# Data as a New Asset Class

- Channel: [Vana](https://streameth.org/vana)
- Date: 2025-03-06
- Duration: 33:41
- Watch: https://streameth.org/watch/67c8fceedfe4387924fa39be

## Description

with Bryan Pellegrino, Casey Caruso and David Phelps. Moderated by Shann Wells

## Transcript

Brian, one of the founders and CEO of LayerZero. Hi. I'm David. I'm co-founder of a contest platform called Jokerate. I'm Casey. I'm the founder and general partner of a frontier tech fund called Topology. Great. And very early supporter and investor in VANA. I guess just to set the stage for the session, you know, why is data an asset class? You know, historically we've seen commodities like gold, oil, digital assets like Bitcoin become investable assets. What makes data a fit or not a fit into that paradigm? Yeah, I think the original thesis of VANA was just hold true. One thing I love about VANA is how early they were to this belief, even before most people saw it. And now with the rise and proliferation of AI and machine learning and specifically unsupervised machine learning, people are really starting to grok the value of data. But data powers our entire world. It is like the model in which our entire internet has been built on in terms of like all the services we use, which we think are free or not free. They're collecting data to make those products better. And I just want to double down on the fact that previously we used to think that data would have to be like clean and labeled and presented in a certain way. And while that still is somewhat true today because generative models have found such PMF in our world, all types of data is so valuable. And that's just increasing kind of like the not all data is fungible, but I think even some more messy data we're finding a ton of economic value in. I think there's kind of two ways maybe to approach this question like historically. You know, like oil to the 20th century is what data is to the 21st, right? And I think one is to think like the 20th century was really like a century of hardware and the 21st century is really a century of software. And I think that distinction often really helps me to think like even you think of the ways that people imagined the future 100 years ago. They imagined it totally in terms of hardware, right? And then the actual future we've gotten is all based on software, right? It's literally, you know, you have the brokest people in the world and the richest people in the world who are literally doing the same thing all day, which is staring at the computer screen and clicking around on it. And so I think like we're in this world where obviously we know data is so powerful in terms of how it's being collected and monetized by corporations and by states, right? And that's obvious. Like, you know, the way that Instagram has profited off our data is just gargantuan. And I think like the question is like, you know, is there a way that we can profit off that data? You know, is there a way that we can have more accessible, transparent, you know, rails for this in a way that actually can like return IP and, you know, royalties, et cetera? This is a very 2021 question. It's not very 2025 questions, I know, but I'm so excited by them. You know, I think the other way to think about it is like when you have that paradigm is actually, you know, data has always been the most valuable thing. Like you go back to the Roman Republic and like the thing that they are most concerned with, right, is data. Like in order to have an empire, you have to be able to standardize all the measurements in order to be able to collect and understand, you know, how much grain people are producing in order to get taxes on them, right? And that's a data question, right? The more that you can standardize this world, the more that you can collect and that seem like a state. I think it's just a really good kind of study of that, right, of the ways that states have always used data. And so again, I think the real question here is like, okay, we've seen the value this has had for states, corporations. Is there a way that this could actually be, you know, powerful for everyone? And this is something that could be fully, fully monetized through blockchain. So apologies for going full 2021 on everyone here. Yeah, I think we're pretty long past the question of is data valuable, right? Because data itself, you know, you're talking trillion dollars of market cap between Google and Meta that are just largely driven by their ability to serve ads through proprietary data that they hold. Apple ovens, $140 billion, $180 billion market cap right now, same thing, just the ability to have some set of proprietary data or tap into unique data. So it's like data is clearly itself is valuable. We see it valuable in terms of the ability to serve ads. You see it valuable when you run out of data in the AI sense and the ability to like reinforcement learning and create your own data, right? So there's no question around that. I think the real question in this space generally is, is there a significant value to privatizing and like democratizing data? And is that a direction that you can actually get incumbents to move in or force their hand? Awesome. I'm not going to start showing Vanna on how we do provide a network for your privatized user data. What market infrastructure is needed for data as an asset class? Yeah, what do we need for data financialization? What do we have today and what's still missing? I know you each have some specific perspectives on that, whether that's cross-chain composability or interoperability. We'd love to hear your thoughts. I think we've gotten really good at collecting data, like very, very good. You know, like especially now with the new automation tools, like it's just so, it's gotten so much easier. But then I think grading or quality control is still like the next era in my opinion. I think the same thing with synthetic data. Like we're really good at both collecting and generating. But then the question to me or like the challenge ahead is how to make sure that we're using it in the most efficient way and cleaning it. Which is kind of funny because I feel like we've been in this position for almost a decade now of just like how it all comes down to clean data. I think programmability is really kind of where I would lean on. Like right now data storage is really good and there's a lot of data out there. But the ability to like turn it into these kind of money Legos, except non-money Legos I guess is how we could think about it, is sort of an unsolved problem. So I think Vanna, Iris, like there's a lot of really exciting kind of entrants here who are like really thinking, okay, we figured out data storage, but can we figure out programmability? Like a way that anyone can then build on top of your data in a way that can be attributed fully, potentially even have royalties, all these sorts of things that, you know, have become these dreams of this space. I think the other one is privacy, right? And really just thinking especially with like DKPs, like are there ways that you can have private data that can still be publicly attested to? And so you can still build on it. You can still get credit for it. But it can be your own private data that, you know, is still fully decentralized. And I think this of course is one of the biggest, hardest problems. I think it's just funny that everyone's kind of given up on privacy now because it hasn't worked for 10 years when we're actually at the point now technologically where it's very, very close to working. And that always happens, right? There's a lot of hindsight bias that can, you know, be really detrimental. So I actually think like private data is really, really exciting as well. Yeah, I agree with Casey a lot here. Like we had spoken about this a bunch in the past. I think obviously the ability to collect data, store data is all easy. I think it becomes very difficult, especially now when data through LLMs and everything else is much easier to create an abundance of the ability to actually figure out, like you look at early GPT and it's just spitting out random Reddit answers that are just like completely unfounded. You know, this is like all of this stuff where you just can't take data at its base value. I think there's a ton of sub problems. And if you're taking privatized data and then trying to train a large model on top of that, how to attribute like what the actual outputs are or how the quality of that data was in a way that maintains privacy but still makes it so it is, you know, some form where more valuable data is better. There's a ton of sub problems to solve. But the fact of like getting the data itself, normalizing the data, using multimodal data, like all this stuff are largely solved problems in a way. Yeah. Can I ask a follow-up question, David? Okay, I just want to talk about the verifiability. Yes, yes, yes. Okay, let's do it. So what, you know, you're saying like the tech is here now. Like can you just explain that a little bit more? And then, you know, okay, we'll stay there. For the privacy part? Yeah, like you were saying, like, and I kind of, I largely agree with you, but just since we're here to make this by day, like tell me why you think like the technology is ready for privacy. Yeah, I think there's been really great innovation even in like the past few months, right? Like you look at what like, take an example, like Polygon Miden is doing with client-side execution. Like that's really exciting. What it means is like that you can actually be executing locally, like on your own device, right? Generate a ZKP, that's attested to on a blockchain. And so what that opens up is like, yeah, you can do stuff like private OTC, right? So OTC or centralized exchanges, those are currently actually like the bulk of crypto and they're fully centralized and they're not on blockchain rails, right? And you can start to think about, okay, that's the kind of stuff could be brought over. But it also opens up this design space where you can think about stuff like intense, right? It's like, if I put out and say, oh, I want someone to meet the following criteria for my transaction, right? And then someone else can come along, meet it, fulfill it, that can all be, you know, private as well. So I think it's like, you know, that's the kind of stuff that I'm looking at, I'm quite excited by. In terms of like, is there a full, fully decentralized like ZKP data rail out there? No, definitely not. And like, you know, we're closer, but we're not there yet. You're supposed to fight me. Well, I think I'm just one perspective. Like, I think, I think there's like two things that are going to happen with privacy. I think like either edge compute really starts to work because quantified, quantified models really work. And then like the need for crypto, I actually think is greatly diminished in that case. We're like, if Olama becomes, it depends. Well, like, yeah, basically it's like O3 mini really starts to work and can finally, you know, have like a better scorecard than it does today. And like evals across the board, then that becomes really interesting. And then I think the other option is what you're saying, which is like, maybe they don't, the models stay too heavy. And so we do need to do things across internet and then we need something like ZKPs to preserve the data, but also pass it along between hardware. Do you agree with that? Yeah, no, I think that's a great way to look at it. Cool. Sticking on data financialization for a minute. What, you know, assuming that we can have private programmable data, assuming we can clean it and it's of high quality, what needs to happen to make it, to make data a liquid tradable asset? And, you know, what are some early stage examples of data being traded, valued, or leveraged as an asset? And Casey, because I can't, because I'm moderating, if you could talk about data DAOs. Oh, yeah. Okay, I'll start there. Okay, yeah, data DAOs are by far the best example of data being tradable. For people that aren't familiar with data DAOs, that's kind of like the underpinning of how VANA works. And people are incentivized to go out and either leverage technology or even people to get that data and then can pool it, collect it, and then create community-owned data pools that then can be sold or done. There's a variety of things you can do with them. And there's like, even somebody, I don't know if he's in the audience, but he's like collecting Tesla data and selling it to enterprises right now. Hopefully I'm okay to share that. Yeah, he'll be on the next panel. Okay, I'll be on the next panel. Okay, I'm like, are these secrets? Yeah, and so those are data DAOs. I think those are a really good example. I think in the Web 2.0 world, it's like Reddit selling their data for like, I think it was $60 million back in 2024 was an interesting case study of how our data is being monetized without us. But, yeah. Is that a good overview? Yeah. Okay. We got Reddit coin though, so. Yeah, meme coins are the answer here. I mean, I don't know that I have a good answer. I think this is a really, really hard problem on a lot of fronts. Like, one is that the way we've built kind of non-financial blockchains is to get rid of financial standardization, right? You don't have to like update state every second to say, here's all the account balances. And so that allows, right, for people to store video or images or, you know, in different ways, but then there's no standardization across the board for what that looks like. So that's one problem. I think the bigger one is like, you know, historically, the way you make money on data is by privatizing it. You have private data and someone wants access to it and they have to pay to get access to that. And so when you talk about building transparent data, you know, you're really undercutting every business model of the past 20 years. And, you know, that's really exciting in terms of, you know, traditional disruption in technology that you can undercut previous technologies to come up with a cheaper and more scalable alternative. But it does raise the question, well, how do you monetize that new thing, right, when it's out there? I think a lot of this will come back to kind of social consensus. Like, it will come back to just, like, blockchains themselves are a social consensus mechanism for people to give value to something. And that is fundamentally at its core a meme coin. Like, Bitcoin is a meme coin, right? And I think we'll tentatively find our way out there. But until we have privacy, it's very, very hard, I think, to build business models. I'd love to hear everyone else's answers for how you do this. Yeah, I mean, in terms of the actual infrastructure, again, the data itself is the easy part. It is the ability to claim and aggregate the data and have it be usable in a way that is not then immediately stolen, right? Like, that's basically what it is. If we're publicly putting all of this data on a blockchain, then the data immediately loses all of its value the moment after that, because someone's just going to fork it or take that data and use it without paying, the same way we saw NFTs with royalties early on just immediately trend to zero. It's very hard to enforce these mechanisms in a way that's just, like, totally open, unless you have a pretty restrictive platform and social consensus around, like, that being the way to access this thing. So I think privacy is a huge element of it. But again, you run into this attribution problem. And so I think when you're talking about actually what does it look like to use data, one, again, most of the – I think it's – there's, like, elements of deep in here, right? Like, why is deep in interesting, deep in interesting to either collect data in a way that is more robust across a broad series of nodes and people than maybe traditional measures is or provide some sort of infrastructure? Like, you have to think about what data is actually easier to get from a broad set of people, like maybe weather data, maybe certain pieces of data. But, like, if you're talking about proprietary data for ad stuff, it's, like, it's very different. Google has every action you do. You have to have a – every user is a huge subset of users willing to, like, opt in to give their data in a way that they can trust to then provide it to other people to be able to use. And so it just feels like it's hard to compete in things. Bottoms up oftentimes from a crowd of disorganized parties. That's the challenge. That's what you have to overcome. I'd actually like to add a bit, I think. Yeah, I think, like, there's kind of two ways to think about this question. One is, like, this specific piece of data itself is valuable, right? Like, this specific data about the President of the United States is really valuable. The other is just data as a commodity, right? And, like, how do we store it? And it's probably important to differentiate those two questions from each other. Like, I think, you know, we hear all the time with, you know, data availability, for example, that this is totally a commodity. And that's true if you have, like, one megabyte. Like, one megabyte per second data availability, like, anyone can provide that. But if you want to go to, like, one terabyte per second, there's very, very few places that you can go. But they can actually provide that. And once, you know, a place like Celestia or ID&DA, they can offer that, like, actually has a stronghold over the space because they're the only ones that you can actually go to to get that. And so that actually gives them a lot of market power and a lot of market pricing, I think, too. So, yeah, I think it's important to maybe distinguish between those two types of questions, which is, you know, the commodity of data and then the actual specific piece of data and how you monetize both of those. I want to throw AI in, obviously, like an AI's role in data financialization. So, like, we've hit the data wall, right? Like, we've run out of high-quality publicly available data. So we absolutely need to figure out how we can access high-quality private data. Like, is that the catalyst for turning data into a true asset class? I just have an adjacent thing to say in terms of data value with AI, which is like this young man having a sandwich over there. If you, like, had five sandwiches, they would be worth five sandwiches. And that sandwich alone is probably worth, like, I don't know, maybe $15. And you times that by five. Data is, like, very different in terms of value, where, like, one piece of data is almost, like, I don't want to say valueless, but it's like the curve is, like, very little, very little, very little, very little, very little, very little. And then all of a sudden it's, like, exponentially higher because of how generative models work. It's not exactly that, like, I'm doing this in a broad stroke, but there is, like, some function there that is somewhat true. And it really has made data financialization so fucking hard because, like, when Anna and I were working on this back in, like, 2020, like, we couldn't give people enough money for their data because them alone didn't matter. And it's, like, only the collective which really mattered. And then once we got the collective, we could pay everyone a lot because it was worth so much. And so there's, like, this chicken and egg issue. And then AI kind of makes this, exacerbates this because, again, like, the generative side. And so, like, why this is such an interesting problem for crypto is, like, there's elements of crypto with incentivization that actually, like, are solutions to what I just described, which is a really, really, really, really hard problem and why I think we don't have, like, I think there's also a chance. Like, I mean, there's a chance right now that a lot of the data is valuable because it's private. And at the moment, it's broadly accessible. It basically becomes commoditized and not worth very much. Like, I think that is a possibility. I think we're already sort of hitting the point where models themselves are getting fairly commoditized. And, like, a lot of the underlying businesses are going to be around the ability to capture the consumer or the ability to have a private set of data to make a model that can perform better in a given vector than kind of all of the models that are training on the same sort of data set. And so I do think there is a world that to the end user, like, the data is extremely valuable now because only small parties have it and they can do a lot with it and leverage that and extremely, like, wield it very powerfully, whereas the moment that it's open to everybody and becomes broadly commoditized, even across the entire whole, there just isn't a significant amount to pay for it because it just gets – effectively, let me put it this way. The moment it becomes public, it almost becomes a race to the bottom from everybody in the same way sort of DeepFleek did with training itself, right, and what actual training costs may look like on a long-term time horizon. I think the interesting company to think about here is probably like YouTube. Like, YouTube is just storing massive amounts of video data and, like, presumably, what, 99.9% of that nobody ever looks at. You know, it's like someone took a video of their cat and, like, you know, it's seen once but it's still there. And they're doing this, right, because the 0.1% of content that people do look at is still immensely valuable. So it's worth it to them to just continue to store all of this worthless data for all that expense. And, yeah, I think that's the question, right, is, like, we are now – I absolutely believe we're entering, you know, just an explosion of data now. Like, one way to think about it is, like, if you wanted to make a movie in, like, the 1940s, it would have taken you two to three months to make a 90-minute movie, right? If you want to make a movie in, like, the 2000s, you can shoot it directly with a digital camera in 90 minutes. So it's 90 minutes to make a 90-minute movie. With AI, it can take you five minutes to make a 90-minute movie, right? You can, like, just go prompt it and it can create the entire thing. And so you think about, like, what that means for, like, the explosion of data, especially, you know, really, you know, expensive data like video. It's going to be – I think the real race, right, is between the massive generation of data on one side and then our ability to create these kind of decentralized rails that can commoditize that data for cheap on the other. But, you know, if anything, like, our ability to commoditize and make it cheap is the value proposition to be able to create, you know, YouTube that's 100,000, 10,000 times the size. So Casey alluded to data DAOs and the fact that my data alone isn't that valuable. But as soon as that gets to a certain scale where it's pooled with enough other people, it becomes highly valuable. So assuming that's true, you know, that we can, you know, have data – we do have data DAOs in the Bono ecosystem, you know, how can they emerge as market makers for data similar to how they operate in DeFi? And I guess I'd ask Brian you that question. And then, David, to follow up on that, I'm like, how can these data economies be structured fairly? Like, if we are entering a world of data DAOs, like, how do we do that well? I don't know. Brian, you can take this one. Brian, take it for us. All right, sure. On how to actually make it valuable from a collective group, it literally is almost like collective bargaining, right? Like, if you talk about collecting Tesla data, right, like one Tesla can do that. But assume Tesla can't do that or you have some layer of data that they don't get that you can provide. And all of the Tesla drivers come together to do that, then you have, like, extremely large pricing power for anybody who's interested in that data set, right? Like, if everybody independently posts their data from their Tesla publicly and says, please pay me for this, it will always just be a race to the bottom because I'll just take the bottom, you know, some X percent of data from the people who are willing to offer it the cheapest, and then everybody else has a zero value to their data because I can just get the majority of it for cheap. And so, yeah, I mean, data DAO is extremely interesting from that side in a concept to get a large aggregate of groups and have, like, collective pricing power. But there is, like, this social dynamic of always being able to potentially dissent. And, yeah, there's a very interesting market dynamic there from how that actually happens. But it's clear that the data, like, again, having structured collections of clean data, which is something that we were referencing to earlier, knowing that, hey, this is from all Teslas verified, this is what the data is, this is how it's structured, here's the miles you configure it by, like, everything else that you want to do, is extremely valuable. But, again, the value is in the ability of those, I guess, who price it and how easy it is to access. There is a world where that is worth, you know, two orders of magnitude less or more. Like, the vol on what that should actually be priced at, I think, is really massive. I don't have a good answer here, but there's an overall trend, right, that anything digital has become cheaper than anything in real life has become more expensive. I think over the past, like, two decades or so, right, music has become cheaper and cheaper to access online, so concerts have become more and more expensive. And so, like, you know, fashion, I think similarly, like, you know, fast fashion becomes cheaper and cheaper, so, like, luxury items become much more expensive. And you start to see this, like, kind of gap that grows where what in the 20th century would have been, like, a middle class activity is now a luxury activity, to go to a concert, right, is like a luxury activity, because why do you bother? You could get that from home for cheap and it's easy. And so I think, like, this is really interesting that you bring up that example with, like, Tesla, because that's what DEMO is doing. Like, DEMO is trying to, I think, do that. And I don't have a good answer here, but I do think the other big question here is not just the financial cost. It's, like, there's another economy going on here, which is, like, when is it worth it to me to give up my data in order to be part of a pool that overall I will learn from? So if 100,000 people are already contributing to this pool, is it really worth me going out there if I can already access their data for free? And you get into these kind of tragedy of a common situations where you probably need to be actually economic incentives to get people to share this. Do you think they're the same actors, though? Like, I think the people providing the data don't actually want to use the data. They just want to be fit. Like, the people using the data and the people providing the data I view as two, like, almost totally separate class of people. Well, that's definitely the Web 2 model, right? The Web 2 model is we're giving up our data all the time and we don't know what it is and it's out there. And then you start to get into this world now where it's like, oh, everyone wants their data about health and they want their data from their eighth sleep, you know. And now it's something like we're living in a world I think where consumers are more, you know, adjusted to a world of, like, wanting data. Whereas 10 years ago I think they were scared of it, right? It was like, oh, giving up my data seems terrifying. I don't want anyone to see this. Now we just assume everyone's collecting our data anyway. No one cares. The moral issue is, like, the moral compass has totally changed on this. And so I think that also changes a bit of the economy in terms of, like, how do I see my own loss in giving up this data? So there's still an effort, right, to go through. So I still think, yeah, you're right, right, that it's like most people probably don't care. And so there might need to be an economic incentive to get people to care as well. And so that becomes really strange because it's like, you know, traditionally you should be paying to get data, but now we're actually paying people to give data too. So, you know, figuring out those economies I think is hard. NFTs I still feel like are the most interesting attempt to do it. They didn't succeed. But, like, I do think the attempt of, like, doing artist royalties and, like, you know, being able to monetize, like, someone putting public art out there is still, like, the closest we've kind of come. But we didn't crack it. Something you said was really interesting. I haven't heard this out loud, but it's one of those things where when you hear it, you're like, oh, that's obvious. But that the physical world is getting more and more expensive to operate within, and the digital world is becoming, you know, more accessible. Are there any just, like, counter examples just to play with this for a second? Because... Yeah, I mean, fast fashion, right, is, like, largely a digital thing, but there's definitely, like, I think it's infiltrated a lot of, like, physical world as well. A lot of fast food stuff too. Like, I do think there's been, like, kind of a growing gap in terms of, like, what is fast and accessible versus, like, what is luxurious that probably permeates the physical world too. Walmart, right, is doing better than ever right now. Costco is doing better than ever, right. It's just interesting to me because, like, because of that, because we live in a capitalist society, like, the fact that the digital world is so much cheaper, it obviously pulls us closer and closer into that as, like, our primary mode of operation or living. And then the downstream effect is that what might be happening with the privacy is, like, in the physical world, we're very aware of what privacy means. We live in homes that, like, aren't glass. We don't, like, share our salaries. There's just, like, a lot of cultural things that we do to preserve privacy. And in the digital world, it feels like we're still, like, just at the beginning of even understanding what that means. And, like, I just remember even a theory on, like, the fact that everything is just private is... I mean, public is insane to me. Like, it's really insane. And there was, like, back in 2019 and even before that, there were all these attempts to, like, private chains and private R1s and, like, Monero, like, all of them, ALEO and new attempts and old attempts. And it's just kind of interesting to me that none of them have really reached the PMF that I think they would have. It's a little bit of a tangent, but I'd be curious your takes on this. I think it's just too early. I think it's going to happen. I think it's unavoidable, but I think it's too early. And also, like, privacy is something you still get from Web2 quite easily. And so, in many ways, what's exciting about building in, like, crypto is... Right. Yeah, it's like a feature, not a bug. Yeah, it's both, right? But I think being able to take advantage of it as a feature gets really exciting for the kind of net new things you can build, whereas if you want to build private stuff that already exists in Web2, right, like that already exists in the centralized exchanges and the OTCs. And so, like, the actual value of, like, privacy on chain is just going to be really boring, which is it's going to be cheaper and more scalable, like, privacy. I don't think it's going to be that the privacy itself is, like, a net new feature compared to what you got before. All right. We are at time. I want to ask one final question for you guys. I'm trying to think of what the spiciest thing is. I mean, like, what's the craziest feature you think we can see around, you know, private? So what do we have? We have private programmable data assuming blockchains work. We have an ability to clean, structure high-quality data through honest proof of contribution systems that the data devs are all working on. Like, are we going to see, like, AI models lease by or collateralize data? Like, Brian, what will we see on the DeFi side, like, assuming that there is, like, healthy tokenization of these data sets? Like, where might we be? What is the craziest world is a really tough question to answer. There's a lot of crazy worlds. It's this one. Yeah, yeah, maybe. What is the craziest world that we'll see? I mean, I think, again, I think the most dystopian and maybe most likely world is that, I mean, actually, probably very much the most likely world is the super majority of all data is produced without human inputs, basically, and that maybe the value of human data goes down somewhat significantly and autonomously generated data becomes more. I think also the ability to track, like, everything, whether it be, like, Web 2 privacy we've all talked about is, like, we're very openly giving our data to basically everything now. More and more satellites, your phone, everything. It's just, like, all of your data can even be, like, almost, like, tracked. You go to look at any property data and you can tell how many phones moved in and out there, how many people stay, how many people work there, how many people move in there, and then you can figure out where they came from and back into all these pieces of data about them without having them opt in or do anything other than, like, have their cell phone on them. And so I would say the craziest, in my mind, scenario is just that none of it actually matters and we're basically just like a cog providing data in a machine. But none of our data, like, is, I guess, collectively important, but independently our data is less unique and matters less and less because the ability to serve ads to us becomes the – right now it's very important that it individualized because all of the value of the data is in serving us ads. All of it is in human commerce. And so I guess if it moves away from human commerce, the lens of what data will look like will probably look a lot differently. I think my dark vision for the Internet is that it kind of is a cure for its own disease, which is the loneliness that it creates. Like, putting people behind a screen all day and then promising them social connection through that screen is kind of this, like, endless loop that people fall into, right, of feeling lonely and then having a cure for it be the same thing that's making them lonely. And I think, you know, the craziest world is that we start moving towards these AI characters and creations and friends and our entire – you know, right now we have parasocial relationships with celebrities, but we can have fully social relationships with non-existing creatures. And I think that almost feels inevitable and is already happening. It also creates this really interesting world where I think every new kind of creator platform tries to go after previous creators and then finds that they don't want – they actually have to get, like, a new type of creator. Like, YouTube couldn't get Hollywood stars, they had to have their own. And TikTok couldn't get YouTube stars, they had to have their own. And I think AI is going to be similar where right now there's, like, a lot of focus on body doubles. You know, it's like we're all going to create our own double. But I think that's going to be, like, kind of weird and scary. What will happen instead is we'll have people create entire new creatures and characters in AI that will exist and have, you know, I think, like, fully social relationships. So, yeah, I'm not saying anything that wasn't predicted by the movie Her. I think, like, the data question is really interesting here because it's like if all of our lives are being led in terms of data and generative data, I mean, yeah, that truly is the oil for the 21st century. Okay, I know we're out of time, so I'll keep this short. I agree. This probably question is not the great question for all three of us because we probably think way too much about this, all of us up here. I think a lot, though, of do people know, like, the hero's journey or, like, the concept of that? Okay, so basically if you don't know it, it's the belief that a lot of our existence or our meaning comes from this, like, arc of having a journey. And I am just so freaked out at how easy it is now to accomplish a journey. Like, you just gave this example of how easy it is to create a video. Has anyone ever tried on that it shouldn't be that easy to make a video because, like, it's the path and, like, the journey that makes the destination worth it? And I'm, like, freaked out that society is making all these things that used to take a lifetime take an hour. And to me, then, if you play the tape forward, it's like, well, what are we going to do? Like, what are we – not in, like, the trite way of, like, oh, we're going to be hooked up to a machine. Not like that. Not like a matrix. Like, genuinely, we're going to have these existences, and I don't know how we're going to be able to find fulfillment if we have it instantly. And so from that, like, the reason this is connected to data is, like, the data enables us to kind of recreate history or recreate on the fly without having to do the individual action. Like, the reason we can get the videos is because we have the data from the other videos. And so it kind of, like, freaks my mind out. I also think brain-to-brain communication is coming very soon in the next 10 years, and so I think that's going to completely dissolve language. Welcome to instant dopamine. It's your journey. Yeah. Yeah, exactly. All right, so we might have data dows talking to each other in their own language or something. Awesome. Well, thank you for taking us down some dark, yeah, futures, but also some hopefully great ones. Thank you, everyone, for, yeah, joining the session.