces or or blobs of biology, how two little cells can actually perform coincidence detection. That's kind of amazing that you have wetware, like just fats and salts and proteins that can detect coincidences in the real world. And if you assemble them hierarchically, somehow you get, you know, visual perception, auditory perception, cognition, emotion, and all the amazing things that brains do. So he was researching this at a very low level, just like how do do how do how do groups of neurons decide to wire together to fire whenever, say, they care about the same thing, like a certain color or a certain type of sound? So heavy and loop is just something I was, like, trying to come up with an like, some sort of a I was trying to invent a Twitter profile for my future academic career, and this was before Musk bought Twitter. I was like, oh, this has to have, like, some lore behind it. You know, I gotta show up creative or something. So I took some time to brainstorm. And, so heavy loop, basically, it's just the idea of auto associative learning, but in a loop, where you're training yourself over time to learn new things. That's how I think of it. So it's like an, an like, if you're gonna ask me to define it, I would say it's an autodidactic percept. So it's like something that you can perceive that teaches itself over time to continue perceiving new things, perhaps. So to to to be an AI, an AI let loose.

Yes. Yeah. It's interesting. I wasn't thinking of AI at the time, but but it's it's interesting though, because there is something like abstract about learning in in these, like, neural systems. There's, like, the wet stuff, and then there's the cognition that happens. And I don't know. In neuroscience, you kind of typically treat them separately. Maybe we shouldn't, but we kinda model them cleanly and try to ignore all the all the messy wet stuff. Right. Right. Right.

Yeah. Absolutely. I'll start off with describing kind of where HoloNim came from. Mhmm. And I'll, briefly describe what it is. We started off actually working on private, identity proofs for decentralized science organizations that were on chain. So if you wanted to use, say, like, the DWeb to store or retrieve data, how do you deal with, say, private attestations on that data? Right? Like, how do you know that the data actually came from a certain person, from a certain lab, has, say, the actual data contained within it? And if you're gonna use a public blockchain for this, you don't necessarily be wanting, you know, wanna put, like, even, like, encoded data on the chain. So there's all sorts of stuff, you know, floating around for the past couple years around this. I was doing a lot of research on, like, you know, things like proxy re encryption and, some of the early work that ConsenSys did with the city of Zug, putting them on chain with a DID schema that never really took off. It was really interesting looking at things like Shamir secret sharing for building data vaults. So but really where I think, the tires hit the road was when I convinced my cofounder, Nanak, to quit, his job at Georgetown. He was also in computational neuroscience lab, kindred spirits, us all. And he, really, really took a liking to, like, immediate, I think, affinity to the zero knowledge proofs. Kind of just all of that.

he, yeah, he really, found a direction where we could start building out actual feasible zero knowledge proofs with, like, for consumers in in in web three for, like, normal dApps actually bringing it to production. And it it wasn't a coincidence either. Like, at that time, a lot of the developer tooling around CK proofs was beginning to mature. You just had to have enough of a bat math background to be able to understand its significance or its limitations to properly engineer actual solutions that worked. So we did it. You know, we built something, won won a couple hackathons. And ETHDenver actually was the first one that we built, the first, version of it. And, yeah. We we, we won a couple of awards. People found out about us. And then we got, like, a contract from Lobby three, which is a political civic advocacy group, started by Andrew Yang to raise funds for UBI and Mhmm. Anti poverty. But they were using crypto for this and needed to verify that their donors were actually US residents before they could put the funds to use for campaigns or for for lobbying. So that's when we realized, hey. Like, you know, there's actually some really interesting real world use cases here in in in civic infrastructure. Or taking, say, the the tools or the policies or or the needs that we have in bringing people together to do things for society, crypto can actually help us do these things in a much more resilient, distributed, and maybe efficient way. So that's really where we got started, really dove deep into CKID proofs. So HoloLens today, it's, basically a software labs foundation that builds out, technology for keys while it's an identity. And we support the establishment of this open framework that we've, that we call human dot tech, which is, a technology framework for embedding digital rights into, distributed web technology. So things like self custody, privacy, security, personhood, for example. But, yeah, it's interesting because we started off really in this, you know, trying to build decentralized science infrastructure towards realizing what we were building is a lot more lower level and touches many aspects. And the most kind of significant it felt really was civics, like how how societies and countries are organized or, can build resilient technology pillars to really, I think, like, protect and elevate the kind of principles or or or values of of that society.

Yeah. Totally. Yeah. I can give you guys the the the big overview. So, HoloLens is a software shop, and we're trying to create something bigger than it, which is this open framework for really putting humans at the center of technology. And so what we've done is kinda similar to, let's say, what, consensus or, you know, theory foundation has done, which is we're gonna build out this base layer infrastructure. We're gonna give it everything it needs to get going, with engineering support, community support, funding support, and bootstrap this base layer for basically embodying digital rights into technology or embedding them into technology. So what does that mean? There's multiple components within the human dot tech infrastructure. At the very lowest level we have, which really roots everything, is something called the human network. The human network is is a bit of a big deal. I don't think many people really realize just how important it is to have, verifiable randomness for key derivation and zero knowledge identity proofs. That's something that a lot of identity protocols that have tried to solve, this problem of fragmented identity with privacy, kind of like falling into the same pit, pitfalls.

Yeah. No. Nice. It's an entropy network. So when we talk about, really, you know, when we talk about embedding digital rights into technology, what we're really thinking of is, hey. Can we give everybody self custody for keys, with so that they can have encryption that nobody can break, without their you know? And and so that people can't reveal data without without, can't show data without consenting to it? Can we give people the ability to create identities that are theirs that nobody else can grant or take away? Can we give them personhood, digital personhood in in in the in the digital world? So to do that, you need entropy. So the network operates in a decentralized manner. It uses Ethereum proof of stake with eigenlayer and symbiotic to secure and offer incentives for nodes to actually run the functions of the network. Each node computes what's called, an oblivious pseudo random function on a piece of data. And what that means basically is they're receiving some blinded piece, shard of data, and then they compute, basically, they transform that piece of data into something that's information theoretically secure, meaning that whatever comes in, what comes out, the probability of any character in that string appearing is the same as any other character appearing in any string. There's no information whatsoever. It's max entropy, which makes it very hard to brute force. Right? Like, if you try to guess the characters one by one by one by one. Guessing one character tells you nothing about the rest. So this is really guessing one character tells you nothing about the rest. So this is really useful for key derivation. For example, if I wanted to create a key from some biometrics like my heart rate variability or my, combination of, like, say, my my face and my voice or my face and, like, a web accounts. All of these pieces of data are low entropy, meaning that somebody, if they know just a little bit about me or have some other information about me, can probably brute force this and figure out what transactions I've submitted, what I've signed, or even crack, like, my private key. So with with an entropy network, you can take this low entropy data that represent human attributes and actually derive secure private keys. And that's what we call human keys. That's something you'll see in our documentation where we're like, hey, private keys are great, but, you know, they're great for developers. What we need are keys that are more user friendly and more human friendly. And by driving them from human attributes, you solve a lot of the crypto UX problems like, recoverability, ease of onboarding, and just, like, ease of use in general with the keys.

Yeah. No. Absolutely. And, you know, there are thing other things as well. I mean, that vulnerability is, you know randomness on its own doesn't protect you from that. There's other bits that we put in to prevent spoofing. But for example, with voices, I think they have, like, 20 or 30 bits of entropy, which isn't enough for for a private key. For a private key, you need at least 128 bits of entropy. How do you measure entropy. That's interesting. Oh, it's a very simple equation, actually. So Okay. If you wanna calculate entropy of anything, you literally can even calculate the entropy of, like, your own podcast if you can transcribe it. Uh-huh. So what you do is, and the entropy will give you the bits of information in that corpus, that data that you're using. So so what you do is you break up whatever it is that you're wanting to measure the entropy of at the level it makes sense, like, say, at the syllable level or the word level or whatever sentence level. In the case of, like, of, say, like, a voice, if you're trying to calculate the entropy of a voice recording, you'd maybe cap, bin the frequencies, so within certain frequencies. And then you would threshold it. So say, like, you only take, you know, amplitudes over a certain amount. And then for each unique frequency, you count how many times it appears in the entire recording, in the entire corpus. And then for each so you create a big table, and then you do this equation called, the Shannon information entropy equation. And you calculate the probability times the logarithm of the probability, and you sum that for each of the bits. So for frequency one, you take the probability of frequency one, and you times it by the logarithm of the probability of frequency one, And keep that, and then you add that to the probability of frequency two times the logarithm of the probability of frequency two. And you do that until you computed that, the summation across all of the features in that corpus. And then what you end up with, which is funny, like, it's hilarious. Like, I don't even know how this guy came up with it. You end up with, the bits of entropy or the number of, yes, no questions you would need to ask to basically, like, decrypt or reconstruct any piece of data. So it gives you the amount of information overall or the amount of surprise within that dataset.

Yeah. Entropy is amazing. It's yeah. I spent a lot of time thinking about that in PhD studies. I wanted to do something with, like, language acquisition. But back to crypto, I think, so so so we have the human network. It lets you derive keys from low entropy data, and it also does some other things too. So it lets you create what's called a secure nullifier. For example, if you wanted to, generate an identity proof in zero knowledge, and you wanted to make sure that that identity proof wasn't used more than once and, was very difficult to reconstruct the original data that was used for the identity proof. You would need to use some sort of, like, secure randomness for creating this. And, like, a huge problem right now too in the space is you have, I don't know, like, a dozen or maybe more identity protocols that are popping up zero knowledge identity protocols that are popping up, and all of them might offer the same type of proof. They might say you might have a world's proof of humanity. You might have the CK pass proof of humanity, an orange protocol proof of humanity, a reclaim proof of humanity. And so if you wanted to multiply yourself, right, you would go to each of these different protocols, generate the same proof, and then any project that accepts them, those proofs, you just present that to cash in each time or do whatever it is that you needed to do. And that's because they don't use, like, a verifiable nullifier. And that's something human network can do because it is a network. It's decentralized. There's minimal trust assumptions. It's not something that we're running. It's Ethereum node operators are running it, which is providing software for it, maintenance for it. And, yeah. And so this way, you can create this kind of nullifier that's universal and and and lets you spend your identity truly only only once, for example, if that's the use case that's needed.

Exactly. Them randomly. Perfect. I wouldn't describe describe it better than I could. That was But, yeah, definitely,

Yeah. Absolutely. So at the very low level, we have the human network, which kind of powers everything that we're doing because we need to have verifiable randomness for anything to do with, keys, wallets, or identity. So on top of that, we built something called the human wallets. It's something that we're really excited about. It's been in development for about a year. It's a very complicated complex piece of software, over I think like a 110, 120,000 lines of code at this point written written by primarily, small small team of cracked engineers and incredible engineers. And, and, yeah. So the cool thing about HumanWallet is it uses this really friendly entropy, you know, entropy power, superpower of the human network to easily create, keys for users so they can basically any dev can create this iframe or embed basically this, like, login portal that allows the user to present any familiar source of authentication to then log in and have an account. So I can come up to, say, like, a crypto you know, like, a a political donation page, for example, say, Andrew Yang's, you know, donation page or whoever. And I can just basically sign in with OAuth and authenticate with biometrics, and I'm able to derive a key that I can then, you know, on on ramp into crypto with or sign transaction with or do, voting with, for example. So Human Wallet is special in its way because it also has something called, it's the first of its kind, actually, of a two PC MPC wallet or a two party compute, multi party compute wallet. So for the for the nerdy ones in the audience, you you might appreciate this, but what we do basically is take that key that we create and split it into two. And we found that this is one of the best ways to secure, any self custody, crypto wallet where one of the keys is basically tied to your human attributes, in a privacy preserving way. And then the second half of the key is, kept is kept behind, some some policies or conditions. So, for example, if you're transferring, say, over a $100,000 out of the wallet or, say, like, $10,000 out of the wallet, that's even though you might have authenticated and presented yourself with the first part of the key, if you hit some sort of predefined condition, the second part of the key won't be reconstructed unless you do some sort of action. Like, for example, I don't know, like, click a link in your email or send a text message, SMS, or whatever could be set, like, assigned with a hardware wallet, for example. So this, we find, is gonna be really powerful when you need to actually interact with complex pieces of software, especially in DeFi, as AI starts to really take root and intense kind of, like, more programmable transactions that are bridging across chains. And there's all sorts of stuff that's happening on the back end. And what's happening in our move to make Web three more, I think, easy for the average person to use is a lot of security has been abstracted away. But this one approach can help us basically take a lot of the best security practices, put them into, say, a back end AI or policy engine that requires, you can rank the level of risk for each of the different transactions and use that to kind of condition or or program how the key is assembled and in which cases can you actually submit, a transaction. And all of that can be use user defined. Right? So if you're, you know, somebody with just, like, a dollar in the wallet, like, maybe you don't care about these things. But if you're, you know, somebody who's running the entire family inheritance, you know, all of world liberty fire something in in your MetaMask hot wallet,

So so we built this for the for the folks that really are tired of getting getting burned, tired of getting hacked, of clicking on, like, phishing links. We built it to be incredibly resilient to a lot of these attacks.

Yeah. Exactly. It's in the end, it's still a self custody wallet because the key is always reconstructed on the client side. There's the keys never reconstructed anywhere else, and only the user has the ability to do so. But you can basically remove some, some of the authentication privileges from the client side device and put it in a server so that you can always have, say, like, formal verification on the transaction that you're trying to execute. So the Bybit the Bybit example is a great, great example. Was it, like, a $1,400,000,000 heist by North Korea? That's crazy.

And it was it was literally, like, you know, two or three guys who had a multisig. It was a safe multisig, which is the industry standard right now. And, they have the hardware. Hardware so they did everything right that Vitalik told us to do. Right? That security experts tell us to do. We have the hardware wallet. We have the safe multisig. You know, we check the transaction. We do the test transactions. So the attack was was very sophisticated. But in the end, you didn't have these policies in place that prevent kind of, like, the suspicious activity because the code that they ran to execute the safe multisig transaction with the hardware wallet wasn't verified anywhere. So on their ledger, you know, they open their ledger and they see some hexadecimal code, and they're like, yeah. Okay. I mean, looks right to me. Not really. I mean, you can't really tell. It's hexadecimal code. Right. Right. Proof. And there goes $1,400,000,000. Right? So that's called blind signing. So what would have helped with this is if you did have, say, another server that handled the authentication in a way that it could be formally verified. So you can read the bytecode, and, decode it, run it through some transaction simulation, and then deliver a human readable output to the user Mhmm. For additional authentication. Say, hey. Look. Warning. You know, you're about to send funds to an address you've never interacted with before. For people who maybe,

Yeah. I mean, if if anything, it's it's forcing the space to if we're gonna be decentralized, we're gonna have to be able to handle adversarial attacks like this because Yeah. You know, what's what's the point of doing this if we can't build resilience and and and security into this?

anywhere close or maybe I mean, there are 100, but, yeah, still not that close. Yeah. I don't I don't think so. I mean, like, if you're not counting, say, like, like white collar crime, but, even then even then, I don't know. They're gonna make a movie of this someday, I I bet you.

approach that. Yeah. That's a good point. It's interesting. Yeah. We don't really think about identity as much as computers do when they're interacting with us, or interacting with them. Yeah. Human Passports are it's our biggest product. About 2,200,000 users last I checked. I think it's pretty much the largest identity protocol in Web three by number of credentials. We've issued over, since starting, 35,000,000 credentials through through Passport. It's primarily used for proof of personhood, so people that wanna prove that they're a unique, breathing, living human being, specifically for, governance or airdrop distributions. So, you know, Uniswap is a protocol that really set kind of the standard for what it means to bootstrap a community by distributing ownership of your decentralized protocol. So back then, they gave 400 UNI to each Ethereum address that you, interacted with the protocol or ever swapped. And then they gave a little bit of extra to maybe a little bit of extra, actually, to other projects that were providing liquidity. So nobody saw this coming. So there was no way really to, like, anticipate or plan for this and create lots of accounts or anything to maximize your your airdrop. I think some people made, like, life changing money on this just because they happened to have a lot of burner accounts that they swapped, like, a dollar with or something once or twice Yeah. Yeah. 10 times. And, so but now there's an entire industry around airdrop farming, a multibillion dollar industry at this point where, basically, there's, organizations in Southeast Asia, Africa, North America, all over the world, really, that we've that we've come across where their solitary goal is to interact with crypto protocols and simulate authentic human engagement. And this is independently. The protocols aren't paying for this. Right? They're simulating real engagement with these protocols. So, like, creating fake GitHub accounts, fake commits, fake pull requests, fake tweets. You know? Literally, if you've seen the videos of these parts, it's it's it's all fake. Yeah. And I don't wanna call any specific chains out, but, you'll have these farms where they have just racks and racks and racks of, cell phone motherboards that are preprogrammed to just do certain actions. It just costs, like, click farming. So it's a real thing. And projects, you know, when they do an airdrop, they're they're thinking of it as a customer acquisition cost, typically. Like, we want to we want people to care about our token. So one way we'll do that or our product or community. So what we'll do one way we'll do that is by giving the token out to a group of people and trying to channel those Uniswap, you know, good vibes from back in the day and use that as a way of, like, on ramping people into the technology. Now the problem is that a lot of airdrops that have been done since then have basically just given their money to a very, you know, small number of well organized, folks that are that are farming these things. So Human Passport really emerged to help projects do fair distribution so that they grow authentic communities and really signal, hey. Like, there's real people using the technology. And one thing I'll mention too is, you know, that's our primary use case today. We've done, like, over $430,000,000 of airdrops have been have used Passport to distribute, which is something we're really proud of. But it really also had humble beginnings. So we started off actually within the Bitcoin ecosystem doing, helping quadratic funding matching to make sure that the same sort of thing, people that are farming the Bitcoin grants ecosystem to drain the matching fund because it's the same thing where Mhmm. If you can simulate people that are voting for projects, you can have a bunch of fake projects, have your own bots voting for your projects, and then you get all of the matching funds. So, whether it's grants or airdrops, and one that we're seeing now as well is is staking. So projects, for example, like the restaking platforms like Lido and, and, like, Rocket Pool, etcetera, they, you know, they wanna be able to really show that it's so decentralization matters for Web three. Right? We don't want all of our nodes in one country or one organization because I mean, that's why we built this in the first place that if an exchange goes belly up or if one person goes belly up, that the rest of the ecosystem can continue on. It's not but a scratch. Right? So Right. It's really important to make sure that if, you know, people that are staking Ethereum actually are many, many individual different people, not just, say, one organization. So that's something we've seen as well kinda coming up in in recently as proof of humanity has never been more important, to really show, like, who people are and to be able to actually have trust in who you're interacting with and some of the guarantees, for example, behind, like, say, a network if it's really just one big organization or a bunch of, you know, real people.

Yeah. I'm I'm happy to. You know, just bringing it back to the earlier point, you know, one thing that really matters more than ever today is, you know, we're becoming more and more reliant on these digital systems, and we need to find ways to take the rights that we assume to be self evident in society because there's law and police to back them up and just embed them by default to technology. And the more resilient we make these rights, the more carefully we apply these rights, I think the better off society will be. One of the most important rights you can have, in any society is personhood. Right? Saying that, hey, you are a person. You can demonstrate you're a person. You have these rights, privileges, and affordances that come with that, right, depending on whatever community or social contract you've entered. But doing personhood in this way is really, really difficult. Right? Especially it's hard enough in the real world. I mean, you've probably audience and you guys have all heard about, like, you know, Social Security fraud or, you know, like, you know, who's really a citizen and who's not a citizen and etcetera. And, yeah, there's all sorts of crazy stories too about, like, people pretending to be somebody else or what have you. But, doing personhood on on chain, completely decentralized, completely distributed is really freaking difficult if there isn't some sort of centralized party in the middle, if there isn't somebody who's controlling all of this and making sure that the standards are being followed and the whole system is is is kept cohesive. It's not impossible, but it's very freaking hard. So Worldcoin, started off, or now world, started off trying to solve this problem specifically after doing tons of research and finding that most of the stuff just wasn't good enough. And they said, you know what we're gonna do? We're gonna go with the highest enter source of entropy to identify people uniquely so that it can't be defeated by generative AI in the future. And we're gonna do it in a way where we're gonna make sure that people can't even, like, spoof it, because they have to show up in person to a trusted node operator, who's gonna scan their eyeballs. And with those eyeballs generate, basically, the way it works is that you have somebody who's a WARB custodian. They get, like, jobs to go verify people at certain times and places and days. And, the user come they sign up. They have, like, a world app, and then they create, like, a session ID within the world app. When they are ready to register, they click the button. And then the session ID is used in combination with the orb to create basically a unique identifier with what's called a nullifier, that's only known to the user that holds that session ID. So that later, if they lose their device or something like that, that they could recover this later. So the orb does have all this trusted execution hardware on it. It has been open source, but public breakdowns have been very highly controlled. So there's a lot of skepticism in exactly what's inside the orb and what's running and how's it's how's it running. And the other thing is that they ship this really fast with very minimal oversight. So I think, it's incredible, right? Like they built something in just a couple of years after identifying a need for it. But also they built something to scan the eyes of every single human on earth within a couple of years and very little challenge oversight as part of that. So Yeah. The ambition, I think, doesn't really match the kind of time in rigor or proving of of what they're trying to do. And there's yeah. It's a Silicon Valley mindset, high scale growth mindset with this hardware that you might or might not trust. Basically, your embeddings, so your embeddings are basically gonna be the iris data that's preprocessed and transformed by the org, are gonna be encrypted. And then there's all these different stages of encryption where they create a key based off of this data. And then that other key is used with, like, a meta like, a wallet key within the wallet, kind of like a MetaMask key. I think it's built on safe now, actually, with a safe key that you then use to create an on chain accounts. And then the encrypted embedded, embeddings, the the machine learning representation of your iris is stored in this other, in this other collection of servers. So one thing to note, I think this is the most important thing to note, is your iris data is uniquely identifying, and it's public. Right? Like, my iris data, I'm walking around constantly with my iris data just on display. Not all cameras can detect, can detect iris data, but a really good CCTV camera, it's been shown, in in, in the literature that you can get a good enough reconstruction with post processing on iris data from up to 20 feet away. So what does that mean? It means that you have the ability to uniquely identify as many people as Worldcoin has scanned through good enough CCTV and maybe, like, state level, state type like resources, which makes Worldcoin a target for state level, intelligence or state level, say, operations that are intended to, say, find specific types of people or track certain types of groups of people for whatever reason. Right? And so World Finance kind of set itself up as a honeypot, potentially, if certain assumptions don't hold. So the first thing is that, because the machine learning model of the embeddings is made open source and public, you know what kind of transformation to go from to get to embeddings, to get to, like, the machine learning representation of it. So if you have an example of an iris, you can generate, kind of, like, the intermediate representation that's used to create the key. And you would need to do a bunch of other things too as well to, like, actually break the key. But that's good enough for state level, say, after. And then second, you need to trust that however the embeddings are transmitted from the orb to the secure, set of servers. It's, I think, three, servers that are running an MPC setup, across three universities. That tech was set up, I'm pretty sure by the by the Worldcoin team, and those universities are just running the the hardware for it. So you need to trust both the setup, the software, the communication channels, and that these MPC systems really do hold up and that none of them can be compromised at any point. And that's assuming that a state level actor is disinterested in this. Right? Because I think that it increases the stakes as well. Right. So that's what we're dealing with. We're dealing with one of the biggest honey pots that's being assembled primarily off the backs of people in the global South to bootstrap a token to get to a billion dollar market valuation based off of traction metrics, that are somewhat I don't wanna say exploitative, but can be interpreted as exploitative and maybe exploitative depending on how time comes to because I don't see any data centers. I don't see any OpenAI models being trained or run within Africa under self you know, under jurisdictions and sovereignty within those countries. I really see them just becoming new consumers for for for western, I don't wanna say it, but techno imperialism. Right. That's my concern. That aside, I do really respect the technology and the work that the teams put into it. I'm just very skeptical that this sort of monolithic approach powered by kind of tokenomics is gonna safely scale to the entire world without having some major hiccups along the way.

Yeah. I mean, one thing our major complete philosophical difference here is that instead of taking a monolithic approach where we control the hardware, we control the software, it's open source, but, you know, it's only our we control who can pull requests. And we're creating kind of, like, this Apple type ecosystem. Right? Like, you know, Apple is built on UNIX. It's built on open source in the same way that world is built on the optimism and Ethereum stack. Right? But there's very little control that you have within the ecosystem, and you just trust Apple that they're they have your best interests in mind. So the world is trying to position itself in in in the same way. And, honestly, maybe their biggest competitor is Apple because Apple could pivot to this at any point. So Mhmm. What we see is a huge opportunity, though, to build truly distributed, decentralized, pluralistic models for identity proving. So that's the approach we take with Passport, where you, developers have kind of full control over how they wanna verify or onboard their users. It's also why we have the biggest and the most adoption within, places where civil defense and proof of personhood is actually mostly needed right now, which is in web three, for proof of personhood.

and, so devs basically have the the the tools that they need. They can choose to verify with on chain reputation, web of trust. They can choose to verify with, say, biometrics, for example. They can choose to verify with government IDs. And there's an a la carte approach that the devs can take that lets them sculpt kind of the specific type of identity offering that they need for their users. So this kind of more tools based approach will let us fill in this difficult gap. So we're actually tackling the more difficult problem here, which is uniting different types of identity, to be interoperable and interusable while still preserving privacy and user data sovereignty and not locking into a specific ecosystem. Whereas World went for the let's scale as fast as we can because we found this one little niche of these sets of assumptions and technologies that lets us have a venture sized outcome in the in the next ten years by Right. Basically, toke you know, tokenize converting people's eyes into tokens, all of the world's eyes into tokens is the mission. Right? We're tokenizing the eyes. Yeah. Token eyes. Uh-huh. Token eyes. Oh, interesting. There's something there.

Yeah. Absolutely. I'll this is one of my, I think, favorite topics to to think about, and we surround ourselves with multiple challenges and difficult problems to solve, whether it's presented or something. But this is a really difficult one that's worth tackling. The first thing I think it's worth mentioning is that the solution is gonna end up being, literally defined by how we decide to organize as humans in civic society. That's what I'll say. This is gonna be a human coordination and governance problem, when it comes to privacy and accountability. I'll start off by saying that. But let's take a look at kind of, you know, why these different philosophies have emerged. So, you know, if we start off with kind of, like, the main criticism of blockchain that it's you know, there's these secret super shadowy coders out there stealing all the world's money. And that, you know, a maximalist might say everything should be open. Everything should be public. Everyone should see what transactions you you're doing. You know, the government should have the ability to, request data on demand to see who you are, what you are, who you do, etcetera, etcetera. Right? That's not that one side of maximalism, and those people definitely exist in in the government and intelligence, services, I'm sure. Now how about the other side? The exact polar opposite. So on one side, you kinda have, like, this, you know, sense of need for security that there should always be backdoors and security guarantees, because anything hidden is a threat. And then on the other side, you have this other world where, you know, identities are completely interchangeable. You know, nobody knows who anyone is or anything is. Nobody knows who what transactions you've had, what you've had for breakfast, you know, what your real name is, what you look like, what your preferences are. When you go and buy bread, you know, you're buying bread from a maxed person. You don't know who they are. They're you don't have anything about their identity. They're interchangeable even. Like, the identities are completely fungible. There is no kind of uniqueness as well. It's max entropy as can be for any functioning society. That's the extreme, extreme, extreme end, where you have a complete fungibility of identity and unique identifiability. It's just not possible. A society like that, while it's fun to imagine in a science fiction setting, I think really doesn't lend itself to kind of, I think, a society I would wanna live in. Right? I wanna know who's baking my bread, or or who I'm buying tomatoes from. And so one huge struggle is to kind of avoid these polarizing argument argumentative points and recognize them as kind of the boundaries, and then start thinking about what's the wiggle room in between. So how do we have privacy with accountability while preserving kind of trust and human interactions, that that that that, that come along with that. We thought a lot about this. So we came up with a schema where, basically, what we do is we think of privacy as a privilege that's afforded, or you could say a right that's afforded within a social contract. Right? So, you could think of it as as kind of a privilege at large. But when you enter into a social contract, into an agreement with, say, a specific community or, country or application or DAO or what have you, you're basically consenting to, you know, have consenting to a set of rules of behavior, of governance, of how you're gonna conduct yourself. And with that comes, say, privacy. However, if there comes a point where you infringe those rules, say, for example, you, your accounts your pseudonymous account is the one linked directly to training the treasury by exploiting, smart contract bug. That would become a condition in which maybe your data would be disclosed or decrypted.

Yeah. Absolutely. So yeah. So striking this balance where you have privacy with accountability. Right? So you consent into some rules, and then as long as you behave by those rules, you're considered anonymous. But if you infringe upon those rules and it's it's a requirement, you your information might be decrypted. So a perfect example of that is, the canonical bridge that we've built to Aztec, which is a privacy first, l two with built in zero knowledge proofs for, basically private state. Like, all of your data's basically generated as zero knowledge proof on the client side before you even submit a transaction. So Aztec's amazing. It's the future of privacy. Now the problem with Aztec is it's a privacy blockchain, which means it's a place where if I'm a North Korean hacker or I'm a bad actor, this is a place where I would go to hide from accountability. This is where I would dump my my ill ill gotten gains. And to prevent that, the team was really conscious about it. The Aztec team reached out to us about a year ago now, actually, and said, hey. We like what you guys have been doing in the identity and zero knowledge space. Would you consider helping us out with this hard problem? So we together, we designed a solution where to bring money into the Aztec ecosystem, you have to pass through a bridge that has this, programmable privacy with accountability built into it. So if you're transferring below a certain threshold amount from Ethereum to Aztec, you just need a proof of personhood proof. So you just need to prove that you're a unique person who's transferring this $1,000 once and that you're not transferring a thousand dollars with a 100,000 different accounts. If you're transferring more than that, you're gonna need to do something a little bit more hefty. You're gonna have to go through, a zero knowledge KYC flow, so where you prove in zero knowledge that you have, a legit government credential and that the information that government credential doesn't show up in an international sanctions list. And we let, and that that sanctions list could be defined as as is. So if you pass that first pass, then you're allowed to to bridge. But you're also shown a little pop up, right, that says, hey. You know, I agree to follow the terms and conditions of this. If, you know, it turns out later that I actually was a North Korean hacker, I show up on a blacklist that's maintained by three or four different, you know, say, Oracle services, then those will be conditions enough for my data to be decrypted to some public key. So what's cool about this this is just one example of it. But what's cool about this is you can actually put in all sorts of programmability around the decryption conditions. You can have, say, different sources that you're listening to that, say, would report adverse events, and then you can even have decryption occur to, say, a consortium or some form of DAO of legal representatives that represent the space, that represent accountability for the space. So the design space is really huge, for for this. Now how do you actually get people to come together and think about this? Like, I started off saying, like, hey. Like, in the beginning, it's all about human coordination and governance that's gonna decide this. Like, in which cases do we implement this, and how do we implement it? So that's why we built this thing called, the covenant as part of kind of the nucleating core of human. Tech, which allows, people to come together with these considerations in mind. The covenant is basically a system of, different principles or, values that are important when building technology. And we've kind of opened up the the the this framework to basically anyone out there, who's interested in these sorts of things and being able to contribute different pieces of written works, academic works, essays, and also artistic pieces like poetry or, we've had some incredible contributions from artists that are Sotheby's curated, as well as, like, you know, that have worked in the OpenAI fellowship and all sorts of stuff. So the covenant really is, a way for people to kind of bring these different vantages or viewpoints or pluralistic forms of, say, structuring social contracts with with different values and giving them kind of, like, this basic framework to align over. So we see it as kind of, like, this really primeval starting point that allows, these, like, really complex human coordination problems to kind of fork off of.

Yeah. Absolutely. Thank you, Josh. It's been absolute pleasure and could, yeah, could definitely go on for hours. So it's great being, getting to talk to you. Yeah. To find out about us, it's super simple. Just go to human.tech, and there's lots to explore and drill down there. So Yeah. Cool.

Alright. See you later, sir.