Web3 Galaxy Brain 🌌🧠

Nicholas: Welcome to Web3 Galaxy Brain. My name is Nicholas. Each week, I sit down with some of the brightest people building Web3 to talk about what they're working on right now. My guest today is Kozuke Hata, a.k.a. July, founder of Faust. Faust is a nascent hardware startup working at the intersection of distributed sensing, blockchains, and self-driving. The team is composed of engineers from automotive and machine learning engineering projects at Google, Apple, and Tesla. On this episode, July shares background about the self-driving and robotics industry gleaned from his years of experience working at Kitty Hawk, Larry Page's flying car moonshot startup. July also shares some of the prototypes Faust has built and gives a glimpse into how robotics, blockchains, and networking will coincide over the next few years. It was fantastic getting a chance to chat with July about everything Faust. I hope you enjoy the show. As always, this show is provided as entertainment and does not constitute legal, financial, or tax advice or any form of endorsement or suggestion. Crypto has risks, and you alone are responsible for doing your research and making your own decisions. July, welcome to Web3 Galaxy Brain.

Kosuke Hata: Thanks for having me. Very excited.

Nicholas: Yeah, I'm excited to talk about Faust. I first discovered Faust on Farcaster, but it's actually a hardware startup. Maybe you can tell me a little bit about that.

Kosuke Hata: Yeah, so a little bit of my background, I come much more from a robotics background. I've mostly worked on software for physical systems. You know, I've been involved with, I think, autonomous vehicles and building semi-autonomous flying cars for Larry Page and Kitty Hawk for about half a decade and built out a lot of sort of hardware and vehicles. Yeah, we started on this path just thinking that, you know, in the longer run, there's just going to be vehicles and machines and robots that are going to be on chain. And what is the sort of physical, what is the hardware and operating system? What does the future look like for that? And for these trillions of machines eventually that are going to be on chain that are going to be driving much more of the transaction volume than humans are.

Nicholas: So there's a connection between robotics and hardware being driven by also a connection to these decentralized networks, which I think we'll get to in a second. But before we do, I do want to ask you about Kitty Hawk. I'm not sure how much you can say, but people, some people have heard of Kitty Hawk and some people have, but not maybe can you tell a little bit of the story of Kitty Hawk where you worked before starting Faust?

Kosuke Hata: Yeah, yeah, for sure. So, yeah, so, you know, a lot of this stuff is public. If you look up, I was a part of the team, this team called Flyer. We built specifically like a single seater, semi-autonomous car, flying car. It's really like a really gigantic drone. We did over 25,000 flights. A lot of it. Most of it was crewed, most of it was unmanned and most of the flight, you know, were, yeah, didn't have a person in it, but, you know, I also flew in it maybe like 10 or 11 times myself.

Nicholas: Oh, cool.

Kosuke Hata: Happy to share that experience as well. Yeah. It was just sort of building this vehicle, I think from about 2016 to about 20, 2020 ish.

Nicholas: Yeah. So, so Larry Page, co-founder of Google, started to Kitty Hawk, wants to build flying cars and you worked on one of the projects. And I guess there were a handful of different streams working at the same time.

Kosuke Hata: That's correct. Yeah. There were essentially almost kind of like a different versions of the, the sort of same vision of, uh, I think solving urban air mobility. How do you solve urban air mobility? What is the, it's like brute force attacking the idea maze by building multiple vehicles at the same time. Yeah. That's kind of what it felt like, that would be the best, I don't know, a one way to describe it almost. Does that make sense?

Nicholas: Right, right, right, right. And I mean, when it started, I think there was, there was Kitty Hawk and there was maybe one other company at the same time that got some VC backing. that was a little bit, uh, in the press at the time. Um, I'm curious, are these projects still going on for, for, for one? Yeah.

Kosuke Hata: So, well, well, there's, there's a lot of projects out there. There's also a lot of like billionaire projects out there. Related, you know, like Sergei has LTA, which is lighter than air, which is like an airship project. Um, there's another sort of Larry page backed, uh, project called opener, um, whisk dot arrow, which eventually, um, I think, I think it's called whisk dot arrow still. Uh, and then, um, yeah, so there's a, there's a bunch of the, the internally kind of like adjacent to Google project and then there's, uh, yeah, a variety of like Joby. Uh, he hang in China, Joby's in Santa Cruz Archer, which is down the street as well. That took on a lot of former whisk arrow people. Um, so a lot of different, um, companies, many of them are actually quite slow around and, and a bunch of them have run out of money because it's costly, costly, right.

Nicholas: And so what were you doing there? So you working on a semi-autonomous, uh, drone that can carry humans, uh, obviously semi-autonomous can carry humans, although maybe not obvious, maybe you could fly from the ground, I suppose, but, uh, nevertheless, um, and what, what part of the stack were you working on? What kind of problems were you trying to solve?

Kosuke Hata: Yeah. Uh, I mostly focus a lot on the, uh, the vehicle software. Initially I was, uh, doing some embedded stuff. Um, some, yeah, essentially a variety of different things. Um, I ended up working for a while on some of the OTA. Uh, I worked on. On the, uh, the sort of screen, uh, and kind of interior screen alerting and design. Um, I also worked on, by the end, I was working on the perception system for like the autonomous landing of the vehicle, uh, camera selection. Yeah. Cameras were the sensors that I kind of was focused on. Um, yeah, a variety of sort of different tasks. They, I also worked a bunch on like the, like the CI, uh, yeah. Just like anything that we needed to do. And it was like a relatively small team initially. So it wasn't like a large team. And then it kind of grew to larger size over time.

Nicholas: Uh, and I don't wanna spend the whole time talking about Kittycock, but I guess you must've made a lot of observations. It must've had a lot of observations made of, uh, sort of what works and what doesn't in this kind of moonshot hardware, but also big software problem. Uh. Can you, can you draw any lessons from it that those of us who haven't had the chance, uh, might benefit from?

Kosuke Hata: Yeah. Um, I think like one of the weird side effects of like working on such, well, so first and foremost, I think, uh, Sebastian and the team at Kitty Hawk did a really great job of recruiting in general. So we just, for lack of better way of saying it, just have the best people, uh, working on this. And, and I think the, the sort of alignment of when you have. A project that really galvanize, um, galvanizes like people to be excited to like work on it all the time. It's, it's like a really, it just makes for a very unique experience. Uh, people don't leave willingly as much. it's, you know, people are willing to kind of spend more time on the project. And that's kind of a special, I think it's like, it's one thing to be, to have a bunch of people that are, um, I've heard of this, like similar. Yeah. Like essentially at places like space acts early on was sort of similar where people were just like, we, there's like no other problem. I kind of rather work on than the problem that I'm working on right now is what a lot of people were feeling because it's so. Interesting. Um, I think that's a, that's a very unique position and, and, and luxury to have, I think.

Nicholas: And if anything, you know, we I I've been sort of chasing that ever since the, the feeling of camaraderie amongst people going for some massive vision, but that it might actually be trackable.

Kosuke Hata: You don't know if it's going to work out or not, essentially, but it's like somehow still you're doing it.

Nicholas: And also, I guess people who are quite experienced on the team as well, people who really do have the chops to be able to achieve it.

Kosuke Hata: Yeah, I think it's like, you know, experience comes in different forms as well. Like you have experience in the form of kind of like spending a ton of time more in academia and have thought through the problem and different ways that it could work. It comes in the form of being a part of like maybe robotic startups before so that you have had experience like leading teams within robotics companies. I think it comes in sort of many different flavors, but it's really like the uniting forces is a vision and the selected mission really does kind of, you know, as the dude says, it really ties the room together.

Nicholas: So are we going to get a flying car out of one of these companies, do you think? Or how long do we have to wait? Or? Or if not, what is in the way?

Kosuke Hata: Yeah, I think it's like. the answer to the first problem is definitely yes. Like we are going to get stuff. It's going to happen at one point. That's like, let's put it that way. It's like definitely happening at one point. I think one of the companies, you know, some of the companies are already starting to put stuff out there. I still think the market is like relatively early, but it is like it is a whole. there will be ecosystem. That emerges out of this for sure.

Nicholas: Do you think that'll be in some place like, I don't know, some part of China where there's like a greater wealth density and people are willing to pay for a luxury autonomous flying short, short throw flights, I suppose?

Kosuke Hata: Yeah, I think there's like sort of initially opportunities to start in locations, whether it's in China or other places to kind of replace something that like a helicopter maybe or something. Like that, that, you know, you can start with and it sort of initially isn't really useful. And then, you know, over time, you know, a car when it was first built wasn't really you had, you know, I think it's easy to. it's not easy, but it's just like impossible or difficult to imagine a world where you have this like steam powered like quadricycle and no concept of a gas station. You know, like single digit percentage of paved roads in the entire country. You know, obstacles of people who don't know how a car work, don't know how to avoid a car, no traffic laws whatsoever. Horses are the norm. It's really difficult to understand how different of a world we lived in, like maybe even, you know, 120 years ago.

Nicholas: Yeah, absolutely. When we talked about this last time, you mentioned that you mentioned that data is the thing that's growing the fastest at these companies. Is that how can we understand that? And which kinds of companies? Which companies have this data are experiencing this growth in data that they're collecting? Yeah.

Kosuke Hata: So one of the things, you know, like a lot of the robotics companies that I've been a part of, you know, one of the fastest growing parts of these companies is really the data. And more often than not, it's, you know, if you're like a self-driving car, it's, you know, hours and hours of footage. You know, it's safe. You have like one camera generates, you know, I don't know, like an hour long ride generates like 10. You know, 20, 30 gigabytes per hour. You have six cameras around the vehicle. Multiply that over the duration of the entire day. And now you have like, like insane amount of video footage. Couple that with, you know, GPS data that's streaming the whole time. You know, LIDAR data that's also pretty dense that you're, you know, this generating gigabytes of data combined with all the other sort of information like, you know, accelerometer. Data or sort of like motor commands and also you just end up with a ton of, a ton of sensor data. And that's just sort of something that just grows a lot over, over the period of, of, uh, the lifetime of like these vehicles and they're in, so yeah, companies are just sitting on these things.

Nicholas: Essentially do companies like a Tesla collect, like what, do you have a sense of what percentage is collected by companies that do have the hardware in the, in the vehicles right now?

Kosuke Hata: So I think it's harder to say for Tesla, because I don't know, I don't know. Slash. I think they're uploading every piece of video footage, but I know for a fact for like some of the self-driving car companies, uh, yeah, it's like pretty much every ride is the data is being collected. Uh, and it's using all the data. So all that gets, you know, uh, it's, it's sort of like there's multiple use cases for it. So you, you, you throw it all into kind of the cloud. You maybe run like you have like an ML model that checks to see if they were like failures. Um, there's also sort of instances that you use that data. You know, it, it. Collects all. essentially, it's like, if you're collecting mapping data, you know, Tesla's making their own mapping models, uh, cruises making their own map. Like we're just doing the same stuff over and over again, because there's zero incentive to share this data. And even though we're kind of like all collecting this data, uh, versus each other.

Nicholas: Right, right, right. And I guess, I guess, uh, as compared to something like street view, uh, or other Apple maps or people who've maybe collected it previously, maybe there's more sensors or different sensors. Higher quality cameras, more angles, LIDAR, things that maybe, I suppose some of these things are in the original, like Google maps, uh, street view, but some of them probably are not for sure.

Kosuke Hata: Um, more often than not, it's usually time series data. So it's like time series data in the sense that, uh, you know, it's like every, there, there, there's sort of like data at certain frequencies, maybe like if it's like 50 Hertz or, um, I don't know, like controller data, you, you're essentially, you're collecting 50 data points over a single second. So, uh, it's much, much. Much more granular data than say something like you're collecting GPS data, maybe at like, I don't know, 10 Hertz or something. So you're 10 times a second or, you know, twice a second, or even once a second is still like it adds up.

Nicholas: And of course, as it relates to the actions, the behaviors of the, uh, autonomous vehicle. So it's not, it's not, uh, it's not just, uh, environmental data. It's it's data about the interaction between the model and the, and the world. Uh, that's crazy. So, so does this lead you to. Uh, have some reflection on, on the fact that the data is growing, but it's all done in a proprietary fashion. Does this somehow lead to, to your thoughts about Faust?

Kosuke Hata: Yeah. So initially we kind of started out saying like, oh, we're not gonna do hardware. We're not gonna do robotics. Um, we actually started more with building a, uh, kind of decentralized. Well, you know, first it was like, oh, well, it's hard to share the sensor data. Companies are making the sensor data. What if it's, there's sort of like a marketplace for buying and selling the sensor data? Would that make sense? Um, and that was sort of like the start. Like one of the first paths that we looked down and sort of started to realize, is there a way to kind of like decentralize and share the sensor data? And, um, so that was kind of the initial foray into why it, you know, it's like, if there was a way to prove, if there was a way to share that data, well, then what do you need to do? You need to prove somehow that that data is in fact, not like just generated data. How do you, how do you prove that? Like data in the real world was generated by these sensors. Someone should do that. Oh, how do you, how do you do that? And that, that, that kind of. Also started the whole path partially as well, because it's like, how do you know that the sensor data, data that's created on the internet is in fact, uh, kind of real or not. Right.

Nicholas: If you made a marketplace, how could you trust the things that you're buying data that you're buying? Right. So, uh, so I guess that's maybe a little bit of a, how you got interested also in EVM and ultimately Farcaster and things like this, or is there a connection there?

Kosuke Hata: Yeah, I think, you know, my, uh, initial sort of foray was just like, I read about, uh, you know, I've been interested in distributed systems and, and. Photography, um, but honestly, like coming more from the distributive side of things, I think robotics surprisingly uses a lot more distributed systems, um, than one would like to imagine. Um, and, uh, yeah, I, I read, I think the white paper and the yellow paper kind of at the time and look through it. And especially the sort of whole thesis around the world computer and how it was like really difficult to take this thing down. Um, it was just like, oh, this seems like this is going to happen, uh, in one form or another. Um, I. Think. I came across some friends who were mining Bitcoin out of milk crates, um, earlier than 2016, but I was like, not as convinced, um, on the sort of, I wasn't sort of like cypher punk pilled and, um, much more into the, the sovereignty side of things, but I was definitely into the world computer. And that was the part that, that, that sold me. and with Farcaster, I think something similar. I, I read Dan's tweet. I got. I started looking into Farcaster. I liked the idea that the client could die and that Twitter could, you know, Twitter, but with APIs, um, that were open seemed to make a lot of sense to me and wanted to give it a shot and DM Dan. And next thing, you know, uh, I'm casting.

Nicholas: Yeah, here we are. Um, so all of this kind of relates back to, um, to the idea of like, uh, developing some kind of open source or decentralized source of sensor data. Uh, and then you realize you have this need for notarization of the data or some way to verify that it's real kind of the, the world coin kind of problem, uh, but for sensor data rather than for individual identity.

Kosuke Hata: Exactly. And it's like, how do you kind of create a ground truth for, you know, and, and, and we kind of approach it, I think in this way of like, how does it make, uh, you know, I think the ultimate objective at the end of the day is it's like, I just really want to make more vehicles and machines. And I think there's a very unique and interesting opportunity. Uh, of sort of doing in that from the decentralization path in the long run, it's a very long run path. Um, it's not sort of an immediate thing that I think is going to happen, but I do think there's this aspect of. Vehicles machines. I, I, I personally think that crypto is perfect for, for robots and vehicles because they're sort of going to eventually have this like local model and AI, they're going to be running around. They're going to have an identity in the form of like essentially a wallet. Uh, they're going to be able to prove things to other things. They're going to be able to pay like all of this done through code. It, it just like makes sense in the long run. I think it's not something that I think everyone's going to be like, there's a market for this in, you know, three to five years, but it is like something that I think will, it's just like, it just makes sense to me that it, it would be a thing.

Nicholas: You were saying that, uh, you anticipate that blockchain transactions will be predominantly machine to machine transactions, uh, rather than human. I think. Yeah.

Kosuke Hata: And, and whether it's sort of like AI, that's, that's sort of on the cloud side or, or. More locally kind of in the real physical world as well. I think it, to me, it, it makes sense. It's something that there's nothing in, um, I, or it's just something that would makes that I could see happening.

Nicholas: Yeah. I mean, it's definitely the case inside of that future inside of a browser, inside of an app. It's it's the human initiated action leads to countless, uh, machine initiated, uh, transactions. So it's only natural that, or it seems natural that the same would be applied to, to blockchains. Um, so, uh. What I, I guess, can you paint a little bit more of a picture about what Faust is doing? Or I, I'm really actually curious about like kind of the prototyping process that you've been through and what you've gleaned through having different ideas and sort of updating your model of what Faust might be. Yeah.

Kosuke Hata: I, I think it's been, um, anything, if not a, just the process of figuring out, like, how do we kind of tackle that future where we think there'll be sort of these multiple vehicles moving around? Like, what is the starting point? I think it's been sort of like different. Investigation. Investigation into building different products that would, would start off, uh, like, what is the sort of wedge, um, we initially looked into, like, one of the first harbor things that we built was like, oh, we could tie a sensor to a smart contract and how we do that. We kind of do that through building, essentially build a smart, uh, not smart, um, uh, a hardware wallet that contains the keys that you deploy smart contract with, and you can tie that to like a sensor. And if you can tie a sensor to that device, we could do something interesting. So like one of the first kind of toy, uh, things that we built was like, uh, um, this hardware wallet, essentially kind of like an NFC that you can kind of like beam securely private keys to. Um, and so you could like load the private keys of like a wallet that contains a thousand dollar USDC or something. And then now you have essentially like a brick with like, I don't know, imagine a passcode or something that you can like give to other people. Like. Transaction was like, would that be sort of the way that we could start, you know, building, um, it's like sort of smart artifact thing. Maybe we can tie a GPS to it. I don't know. It's like, what if you could have a, um, thousand dollar USDC that was only usable in this, like GPS poly on, uh, is that a good idea? I don't know, but that's like, we could do that. That's interesting. Right. And so it's kind of like looking and thinking about the possibilities of what we could do with that. Um, we, then we. Sort of built a device that had more of like a Linux machine attached to it. Similar to kind of how we built previously built robots, but also had a hardware wallet. Um, and we realized that we could like build that, that, that could contain programs that could run on top of this environment that we built. Um, which, you know, could utilize this kind of notarizing aspect of kind of timestamping sensor data with this private key. Um, we looked into.

Nicholas: Uh, so in that, in that example, you'd be sensor would be sensor would be collecting data, hardware, assigning it with a private key. So you have a kind of, uh, harbor assigned data collection. Um, does that prototype or does that direction, I guess it still lacks application. There's it's, it's, it lacks a, it's not, it's not yet at the level that someone would be able to pick it up and do much more with it than think it's cool for, from what I understand.

Kosuke Hata: Exactly. Exactly. So, yeah, then we kind of went on the path. Um, okay. Can we use this for, uh, climate tech? And we started looking into like ways to, let's say, uh, notarize sensor data and how much kilowatt hours were being used for like large machines that were, uh, extracting carbon dioxide from the air. And so like. how much of that could, you know, if, if there was a way, you know, and then, and then we like went this whole down this whole path as well for a while. Like, I, I think we kind of go into this, this pace of. Um, figure out some markets or, you know, do a little prototyping, figure out the, the sort of market that we think would fit this thing, start doing some investigations into customers and see if they would be interested and then see if we can kind of adapt the solution slightly to fit their business needs. And if so, and then we kind of evaluate whether that's the company that we'd want to become or not.

Nicholas: We did this like a couple of times. Got it. As, as that process evolved, you realize. That it would be interesting to, I guess, is there a part of this that we can think, I, I know you've, you've sort of, uh, shown some images of the, the rock, um, how did that experimentation lead to the rock? And maybe you can explain what that is a little bit. So

Kosuke Hata: we kind of went through that process and, um, over time we, we kind of got a pull in, in one direction, which was camera, um, you know, essentially do, you know, sort of notarizing camera data and, and sort of, you know, if you can notarize camera data, you can take. You know, almost deep fake proof photos. You can kind of, and there's sort of interesting things that we got from people that we were talking about who wanted to build stuff on it. So we kind of packaged it together. Uh, we called it rock, which stands for reality, Oracle computer. And, uh, we kind of made a very basic version with some very basic software, honestly, that we wrote. And, uh, we kind of pitched folks on this idea of what, you know, what it is. And, you know, in a, in an. Early enough stage, go on and kind of take a stab of, of what you want to build on it and, um, essentially have people participate slash kind of buy into this program. Um, so we, we shipped about 10 units, uh, relatively early on, I thought, or about September ish last year. Um, and we had 10 units that we sent out and, and we, we sort of started that like kind of. Testing program.

Nicholas: So this is a device with some cameras that are taking notarized. Sensor data. So you can be guaranteed that these cameras at this location at this time, according to this private key, which you, I guess, choose to trust or not, um, took these, took these images and then that sensor data is fed into like a, I guess it's, it's just for the personal consumption at this point of that person, the owner of the, the device. Right.

Kosuke Hata: It's, it's, you know, primarily tied to the, you know, so in the future it would be, there'd be a little bit more kind of like. There there's further advancements and things that we could build, which would be like more building it, you know, hyperspace. We know we want to do it with this trusted execution environment. Um, we kind of prototype separately, this whole running within an embedded system that runs this real time operating system. We could build a trusted execution environment that sort of separated, uh, from, you know, we, we, we kind of built a separate prototype, but it was a little too complex to like ship it out to everyone at the time. That was like, okay, we'll just run it all in Linux and just call it a day so that we can just kind of test. We, we wanted to test the idea out more than, um. Um, spending months building something, uh, just so that we knew we were kind of on a path. Um, and yeah, it, you know, it sort of like, wasn't, you know, because it's sort of this Linux environment and we built this sort of meta OS on top of it. Um, people could sort of write programs for our environment and they'd be able to leverage the camera data, the signing, and also like the pushing to a smart, like calling a smart contract all from within this environment.

Nicholas: So altogether, what did these ingredients give us? If we. Sort of. Sort of squint and think about what they might look like in a more evolved form, what is the advantage of having notarized sensor data tied to a blockchain?

Kosuke Hata: Yeah, I think it's like right now, to be perfectly honest, it's cool. That's really it. Like the market is not quite there is my, my feeling. And if anything, it's, um, you know, there, there's sort of markets and directions that we can go into. that I think, um, are it's becoming clear. We either have to take a lot more of a. Um, also this is like ideas that I honestly have been thinking about since the last time we talked more, um, which I guess was like, not even that long ago, but I feel like a

Nicholas: couple of weeks, I think,

Kosuke Hata: um, kind of come to this point where it's like, it really, like, we need, we need a market and we need something that is really going to drive this forward. It is sort of either, either. We go further into kind of more of the real world assets, or we play the more deep end game and become. It becomes like a helium like network, um, or it's a sort of general purpose dev tool for notarizing. That's, that's kind of the third route that we initially went down and that really isn't working. We're not getting like good. It's not, that's not the use case for this device, you know? Um, so it's either like, we either have to commit to the RWA or more deep in more like help, you know, other smart contract folks who want to put already building protocols, essentially add more sensor data, kind of like more real world. Data to their thing. Um, but I think it has to be for something specific like that. And, um, in its current form, it's more of a novelty to begin with for sure. But there are other ways to kind of make it more specific that we're thinking of for sure.

Nicholas: But I know last time we talked, we talked a little bit about, um, the applications that might be useful for self-driving vehicles. Um, given that they currently navigate the world with like a, basically a POV amount of sensor data, but not much environmental data aside from maybe. Traffic conditions or something kind of rough like that. Um, but I suppose those applications are difficult to address quickly because, uh, you need somebody who's going to actually integrate it into an autonomous vehicle, which there aren't that many of, uh, and they're all pretty proprietary.

Kosuke Hata: So I think, um, we've definitely, we've been looking into that path a lot more recently. Um, so that's the sort of this newer path that we've been going down. Um, we have been thinking through of, so we, you know, we, we've spoken a little bit to like AV company. Folks at AV companies, and if anything, it's like AV companies are still in that kind of initial camp of there's no desire really to like decentralize any of their sensor data, because even if the incentives are high, they're a company, they're not like going to sell their data for pennies or they don't really need, you know, they're not going to sort of like bend around to, uh, you know, accommodate, um, us in, in a way, because what, you know, what is this device? Like, it's not, you don't know if it works or not. Yeah. Um, so then we, we kind of also started thinking about, well, can we also build a vehicle for it to kind of start to test things out and see if we, again, we're, we kind of come the, uh, the rest of the team is, is mostly like ex Kitty Hawk engineers. So we come kind of come more from this background of building vehicles. Like, can we just build the whole stack? Can we build this like ground truth network that sort of senses data on behalf of the vehicles? And that improves the, the quality of the information that's available for. For these vehicles. And is that a path? So that's, that's another sort of like direction that we've been. Been exploring as

Nicholas: well, but in that example, you can imagine that the sensors are the value of having notarized sensors tied to blockchain reputation is that you can then trust what the, I don't know, the sensor at the corner says about what kind of traffic is coming in a way that you might not, if it was just a decentralized, I don't know, something like a forecaster of, of data where there's no underlying connection to no hardware signed, notarized data. You might not be able to trust it as well. Um, so yeah, interesting. Anyway, sorry, go ahead. Yeah.

Kosuke Hata: So, um, I think we, you know, we we've, we've gone through and, and long story short is, is I think we were, you know, especially before we, we started kind of committing more to like the, this AV space or at least exploring the AV space. We have, I don't think we've fully committed yet, although we've spent a bunch of time is, um, we've been trying to think of like, how do we position ourselves to grow beyond, you know, if we, if we just build a novelty thing and we ship. It, it's cool, but it doesn't, you know, do anything specific. Um, so can we build it more for this AV path is, is like one area that we've been looking at. the other sort of, again, sort of the, the, um, it's like, I feel like, you know, I've, I've been, we've been going down to like a lot of different idea maze paths to see if like one of these, what are these things that are going to work? Another idea maze path that we went down, um, was more like, can we build it? Specifically more forward, uh, devs, can we build like one of the, the paths that we were going down was like, can we build kind of a, a camera device that's specifically for more for protocol devs and these protocol devs would make it so that they could write their own software on top of our device in a very easy way. And they could sort of like, they could add, you know, more data into their protocols. so that have like, they have notarized data, uh, you know, but really the end game of that. Is we help that protocol number go up and is that the sort of like, is that going to be our business model? You know, so it's like, is that the way that we're, you know, how many of those people around what's the market size for that? And thinking backwards, we're like, okay, well, that's not that big. We don't, we're not going to, unless we sort of like build our own network. And then if we build our own network, kind of like helium, then it's like, we sell that data. But with the data that we're selling or the service that we're selling is like what those devices and machines would, would sort of like, you know, do. And if that's the case, then we also have to incentivize people. So we would have to probably make some sort of token, um, though we're not like interested at this moment in like kind of going down that path as much. So, yeah, it's been kind of like trying to figure out which, which, which of these are the paths that is most suited, both for our team and for kind of generally this future that we want, which is, we just want to build a lot of these vehicles. And we think that they should all, they should be decentralized and that they should, they should be on chain. And so we've been figuring out how to get there.

Nicholas: Are there converging trend lines that are important to this, uh, area of interest, I guess, things like cost or changes in the supply chain over the last few years or advancements in AI, are there trend lines that kind of, uh, are coming together in the future in some point that inspire, uh, the types of ideas that you're thinking about when exploring the idea maze?

Kosuke Hata: Yeah, I think with the proliferation, uh, initially of smartphones in the 2010s and now more essentially, uh, robotics and, and more software interests of self-driving cars like sensors. Something like even LIDAR has become order, order of magnitude cheaper than it was a bunch of time ago. Um, the development costs of building hardware is also going down. There's also emergence obviously now with more embodied interest in embodied AI. So transferring, um, transformer architecture, um, type models to, uh, being used for embodied AI and stuff like Palm E. Uh, kind of renewed interest in not only being able, you know, so it's like models, especially like deep learning has been really commonly used. It was like one of the first things that happened with self-driving cars. People took the perception part of the stack. So if you're sort of building like a self-driving car, you need to see what's going on in the world. And then that information needs to be, uh, essentially compared with what you want to do or what you, where you want to go. Um, that's sort of fed into a controller that spits out essentially like. Uh, steering command, like a steering angle for your wheel and also like axle or break, you know, it's like vastly simplifying it. And then you do that thing in the world. And then the center has to like the sensor in this case, like the camera or the GPF and other things have to pick up what's going on in the world to their best guesstimate, which is really just a probabilistic model of the world. And then compare that with what you want to do, and then just do that over and over and over and over again. So, um, in that sense, it's like deep learning was really great for that perception part, like being. Being able to just guess what, you know, essentially taking a bunch of, you, you train a bunch of images, you make sure that it works, that, you know, it recognizes these kinds of images and, uh, and then you can like do stuff against that. But the rest of the stuff was hard coded. And now that sort of like LMS that, or not LMS, but, um, sort of transform, like architecture has become popular and will increase interest and research into it. I think it's also like a new kind of unlock that. And, and also just like 3d printing, these things become. Become cheaper. Um, there's a lot of like kind of conversions in that sense as well as like hardware gets better. And also, you know, the other part of it is like embedded, um, embedded meaning like low power GPUs are, you know, Nvidia, this is another thing that like a lot of people, I guess, maybe outside of the robotics industry don't know is like Nvidia make pretty much all of the embedded GPUs that are used in robots. Like in the world, there's this whole entire series called like Jetson chips. Uh, Nvidia has these, uh, Tegra chips that they're called Jetson boards, their system on chip. Modules that like, this is like the default standard for building a robot. So you look at probably like most of the stuff that's out there, whether it's from self-driving cars to humanoid robots, it's like everything is using an embedded GPU for inference on the edge. Uh, it's completely dominated by that.

Nicholas: And the, and the price is coming down.

Kosuke Hata: It's not coming down, but it's getting better. So it's like the, the tops per dollar is coming down, right? Trillions of operations per second per dollar is definitely coming down. Yeah. So you're, you're seeing. Like exponential increase.

Nicholas: So improvements in AI application of transformer type architectures, uh, that, you know, great, more, more compute in an embedded computer, uh, than previously, uh, for the same price. So essentially this we're. you're envisioning a world. I mean, it's, it seems like kind of obvious enough that, uh, the compute just becomes cheaper, smaller, faster, and the software exists to drive it. Do you foresee, uh, is it, or. Based on what your experience are actually building this stuff, how, um, tailor made must the autonomous agent operating a vehicle be to that embodied situation versus how long is it until we have a single chat GPT that can drive any vehicle?

Kosuke Hata: Um, I think like the models, I, I, well, the, the latter I think is like really, really far still. Like, I don't think it's going to be done with transformer architecture is the long story short for, for sure like that. I'm pretty confident. Um, I think there's, yeah, there's emerging sort of models that are good at certain types of things. Um, but it's, it's, it's yet to be seen whether that, uh, Trent, that sort of how good chat GPT is at generating language transfers across to different sort of domains or how well it does with multimodal. So I'm also not like an AI expert, so I'm don't, I don't know enough. To be able to like, definitively say these things. to be perfectly honest with you. I know in the context more of like robotics, but at least that's kind of the interesting parts that I see coming down the line.

Nicholas: Right. And I'm curious how, yeah. Have you figured out a way in which like the notarization of data really is important for training processes or, uh, in some way else relates directly to robotics? Like, obviously, I don't know. I suppose the maybe lamest version of it that I can imagine that's relatable is like. An iPhone detecting when you've replaced some component, uh, and it being aware that each component comes from the original, whatever is, is an Apple component or something like that. There, we already have some amounts of, uh, maybe it's a stretching the idea of notarization, but at least the authentic identity of components are in certain devices. Um, you know, the, the software is at least aware of whether the device components are authentic or not.

Kosuke Hata: I mean, I, I, again, I think the, the thing that is still really powerful in the long run. Is that I think there's, there's sort of two components of it. One is this idea that, uh, notarization is done automatically increasingly. So as these sort of sort of sensor, essentially sensor driven notarization, I think in the longer run versus sort of people clicking on a metamask that, you know, a camera would see an image that for the model that's running on the edge and it would see that that's happening. And then it would make that sort of notarization is there just seems like it would, that would, that would happen more. The other part is. The, I think the provenance aspect of it as well, where the chains of these notarizations create for more rich information about the provenance and history of certain data in their lineages. Uh, and being able to track those over time in a way. Um, and again, I think the consumer and creator for these are primarily going to be, uh, if anything to AI, like, I don't think it's going to be people as much. And the reason why is because there's just going to be way too much. Much data to traverse by human.

Nicholas: And do you think AI, or do you imagine that systems that propagate AI or the systems that the, the financial organization that retains, uh, and perpetuates an AI into the future would have a financial incentive to purchase notarized data, like live notarized data? for that reason, in the same way that I suppose Elon is monetizing Twitter because of the value of that data set that, uh, sort of trustworthy physical sensor data. Yeah. Might have a similar. Trajectory.

Kosuke Hata: That's kind of like at some point. that's the hope in, in, in many ways, though, I do think like just kind of taking a step back, there's sort of two types of data that we initially have been sort of been interested in. And, and, and one data is, uh, more of this kind of like long historical, like all, everything that has happened in the highest fidelity form possible data sets. This is kind of the, the type of data that we initially started with when we were looking at decentralized sensor data, where it's like every historical. Data combined. also, there's a, there's this whole other angle where the, um, you know, one of the, the successes of chat GPT is because there was tremendous amounts of language information on the internet already. And so you could crawl through Reddit. You could crawl through all this sort of web, you know, all these web pages on the internet to train a model because there was just so much data when it comes to a lot of robotics information, partially because a lot of these robotics companies are, uh, you know, kind of hoarding their data. But really it's because there's just not enough data period about robotics. It's really hard to still train. Large models against training data, because there just isn't a lot of it. So that's another thing that could potentially be interesting and useful. So that's, that's the one type of data that we initially started looking at with like high fidelity, long, you know, lots and lots of data. Um, the other side of things is more of this kind of, uh, what's going on around in the world, real time ish. It's sort of like there's a half. We've noticed there's a half life of data. It's almost like you want to know the telemetry data of everything that's going on around you. So if there's like a bunch of vehicles, what are they doing? Where are they heading? What's their location? You know, all this information is critical and sort of like a multi-agent autonomy environment. And that's more, that has a half life. So if you have data about where they were an hour ago, it's not really as useful as where they were 20 seconds ago. And if there, if you have data that was like five seconds ago, that's even more useful than 20 seconds ago. So there's kind of this very, it could be. Low fidelity. It's like, it doesn't have to be super high resolution, but it sort of gives you a sense enough about where the vehicles are and what they're doing. And that's also important in, in another way. So those are the two types of data that we were like specifically interested in as we were kind of approaching this, like multi-vehicle or multi-agent network, as we call it agent more in the physical fence, less than the, uh, the cloud.

Nicholas: Um, who of the companies out there today, or maybe other entities too, who has enough data to achieve self-driving? Do you think?

Kosuke Hata: Um, I think there's like. Interesting routes of until at least like recently, again, I'm not self-drive fully self-driving person, uh, at this point. Uh, so it's like hard to fit, like a lot changes, as you know, you know, there wasn't even chat GPT and well, there, there wasn't like a, uh, it wasn't popular until like the late 2010s. So there were a lot more software and car companies that started before even, uh, attention is all you need was published in 2017, I think. So it's hard to say exactly. But like with, I think transformers kind of makes it so that it's a data problem to a certain extent, right? Like if you flood the, you know, if you make the training size, the data set big enough, you will be able to kind of all these like skills will emerge is the idea. But the amount of data that you need, does it sort of like exceed the amount of data that exists in the world? I don't know, um, exactly. I'm sure like other people may know that answer to that. Um, there are. Also kind of problems that are unique to self-driving, um, that I think some of the companies like wave is another one that's emerged recently that are taking a different approach to solving. So, so there's like autonomous car companies that are building more kind of like robo taxis, uh, specifically for, um, you know, communities like, um, what's it called? Uh, like cruise and Waymo and Luke's. Luke's is owned by Amazon. And so they're, they're like trying to serve. A consumer at the end of the day, whereas companies like wave, um, they're, they are really focusing much more on, uh, kind of building a found, you know, like a foundational model or a model that could eventually be supplied to even like car manufacturers to use it for their vehicles in a way. It's like a general purpose model that could then be installed or integrated. So it's like the pitch to car companies would be something like they would save time on not having to hire and catch up. In terms of R and D to build an autonomous vehicle. That's like level three, level four and beyond. I think, you know, Waymo and cruise are just approaching a straight on it, like level four ish. Um, whereas Zoox is just going straight to level five. and level five is like no steering wheel. Level four is like a steering wheel if you need. And level three is more like Tesla where it's stealing steering wheel by default, but you can turn on autonomy.

Nicholas: I guess also you can imagine something like level five in San Francisco without. Achieving level five in Pakistan or something. There maybe are different geographies where the regularity of the experience is such that it is a kind of constrained enough demo. Of course, there's going to be situations where even they get into trouble and unforeseen situations, but where there's clean, like typically where you have road markings, et cetera, where people generally follow some set of, uh, like rules rather than, I don't know, off road into the mountains somewhere. Perhaps that's a different, uh, set of problems. Or when you think of like the levels of autonomy is level. Level five, truly it can, it can drive anywhere.

Kosuke Hata: Oh, I, I think it's more, it's more like not having a steering wheel and being picked up and, you know, it's sort of like changing what a car really is, you know, people there's like ownership changes as well. Uh, I think it's not like, it's like completely not owning a car. It's more, there would only be, you know, ride tech, ride sharing and especially in city areas. Um, and you would be picked up and brought to a certain location and, but the cost is so, you know. Cost effective that you would choose to do so, or that would replace cars, at least as a sort of idea. I think that's more like kind of the promise of level five. And you start to see that as well with a lot of car companies like Toyota that are sort of positioning themselves more towards mobility, um, as kind of the overall thing that they want to do more so than just sort of like building a car. Because I think they're missed the boat slightly on the whole EV thing. And they're kind of trying to leapfrog into level four and level five for the future. Where, especially since. I also have this random obsession with, uh, Chinese EV companies and, uh, how, like, there's just like hundreds of, uh, Chinese EV companies that I never knew until about like a couple months ago. And there's a whole nother interesting. Yeah.

Nicholas: Tell me about it. What's what's going on. Yeah.

Kosuke Hata: Uh, so there's like this whole plan that China has of, um, trying to have like 50% of, I think cars by 2035 are all going to be EV. Um, and also that they're going to have like smart cities and this is sort of their larger, like. Yeah. Plan to both offset carbon dioxide, as well as, um, build out an infrastructure, um, that's suited for autonomous vehicles. So part of their step one plan was to flood the market and become kind of a leader in EVs, um, you know, leapfrog combustible vehicles and become the leader in building EVs. So they had these like huge incentives, uh, to incentivize car manufacturers in China to like build a bunch of EVs. So a bunch of people went gang gangbusters and just. Just build out just like there's, if you, if you Google like Chinese EV car graveyard, there's like just hordes, you know, I don't, I don't know if you're familiar with like the EV, uh, the, um, the bike sharing thing, that craze that went on for a while. And like in China, they were just like way more bikes built than there needed to be. And then there were these like mountains of, uh, bikes, same thing with EVs. It turns out there's like graveyards of just like hundreds of thousands of EVs that have been built nowhere to go. And they're like sitting in a field, um, but there's been, there's like over a hundred brands of EV cars, smartphone manufacturers are trying to get into EV sales partially because they're also looking at this as an opportunity to get into the EV market. Uh, through like where, you know, Tesla is through this like model, you know, they know software, they know autonomous, they can get more into the autonomy side of things. They have the talent to like, they have software engineers. So they're building like the, the Tesla to get, to compete with like a luxury brand. Like Tesla in China. Um, then there's of course, like the pure, pure, like software companies like Baidu, which is like the Google of China that are, that have been trying to build like an AB, like an autonomous vehicle. That's level four, level five for four years and try to kind of compete with Waymo, um, and cruise. So there, there's like this whole, anyway, there's like a whole world of EVs that we've never heard of. And we don't get to have, uh, for a variety of different reasons.

Nicholas: Right, right, right. I wonder if they will, I mean, presumably they will apply. Pressure to, uh, the price of a Tesla, uh, or other EVs that are available, right? There's very low cost Chinese EVs right now.

Kosuke Hata: Uh, so, so one more thing about the, the, the, the sort of plan overall was that, um, the, the first step was building EVs for China. Then the second sort of phase is like building, uh, kind of, you know, more autonomy built into the cars. But the third step really that they're trying to do is to, to kind of create this, uh, you know, they're taking a much more top-down approach, but to create these smart cities that have sensors and vehicles that can communicate with each other. You know, solving this vehicle to everything problem, but a very top-down approach to the problem by saying, these are the companies that are going to do it. Everyone's going to adhere to these standards. We're going to make level five more possible by building a smart city from the top down. And also this is our path to reducing dependency on, on sort of coal and, and oil and, and sort of investing significantly more in, uh, electricity, uh, EV cars and, and sort of like building EV car related batteries and other things in infrastructure. Right. So there, there's this like whole commitment and you can, you know, you can search this stuff. Like they, they, they announced this like years ago and they like started this plan, which is kind of nuts.

Nicholas: It is nuts. It is nuts. I wonder to what extent it will affect the market for EVs outside of China. Um, presumably they will, they will sell outside of China as well. And it will make it much more affordable to have at least the basic EV, if not an autonomous, uh, top-down smart city.

Kosuke Hata: A hundred percent. So, um, yeah. And then the interesting part, I think, you know, in the context of what we're doing is it's like, it's very difficult for them. To imagine in the U S or even sort of like in some Western countries that, um, there's gonna be a top-down approach to saying, this is the sort of requirements for what you need do in a smart city and the infrastructure to build it. So. I think in the longer run, it has to be this more grassroots, the decentralized approach of like incentivizing people to do it to like essentially decentralized incentivizing people to make the city like significantly smarter. But in order to do that, you need to build these machines and devices that are inherently able to trust each other. And open up the network in a way that other sort of participants can enter in a permissionless way. And, and, you know, prove to each other what's sort of happening around them is real. And if that's gonna be the case, it just seems that the best sort of approach or solution to that is doing it on chain.

Nicholas: One other application that came to mind when we were talking about sort of sensors in the city, uh, and what models might sort of incentivize people to participate in them is, uh, the, like the ring cameras are another example of. Camera based sensors in the city that are, whether it really is a real market and where you do want to know, I suppose that the data, if it's borrowed from somebody else's device is actually authentic data. Um, have you considered, uh, anything along those lines?

Kosuke Hata: Totally. Yeah. Um, so one of the, another really big company, like, I feel like this is kind of becoming like a China conversation weirdly, but, um, I don't know if you're familiar with sense time. No.

Nicholas: Okay.

Kosuke Hata: Since time is like. Again, sort of ideologically kind of the opposite of crypto. It's the manufacturers of the company that build pretty much like the Skynet equivalent through cameras that's for the, for the government. So they have like, they, they're, they've built out the entire infrastructure that essentially takes camera footage from every, you know, street corner or whatever to essentially be able to detect and like gleam kind of like what's going on in every possible sort of like location. And it's just, you know, a very different approach than how maybe like a Western. More kind of democratic government. Or not democratic. Let's say like liberal with a capital L government would approach, uh, you know, people's rights and stuff like that. Um, in terms of how to detect or sense when things are going. But if you're sort of like approaching it from, Hey, we just want to keep the safety and we want to be able to detect what's going on in the city. This is like the largest company that's doing that in the world today. And within an environment that's, I would say much more kind of in a Western world, I think you'd want to do something like that, but you don't want to violate everyone's rights. All in one go. So you would want to have some sort of privacy, uh, you know, or I guess no one would really let you, not that you should, but like you want to kind of do that in a more, maybe you prove to other cameras that you want to do this almost in a more decentralized way. I think, I think you still want to do these things like, you know, but you want to do it in a way that preserves people's rights and prefers people's privacies and preserves that sort of fits significantly much more within the kind of ideals of what a liberal democracy is. And so, first of all, there's no way that it would happen at this top down approach in like somewhere like the U S. it just, I mean, can you imagine it would be a, it would be a total PR disaster and, and also just, uh, people, some people's worst nightmare. Yeah, absolutely.

Nicholas: I mean, I, what comes to mind is the Amazon ring, which doesn't ask anyone on the street for permission to continuously record what they're doing or your neighbors, et cetera, who get caught up in it. But, uh, nevertheless, it's pretty popular. I wonder if there's some application here. That's, um, sort of unbundles the hardware provider from the backend services such that it's just, uh, maybe it opens up to hardware manufacturers who are capable of making cameras, but aren't that qualified to create quality software to, uh, plug into some decentralized network that is a market for their hardware. Uh, but it does take some, some application development and convincing to figure out what exactly the consumer wants and, and then also what the, uh, what the OEMs might want. Yeah.

Kosuke Hata: And, and again, I think you see this, I think you, you, you see this as a response from kind of, you know, camera manufacturers and even Adobe, Canon, all these folks that are making like the C2, C2P standard, or I can't remember the exact name off the top of my head right now, but essentially it's like, you know, a sort of agreement, um, consortium of people coming together, companies coming together thing. This is the sort of standard that we should. Uh, adhere to when it comes to proving that. The content authenticity, like, is this image actually kind of taken by this device or not? And that, that's sort of something that started already and there's, there's tremendous interest around it. I think the difficulty is, um, it's, there's not a ton of momentum behind who is going to really need this, right? Like it's something that we all kind of generally know we need. We just don't know what it's sort of useful for quite yet, or, you know, everyone knows it's important. It's just. How is it important? You know, this is like a really weird problem.

Nicholas: Absolutely. I mean, it does seem in a previous episode, I talked to a team who did a hackathon project at an eighth global event called ZK microphone, where they were doing hardware signed, uh, audio recordings, and then using a ZK tech of their own brewing to allow you to then even edit the recording. And without revealing the, the deleted portions of the recording still verify that it was derived from, uh, a notarized, uh, or hardware signed original recording. Uh, which was pretty cool. And they talked about this C2PA, uh, standard as well, but mentioned that it was sort of very centralized and not fully kosher from a cryptographic perspective, at least. Um, but it does one other application. I was talking to a developer in the space, uh, Ellie day, who I think had the great idea. I mean, I, I think it's actually worked on it in previous iterations of projects, uh, sort of, uh, notarized cameras for an only fans, uh, application, you know, like where you want to. To know that the PR who, in what applications do we really want to know that it's a real person on the other side? Maybe use world coin applications, these kind of, uh, stable coin applications, but maybe also something where you actually do want to know that the photo came from a real human in a, in a post generative AI world. Uh, and then of course, you know, these like more journalistic, uh, applications where you can imagine, uh, some kind of web standard, uh, that allows you to verify the public key. Of the photo is the same public key as is always used by that photo journalist or something like that. Still holes in the holes. And even those solutions for, of course you can, the photographer could falsify the hardware key could potentially be removed from the device used to sign some alternate image or, uh, you know, there's different ways that the system could be messed with. But as a form of like an extension of a human's reputation or a human signature with a cryptographic signature tied to a device that the recording was created on, you can start to picture some of them. Although it's hard to see how any of them is like in itself, a really great business there. They seem like technologies that should. Merge, but maybe we'll be dependent on like, uh, the distribution that an existing business already in that space is able to sort of add that tech too. Um, but, uh, yeah, I'm not sure if ring ring cameras have the exact same problem, but it is interesting to think about.

Kosuke Hata: I think, I think the, the, you know, the, that's definitely something that we've been struggling with internally, uh, as a company and a product as well, which is the fact that like, what is the, what is the market beyond the technology? So the technology is interesting. Yeah. I don't know that it will be valuable in the long run, but sort of what is the immediate use cases that, you know, if you, if you think of even sort of like a ring product, it still solves the problem of like being a doorbell or, you know, letting you know who is outside your house. You know, it's sort of solving specific and immediate problems that are going to be aiding it without, um, maybe touching in on the larger vision of, you know, sort of being kind of a watchdog for your neighborhood, or now you have a collection of all these rings in the world. Yeah. And, and they can, uh, you know, inform each other about what's going on. There's a very sort of immediate problem that they're solving that allows them regardless of centralized or decentralized, you know, for a moment here, uh, that they're, they're solving and they're able to kind of scale up to that. And especially coming from the hardware side. And I'm kind of realizing this as I, as I, as I talk about this with more people is, um, when it comes to hardware, we, we need, you know, if we're not going to do a token, for example, or we're not going to sort of like build a network. Based on, on, um, doing something with tokens, whether it's not, you know, even if it's sort of not our token, but we take a cut of a token or, or some sort of thing where it's just, you know, we're building hardware. That's kind of what we've been sort of focusing on. At least, um, we have to kind of scale up the production. Like we have to build a lot of these things and, and these things have to be sort of inherently useful or valuable in order for us to take a cut of the revenue that, um, we're getting. So. If that's the case, it has to sort of either be really built to scale, or we have to kind of like focus on some specific use case that makes it valuable for businesses or maybe even government in the long run, or, or something that allows us to start. Because I think without that, you can have these cool things that we know they're going to be valuable, but they just, uh, don't quite pay the bills. As far as sort of like the product itself.

Nicholas: Yeah. It's not clear that like, you can know that some technological, uh, convergence point will happen in the future. And. And maybe even take a guess at when it will be, but that doesn't mean that any old person could, uh, capture the opportunity. It may be more available to people who are, uh, more situated already in a relationship with the end user or whatever, whoever the consumer might be. the business opportunity. I, to me, world coin is one of the most interesting ones because they perceive it to me. It's a very, whether or not the politics of the actual organization, just as a gesture in business. It's. Uh, fascinating to predict. With confidence that some problem will arise and there will need to be a solution. So let's reverse engineer from then what steps we might take to arrive at the solution that will solve the problem. And actually thinking about it in terms of the things that Faust is looking at, it does seem to me like inevitable that cities are blanketed in cameras, uh, that are constantly collecting data and providing it to entities that have a legal claim to it. Perhaps. Uh, and also entities that are willing to pay for it. Uh, so you could imagine, uh, kind of ring network, uh, that is a decentralized network where you are financially incentivized to provide your data, but the data is not maybe directly, you know, it's not a public API. Instead, you have to pay the network in order to gain access or something like this. I, I'm sure there's lots of problems to be dealt with around privacy, et cetera, but, uh, it does seem inevitable that in 200 years, uh, develops.

Kosuke Hata: Yeah. Yeah. I was going to say decentralized sense time. Like I, I literally think about sense time, uh, in that sense. Yeah.

Nicholas: It makes sense to me. Like, why should helium be restricted to just serving 5g or 4g or whatever it is, uh, or the prior ones that have done the same with wifi? Surely they should be read, right. And, and maybe actually in a way it is a little bit like, uh, it is a bit of a colored coin, an ASIC of a network to have something that only does 5g when you're distributing networks of nodes that are internet connected and have decent uptime. And can potentially have trustworthy sensor data as well as, uh, broadcast. Um, it seems like that should be a more generic network, not something that's, uh, stuck with a particular set of, uh, input and output. Yeah.

Kosuke Hata: So, um, I, I, I agree. I think like the world coin approach is a very interesting sort of like alternative approach to approaching this less from kind of the like traditional hardware route side of things and thinking of, you know, leaning fully into. Essentially tying the, you know, I, I think they, they've pretty much made the hardware costs negligible and they're sort of building to me. Uh, they are a token company. They've created essentially a value out of notarizing people's irises. And that value is essentially tied to those tokens. And so fundamentally now it's like they're under the guise, obviously of UBI and all these things. But fundamentally it's the fact that they've been able to notarize people's irises. Regardless of the story is the thing that's sort of like iris back tokens.

Nicholas: That's a good point. I'm very curious if they will suffer the consequences of the audiences that they chose to onboard first. Um, me, I, I, I don't actually have numbers on this, but the anecdotal, uh, which could easily be very inaccurate, but the, the, the anecdotal retelling of it is that there are many people who are. Giving up their accounts. Immediately in order to get cash in their local currency or something more useful to them because they don't perceive that the value. And so you can imagine if you onboarded, if everybody onboarded to a crypto wallet. That was biometrically linked without understanding why they would have any use for crypto. Is it really a useful set of humans? If many of them have potentially sold their souls, uh, like within minutes of, of registering, uh, not to mention the fact that those audiences. Well, anyway, it's, it's a global audience, but. Not a crypto native audience, particularly not that it has to be crypto native, but do you, if you erode the quality, it's like, if you advertise an airdrop. Uh, as the first thing you do with your company, uh, uh, usually, usually that has negative consequences for the long-term trajectory of the company. So I'm curious. or the project. So I'm curious if they managed to survive that.

Kosuke Hata: Yeah. I think it's like the biggest airdrop. You know, for, I mean, it, it, you know, it's, I guess there's a sense of irony about it. That it's, it's in a way more Faustian than Faust itself. You're literally giving your, your eyeball away for tokens. Um, I can't imagine a better, better sort of story for something like that.

Nicholas: Well, I can't, what I can't understand is why they chose to make it look so creepy. Well, there's no need for it to look so creepy. Uh, I was just gonna say, why don't they just make it look like an innocuous object and do it like, uh, you know, at the checkout of a grocery store, every country in the world has grocery stores. At the checkout of the grocery store, there's always, or in some places at the more industrial scale, uh, commercialized groceries. There are credit card signup schemes where they give you 50 bucks. If you sign up for a credit card, they could be doing the exact same thing. The device does not need to be so prominent. Uh, so it's very funny that they chose to do that. But yeah, anyway, we'll see. What were you going to say?

Kosuke Hata: Yeah. And again, there's a lot of interesting and cool technology that the bill and a lot of the things that they've built is, um, you know, I, I think is. Valuable. And, you know, honestly, I've learned a lot from what they've done as well. And it's sort of really interesting use case. Um, I think the, the different path that, that, you know, sort of we're taking is, is betting much more, um, kind of been thinking more about it as like, you know, being fully, uh, on chain first with the sort of incomplete. Kind of stack of hardware that we're building or just sort of camera only instead of maybe like a full vehicle or robot. Versus maybe starting more with like building a full vehicle and then, you know, sort of including more of the on chain part as it becomes more mature. And, you know, we're certain about how to do it. So I, I do think time will tell how it will play out in terms of this, you know, air dropping to the entire world, $50 or however much it is, how that plays out. But if it works then, then great. Um, but I do think there's like the likelihood that it doesn't work as high because you just don't know what idea maze is going to work out in terms of these. No, I still think brilliantly. Right. In, in, in, in the early innings of figuring out, you know, incentivization mechanisms. I mean, you just see this all the time with like airdrops and tokens, you just give them out in different ways and allocate them. And, um, it just doesn't quite last beyond a certain point because I mean, you're literally giving away tokens and the mechanism of that is, I think like people are still figuring out how all this works.

Nicholas: For sure. I do think there are some that have, you know, I don't know, Ethereum, et cetera, Bitcoin. I mean, not give, I guess even Bitcoin. coin you could say it's not an airdrop but uh the coinbase transactions are you know? i think there are. there are systems in the wild now that are working and will last. uh but i suppose there's a large part of the market that recognizes that there's a shorter term. but uh there you can extract. you know you can achieve greater leverage by just dropping the token skipping the utility altogether. there's a kind of gradient of how much justification labor people are willing to do into crafting a narrative to uh create some kind of token. and then at the the pinnacle there are some that are actually genuine technological uh technologically interesting innovative projects that are trying to get to mass adoption through token drops. it does make me think i don't know sort of i have this image of like a almost pokemon go or gardening game where you're installing cameras in order to uh you know enrich the data for a certain hexagon of space. it does seem like i can actually imagine. that being i don't understand exactly what the model would be around accessing it. uh. but you can't imagine people happily decking up the token out their houses and any kind of property they have access to legally put cameras and doing so in order to collect tokens and participate in a network.

Kosuke Hata: closest one that comes to mind is like hive mapper is one of the deep end companies. that comes to mind. demo is another one. that comes to mind. hive mapper specifically you put dash cameras onto uh your car and then you drive around and collect almost like google street view data to kind of create a map uh with and you earn tokens based on that. and you know i think the whole thing is kind of like. a kind of like a kind of whole thesis is that there are people who will buy this map data and that will be valuable in the long run. um but also again still kind of yet yet to be seen if that's going to work or not. i'm pretty sure they're you know supply side data wise there's people who are driving and doing this. well um driving and earning money i presume. um you know. but but then again it's also like once you start doing a kind of token thing it's more like what is how do you do it? what's of it. I mean, Helium for a while had their own blockchain, had both a token that incentivized people to use Helium as well as to pay for the data. There's been multiple models and I still don't quite think anyone's quite cracked it. yet as far as like, oh, this is the path forward to kind of create these. I think Deepin is also in its very, very early stages and it's not quite a big narrative quite yet. It's been mostly speculative as well. Outside of, I think the closest one is only Helium, but Helium has taken eight, nine years to build a mobile virtual network operator carrier on top of their sort of like network of Helium devices, which I mean, again, is like an amazing feat, but it's also like, it hasn't been really clear how to kind of do that yet. Right.

Nicholas: And then ultimately it still has to be better than the alternative, centralized alternative in some dimension for users to actually care about it. Exactly. I was just going to ask a very, a question in a very different direction. Um, I enjoyed your essay on your website about Sorastro's Aria, uh, that you wrote, I guess, earlier this year. Um, and in it, you tell the story of these characters, uh, who are on an adventure and discover that there's more to what they've been told as evil than they originally understood. And ultimately in the essay, you focus on the kind of encouraging push that the characters receive to travel forth into the unknown, even though they can't quite know what the, what they're risking or what the, what the value is of what they bring to the world. Um, and you note that we need this kind of push because we don't know the repercussions of our actions or the value of what we choose to do today in the longterm, uh, certainly beyond our lifetimes, how, how our actions might be reconstrued or have impacts we didn't foresee, uh, or no impact at all. Um, and it reminds me a lot of, uh, some of the sci-fi stuff I've been reading lately, which often involves, uh, hibernation and time dilation, uh, where the consequences of one's actions today, when you set off into the world might be, or you might set off into the world, with one sort of worldview today, and then arrive in a place that's radically different from the world that you came from. Uh, and so, uh, I think this is an interesting perspective to bring to building technology. That's a little bit higher level than what a lot of the, in the weeds, uh, issues people are thinking about. So I'm curious, um, yeah, maybe can you, can you reflect at all on the content of that essay and how it affects, uh, what motivates you or how you're choosing what you'd like to work on beyond what might just be good. This choice, but also what might be worth your time.

Kosuke Hata: Yeah. I, I feel like, um, I could talk about that honestly for, for a while. I think that's a very like, uh, you know, if we want to go in this direction, I'm more than happy to talk about kind of the meta process. I think of, well, first of all, I've been thinking a lot about how building companies is really dumb mostly because it's so painful, the process of creating something like, why would you do this? This is like pretty stupid to do. Right. So if you really do something like this, you're really not only risking yourself, your reputation, all this other stuff, whatever, but really you're, you're, um, you know, putting yourself in a vulnerable position. and you're doing this mostly because you think that you have something to express or something to do in the world that's worth doing either for yourself or others for a larger hole, something that's bigger than yourself, however you might put it. Um, good point. I think if it's harder to deny doing that, you should do it, but I feel like there's also different kinds of companies to make. I mean, you could make a agency, you could make a pop, you know, more lifestyle business. You could make a incremental innovation company that solves a specific problem with that's maybe like a B2B SaaS in some specific area and very profitable, low key business that does really, really great. Um, and you can also kind of like take radical jumps into, you know, like, you know, like, you know, like, you know, like, complete unknowns. Uh, and I think a lot of often, like for me personally, I think building companies I've realized has been a exploration of sort of, it's been more self-discovery than anything else. It's a process of like learning what I want to do with my life, who I want to become and what I really want to be doing. If I didn't really care about what other people think and, uh, what other people think that I should be doing or all this other stuff, you just kind of commit to that. And, uh, you know, sometimes I don't even like what I hear back in many ways because it's hard to do that thing that you actually want to do, but you just kind of, yeah, I think, you know, one of the first things I remember when I left, uh, or after Kitty Hawk was like, I'm never, I'm not going to do hardware. I'm not going to do robotics. It's hard. It's a stupid thing to do. And here I am doing it because it's like something that actually do it. I genuinely am excited about it. And I think there was a moment, especially like, you know, where I was like, Hey, I think I'm going to do this through blockchain. And part of, you know, some of my friends were like, that's stupid. Like, why would you do this? Like, no one cares. I thought this was a scam. Like no one cares about this stuff. And it's like, I don't know. My gut says this is, there's something interesting here and it's worth spending my time. And I have to kind of like, we all have different paths and I have to trust my gut. So it's like just kind of following the path, regardless of what, what you think the best outcome is or what it is supposed to be, or what other people tell you, you should do is, uh, kind of. in some ways it's, it's difficult. I'd like to think it's rewarding in the long run, but I'm not even sure yet. So it's, it's, uh, it's kind of, you're just like an echo into the, into the void, but you just got to do it.

Nicholas: Yeah. I think one thing that resonates, I just finished reading the third installment of the three body problem series. And, uh, uh, with that series, by the way, great, great series. Um, I had read the first two a long time ago and then reread them and finally read the third one, which for anyone who's read, maybe just the first of the three, the third novel, especially at some point, the pace picks up in terms of how much speculative sci-fi and we get a lot less, I'm

Kosuke Hata: getting goosebumps just thinking about the third book.

Nicholas: Yeah. It's sort of, in a way it feels like he had enough material, for many, many more books. And then it was just like, all right, well forget the narrative. Let's just, let's just shove in as many mind bending ideas as possible, which I think is what draws drew. It's what drew me to the series in the first place. Uh, and, and you know, there are some good characters, but really I'm there for the, uh, in this particular series, I find the, uh, the, the sci-fi thinking is, is, is the stronger part. Um, and as people, I'm also reminded of like forever war, uh, another sci-fi series that where they're, uh, it's sort of an agalactic battle. and the travel, the tech, the technology for traveling, uh, accelerates humans to such speeds that by the time they arrive, hundreds of years have passed thousands of years, sometimes from at the point at their point from their point of origin. So as they travel and do different sorties to do battle or to travel, et cetera, the cultures are changing radically. And so the politics that one left with, uh, as a soldier, maybe have nothing to do with the political situation when one arrives. Uh, which I think can in some ways put into context, the sort of vertigo feeling of even today's geopolitics, where people have completely different perspectives of what's going on in the same situation. And they may even all have some legitimate, uh, you know, truth or truth is a bit of a big word, but there may be some legitimacy to various perspectives that are not, uh, coherent and cannot easily be synthesized, uh, into, into anything more than a theory about why the relations unfold that way. And I think, especially in crypto, as we look at things like, um, I see the contrast also in things like Farcaster where the team building Farcaster is building, uh, with like a 10 plus year horizon and are interested in long-term. I mean, if you consider 10 years, even long-term, I suppose it depends who you ask. Uh, but certainly compared to the majority of the crypto activity and even the activity that happens on Farcaster, which all of the application development, et cetera, is much shorter timescale thinking often even, weeks, days, uh, timescale, uh, trajectories for, for their projects. Um, so I think it's something we're all, all dealing with. And I, what I liked about the story in the blog post that you wrote is that it kind of reminds me a little bit of the third, uh, of the death's end book, the third installment of three body problem where the characters are, even though they're in a kind of, so occasionally they're in a dejected, depressed, uh, stuck staring into the middle distance, not knowing what to do with themselves anymore. Whereas the most horrific things they can imagine happen around them, um, and to them, but they also spontaneously in the way that, uh, novels characters do come up with their own, uh, new motivations in each situation they find themselves in and, uh, find themselves drawn to enact their morale, moral compass, their morality through some missions that they define for themselves or that the world imposes on them in some way. And that is what keeps them going forward. And with the compounding upending of the techno social surroundings that we live in that AI and et cetera, every new technology is bringing and compounding upon the changes of the last. Uh, I think it's necessary that we find this kind of motivation that you're alluding to in the blog post to feel capable of setting some kind of goals for ourselves and taking up the tasks that may not be fun all the time, um, in order to create meaning in our lives. And to find out what's out there through exploring rather than sitting still and, and being worried about it.

Kosuke Hata: I mean, I, I think of like in the context of without like spoiling too much, you know, I think there's elements in the book where it's like the characters feel if anything, that there is no future because of the circumstances and situations that are in place. And I kind of think of something that's akin to us, like, you know, maybe graduates that are going into computer science, thinking that because an LLM or chat GPT exists, like there is no more work to be done and losing hope that humanity doesn't have a future beyond, you know, there's no more code to create because everything is createable. And so why even do anything? And that, you know, in, in any ways it's, that couldn't be further from the truth that there's so much, you know, with, it's not to say that that's not the case. Like, I think most of the jobs that exist today will go away, but also kind of looking back in the past, you know, um, 90, like most of our waking hours, pre industrialization, pre, uh, fertilizers. We used to spend like all of our waking hours and all of our offsprings hours were spent towards making food because we couldn't feed enough. Energy for ourselves to survive. And we don't have any of those jobs anymore. So it's like, yes, everything is going to be reified, but it it's also kind of been, it's, how it's happening faster, but it's, it's been the case and we have to find ways to kind of. Overcome our. Own. own dispositions and grow through them. so yeah just kind of referring to that book i i think of the circumstances that the characters are put in is like it feels like there's no way out you know but it's that's. it's like saying that you think everything in the world has already been built and there's nothing else to build anymore because we have everything and that's just like that's. that's a limitation that you've kind of built on yourself and and you can you know some people may choose to just say that's that's the end. or and i think every era has this every time has this and in many ways but it's about kind of like venturing into the great beyond and having the. yeah it's like the courage to do so is what is a lot harder than the doing sometimes the technology and all the implementation. at the end of the day once you have the courage to do it i think you're like 90 of the way there. it's the courage to find within yourself or others or whatever it's like. that's the hard part.

Nicholas: i'm reminded of something you said about um offline. you told me about sort of uh distraction or trying too many things in the uh. flying vehicle space uh and like a lack of long-term conviction hurt many companies. uh made it difficult for them to achieve their visions. at least yet um and i think it's interesting to think that even billionaires you know must have conviction in order to achieve their ends. it's not. i wonder if part of the reason we like i just saw uh earlier today. um uh i didn't read it but i opened a tab on some. uh the end of software uh proclamation and maybe so but we maybe we struggle also because our sources of information are the same uh sources. for i mean if you can even think that we are able to collect any kind of raw information without a perspective our sources of knowledge uh which have the implicit bias of the authorship and the intended audience. uh and and and and and and and and and and and and and and uh are the same places that we get what we think of as information and so it's difficult to separate hearing about. you know the place that you go to hear about the latest developments in ai. probably twitter is also the same place where you get these kind of overwrought uh gladiatorial sometimes vc sometimes founder origin just proclamations about what is and what isn't. and they're very convincing of course because like comedians testing their material in uh you know comedy sellers uh you know comedy venues. uh these characters are the ones who are doing that exact work all day. uh. and so you know you can encounter things that are in the in the single scroll. you encounter completely opposite opinions that are both rather convincing and without any uh deep conviction or attachment to either outcome or some third thing. it can be difficult to not be swayed by them but at the end of the day uh most technologies that are popularized that i'm aware of are popularized with some idea of what their application will be. that ultimately is a minor note or not even relevant application of that technology and in the end some completely different use cases emerge that are native to the technology. and i feel we're struggling with that with crypto a lot because the native applications that are um the most successful in the eyes of end users are speculative ones and so it makes people wonder if there are any applications beneath that. and yet there's this whole community of people who are doing much deeper technology and they're not just research uh far from the application layer but who have trouble sort of rationalizing it in a way that makes sense to an end user. and uh it'll be interesting to see. i have no doubt that blockchain technology for example or the variety of crypto technology cryptographic technologies that have emerged from the financial engine that blockchain has put behind cryptography will completely refactor the world. but uh i think the big thing that i'm recognizing looking at a lot of this stuff is that there's no shortcuts to building really impactful things. and uh while we can productize things that are low-hanging fruit like existing ercs etc. and turn them into something and maybe that becomes a brand that can ultimately deliver a product that is more than the initial uh foray that established that community at the end of the day like a lot of these projects are really huge projects that need to be undertaken and they're not going to be finished in one day and there might be other people who are competing with you uh who maybe have a better team or more money and what have you? so that's why i come back to things like world coin because it's like that's a situation where you identify a problem or some converging trend lines far in the future and or open ai is another one. ironically both maybe not ironically both sam altman companies but where you you know something will happen or you believe it will happen and then it's so ambitious that very few people are actually willing to compete with you for years until you're so far ahead of the competition potentially that no one really can catch up with you. um and i think we're aware of that but i think many of us are. um either it's difficult to come up with what those inevitable things are and also it's scary to commit to a mission that might take 15 20 years to accomplish. it's a it's a long trajectory and you look at companies like apple that everybody looks up to and really there are a lot of people working on some very mundane things at apple in order to make that machine function and there's hundreds of people probably working on the airpods team. but in 100 years you know your grandchildren aren't going to care that much about airpods. you know they're not certainly not this generation. uh. and yet it takes it takes a lot of human labor human effort to make those projects continue to function to to to make sure that icloud keychain doesn't lose everybody's passwords to make sure that uh you know your photo library isn't leaked. there's a lot of not glamorous jobs that that must maintain once the uh this kind of empire is built.

Kosuke Hata: also yeah i think like just kind of touching on a few things that you know i've definitely with doing the best that i can with files to focus ourselves on something that's going to be very long term and you know something that's much sort of bigger. and yeah and i think as you said we've taken uh kind of an assumption that it's a no shortcut approach to building some of these things. and and if anything um you know i'm reminded of something like berkshire hathaway. and um you know um the way that they kind of approach think about investing is you know always i think throughout the the pretty much um history has been to bet on make essentially like one like small number of bets overall but just very very high conviction and not listening to sort

Nicholas: of the noise but to

Kosuke Hata: kind of like exactly you're just betting on one or two things that you think are really worth betting on. but you've done sort of research or thought through as much as you possibly can and i think honestly the similar. you know what is a? what is a uh startup in a? in a way it's it's a levered bet. you're you're especially venture-backed companies are if anything you're betting uh time and your livelihood but also like a certain amount of capital that this is going to be uh big in the future. but you know you have certain insights or certain things that you think are going to make this version of this work and with interest and also like outside people that get involved uh over a long period of time to like make a bet to like make these things an outcome. and you know really i was i think i was casting about this the other day but like you realistically don't you have like a few decades really? uh you know considering that we don't get hugely expanded uh lifespans in the next couple decades we don't have like you know at least what i feel like this generation doesn't have like a significant. you know we're not going to live to like 300 years. probably it's it's my guess like we're really going to get a few decades to make a difference. and how many like startups can you do in that? how many bets can you make? and you really can't make like that many bets. so if you really make a bet and you you make a high conviction bet you're. it's not about like other people's money or not. you know your money or your time or whatever. it's. it's really just that there's only certain like certain amounts of limited number of moves you can make on the board. uh so they better be worth it. essentially is kind of my feeling. it's like sure you mean you know there's side quests and other things to do and all these things. but but i do think startups are are sort of these like investment you know similar to investment decisions like you know warren buffett or and and other people off out of berkshire hathaway are making. is it's like? these are decisions. these are sort of like almost time investment decisions that you're making and in order to do that i do think you need to the ones that at least you know. in my mind i guess my sort of model is the ones that are successful are the ones that have been as thoughtful as they possibly can and have taken the slow approach to determine what the bet is. and then when they determine what that bet is they just strike the hardest. uh and fast. and you know that's all. mostly people see but people don't see the part where it's like you're. you're doing the research and you're sitting and thinking essentially uh as as kind of like. you know. um charlie bunger would say it's like warren just sits around reading books all day. you know that's all that's his day job you know. um it doesn't look like he's doing anything. but and then they just like make one decision and then they're like oh that's it you know yeah we're gonna buy this and we're gonna hold it. and also you know their investment decision was never we're gonna sell. they don't they don't want to figure out when to sell. so what do they do? they make really good bets that can be sold at any point later in time right because they are good investments and that's their thesis right. so it's like just very long-term thinking as far as i think you have to take that. but again that requires kind of whatever you want to call it conviction balls. you know it's sort of like a. it requires a certain amount of i think um i think it's monger called it like chutzpah. you know the to to say like. this is the thing and in order to do that you need this kind of time and sort of maturation of the of the idea to come together. in order to to essentially you need to be able to you know kind of tying it back to the the um the rest was already like. you need to be able to to have something or someone or yourself be able to tell you this is it this is worth doing? yeah and just make you never. you don't know obviously when you make that decision but you got to cross that rubicon.

Nicholas: yeah absolutely do you think that there's.

Kosuke Hata: it requires a ton of work.

Nicholas: yeah ton of work to figure it out. do you think that there's people who because it's it's possible to to be convinced that something will happen in the future and even maybe that one is you know that it's meaningful enough to your self to get involved in that to stay involved for a long time. but do you think that there's? it's it's. maybe i wonder is it another thing to be correct like? is there some other skill in being correct? it's it's one thing to bet one's life on something but uh i wonder it does seem like some people are very good at betting uh in this way. uh that they understand something about the game. that is not obvious. i've always felt this watching elon interviews. it's like but the one thing he won't tell us is what his actual secret is because and of course it's hard work etc. but this is the best the best delegator in the world. and yet he never talks about delegation. really he never really talks about what it means to manage 17 companies at the same time at like triple a level uh. or to be able to play social media on the knife's edge uh fighting with regulators etc. and pull it off. and i of course there's luck there's charisma there's accumulated knowledge through experience uh. and network i presumably i mean you know especially in these kind of celebrity ceos situations often a lot of the attention is focused on the ceo but the reality is that they're. maybe their greatest skill is being able to assemble a competent team behind them that doesn't need the spotlight and it's actually executing which i think is the case at almost all of his companies. um so i'm curious yeah is there something else that is even beyond choosing what one cares about most for oneself? is there some other kind of validation of if you're actually going to be able to make it that we should be? uh we should be attempting.

Kosuke Hata: i mean i think like you know elon is is definitely you know whatever people think of elon. um you know other folks i think in the past from you know folks like gates jobs all these folks that have come before at least in the technology field um and again like not just with technology just in general. i think it is this sort of combination of uh tremendous amount of like self-belief uh but also like i think part of it is you know i i think of something like napoleon or something and part of it is just uh what's kind of the the you know sort of i guess a weird and naive way of saying it is they're just kind of built different it's uh and part of it i think is just the you know i i think of like kind of. this is a sort of completely separate topic but i i often think about how like energy um so earth is like a closed system so like all the kind of most of the elements that are on earth. it kind of circulates right like carbon like water circulates into the air it rains and you know there's a hydrological cycle. most of the elements are are closed. the only open loop part is the open part is is the sunlight. so the the energy generated like everything is really downstream from the kind of like. as long as the sun lasts there's energy coming in and so all the growth all the like kind of cycling all of it is downstream from energy that's generated by the sun. uh of all the sort of everything that happens on earth is just kind of like generated by energy. and um i don't you know i i think it's like possible to also interpret this in a more kind of like lulu way as well where like people have a certain amount of energy uh and they're able to kind of like expand that energy as well. but really like these are just like it's an open system of like energy is really downstream from everything whether it's like how much energy in kilojoules that we can generate as well as energy that we can generate as a civilization. um economic output is very linked to how much electricity we generate. so one of the reasons why norway is a tremendously wealthy company is because of all the the the sort of unique geological nature of having fjords allowed them to create very inexpensive um uh hydro hydrological power like pretty early on and that allowed them to generate more energy and then be create a timber industry. and then you know. that's why they're like one of the wealthy they have like a million people and they're just extremely wealthy per capita because they have a ton of energy and and if anything it's like i just think of just some people you almost have more energy for for kind of like hours spent in a way for a variety of different reasons whether it's emotional or just intrinsic. i mean you look at like michael. jordan is another one. it's just sort of like an excess of energy that is way more there. i don't think there's a secret sauce. i think it's just more just more per not more redundant in terms of quantity but like quality and quantity. so some people it's more quantity some people it's more quality. um but it's. it's really this like insatiable amount of energy that just gets driven through these people that you just it just makes. most people don't really care to do like could talk about building companies like building companies is a stupid idea that requires a ton of energy for absolutely no reason for most people do you know based on people's risk tolerances and just lifespan in general. but you just kind of don't know what to do though.

Nicholas: yeah yeah i i i wonder we were talking about comedians.

Kosuke Hata: i don't know if that makes any sense but that's kind of what i think about for sure.

Nicholas: some people say that comedians all have some kind of trauma that drives them to try to be funny so much and i don't think that's actually true. but i do think there is a certain kind of founder um messiah that

Kosuke Hata: uh

Nicholas: the part of the culture that we're both familiar with is is invested in in finding and i think those people often do have something that they are trying to prove that is unprovable through work even. i mean uh i saw who's at 4156 talking about uh elon's sort of um ability to delay gratification uh or you know. but obviously his commitment is just to build he's trying to be everywhere that the technology and capitalism are converging and be there first and make the greatest mark uh and be the most ambitious. and it is competitive it's not. strictly i'll do things that no one else is willing to do. uh often it is i'll do things that others are very willing to do but i want my own version of it. uh so i think that that kind of commitment to working so much um and never being satisfied fully by the fruits of one's labor is a personality. uh i mean a trait of personality or a a world view that the person has adopted for some reason according to something that happened in their lives. some influence is that they only thought we literally have to give up something where we don't have to be productive forever. um i think to me it's interesting to think about why. because i think that includes both on the bottom subscribe and um you know thinking about black wide open situations. um you know i think i mean i.

Kosuke Hata: i agree with everything that you're saying you know. But I think we're also kind of like, yeah, there's. I'm more than happy to also just like talk about this more. I think we're kind of in the middle of like an evolution as well. And but I think for now, this is I feel like I've mostly talked about the main things that I would. I would like to chat with you. But I definitely can keep going for sure. And if anything, the story isn't done.

Nicholas: Yeah, for sure. If people are interested in the kind of technology you're working on, where's the best place to reach out or to learn more about what Faust is up to? Yeah.

Kosuke Hata: So we have on Farcaster or on Warpcast slash. Faust is the channel that we have. And if not, feel free to DM me. I'm at July on Farcaster. And otherwise, I think we'll keep folks posted through that. So you can always DM me or you can post stuff for the channel. But if not, I think it's more. We'll keep folks posted.

Nicholas: Amazing. July, thank you so much for coming on the show. It was great to talk to you about all these subjects, Faust, sensors and the meaning of life. It was a great conversation. Thanks so much. I really appreciate it.

Kosuke Hata: Thank you so much for taking the time. It's like really awesome to thank you for letting me talk. I know I can talk for a really long time about things that I have probably no right to talk about. But I can. I have opinions that are probably wildly wrong. So thanks for humoring me. And hopefully it's entertaining.

Nicholas: It was a pleasure. It was a pleasure. Hey, thanks for listening to this episode of Web3 Galaxy Brain. To keep up with everything Web3, follow me on Twitter at Nicholas with four leading ends. You can find links to the topics discussed on today's episode in the show notes. Podcast feed links are available at Web3GalaxyBrain.com. Web3 Galaxy Brain airs live most Friday afternoons at 5 p.m. Eastern Time, 2200 UTC on Twitter spaces. I look forward to seeing you there.