Bittensor creator Const on Affine, dTAO, "mining reasoning," and more | E2326

This Week In Startups is made possible by: Northwest Registered Agent https://northwestregisteredagent.com/twist DigitalOcean https://do.co/twist Odoo https://Odoo.com/twist Today's show: We've been going deep on Bittensor and how it works all year on TWiST, meeting with the creators of some of our favorite subnets, and exploring how the system decentralizes inference, compute, and storage. Now, on this very special episode, Jason and Lon welcome Bittensor co-founder Jason Steeves (aka "Const") to the show. He gives us a quick tour of the project's first principles, before diving into how subnets incentivize miners to contribute, how dTAO turned the creation of subnets into its own competition, how Const says he employs "the internal machinery of capitalism" to rank subnects, and his latest project, Subnet 120 (aka Affine). Find out what the OpenTensor Foundation actually does, go inside the Templar controversy, and learn why Const is trying to make himself increasingly irrelevant. Guest Jacob Steeves on X: https://x.com/const_reborn Bittensor: https://www.bittensor.com/ Affine: https://affine.io/ Relevant Links Original Bittensor whitepaper: https://bittensor.com/whitepaper Bittensor governance and documentation: https://www.bittensor.com/docs Beginner's Guide to Dynamic TAO: https://www.tao.media/the-complete-beginners-guide-to-dynamic-tao-dtao/ Prime Intellect: TOPLOC: https://www.primeintellect.ai/blog/toploc Engy (SN53): https://engy.ai/ Lium (SN51): https://www.lium.io Targon (SN4): https://targon.com/ Templar (SN3): https://www.tplr.ai/ Hippius (SN75): https://hippius.com/ MakerDAO: https://makerdao.com/ Stillcore Capital: https://stillcorecapital.com/ Timestamps: 0:00 The origin of Bittensor (and the name Const) 10:48 Northwest Registered Agent - Get more when you start your business with Northwest. In 10 clicks and 10 minutes, you can form your company and walk away with a real business identity — Learn more at https://northwestregisteredagent.com/twist 13:55 From Bitcoin mining to mining intelligence 19:33 DigitalOcean - Head to https://do.co/twist to start building on DigitalOcean's AI-Native Cloud today — and cut your AI workload costs by up to 50%. 30:17 Odoo - The all-in-one business platform. Get started for free at https://Odoo.com/twist 32:37 What is a subnet? 33:35 Forcing honest inference from bad actors 37:14 Building a subnet is harder than it looks 40:42 Understanding dTAO 51:57 Where the registration TAO actually goes 1:02:57 All about the Conviction "rug pull" 1:15:19 How Const is making himself irrelevant 1:17:33 The "Buy One TAO" philosophy Subscribe to the TWiST500 newsletter: https://ticker.thisweekinstartups.com Check out the TWIST500: https://www.twist500.com Subscribe to This Week in Startups on Apple: https://rb.gy/v19fcp Follow Lon: X: https://x.com/lons Follow Jason: X: https://twitter.com/Jason LinkedIn: https://www.linkedin.com/in/jasoncalacanis Check out all our partner offers: https://partners.launch.co/ Great TWIST interviews: Will Guidara, Eoghan McCabe, Steve Huffman, Brian Chesky, Bob Moesta, Aaron Levie, Sophia Amoruso, Reid Hoffman, Frank Slootman, Billy McFarland Check out Jason's suite of newsletters: https://substack.com/@calacanis Follow TWiST: Twitter: https://twitter.com/TWiStartups YouTube: https://www.youtube.com/thisweekin Instagram: https://www.instagram.com/thisweekinstartups TikTok: https://www.tiktok.com/@thisweekinstartups Substack: https://twistartups.substack.com

• 0:00Speaker 00

  I've been looking for a crypto project that would solve problems in the real world. Instead of just saying, here are all the tokens, have that and everybody starts speculating it's a store of value. You have to make them by solving this mathematical equation.

• 0:14Speaker 01

  It's not an understatement to say that we are up against nation states because this is truly like China versus the United States. Order for the rest of us to take on OpenAI, we need to come up with a way that we can work together. For some people, mining the tensor is like the most fun game they've ever played, ever.

• 0:32Speaker 00

  I predict I'm going to make tens of millions. That's my goal.

• 0:35Speaker 02

  This week in startups is brought to you by Northwest Registered Agent. Got a new business idea? Northwest Registered Agent helps you bring it to life. Get a free domain, email, phone number, and more with no purchase required. Learn more at www.northwestregisteredagent.com/twistdomain. Digital Ocean. Want to see what building on a true AI native platform looks like? Head to do.co/twist to start building on Digital Ocean's AI native cloud today and cut your AI workload costs by up to 50%. And Odoo, the all-in-one business platform. Your first app is free. Get started today at odoo.com/twist.

• 1:13Speaker 00

  All right, everybody. Welcome back to This Week in Startups Twist. Man, do we have something special for you today. I have been tau-pilled. Yeah, I think that's right. You're tau-maxing. Bit tensor maxing. There you go, yeah. All of that, because I, all of that, I've been looking for a crypto project that would solve problems in the real world, Lon. Yes. And I felt like we got it on Bitcoin. I made some purchases when it was under 100 bucks. I got hacked. I lost it all. My wife bought a bunch, under 100, 200. We made millions. Fantastic. Wife's got like, if you put my Hall of Fame investments on a leaderboard. Your Ubers. Two of my wives are in the top seven. Wow. Look at Jade.

• 1:56Speaker 00

  Top five, well, dollar amounts and also on a percentage return. When is Jade raising her first fund? I think that's the question. Exactly. She's got half of this one. She's doing okay. Oh, all right. Well, there you go. Yeah, and rightfully so, having to deal with me for 23 years. But the second bet I made was on BitTensor. Why did I make it on BitTensor? I just saw these subnets actually solving problems in the world and I said, this is what I've been waiting for. Yes. And there's a guy named Const and he's here with us today to talk about BitTensor. We've been talking about it on this program all year long. It's one of our themes, Lon. I've learned a whole lot about it this year, let me tell you. Now, let's get into it. Let's bring on Jacob Steves. He is also known as Const, C-O-N-S-T, Const.

• 2:39Speaker 00

  We'll find out why. I have a guess. Hmm, I'm curious. Oh, here he is. I'm curious as to your guess. Do we call you Const, Sir Const? I'm not, I'm not calling him Sir Const. That's not happening. Jacob?

• 2:52Speaker 01

  Just Jake. It's fine.

• 2:53Speaker 00

  Jake, okay, Jake it is. Is Const because you're constantly shipping?

• 2:58Speaker 01

  Uh, or the protocol is constantly changing and annoying people.

• 3:02Speaker 00

  Okay, there's that.

• 3:03Speaker 01

  That too. Actually, it comes from Constantine, the emperor.

• 3:08Speaker 00

  Sure. Got it. Makes sense. You are the emperor of BitTensor. Let's start off with, why did you create BitTensor? What's the history here? And then I want to get into, you know, all the subnets and the economic model, which I find so fascinating. And we're going to break this down if you're, if you're a neophyte to crypto or you think crypto is a scam. I really want you to pay attention to this one because there's two times in the history of crypto I said, definitely not a scam. Definitely something going on here. It was Bitcoin. I made millions. And it's tau. I predict I'm going to make tens of millions. That's my goal. So Const, tell us, where did this all come from? BitTensor and tau?

• 3:50Speaker 01

  Well, I think there's, there's a project you're probably missing there, which is Ethereum. And, and a lot of people have said that themselves. And Ethereum took from Bitcoin the ability to write these immutable contracts. And they were like, oh, wow, let's abstract that quality of Bitcoin where you can do a transaction. But why don't we make the programming language that allows you to make any type of complicated transaction? Let's build MakerDive, which is a very complicated system of, you know, collateralized lending pools and, and stablecoins. That's all because of the abstraction of the contractual nature of Bitcoin. Hey, let's go beyond transactions, basically.

• 4:34Speaker 00

  And that's what people refer to as smart contracts. Yeah, that's the category. Yeah.

• 4:38Speaker 01

  Smart contracts. So Vitalik saw that and went, there you go. You have Solana and you have Ethereum and all the other L2s and L1s of the world that, that build smart contracts. And we do that on BitTensor as well, right? So we can also have smart blockchain platform. But Bitcoin was, I think, two very incredible innovations. One was this contractual layer that Ethereum spun out. And then the other side would be proof of work mining or the computational side of Bitcoin, which goes, hey, we can aggregate all of these contributors from across the globe together to solve this one very difficult computational problem, which is to just stuff the blockchain full of SHA256s

• 5:23Speaker 01

  and bury the transactions in this immutable time chain so that it can never be taken out by the American government or nobody can do a revision. But the consequence was of being able to build this, this market for a digital commodity. I would say it's like the first digital commodity, something that you could mine. It's digital and it was Bitcoin. It's, it's, you mined it for producing hashes. They created this, this computational network called Bitcoin, which turned out to be incredibly large, like insanely large.

• 5:56Speaker 00

  And this is a key innovation. Instead of just saying, here are all the tokens, have that and everybody starts speculating it's a store of value. They said you have to make them and you have to make them by solving this mathematical equation, which requires Nvidia cards and a computer, a server on the network, or just a desktop computer even. And that meant there was a cost to participate. The cost was compute and electricity. Am I correct in like that framing of the innovation?

• 6:26Speaker 01

  Yeah, exactly. And what, what makes it digital is the way in which it's defined. It's, it's, it's, it's computationally defined commodity, right? Oil is not computationally defined. It's, it's physically defined. But a Bitcoin hash is, is actually computationally defined. It's whether or not you have solved the SHA256 algorithm with the inputs. And, and how many you can solve is hashing power. And that's actually something that you can now trade, by the way. You can trade Bitcoin hashing power. You can sell it. People buy it, which is quite incredible. And so Bitcoin birthed, you know, the contractual basis and then also the creation of the first digital commodity.

• 7:05Speaker 01

  And digital commodities are very interesting because when very well-defined mathematical computational primitives that you can create anywhere in the world by combining things like hardware and electricity together, you build these hyper-competitive markets. Bitcoin is probably the most efficient market we've ever seen in history. Anybody, anywhere can buy and sell it. Well, anybody can contribute to it from any place on earth by plugging something into the wall. And those are Bitcoins. That's what Bitcoin miner machines are. And, and as a consequence of having this permissionless, a hyper-competitive market,

• 7:47Speaker 01

  you know, the efficiency of producing this digital commodity has just gone exponential. Bitcoin's chart is, is like this, but the pow, the, the proof of work power of Bitcoin is, is just purely exponential. It just is never goes down.

• 7:59Speaker 00

  So the price is variable because of speculation, regulation, a different market like Korea embraces it, bans it, and then re-embraces it. There are so many outside factors that determine it. But the amount of hashing, the amount of computers and energy in the network has just gone straight up. And this is important because there's another term of art we should define here, which is permissionless. And, you know, peer-to-peer, something is peer-to-peer and permissionless. Explain that in plain English and why that's so important in the history of Bitcoin. And then, you know, how that impacts BitTensor as we get there. And I want to really take our time on this because there's so many people who think they understand Bitcoin because they understand the price and they may own some of it,

• 8:44Speaker 00

  but maybe don't understand these core fundamental principles.

• 8:47Speaker 01

  Well, there's two concepts there. So there's permissionlessness and, and peer-to-peer. And, and so peer-to-peer is I can send directly to you without an intermediary. That's actually what Bitcoin wanted to solve. They were like, let's make a monetary system. It's the first sentence of the white paper, right? You know, sending, I developed a mechanism where I can send a transaction from me to you without an intermediary, without the need for a bank, right? Like, that's the whole purpose. Because we need intermediaries, then the intermediaries can censor us. Which leads us to the second point, which is permissionlessness. So peer-to-peer is the thing he wanted. Permissionlessness was a quality that needed, he needed to get to that. Um, and.

• 9:29Speaker 01

  So permissionless means that anybody can contribute, doesn't matter from where or who they are. Think of it like the ultimate form of non-bias. You know, like people are always trying to build these organizations, oh, like we don't care about race and gender, et cetera, et cetera. Well, a permissionless market is purely blind. You can contribute anonymously from anywhere in the world, and it doesn't matter if you are a bad actor or a good actor.

• 9:56Speaker 00

  It's purely a meritocracy. We don't care if you're in a communist country. We don't care if you're in a democracy. We don't care if you're 12 years old or 72 years old. We don't care what computer you're using, as long as you've got compute. You could be taking solar energy and converting it into Bitcoin. You could be, you know, at some nuclear power plant, and you have a closet, and you put a couple of computers in there, which people did. And or, you know, you got some city street lamp and ran it to your tent and put a, you know, that was when I knew this thing was truly permissionless, when people started hijacking, like, New York City public lights and running a cable into their apartment to make Bitcoin. This is what permissionless means. It doesn't mean

• 10:39Speaker 00

  necessarily breaking the law, but it is a fundamental breakthrough in how the world works. Yeah?

• 10:47Speaker 01

  Totally.

• 10:48Speaker 00

  Regular listeners already know that if you've got a great idea for a new business, our friends at Northwest Registered Agent want to help you bring it to life. They're going to be the most amazing partner you've ever had. Even if you're not ready to form an LLC, you still need to take care of some basics. So Northwest Registered Agent is now offering free identity services. That means a free domain name, open source website hosting, a business email, and a phone number, and everything else that's going to make your new startup look and feel like a professional company, all with no purchase required. And you know you can rely on Northwest because they've been helping people like you start businesses for nearly 30 years. If you have an idea, you can't get out of your head,

• 11:30Speaker 00

  or even if you're already building something amazing, you already got started, Northwest Registered Agent is the best way to establish your new company. Learn more at northwestregisteredagent.com/twistdomain.

• 11:43Speaker 01

  And it means that we just measure the output. And when you just measure the output, it means you can optimize the output. And there's nothing hindering our ability to get the maximum amount of that output, because we don't have these boundaries. And there's no permission. There's no entrance gate. And it's very difficult to make permissionless systems, because anybody, anywhere can try to cheat. And, you know, we don't classically do that. It's actually very difficult to build a permissionless system, because up until really Bitcoin, everything was manually done by humans, and humans have biases. So there's going to be permission involved with any system that's, you know, even like, let's say, an immigration policy will try to be permissionless in some sense, like it will

• 12:25Speaker 01

  be, you know, blind to certain invariable qualities of humans, like their race, et cetera, et cetera. But it's still, it's still very difficult, because there's a human in the loop, and the humans are going to be, you know, have their biases. Bitcoin is permissionless still to this day. And as a consequence of having, building this pure market, pure permissionless market, you have people contributing hash and power and computing power from all across the globe. And they couldn't, and they couldn't have done it unless there was, you know, there and now, in the outer rims, which, you know, is really says something about the fact that, like, there's, there's excess qualities out there that these permissionless markets can, can take advantage of. And so anyways, Bitcoin

• 13:09Speaker 01

  invented that, first example. And it still is growing day to day. And well, where Bit Tensor started from was understanding that, hey, well, a really powerful computer should be applied to the most important computational problem of 21st century, which is artificial intelligence. That's where the idea truly started. I began as a Bitcoiner. So I was highly interested in Bitcoin, and I was also studying artificial intelligence, and thought, well, okay, how do we connect these two things? How do we connect the most powerful computer

• 13:49Speaker 01

  in the world to the most important computational problem in the world? And that's the founding raison d'etre.

• 13:58Speaker 00

  I'll jump in. We got Mark Jeffrey here in the, in the comments. He says, Bitcoin showed us a new way to do a company. Instead of hiring, you post a coin reward, miners compete to get the reward, miners join, miners leave, anyone, anywhere can compete. Is that part of your vision for where you see this going, that this is going to be eventually a engine for starting a company without, you know, doing all of the build out a small business rigmarole that we think of?

• 14:23Speaker 01

  Well, let's, let's, you know, pull a thread between these two concepts. First, so there was first, okay, Bitcoin mining. All right, let's see if we can, we can build the same type of computational primitive for artificial intelligence. What we needed to invent in order to do that was a very, I use this word a lot, abstract consensus mechanism, which is that we needed a way, Bitcoin measures something very, very, in a sense, it's easy to measure. A hash is just binary. It's just true or false. You don't really need to have any complexity there in a consensus mechanism. And, but in order to measure something like artificial intelligence, which is very high dimensional,

• 15:04Speaker 01

  it's like, let's say you're measuring the informationally, informational significance of a 1,024 dimensional vector. Okay, how useful is that to a machine learning model? And that's not as simple as checking a hash. And so we,

• 15:19Speaker 00

  So Bitcoin, just to summarize that just concisely, Bitcoin had to solve one problem. Therefore, they just wrote it into the protocol. Like, did you solve the hash? Bit Tensor has a much bigger mandate, which we'll get into now. And the one criticism of Bitcoin was, hey, beyond speculation and money store transfer, which are valid things in the world, this is a giant energy sucking machine that doesn't provide any other value. So what is the point of all this compute? And it was during a time of excess compute. Bitcoin was formed in a time when there was plenty of compute available. In fact, people were, you know,

• 16:04Speaker 00

  trying to figure out ways to get people to consume more. And they were just desperate for you to fire up something new on AWS, or Google Cloud, or whatever it happens to be, Rackspace, all these great cloud providers, Digital Ocean, et cetera. But you did get into this, like, is this really worth it? We'll put that debate aside. Now we are in a compute constrained environment, and Bit Tensor is kind of having its moment. So let's go to that origin of Bit Tensor, and how it paralleled, and what you built on top of it, yeah?

• 16:35Speaker 01

  Yeah. So building this way of, of creating a proof of work network for anything was originally, the original intention was to train machine learning models, which we do. But it turned out that, that also, there was a lot of different things that we could apply that primitive to, that was not just training machine learning models. We could inference machine learning models, as example, which is, you know, when you call them and you get the outputs. And so there's a couple of subnets on Bit Tensor, I think you've talked about this recently, the NG subnet, where you can, you can talk to the miners, the miners contribute the compute, and they, and they run the outputs of the model. So,

• 17:14Speaker 01

  you know, that's a broadening of the scope of this primitive. You know, just like how when they abstracted the contracts from Bitcoin, you know, make or die, it took a couple of years for people to invent make or die. At first, it was just decentralized autonomous organizations, and they all failed. And then we actually got really good at decentralized autonomous organizations, and we have everything from Athena, et cetera, and on to the future. At first, we tried to do information mining, which was cool, but not enough to push AGI. And then we got really good at some of the base primitive stuff, like, like doing inferences and aggregating compute. But also, and this is one of the points that I know Mark Jeffrey

• 17:59Speaker 01

  has made, and the way that he frames is very interesting, you know, okay, we can, by building these permissionless markets that anybody can contribute to, or anything can contribute into it, any type of computer contribute, you can get storage into it. You can also get the talent. And, and perhaps, you know, that's, that's something else that needed to be mined permissionlessly, that hitherto was not, we weren't able to do that. And when you, when you have something simple like Bitcoin mining, you just plug it in the wall, it's done. But, but for these higher order commodities, I like to describe them like higher order, because they require like hardware to software, and then ingenuity.

• 18:41Speaker 01

  They actually require somebody going on and solving a problem. Like I'd say that you're, you're creating an algorithm that runs inside of software, that runs on top of hardware. And, okay, the hardware is commoditized, the software is becoming commoditized because through artificial intelligence. And then you have the, you know, the innovation, and the algorithm, and the intuition, and the creativity, all that stuff.

• 19:01Speaker 00

  The application, in a way, the network layer, and that requires essentially on Bit Tensor, creating a subnet with a new application. It's almost like Bit Tensor is, I'm not sure how many subnets there are. I know that's been, we'll talk about that. But this is where, I mean, I think maybe explaining what the subnets are, and what problems they solve, because you're essentially creating what I've called the Y combinator, the tech star is the accelerator. Of AI services built on an open platform.

• 19:32Speaker 01

  Precisely.

• 19:33Speaker 00

  Truth is, most of the big popular models can handle your AI workloads. Now, your performance, not to mention your costs, that really comes down to your infrastructure, not your prompts. That's why I want to tell you about Digital Ocean's inference engine, the best way to serve your outputs fast while keeping things affordable, which is super important, right? Digital Ocean is offering you three ways to run your workloads from a single connection point. First, serverless for real-time chatbots and agents. Second, batch for big jobs you're running in the background. And third, dedicated for those heavy, always-on workloads where you need more control. It's all running on ChIP's tunes specifically for these kinds of jobs and comes with built-in monitoring, so you can always keep a close eye on performance.

• 20:18Speaker 00

  If you want to understand what building on a true AI-native platform looks like, go to do.co/twist. Start building on the Digital Ocean AI-native cloud today and cut your AI workload costs by up to 50%. That's do.co/twist.

• 20:38Speaker 01

  Ethereum is a blockchain where you can create lots of smart contracts. Bittensor is a blockchain where you can produce a different type of mining network. Think of it like a contract, but it mines a particular commodity. It produces something of value from storage to inference to training machine learning models, to scraping the web, to stealing API keys and selling them like GM, or pulling innovation from people across the globe. So Bittensor is a platform for building these contracts, like Ethereum is for maybe more classic smart contracts that people understand.

• 21:19Speaker 00

  Great.

• 21:20Speaker 01

  These mining networks. And then into that platform, we actually have those projects mine, they all have their own token. There's some standardization in the way that they're built, and they compete against each other to attract investment from other holders in the network, for people to hold Tau. And to those projects that perform well, we actually mint the inflation of Tau. So Tau has a 21 million cap. It's just like Bitcoin, there's only 21 million coins ever. There's one produced every 12 seconds. Actually, what's less than that's half is produced every 12 seconds because we've gone through our first halving event.

• 22:04Speaker 01

  And that token, that new limited token gets distributed to these projects as basically additional liquidity.

• 22:10Speaker 00

  So each subnet has a reward system built in. Instead of getting a Bitcoin for mining the hash and building out the Bitcoin network, in Tau, you could work on one of 128 different subnets. You have to stake, you have to put up some Tau to make one of these, and then you can earn Tau, more or less. So explain the subnet concept and maybe highlight the top two or three in terms of actual usage and engagement and what problem they solve. Because that really helps people take this from being an abstraction and a philosophy to being, you know, a startup providing a service.

• 22:52Speaker 01

  Well, if you don't mind, let me try to just explain how a core subnet works in the first place. So because Bittensor is now a meta subnet. It's actually an abstraction on top of itself. So you need to understand the primitive before you can go to a higher level. So the core, a subnet is basically an open network that you can join. And by join, you burn a little bit of a token to prove that you're willing to play the game. And think of it like paying your entrance fee or, you know, competing with other people to join the state, to buy a lottery ticket.

• 23:33Speaker 01

  But it's not a lottery ticket because inside this network, your computer that you registered with is going to do some sort of work. Now, a good example of that type of work is that you're going to get, your computer is going to get queried by clients like yourself, Jason, who's using Claude code and wants to talk to a machine learning model. And you're going to be running a machine learning model on your machine, and it's going to answer those requests. And it's going to respond with, yeah, the capital of Texas is Austin, and you know, the next step in this agentic thing is to, you know, LS into your folders and pull this file. Like, all of that is basically talking to an LLM, and it's competition expensive. And so you can join this network by paying this little fee and running the software on

• 24:16Speaker 01

  your computer, which is a machine learning model, and people will talk to it. And inside of this network, you're going to have another set of participants, what we call validators, that are going to check to see if the computer that you added to the network is doing the job faithfully, right? But not just faithfully, perhaps we're also going to check to see if you're doing it fast and faster than the last guy. And what happens in all of these networks is that people joining, there's a continuous role of people joining into the network, and they're measured based on what the network is measuring. In this particular design, an example would be, I just described what's called inference speed, right? How fast you can inference a machine learning model? It's measuring how quickly you can answer these questions from the clients, and it's

• 25:01Speaker 01

  paying you more and more based on how quickly you can answer these questions and if you're meeting some sort of bar, right? So think of it like a very well-defined, written-to-code description of how we're going to check to see if the computer that you added to the network is faithfully following the rules and doing the thing, and an access along which you can perform better, right? So perhaps you can combine and add more computers to your cluster, or you can speed them up, or you can improve the software yourself and make it faster at serving these inferences. And if you can do that, you can serve more requests, and over time, you can make more money in this network. And so each of these networks are usually, the way that we visualize them is like a curve.

• 25:43Speaker 01

  And so along the X-axis are all of the different participants. Usually, it's about 256 of them. There's 256 computers, which is often more than enough, and I'll explain why if you're interested later. And then there's the amount that they're getting paid, which tends to go up, goes up to the top. And then there's the ones that are being cycled out at the bottom. And think of it like a league, like the Premier of them is a good example.

• 26:08Speaker 00

  Love it.

• 26:08Speaker 01

  So each of the networks on Bittensor has that, this relegation system, this Premier League of the thing that the miners, the contributors to the network is contributing. So some of them may be inference. A good example, like a sub-sample of those projects would be you bring GPUs to the network that people can use. You provide computing power that people, you inference machine learning models. So when people are talking to your endpoint, that you respond. Or it's an example where you actually train a machine learning model. You produce an AI and contribute the AI itself to the network and get paid if your AI is better than the other AIs in the network.

• 26:52Speaker 01

  That's what we would call in model competition. So that's a little subset. That would be like Liam, NG, and Affi, with some of the top subnets on Bittensor. That's what they do. The whole premise of one of these projects is that they're able to use this permissionless, hyper-competitive market to produce this digital commodity, inference compute or models, faster and better than anyone else in the world because we're using the power of Bitcoin. And if that's not their premise, it doesn't make much sense to build on Bittensor. And there, then, that network has its own token. So it's, think of it like a sub-token to Tau, which is like a derivative token, we call it a staking token.

• 27:34Speaker 01

  The name we use is an alpha token, which is a secondary cryptocurrency, which also has 21 million cap, that only exists within side of that network that we just described. You need to pay it to enter, you get paid in that by doing well, and it's, if you sell it, you sell it into Tau. That is what a subnet looks like. And that commodity, we call the alpha token, that token that represents that network, it only has value if, at the end of the day, people are going to want to buy that token to get access to the computers in this network, right? So everyone is evaluating these subnets based on whether or not they actually have some sort of long-term potential for producing value.

• 28:16Speaker 01

  If it's just, oh, it's a network where you can just join, and if you just break random numbers, no one's going to buy your token, because there's no way to thread any, like a relationship between that thing having value.

• 28:28Speaker 00

  And in startup speak, for me, that would be a product with product-market fit. So for me, using energy or E-N-G-Y, for me, that was, wait a second, these tokens for Quen, Kimmy, and GLM52 are half the price of other places. And it's like, well, wait, how come those are half the price? And now I'm looking at it, you know, using OpenRouter, Claude, Fable, and hey, maybe I just plug in my API key for energy and I use that in my IRMEs agent.

• 29:01Speaker 01

  Right. Right, and you know, and, and, and behind that product is this liquid swarm of anybody in the world that can enter and try to, to reduce that price over time. So right now it's half the price, but it doesn't necessarily just stop there because whenever this excess compute, that person can plug themselves into this permissionless network and, and it becomes quite liquid, right? So, so, hey, I, I was, um, I'm not using these computers anymore. Well, I just run the software and I just sell the inference to NG. And as a consequence, um, you know, we can dramatically reduce the, the, the cost of that commodity.

• 29:36Speaker 00

  And this dovetails with your previous statement about permissionless and anybody can join to make a little bit of extra Tao or whatever the subnet's token is. So if I'm sitting there and I was, I don't know, um, providing servers to startups and other folks and I happen to have, you know, a rack that I haven't provisioned yet. And it's going to be provisioned, you know, in a hundred days. I have a client who's coming online in a hundred days. What do I do with that for a hundred days? Well, I could give it to NG without asking anybody permission and start making money from it. So the downtime would turn into productive time, would turn into revenue generation time.

• 30:17Speaker 01

  Precisely.

• 30:17Speaker 00

  If you've got a company, you probably have a tool for invoicing and one for running your website. And then finally, you got one for your CRM, of course. And somehow, even with all these time savers, you're still moving data around by yourself at 10:00 PM? That's why here at Twist, we recommend Odoo, O D O O, the all-in-one management software that's already serving 16 million users across more than 170,000 companies. Think about that. They're bringing everything you need into one platform. That means your CRM, your sales tools, accounting, manufacturing, your website, inventory, and of course, your point of sale, if you have a point of sale, all right where you need them. And they're all in constant communication with each other. No more logging every change order or new sale across three different apps. And the dreaded spreadsheet

• 31:00Speaker 00

  as a database. No, you make a sale and your invoice gets created instantly while your inventory gets quickly updated. So if you're still combing together your back office across five different tools and apps, get started today at Odoo.com/twist and your first app is free. That's O D O O dot com slash twist.

• 31:19Speaker 01

  It, it took quite a long time. It's, it's important to put an asterisk here. It took quite a long time for us to figure out how to build these mechanisms. Like, just like how, you know, the first smart contracts on Ethereum, like the Dow didn't work, right? And you had CryptoKitties and that was cute. But now you have production level stuff and, and all of these side chains.

• 31:39Speaker 00

  How long has Tao been around? And, and how long did it take you to get to what I'll call?

• 31:44Speaker 01

  Yeah, people have been building mechanisms on BitTorrent since, for two years. So the second year that it's been open for people to build mechanisms on BitTorrent, but actually we're 2021. So the network launched with us learning the art of what we call incentive mechanisms. We call subnets, the, the core design, how like this structure of what a subnet is in, which I can talk about at length. It's, it's actually quite a, it's quite an interesting field of the study, the economic study of how do you build a permissionless adversarial game where even the worst person in the world, if they, if Hitler and the devil were to have a child and that child were to mine on your subnet, they would just produce value, you know?

• 32:28Speaker 00

  So you can take away intent, but you need to have some validator that says, hey, this Hitler Idi Amin putting servers on here is not doing so to kill people or to do harm in the world for being a little bit, uh, uh, you know,

• 32:45Speaker 01

  Dramatic.

• 32:46Speaker 00

  cheeky here. Yeah, but they are being judged on a specific criteria, which is, does this server provide inference and provide Kimi or GLM-52? And, and there's something there with these validators that do this, yeah?

• 33:01Speaker 01

  Exactly. Yeah. So, so we built, we built a mechanism that would allow a distributed set of computers that could take any code that you wrote, Jason, for, you know, describing whether or not the computer that joined the network is actually, you know, if Hitler and Satan's, you know, a love child was actually doing inference properly, um, you wrote code that would check that. And these computers, which we call validators, um, you can elicit them to all run the code at the same time. Um, and if they do so, they will reach consensus if more than 50% of them are running the right code. It's very similar to the way in which, you know,

• 33:42Speaker 01

  Bitcoin works, right? If more than 50% of the network is running the correct code, code of Bitcoin, then it doesn't matter if 49% of them are cheating. It's irrelevant. The, the person, the, the, the majority, the honest majority will determine the, just the direction of the incentives in the network. So the, the core mechanism of BitTorrent is that we can build mechanisms like this so that, um, so that, yes, we can build a network that's permissionless and, and verifiable and auditable, right? If you know that this distributed set is, is, is all running this code and it's going to take more than 50% of them and, and the decentralization of that set, which is measured in proof of state. It's not like Bitcoin. Bitcoin is proof of work, which means that the

• 34:23Speaker 01

  weight of a node is based on how much compute they provide. In Ethereum, it's proof of state, which means that the weight of a node in consensus is based on how much economic value they have. It's the same thing in BitTorrent.

• 34:34Speaker 00

  In this case, if you were, you know, NGL, I'll, I'll go to that one because I actually use it. If somebody got on the network and said, yeah, I'm providing GLM-52, you know, this open source model from XAI in China, uh, but they were actually using like some old DeepSeek 3 and they were passing it off as GLM-52 to use less compute and they were giving wrong inference. The validator would say, uh, uh, uh, we're checking that you're actually using the right GLM-52 code, not 5.1, not 4.0, not some other hack.

• 35:07Speaker 01

  Not quantized, yeah.

• 35:09Speaker 00

  Yeah. And we are going to make sure in a way that there's an SLA, a service level agreement here, which you might have with Google or, uh, Amazon Web Services. In a way, you've smart contracted or built into the system an SLA, that the service level is mechanically and architecturally built into the system. Am I understanding it correctly in layman's terms?

• 35:34Speaker 01

  Yeah. And, and instead of, um, a contractual SLA, it's a, um, programmatic SLA, right? So you, you define the way in which you can check to see if they're not, they're doing the right thing. So for an inference subnet, it's actually very difficult, but the, the technology is there. You, the way you do it is you, you query the endpoint, the endpoint gets a, gets a result. Um, and they have to pair that result with a very cheap, what's called a ZK proof. In case of NG53, it's um, Topolock, which is, um, some, an algorithm created by Prime

• 36:15Speaker 01

  Intellect, which is, um, basically hashes the hidden states of the model. Hidden states is like halfway through the machine learning model. They, they take those basically intermediate representation, uh, and then they, they project them onto a different mathematical space and, and then they, you don't need to know too much more about it. But anyways, that's

• 36:34Speaker 00

  It's basically proving it is, it is what you say it is, is a way to say it. Yeah, I'm playing.

• 36:39Speaker 01

  Precisely. And, and, and on BitTorrent, these things get, like, it's one thing to write a paper. It's a completely other dimension to, to write on BitTorrent because we have the smartest people in the world and we have the smartest hackers in the world and they're just right next to each other and they're, they're going to break, they're going to break your system. And, and if you can build a system that on launch doesn't break, you know, it's like, oh my God, it's unbelievable because

• 37:06Speaker 00

  So this is critically important, Consta, um, you know, if it's permissionless, you can have bad actors and then the bad actors can be forced to act well and be good actors by the architecture. But you're constantly being stress tested. You're constantly being attacked because it's permissionless and global. You have attackers who say, hey, there's something at stake here. I can get tokens that are worth something. So there's a value in hacking the system. Therefore, the system must be architected properly. And that's a big part of your job and the BitTorrent Foundation's job is to make sure the rules and, and the architecture and in a way you're the

• 37:47Speaker 00

  police officers or something. You, you're the, you're the Jedi Knights of this, um, system, making sure there's peace and that it's being done properly.

• 37:57Speaker 01

  The, the, the, this is the, the role that we play reluctantly while it's still required. Um, and, you know, as all subnet owners do themselves. So at the, at the level of the individual subnet, they're, they're, they build the mechanism and it breaks and then they fix it and then it breaks and then they fix it and then they get it right and then it starts working and then they make a million bucks. And, and, but in that time, it often requires a lot of massage and, and realignment, you know, um, to, to, to, to, to fight against effectively, you know, nation state militias, uh, you know, there's a level of the people that are trying to destroy your project. And for a lot of people starting a subnet on BitTorrent, they, they don't really

• 38:40Speaker 01

  understand that that's what they're up against, which is I think kind of cute. I often have these phone calls with people. They're like, I want to start a subnet. And I'm like, oh. That's so nice, but you're really not coming at it with the right level of intensity because your adversaries on this network are very serious. And so

• 38:57Speaker 00

  They could literally be a nation state. It could be North Korea, which like North Korea is so desperate for revenue. Lon, I don't know if you remember this story, which we covered. North Korea was placing developers

• 39:08Speaker 03

  Sure.

• 39:08Speaker 00

  as remote developers in companies, not because they were hacking the companies. They wanted the six-figure jobs.

• 39:14Speaker 03

  Yeah, yeah, no, we talked

• 39:15Speaker 00

  They wanted the salaries, how those companies had to develop like whole protocols to identify when a North Korean was applying for a job, pretending to be a different kind of remote worker. Yeah, I remember.

• 39:25Speaker 03

  Yes.

• 39:25Speaker 01

  Right.

• 39:25Speaker 03

  Crazy.

• 39:26Speaker 01

  Building that is difficult, but we have now, I would say like the top 10 to 20 subnets which are premier and they're experts at this. And they can do things like, you know, subnet 51 figured out GPU attestation, you know, trusted execution without trusted execution, which is quite incredible. They don't even have TEs, which is what, you know, Intel does. They built the algorithms that could go in SSH into the GPU and check everything in the world to make sure that that person has that computer and is not cheating. And they built economic primitives to make that make sense. And so, you know, now those churn, but it took a while for us to build those things. And so this is all just one level. That's just a level of the mechanisms on BitTensor.

• 40:10Speaker 01

  But it in an absolutely insane, but it turned out well decision by us. We decided to make the actual creation of the mechanisms themselves on the BitTensor blockchain mechanism itself. So we went, okay, hey, let's apply ourselves to ourselves. We're really good at building these permissionless mechanisms. Let's build a permissionless mechanism, which selects permissionless mechanisms. And so that was what we launched just over a year ago, which we called Dynamic Tower, which is where all of these subnets got their own token, all those tokens got paired to

• 40:49Speaker 01

  the tower, and all of them are trying to push the metric of success inside our ecosystem, which is increasing their price while staying on, basically. And so we, our thesis at the like the highest level of BitTensor is that we want to like, hey, this network is going to incentivize and try to optimize people for bringing inferences. Great, well, we're going to optimize people bringing projects into this ecosystem that produce a lot of value, which we can measure with price, more or less. And, and, you know, with a couple of knobs here, and obviously it's complicated and we can get into it, but that is what we think it will pull the most amount of innovation

• 41:34Speaker 01

  into the ecosystem and also drive the most amount of value into BitTensor, which I believe does work. When you look at the way in which like the, you know, the hand over head crawling towards performing well in this ecosystem, the more competitive we make it, like the more badass the teams have become. And it's quite impressive, like it's truly, it's really reached a point now where the entrepreneurs in this chain are so much better than I am and many times smarter. And I listen to as many of them as I can because they all often are mentors and guides for us at the ecosystem.

• 42:15Speaker 01

  You know, how could we improve the core incentives of the BitTensor blockchain?

• 42:20Speaker 00

  So this is a good pausing moment. Again, we'll just stay on engy.ai. We've been talking about the person running that is a guy named Ning Ren, N-I-N-G-R-E-N. This person came to BitTensor, how and why? What, who are these people that come to you or to the BitTensor Foundation? And we should understand what that is and how that works. Who are these people who are attracted to creating these projects? And what's their goal? Are they like freedom-loving individuals? Are they entrepreneurs? Are they hackers? Are they some combination of these?

• 42:59Speaker 01

  A good number of them are scammers. And, and you kind of can't avoid that because, you know, you have the doors open. And that's part of what we're doing is where we're saying, hey, we can, we can like open our eyelids the most and that's what is going to make us win. But, you know, we're staring right at the sun. So you're going to get a lot of stuff. And then, so there's a lot of bullshit. And, and, but then there's also the best stuff and the highest quality and the most intelligent people that if they can get across the stigma, because there's a lot of stigma in crypto and well-earned, honestly.

• 43:41Speaker 01

  And if they can get through that stigma and they can look at the technology and they can see what they can build here and they understand technically and philosophically like why this is such a powerful primitive. Yeah, those people that see it, they come and but you need, you need a level of openness for sure, because it's, it's, it's not the, the most dreaded path, right? And, but so someone like Ning, like he, he was brought in by, by Algod, I believe, who built a team and said, he knew this guy who had worked at Google Brain, where I'd also worked.

• 44:27Speaker 01

  And he didn't know about it and but he was told about it and he thought it was super interesting as it is. And, and so he got fascinated and obsessed with it, you know? So there's those people, those types of people that come in. And, and then, and then there's, and then there's people that have absolutely no academic experience, but they're just raw entrepreneurs that, that go, hey, wow, this is novel. This is the thing that will allow me to build something that can break down that glass ceiling where, you know, can you really break past these fiat-funded companies by just doing the same thing that they're doing?

• 45:10Speaker 01

  And I think that the, the thing that compels a lot of entrepreneurs in our ecosystem should compels me is that we need to do something different and better and more powerful than what they're doing. Otherwise they'll just beat us with more cash, right? And, and so

• 45:28Speaker 00

  So this is also mind-blowing. The fact that BitTensor exists and it's relatively stable and providing functionality to the world that some number of consumers and enterprises are dependent on is extraordinary as a technical achievement, but it's also extraordinary, as you're pointing out, hey, there's frontier labs, there's open source projects out there, there's a lot of competition. This is the highest degree, like what is the chess rating, ELO or something? Like if this was chess, you're going and playing with the grandmasters. The grandmasters being, you know, Elon Musk, you know, Claude, Sam Altman, OpenAI.

• 46:11Speaker 00

  I mean, these Gemini, Sergey Brin and his team, like these are the most elite and you have to beat them in the offering, if it was inference in this case. So it's, it's an extraordinary achievement on so many levels right now.

• 46:27Speaker 01

  And, and yet the largest supercomputer in the world is Bitcoin. So like this is the technology that has the example of the only thing that's beat them in size. So, you know, there's something to be mined here. And, and like it's, it's, it takes time to build the future. And but now we're really starting to see like what's so exciting right now in, in BitTensor is that these primitives just, they just make perfect sense now. So, you know, we just, we turned up the emissions for a lot of subnets. We said, hey, you got to turn up the amount that you're paying miners. And, you know, some of the mechanisms like, like 51, they just, the revenue just scales with the more, the more money you put in because it's permissionless.

• 47:12Speaker 01

  But there's the companies don't act like that, right? Oh, here's a bunch of money. Oh, it's not necessarily that you can just distribute that immediately and scale efficiently. There's so much inefficiency when you have that, you try to scale a billion dollars through a human organization. You have to hire people. We were talking at the beginning of the call, you have to fire people, you have to give them contracts, you have to get spaces, you have to find all these things, put them all together. It's not easy, but, but a mechanism that's just described by really just a code base and a permissionless market, you just pump more money through it and it just scales. So, you know, because they've, they've tripled the amount of money they're paying miners, they've tripled their amount of compute in two months.

• 47:52Speaker 01

  And, and, and so like these things are really beginning to work and it's really exciting to see how they are coming together and meshing together in an ecosystem. But our goal is not just to do more compute and more inference. That's fantastic. Our goal is actually to truly head to head with, with the elite centralized labs at intelligence. This is, I think, the pinnacle commodity that we can, we can measure inference and compute and storage. These are sort of predicates. But, you know, this is, this is where I think that, well, this is where we're going to go. And this is, and this is what we're trying to build right now. And it's much more difficult to measure intelligence.

• 48:38Speaker 01

  A computer is very difficult and inference is extremely hard, but measuring intelligence is still possible. It's just very, very, very abstract and, and, and hard to get at. So this is what, you know, my personal mission is, and I actually run a subnet on, on the tensor right now. It's for the call, I was working on it right now. It's called Affine. So we're, we're building the, the code, the mechanism that can actually measure that pinnacle element. Like what does it mean to actually grasp in your hand, intelligence as a, as a commodity? So that's, that's the thing that, that

• 49:09Speaker 00

  And how would one measure that? You, we do have like humanities last test. We have all kinds of benchmarks from LM Arena. Is it as simple as saying, here are intelligence tests? And I want to know more about Affine. A-F-F-I-N is

• 49:26Speaker 03

  Affine.

• 49:26Speaker 00

  Affine. A-F-I-N, A-F-F-I-N?

• 49:29Speaker 03

  AFF.

• 49:30Speaker 01

  AFF.

• 49:32Speaker 00

  I-N, okay.

• 49:32Speaker 03

  I-N-E, I-N-E, Jason. AFF-I-N.

• 49:35Speaker 00

  I-N-E. And that's subnet number?

• 49:38Speaker 01

  120.

• 49:40Speaker 00

  120.

• 49:40Speaker 01

  There's 100, there's 128 of them right now.

• 49:42Speaker 00

  There's 128. There's been talk about 256. What does it take to start a subnet? How does that work? Do you go to a board of subnets? You go to the other ones and say, hey, I want to do this. How does it work?

• 49:53Speaker 01

  Well, if we had a board, then we wouldn't really be permissionless. You, and

• 50:01Speaker 01

  So anyone can can register one. And the way that you do that is exactly the same fractal like design. So it's it's the same way that you would register into a network of a subnet. You basically burn some token to pay an entrance fee into the into the league. In this case, the league is 128 in size. A normal subnet is 256. So we'll probably get to try. We're trying to get to 256 at that level as well. So you enter into the network and you start building your system and people show up and go, hey,

• 50:35Speaker 00

  How many tau do you have to burn or contribute? This is like buying a franchise, right? Like buying a team in a league. Yeah.

• 50:44Speaker 01

  Let's look it up.

• 50:45Speaker 00

  Oh, okay. And how is that determined? Yeah.

• 50:48Speaker 01

  It changes every block on on on BitTensor. And so what we we use is a it's kind of like a Dutch auction. So the current rate is 688 tau and we're trying to lower that. So it's it's about it costs about $121,000 to have one of these 128 slots, which is a lot of too much and we want to lower that a lot because but it's a Dutch auction. So so the price decreases until somebody's willing to pay for it. And then when someone a subnet registers, we double the price and then we lower it again.

• 51:22Speaker 00

  Well, I mean, it's essentially the cost of joining the money you would get if you joined an accelerator. It's classically been 125K. So it's not, to my mind, like a crazy number.

• 51:35Speaker 00

  But it's certainly not nothing. So where does that tau go? It just gets burnt and it lowers the amount of tau in the network or it gets into the foundation. Where's the tau go? It goes to you? Who does it go to? The other subnets?

• 51:49Speaker 01

  It actually goes to create liquidity in the initial pooling between tau and your what we call alpha token. So the subnet token, it creates liquidity. So all of these subnets in BitTensor, all 128, have in the liquidity pool a V3 automatic market maker. It's basically somewhere you can just buy the token through with a little bit of slippage. It's a it's a smart contract itself. And so each one of them has a pairing with tau. And so you have to buy tau or to buy the buy those subnet tokens. And this this is, you know, one of the ways in which we we build demand for the underlying token, right? It's like the US dollar, right? It has all of these amazing companies inside of the US in the US system and it has a massive network effect.

• 52:34Speaker 01

  So we're creating.

• 52:36Speaker 00

  Are people speculating and just buying five of them and sitting on them? If I was like an investor and a speculator, could I just buy four of them for $500,000 and sit on them? And be like, I'm a speculator?

• 52:48Speaker 01

  You know, not financial advice. Yes, you definitely could do that. And people do do that, right? The, you know, we, we had this

• 52:58Speaker 01

  question, problem, which was, how are we going to? We built the platform so people could build these systems. But we wanted to know, how are we going to incentivize the teams? Okay, we still have, you know, 10 million tau to distribute. And, okay, well, what we're really good at is is optimization mechanisms. So let's build an optimization mechanism that these teams can can compete in, right? Where they get they come in, they get relegated and they go to the top. And so we came up with this idea of using the price of a paired token as the thing that we would measure. And, you know, that allowed us to build, allowed us to distribute the inflation of

• 53:40Speaker 01

  tau itself into a network which people were mining by creating subnets. So people create subnets and they mine tau by creating the subnets. That's the that's the meta system in BitTensor. And and the idea is that, you know, as a whole, this network, if we optimize for all those projects to produce value, which is measured by their price, we can actually elicit the internal machinery of capitalism, which is thousands of individuals from D gens to to to scientists and the like, to long term investors. We can elicit all of that swarm intelligence to properly order rank these projects, right? Like, you know, hey, if you were to sit down right now, Jason, and go, hey, how can we

• 54:23Speaker 01

  order all of the companies in the United States? And you didn't have the the NASDAQ? How would you do that? It would be an impossible job. So so what we do is in society is we use markets to do that ranking order for us in some way. So we did the same thing inside the internal market of BitTensor. We elicited this this market mechanism, this this trading system to to to order the subnets and and push the the the cream to the top, which is which is what you see. So, you know, when you're entering into the BitTensor ecosystem, I often tell people, you know, be careful because there's there's some rotten milk and there's some cream and and the cream is more expensive and there's raw milk.

• 55:05Speaker 01

  And your job as for playing this game is to actually help us contribute to the the the actual movement of this internal system, right? Like, you're actually governing BitTensor. In that way, BitTensor is is highly decentralized. They're probably one of the most decentralized networks ever produced. Like, it's not like 10 nodes, it's not 1,000 nodes, it's tens of thousands of nodes contributing and making making informed bets on the truth. And so that's how we we built. Yeah, that's how we built it. And that's like, that's that's holistically, I think, what it looks like.

• 55:43Speaker 00

  It's extraordinary. Lon, you had a question?

• 55:45Speaker 03

  Yeah, we got to talk more about A5 before we run a full on chat mutiny. They really, the folks want to want to hear about this. So I'm curious, this is the idea is you're building models? And if so, how are you setting up the competition or how are you setting up the landscape to ensure that your miners are producing the next generation of models, like of the kind of model that you want to train and design?

• 56:08Speaker 01

  Last year, we did produce a model that was better than Flynn's best model at the 35 billion range. And just before we launched it, they launched another 35 billion model that was better than ours. And so, you know.

• 56:23Speaker 03

  Everybody's in the race now.

• 56:25Speaker 01

  It's an unbelievable race. And like, you know, it's it's not an understatement to say that we we are up against nation states because this is truly like China versus the United States. And there's that level of funding. So it's it's it's in no it's in no ways easy, but we did produce a very good model, but it didn't take us to to the level we wanted.

• 56:47Speaker 01

  A5's logo is is mining reasoning. And reasoning is the way in which a model thinks to itself so that it can answer a question properly. And so what we have the miners on the network do is we have them produce machine learning models that can produce reasoning, that they can reason in a way that makes other machine learning models answer the right question. So it's sort of indirect. It's like, imagine if I were like, I, you would think that I'm smart if whenever you talked to me, you felt a lot smarter yourself.

• 57:30Speaker 03

  Right.

• 57:32Speaker 01

  Like, I made you better, right? That's a really good sign. It's like, it's actually reflective, right? It's like, you you feel you're you're you know that I'm smart if when you talk to me, like, things make sense to yourself. And and so we, right now, this is this is what we're we're honing in on as the like the core mechanism for for for that network as like, we, even if it doesn't necessarily produce models that are like stylistically perfect, they they can, at the very least, if this is successful, be adapters to other machine learning models. So you could run GLM. And instead of GLM thinking, you would just talk to the smaller model and then be able to go like 30 times faster.

• 58:17Speaker 03

  Got it.

• 58:17Speaker 01

  But that's not that's not the end goal. Actually, the end goal is that we think that we think that mining that that latent space of thought is is going to be the prerequisite for us training incredibly good models. And so that's what we're doing right now. If you go to A5.io, you can you could participate. For people on the call that are interested, like, we're in this really interesting period of time where you don't need to be a machine learning engineer and you don't necessarily need to be a computer scientist to participate in these networks. Like, you could literally just get your open claw and send him the website and he will tell you or it will tell you or whatever you want, the gender you want. They will tell you.

• 58:55Speaker 00

  They them.

• 58:56Speaker 01

  They them, they them, them.

• 58:58Speaker 03

  Gaff is a he. I don't know what you guys are talking about. He's a he and he is my friend, so.

• 59:04Speaker 01

  He's a he. Okay, great. He will tell you what is needed to participate in these games. And and if you you can just basically send your AIs to to participate in these networks, which is really, really interesting time to be alive where, and this is what I was talking about before the call started with you, Lon, about how we work remotely. So I'm I'm nowhere near the rest of my team right now. But we can all work together through these these games. We don't need to be in the same room. We just build these really well defined computational and innovation games. And then anybody in the world can contribute almost without communicating by just sharing the best work they can they can.

• 59:49Speaker 01

  And if you're not good, if you don't come to work, it doesn't matter because you're just going to get replaced very quickly in these hyper competitive systems. So, yeah, the the call out, I've always said from the very beginning, like, I don't actually really sell tau, the token. I I I sell I sell that you can be part of this. And and you should because there's it's very exciting and it's very fun. It's like, for some people, mining BitTensor is like the most fun game they've ever played ever, like, most dynamic, most interesting, most in rewarding thing they've ever done.

• 1:00:22Speaker 03

  Along along those lines, one thing I'm really fascinated about, we've talked to other subnet or we've talked to subnet owners about this. Like, when you're first setting up these competitions, how much are you sort of thinking about the 4D chess of it? Like, I have to create a competition or a system that's so tight that no bad actor can come in and like game my system? Like, how how much a part of the project is that? That sort of adversarial thinking about?

• 1:00:49Speaker 01

  I would say that that is the entirety of the project.

• 1:00:52Speaker 03

  Okay, fair enough.

• 1:00:55Speaker 01

  The the all the other stuff is fluff in some sense.

• 1:00:58Speaker 03

  Yeah.

• 1:00:59Speaker 01

  Because you're you have to think about your.

• 1:01:04Speaker 01

  You can start with all the marketing. You can start with a website and that might help you and people will maybe invest in your project because they're like, oh, this guy's got a good sense of sales. Um, but i- if you, if you can't solve the adversarial problem in the network, then the network produces no value.

• 1:01:20Speaker 00

  Right.

• 1:01:21Speaker 01

  And so, you know, it's, then it's just lipstick on a pig and the, you, you, you're, you're wasting your time, you're wasting my time, you're wasting everyone's time. And that's the most fun thing. And when you get down to, you know, how do you resist adversaries, you often find that there's this very like compressed idea at the very core, um, very simple compressed idea that, um, you're, you're truly building that is, is like a elemental. Right? Okay. Oh, interesting. So inference verification is, is actually information checking, information parrot. You're checking that the information produced by this thing is similar to this thing. And so they're, they're actually, they're producing inferences, but underneath the hood, they're producing information.

• 1:02:06Speaker 01

  And they're trying to produce the information as fast as they can. And, and so that, that is actually the commodity of an inference network where people don't talk about. Um, so anyways, that's like the, you know, the inside baseball philosophy of the stuff, but I, I, I, I love that the most.

• 1:02:22Speaker 00

  There was a, there was an interesting moment when this, um, language model got created. Uh, we were on All In and Chamath brought it up with Jensen. You probably saw that clip where he was like, this is so impressive. Uh, talk a little bit about not only that moment and like, what you took away from it, but then the controversy with that subnet and it imploding. And what we learned from that. Yeah.

• 1:02:50Speaker 01

  The reason why Templar is called Templar is because when we were building this subnet, um, we, we talked about how one of the holy grails of, in the artificial intelligence field and the psyche and new sphere, whatever you call it, um, is not decentralized inference or decentralized computing, or even what I'm talking about with, with affine, with the model training. It's, um, decentralized training, which is where you have a computer and I have a computer, Lon has a computer and they are making the same machine learning model at the same time.

• 1:03:30Speaker 01

  Because in order to train a trillion from our machine learning model, you need a hell of a lot of compute. So in order for, for the, the rest of us to organically, you know, uh, self-organize and to take on OpenAI, we need to come up with a way that we can work together. Um, and because we only together, we will have enough compute to compete with the big guys that have the, the billion dollars investments in, in, in, in infrastructure. So, so, you know, we need to figure out how we can train together. But in order to do that, we come up with some, I come, I come up against some like fundamental limits of physics, which is that the way in which these machine learning models are trained is that they merge their weights

• 1:04:14Speaker 01

  every step, more or less, they merge them. So you do some work and I do some work and then we just merge them together. And you, if, if one of the parameters in your network is pointing this way and the other one's pointing this way and this one's pointing this way, we find this middle point, which would be where we're, the model is the most intelligent because it's we've, you've trained on some data and I've trained on some data, right? Makes sense, right? So, but in order to move these models across the wire, it's like heavily expensive in terms of bandwidth. We're talking like a terabyte of data more, right? And if we want to do hundreds of thousands of steps, well, hundreds of thousands of terabytes that we need to communicate, which is more than any of us have at our home connections and, and it's certainly not what an average person has at their home and not on their laptops. And so

• 1:04:59Speaker 01

  the, how do we aggregate together in the first place? If we can't even do it with the internet connection we have and and maybe we can do it, but it's going to take 10 years, right? Which then is no longer important, right? So, um, and then on the tensor, to put it another, uh, from another perspective, we have all of these, um, compute aggregators like Liam and Targon and Cube, which people actually bring in, they plug in their GPUs, but these GPUs are all over the world. So for us to use them, we need to come up with a decentralized training mechanism. So, so in some sense, like training a trillion, trillion pair of models sort of beyond what bit tensor can do, until we can come up with an algorithm that stitches together all of the compute

• 1:05:38Speaker 01

  in a way that gets a, a, around this bandwidth problem. And it's also the holy grail because, you know, a lot of people are thinking about this problem and people have thought about this problem for a very long time, including myself. So, um, that's why it's called Templar because the Templars were trying to find the holy grail or they're protecting the holy grail. And, and, and so we came up with that name and, and we started working on, um, the, the way in which to do that, which was, um, to take advantage of some of the tensor's primitives, like, um, we, we have hippie's bucket storage and so where all the computers can upload to single places and then download, um, which is kind of cool as we built this out. And then we also built, um,

• 1:06:17Speaker 01

  uh, the algorithm where we could, what would we be measuring that the miners are doing in the first place? So as I, as I talked to you about, like the, the core of the, the, the problem is, okay, well, how do I know that you, you did the inference, right? So in this, it would be, how do we know that you did the training? How did you, how do we know you trained on that particular subset of the data that we needed you to train on in order for you to merge with us? And how do we know that you don't just contribute bullshit that destroys our model while we're training? That's really difficult. And training machine learning models is a very fragile thing. You, if you have one person there that's fucking around with my language, it can destroy the whole thing. So.

• 1:06:55Speaker 00

  This is why, like when Elon talks about, hey, the next version of Grok's coming out, he has Colossus. It's a some number of day training run. You throw some, you know, wrench into that machine, you got to start over. And so that's perfect. So what happened when Templar and the rug pull? I know this is like the one thing people use as an attack, attack vector on bit tensor and the subnet and the architecture. Just candidly, what happened? Did somebody run off with essentially all the tau in their subnet and just tell everybody to fuck off? And, you know, it's just part of the, the system?

• 1:07:33Speaker 01

  I mean, if effectively, you, you know, in,

• 1:07:38Speaker 01

  in, in, in, in any company, you, this, if somebody works in the company, they CEO starts a company, they can just leave. It's actually, usually there's no contract. You invested me, I, I just, you invested my company and I just be like, I don't want to do it.

• 1:07:54Speaker 00

  Right.

• 1:07:54Speaker 01

  Um, this happens in this early stage startup because some people look at what it's like to be an entrepreneur and they go, well, shit, that's a lot of work.

• 1:08:05Speaker 00

  Yeah, not for me.

• 1:08:07Speaker 01

  Yeah. And, and not only is it a lot of work, it's a lot of stress, stress. And, um, I just got some investment into my company and maybe I think I'm just going to take that. So, so what this person did was they just sold their shares. More or less, they sold their shares. And, uh, while they also sold their shares, they wanted to cover up that they were leaving with a, a, a reason for leaving. Like, it's not that I'm a bad guy. It's they're bad. And so I, I look good. I'm going to cover my ass as I do something that's really shitty, which is take a bunch of people's investment and then just walk away, which is a pretty standard thing and in, in crypto, right? Because it's,

• 1:08:47Speaker 01

  you know, one of the things about permissionless markets and us allowing for early stage startups from non-incredible investors and stuff, this is, things like this is, is that basically people get burned and, and, and because, because most of the time, or like a lot of the time, this kind of stuff can happen, right? And so, um, so they, they wrote an article that basically said, hey, we're not, we're not bad for leaving. It's, it's the mean network and, and cons for being an evil dictator. Um, because he sold some of our token and, and that was mean. And, um,

• 1:09:19Speaker 00

  And then what did they go do? They did another startup or they want to?

• 1:09:22Speaker 01

  I know they will. No, the project fell apart immediately. Um, and I mean, of course it does because you, you, you, you, you kind of like, you can't burn your reputation like that. You can't.

• 1:09:32Speaker 00

  Yeah.

• 1:09:33Speaker 01

  Take money from people to, to, to.

• 1:09:35Speaker 00

  This is a rea, it's a reoccurring issue for, uh, Y Combinator, where somebody goes to the Y Combinator program. If you get accepted, not only do they put that hundred and 25K in, they will give you Elon $375,000 in an uncapped note. Then you get a bunch of people excited and you might get a bunch of people on demo data. Let's just theoretically say, put in another million. Well, I had, this is, I've had multiple people contact me that say, Jake, how, you know, you wrote the book Angel. What should I do here? That I'm not getting updates. The project's not active. They raised 1.5 million and they're just using it to live off of. And we don't see any updates and they're not shipping any product. What do we do?

• 1:10:18Speaker 00

  And I say, that's the cost of going to the casino is you could have bad actors. Now you could file a lawsuit, you could do all these kinds of things. The, you put 50K in to file a lawsuit would be 250K. To then pursue it would be a million dollars. And then the outcome is maybe you get back your 50 in the best case scenario. So talking about trust, the startup ecosystem is largely based on trust. We've literally had people take the money from their bank account when we give them these small checks and just YOLO it and go crazy. And it, it's like you, it's kind of like credit card, bad credit cards. You know, you get like a 2% or 1% fraud.

• 1:11:02Speaker 00

  And you're like, okay, whatever, 50 basis points of fraud is what it's going to be.

• 1:11:06Speaker 03

  I mean, this happens in every industry. You remember there was that Netflix sci-fi show they were creating? And the guy just took the budget that they gave him and just went and bought a bunch of mattresses.

• 1:11:16Speaker 00

  Did that really happen?

• 1:11:16Speaker 03

  That really happened. I think he's in jail now or he got, yeah, he got sentenced already, but.

• 1:11:21Speaker 00

  I remember this.

• 1:11:22Speaker 03

  He bought like watches and yeah.

• 1:11:24Speaker 00

  Fraud can happen in Hollywood startups and on the BitTensor network, I think is the. But you did tighten up the Tensor network a bit based on that?

• 1:11:35Speaker 03

  Related question from the chat room. Oh, sorry, go ahead.

• 1:11:38Speaker 01

  Well, I just want to make a point here, right? So it's a double edged sword to allow anybody permissionlessly to have access to early stage startups, right? Like this is one thing that crypto does, right? It's like, okay, you can be on the ground floor potentially where you can't for OpenAI. And so like this, but as a consequence, we've learned that there's a lot of issues. And so crypto has become a lot more prickly and has thorns now because people have woken up to the realities of this network now. So it's double edged sword, but the benefit is that people get access. And, and I would say that overall in the BitTensor ecosystem, we've made people a lot more money than they've lost. So, so that's, you know, I hold on to that now.

• 1:12:23Speaker 01

  But the reason why it's particularly bad in crypto is because in early stage startups, you would invest in, they can't just go and sell their equity into a market automatically. Because there's no, there's not, there's no market for that. There's no, there's no like.

• 1:12:42Speaker 00

  There's no secondary market of startups, although people have tried, but even for the nascent ones, there would be no buy side.

• 1:12:50Speaker 01

  Exactly. And so, but we, we in crypto build those, the buy side and we let these things float. And as a consequence, it's part of our technology, right? I spoke about how like this internal market is how BitTensor works. It's actually a functioning aspect of the machine is that we have individuals that govern the network through market dynamics. It's not something that we're not, we're not just creating alpha tokens for nothing. We're using them to move a, like a computer through space. And, but so the reason why it's particularly pernicious in crypto in general is because anybody can just go and sell on market. Hence, rub pulls, right? It doesn't exist in startups because you, you can't do that. There's, there's not going to be a second market, secondary market.

• 1:13:34Speaker 01

  And if you just go to your investors and go, hey, quickly, give me my $10 million, I, I'm out. They'll go, actually, I don't know if I want to buy that because you're selling. So what we built into the, into the chain was basically the best that we could do is go, hey, let's build sort of an ability for, for the, the submit teams, non enforced, that allow them to, to express their conviction, call it conviction, by locking their tokens effectively. And in a way that if they want to go and sell them in one big sell, that's a public event. And, and anybody who invested before them could be like, you know what? I think I want to get out of here because you just did that.

• 1:14:19Speaker 01

  And so that's actually led to this. It's been very beautiful actually to see a lot of the teams, you know, go up to the plate and be like, yeah, you know what? I'm locking this thing perpetually for years and.

• 1:14:33Speaker 00

  Which by the way, we call that vesting in startups where you vest your shares over time and then you can't sell them. And if you do want to sell them, there's a board decision, hey, we're going to do a secondary offering. As we wrap here, tell us about the foundation and its role. We had a bunch of questions Maria 123 asked as well, not Maria 678. There's Maria 123 Lon.

• 1:14:56Speaker 03

  You've got to keep your Maria's distinct, Jay.

• 1:14:58Speaker 00

  It's so separate here in the chat, but what's the role of the foundation in all of this? And then what's your stake in all of this? Like as the creator of this, do you have like a gazillion Tao? And how do, how do you stay motivated to keep this whole thing going? Cause my gut tells me it's, you know, Tao isn't at the point yet where it would keep going if Const went away. So how far are you away from making yourself irrelevant in this equation? And then what does the foundation do?

• 1:15:26Speaker 01

  I think it would definitely continue if I went away. Most certainly it, it would be, I think, I hope that people would miss me. But, but there's a, there's a vibrant community of people that have, that want to really participate in this network. And they do, and there's an open source community and those people contribute. And there's 128 different teams that understand this technology really well and want this system to go along. And there's the foundation in Canada, which is no longer involved with development. And there's now another foundation, which is more shielded, which is purely about upgrading the chain. So we, we push a lot of changes to, to tweak and improve the, the mechanism.

• 1:16:10Speaker 01

  And that goes through our, you know, internal governance system on the chain, which we called the triumvirate. And that, which is this, this year going to be expanded, basically because we have a chain that we can build government systems directly into it. We, we intend to build all of that this year. So that's, that's, that's, that's the, the two foundations, you might call it. And there's the Open Tensor Foundation and there's the RAL Foundation and they, they hold different sides. One's more marketing and outreach, et cetera. And one's more development and upgrading the chain. I'm no longer involved with Open Tensor. I, I work just, just in the RAL Foundation and I program, that's my language. Actually, this is very unfamiliar for me, like I am a programmer at heart. That's how I speak.

• 1:16:54Speaker 00

  Awesome. Well, listen, continued success with it. We're going to be monitoring all these subnets. As I told everybody, like I think buying one Tao, just buy one Tao is what I've been telling folks. Why? Because I think it is a ticket to watch something spec, some, a spectacular experiment occur. So if you think of it Lon, like going to see a basketball game, just your 200 bucks or whatever it's trading at.

• 1:17:20Speaker 03

  If it's the finals, you're, you're $15,000 or whatever.

• 1:17:24Speaker 00

  Okay, sure. But who's counting? It's, it's a front row seat to like bet and learn and it's not financial advice. But it's also like in my, in one way, a vote for me, like if you were giving a, you know, GoFundMe. Like I almost see it like as a vote of confidence that maybe there could be a decentralized AI intelligence platform out there. And that's good for humanity.

• 1:17:46Speaker 03

  Yeah.

• 1:17:47Speaker 00

  Part of me, so I see it as like, and then, oh, and there's a third. What if it is Bitcoin? And what if it goes from 200 to 60,000 or 120,000? Hey, that could be like a great bet.

• 1:17:57Speaker 03

  Don't sell too early. That's what I learned about Bitcoin. I had a few and then I sold it when it hit like a thousand, like I didn't think it was going to get.

• 1:18:04Speaker 00

  Dummy, you're a big dummy. You gotta ride your winners. Anyway, how important is it to get people promoting this and getting involved in it? Or is that like actually a distraction? The fact that like people like me are speculating on it now and interested in it. And I'm a, obviously a venture capitalist.

• 1:18:20Speaker 03

  It's a very passionate fan base. It's a very talkative, passionate ecosystem. Anytime we talk about Tao, yeah, the, the video goes crazy.

• 1:18:29Speaker 00

  How do you think about the pumping in crypto or, you know, non builders like myself saying, hey, I want to invest in this because I think there's something here and I'm fascinated by it and I like to make a return. And I think this is like risk adjusted for me, like a great, you know, hey, maybe this thing, you know, I look at and go, hey, maybe this thing can go 100 or 1,000X. That's why I make the bet. It's like a real long shot kind of bet in my mind, but it's also fascinating. So how does speculators like me and then pumping and all that impact these projects on a practical basis, if at all?

• 1:19:04Speaker 01

  Well, I mean, I think that it's inevitable and, and we can't really avoid it. It's the nature of markets, but it's not the, the goal. Like the reason why we have Tao in the first place is so that everyone can have some and that everyone can join and that it can be split up. And, and in the first place, they can be split up. It turned into a whole 21 million of these things. So like, I love it that there's people that are excited and that's amazing. And, and the, the ones that are in authentically promoting it, they come, they go. There's always people that are going to say shit and on both sides, I can't stop them.

• 1:19:45Speaker 01

  I, I, I'm way more interested in, in people that, that are, that see it the same way that I see it, which is more as a really, really powerful technology and that are super excited about that. And the potential for us to build something that's novel and unique and, and competitive. And the third path for what artificial intelligence can be born out of. So the, the.

• 1:20:08Speaker 01

  Yeah, but the market dynamics were also really fun. I'm in all the press chats as well.

• 1:20:13Speaker 00

  All right, there you go. We got over an hour with the man himself, constantly shipping. You can follow him on Twitter, Con.

• 1:20:22Speaker 01

  Thanks, Jason.

• 1:20:23Speaker 00

  Reborn. Great follow. And I think it's great that you're going out and talking. So I appreciate it. I know you got to get back to coding. But I think it's important for people to understand it, like just from first principles. He did a great job of sharing that with our audience today. All right, let's drop off.

• 1:20:37Speaker 03

  Yeah, thanks so much for being here, Con. That was great.

• 1:20:39Speaker 00

  Here's the thing, Lon, you know, these projects are so, they have so much potential and the intention is super important. And when you spend an hour talking to him, it reminds me of what I saw in the Bitcoin early True Believers. And I had somebody on, I think in 2011, and I saw it then. I didn't make a big enough bet. And now I see this and it just, all the signaling is going off. So I was like, all right, let me put like, I don't know if I put like half a million or 750 into this, something like that. Not a lot of money for me, be the equivalent of like, you know, maybe somebody putting in, I don't know, a couple of thousand dollars, right?

• 1:21:21Speaker 03

  Five grand.

• 1:21:21Speaker 00

  Or 10,000 dollars, whatever. You know, it's a smaller bet for me. It's not out of my funds. It's just personal. Because I think there's something here that's notable and my signaling goes up, just like I had the open claw signaling. You remember when that happened? I was like, guys, we got to like pay attention to this because this reminds me of the mobile cloud computing era when like local, mobile, GPS all started coming together. I got that signaling for Uber and Robinhood. Hey, what would mobile do to trading and getting a car in GPS and all that stuff. And being early on a lot of these things, same thing I saw when I bought the 16 Tesla. Now, that doesn't mean you're going to be right.

• 1:21:58Speaker 03

  Right, yeah.

• 1:21:59Speaker 00

  But you do have to place the bet, is what I've learned.

• 1:22:02Speaker 03

  Yeah, I mean, and sometimes you are right, but it's just not the right bet at the right moment. I mean, there's no way, you know, there's, there's no way to know which bet is the correct one, but it, you know, it keeps things interesting.

• 1:22:14Speaker 00

  Yes. And, you know, like, the interesting thing for Tesla was, you know, I bought the two cars for 300,000. I wish I just bought 300,000 in shares and just let it sit forever, you know?

• 1:22:23Speaker 03

  Yeah.

• 1:22:24Speaker 00

  And I always go, oh, you know, I did have shares in it, whatever, but I, and I did fine. So I'm not complaining.

• 1:22:28Speaker 03

  I mean, hindsight on this stuff is always 2020. Like, if I could go back in time and tell myself not to sell my few Bitcoin when they were in the thousands, I was like, I can't believe how much money I made. What a windfall. I'm going to go get myself, you know.

• 1:22:41Speaker 00

  And the mistake was, you should, you could have sold 10% or 20%, but you want to keep it, if something's accelerating, you have to ask yourself, is it going to stop accelerating?

• 1:22:51Speaker 03

  Is this just going to keep going?

• 1:22:52Speaker 00

  And so when I look at this, yeah, I, I, I think I might actually be down right now in my BitTensor bet, which I made this year. And I like to be vocal and clear about this so that nobody thinks I have some nefarious reason. And that's why I told folks, just buy one, because I realized people were starting to take my tweets online, Lon. And, you know, there's all these Pelosi tracker, now there's a J Cal tracker. So when I bought Figma, or I said Uber was at a bottom last week when it was like 67, when it, you know, people were panicking.

• 1:23:21Speaker 03

  They think you're Michael Saylor-ing. They think you're, you're.

• 1:23:23Speaker 00

  I'm not YOLO-ing. I'm not trying to influence anybody. I just like to be honest about it. And when I J trade something, I just take a screenshot now. Like, I bought Figma when it was at 20 because I was like, you know what? Dylan's like a beast. We should have him on the program. It's like Dylan's a beast, you know, like, he's not going to sit here and lay down. He understands designers better than anybody. He understands how to make a great tool, build a great brand, get people to pay for it, all that great stuff. I like to make a bet on him. It's now up to $25 a share, whatever. And now they're doing the same thing. They're retweeting it. J Cal called the bottom on this. J Co called the bottom on this, you know. My promise to you, as the people who listen to this specific podcast, when I do it, I'm going to try to be transparent about it to the extent I can.

• 1:24:04Speaker 00

  I can't do that with private companies because it's not my choice to make the funding announcement.

• 1:24:10Speaker 03

  Right.

• 1:24:10Speaker 00

  That's the founder and the board's choice. So sometimes I will make a private bet. I'm not at liberty to talk about that. But I try to be transparent with the public.

• 1:24:18Speaker 03

  Yeah, the J trades.

• 1:24:19Speaker 00

  Only because, and the J trading is all, I'm, I'm not day trading, I'm J trading. J trading is hold for a decade. That's the nature of J trading. Now, you may want to quit a stock if you get new information, but I like to find stocks that I'm comfortable holding for a decade. That's my hold period I'm looking for.

• 1:24:37Speaker 03

  Sure. You got to wait. You got, in tech, the, the strategy is always wait for there to be a bad news cycle. The stock dips, you get it at a little discount, and then you just hold on to it forever. That's, that's the play.

• 1:24:49Speaker 00

  That's your This Week in Startups. We'll see you next time. Bye-bye.

• 1:24:52Speaker 03

  Bye-bye.