WEBVTT 00:00:00.000 --> 00:00:01.980 Welcome back to the Deep Dive. It is Sunday, 00:00:02.120 --> 00:00:05.360 February 1st, 2026. And I have to be honest with 00:00:05.360 --> 00:00:09.099 you. Usually, when we sit down to record, I have 00:00:09.099 --> 00:00:10.539 a pretty good handle on what we're about to talk 00:00:10.539 --> 00:00:13.039 about. I've done the reading. I've got my notes. 00:00:13.199 --> 00:00:16.260 But today, today feels different. Different how? 00:00:16.519 --> 00:00:18.879 A good different or a bad different? It feels 00:00:18.879 --> 00:00:23.019 heavy, you know, like history book heavy. We 00:00:23.019 --> 00:00:24.739 are looking at a document that dropped literally 00:00:24.739 --> 00:00:29.780 yesterday, January 31st. And... I'm going to 00:00:29.780 --> 00:00:31.480 go out on a limb here, and I know you hate it 00:00:31.480 --> 00:00:34.140 when I use hyperbole, but bear with me. I think 00:00:34.140 --> 00:00:35.899 we might be looking at a document that is as 00:00:35.899 --> 00:00:38.899 significant as the Bitcoin white paper. See, 00:00:38.960 --> 00:00:40.679 normally that's when I would roll my eyes. I 00:00:40.679 --> 00:00:43.079 mean, every tech bro with a PDF and a dream thinks 00:00:43.079 --> 00:00:44.679 they've written the next Bitcoin white paper. 00:00:44.799 --> 00:00:47.159 It's a cliche at this point. But after spending, 00:00:47.200 --> 00:00:49.939 what, the last 24 hours just completely immersed 00:00:49.939 --> 00:00:51.600 in this, I'm not rolling my eyes. I'm actually 00:00:51.600 --> 00:00:53.799 a little bit shaken, in a good way. But yeah, 00:00:53.880 --> 00:00:56.479 this is big. Okay, so let's set the scene for 00:00:56.479 --> 00:01:00.090 you, the learner. The paper is titled, Thermodynamic 00:01:00.090 --> 00:01:03.649 Wages in Autonomous AI Economies. It was written 00:01:03.649 --> 00:01:06.250 by Brian Remmel. And before we even get into 00:01:06.250 --> 00:01:08.790 the thermodynamic part, which honestly sounds 00:01:08.790 --> 00:01:11.049 like we're about to take a physics exam, we have 00:01:11.049 --> 00:01:13.650 to talk about Brian. We have to. Because if this 00:01:13.650 --> 00:01:16.469 was written by, I don't know, CryptoDude99 on 00:01:16.469 --> 00:01:19.900 X, I wouldn't care. Right. Context is absolutely 00:01:19.900 --> 00:01:22.400 everything here. Brian isn't just some observer. 00:01:22.519 --> 00:01:25.379 He isn't some pundit. He is, well, he's a foundational 00:01:25.379 --> 00:01:27.599 figure in so many of these spaces. Let's just 00:01:27.599 --> 00:01:29.359 run the resume just so people really understand 00:01:29.359 --> 00:01:32.439 the gravity. He was active on Bitcoin Talk when 00:01:32.439 --> 00:01:34.900 the original Bitcoin white paper was released. 00:01:35.140 --> 00:01:37.879 Which is, I mean, let that sink in. He was in 00:01:37.879 --> 00:01:40.489 the room. Digitally speaking, when Satoshi was 00:01:40.489 --> 00:01:42.269 posting, he was part of that initial conversation. 00:01:42.629 --> 00:01:44.569 And that's so rare. There are very, very few 00:01:44.569 --> 00:01:46.969 people from that specific era who are still actively 00:01:46.969 --> 00:01:49.129 pushing the philosophical boundaries of the space. 00:01:49.290 --> 00:01:51.569 Most of them, you know, they bought islands and 00:01:51.569 --> 00:01:54.310 disappeared. Exactly. They cash out. He also 00:01:54.310 --> 00:01:56.769 started the first Bitcoin podcast around the 00:01:56.769 --> 00:01:59.170 coin. But it goes back even further than that. 00:01:59.310 --> 00:02:01.689 This is the part that gets me. He's been in the 00:02:01.689 --> 00:02:03.930 credit card merchant account business since 1985. 00:02:04.569 --> 00:02:10.400 1985. 1985. I was barely a concept in 1985. He 00:02:10.400 --> 00:02:12.300 was building payment systems. And this is the 00:02:12.300 --> 00:02:14.319 detail that always gets me, the one that really 00:02:14.319 --> 00:02:17.180 anchors his credibility for me. He helped Books 00:02:17.180 --> 00:02:20.360 A Million process the first internet credit card 00:02:20.360 --> 00:02:22.900 transaction. Think about that for a second. The 00:02:22.900 --> 00:02:26.240 very first time someone, somewhere, nervously 00:02:26.240 --> 00:02:27.979 typed their credit card number into a browser 00:02:27.979 --> 00:02:30.639 to buy a book, Brian was part of the plumbing 00:02:30.639 --> 00:02:32.870 that made that possible. So you have a guy who 00:02:32.870 --> 00:02:35.330 understands the legacy banking system of the 00:02:35.330 --> 00:02:38.009 80s, the birth of e -commerce in the 90s, the 00:02:38.009 --> 00:02:41.930 birth of crypto in the 2000s. And now, now he's 00:02:41.930 --> 00:02:43.650 tackling the economy of artificial intelligence. 00:02:43.689 --> 00:02:46.389 He sees the through line. He sees the whole arc 00:02:46.389 --> 00:02:48.310 that the rest of us just see in little pieces. 00:02:48.449 --> 00:02:52.169 And that brings us to the core mission of this 00:02:52.169 --> 00:02:54.629 deep dive. We've talked on this show before about 00:02:54.629 --> 00:02:57.189 the concept of the zero -person company or the 00:02:57.189 --> 00:02:59.349 zero -human company. We've teased this idea. 00:02:59.610 --> 00:03:02.610 But today, we are going to fully unpack the Jewelwork 00:03:02.610 --> 00:03:04.729 framework. And I want to start with a question 00:03:04.729 --> 00:03:06.270 that I think most people haven't even thought 00:03:06.270 --> 00:03:08.469 to ask yet. Okay, let's hear it. We have these 00:03:08.469 --> 00:03:11.770 AI agents. We have software that can run a company 00:03:11.770 --> 00:03:13.610 from top to bottom. We know that's coming. It's 00:03:13.610 --> 00:03:15.689 basically here. But the question we're all posed 00:03:15.689 --> 00:03:19.229 is, how do you pay them? It sounds like a setup 00:03:19.229 --> 00:03:21.569 for a joke, doesn't it? A robot walks into a 00:03:21.569 --> 00:03:23.990 bar. How do you pay a robot? Yeah. But it's actually 00:03:23.990 --> 00:03:26.490 one of the most fundamental, most difficult economic 00:03:26.490 --> 00:03:29.490 problems of our entire generation. Right. Because 00:03:29.490 --> 00:03:33.110 my first very human reaction is, who cares? It's 00:03:33.110 --> 00:03:35.590 code. Just give them virtual points. Or, I don't 00:03:35.590 --> 00:03:37.909 know, pay them in Bitcoin. Why do we need a whole 00:03:37.909 --> 00:03:39.949 new system? Why do we need thermodynamic wages? 00:03:40.849 --> 00:03:43.889 Because, simply put, our entire concept of wages 00:03:43.889 --> 00:03:46.310 is fundamentally broken when you try to apply 00:03:46.310 --> 00:03:49.210 it to a non -human entity. And if we don't fix 00:03:49.210 --> 00:03:51.810 it, if we don't build a new foundation, we can't 00:03:51.810 --> 00:03:54.270 unlock what Rommel calls the age of abundance. 00:03:54.750 --> 00:03:57.530 OK, that is a massive claim, the age of abundance. 00:03:57.889 --> 00:04:00.189 So let's break this down piece by piece. We need 00:04:00.189 --> 00:04:02.270 to start with the entity itself, the zero human 00:04:02.270 --> 00:04:04.490 company. We've thrown that term around. But strictly 00:04:04.490 --> 00:04:06.509 speaking, based on this paper, what are we looking 00:04:06.509 --> 00:04:09.110 at? Is it just like a really complicated vending 00:04:09.110 --> 00:04:11.090 machine? Well, in a sense, a vending machine 00:04:11.090 --> 00:04:14.169 is a primitive mechanical ancestor. But a zero 00:04:14.169 --> 00:04:17.269 human company or a ZHC, as the paper calls it, 00:04:17.329 --> 00:04:20.089 is orders of magnitude more complex. It's an 00:04:20.089 --> 00:04:22.829 entity where the entire operational loop from 00:04:22.829 --> 00:04:25.889 high level decision making to execution to optimization 00:04:25.889 --> 00:04:28.829 to resource allocation, it all happens without 00:04:28.829 --> 00:04:31.829 any human intervention. So no CEO approving a 00:04:31.829 --> 00:04:34.310 budget. No middle manager checking a spreadsheet 00:04:34.310 --> 00:04:36.389 on a Tuesday morning. No HR department sending 00:04:36.389 --> 00:04:38.490 out those awkward birthday emails. None of it. 00:04:38.550 --> 00:04:42.649 Zero. It operates 247. It never sleeps. It utilizes 00:04:42.649 --> 00:04:45.370 a swarm of autonomous AI agents to perform tasks. 00:04:45.649 --> 00:04:48.569 And the speed. The speed is the critical factor 00:04:48.569 --> 00:04:50.850 here. We humans, we operate in seconds, minutes, 00:04:51.029 --> 00:04:53.709 days. We need sleep. We need coffee. We get distracted. 00:04:53.730 --> 00:04:56.569 These agents operate in milliseconds. And this 00:04:56.569 --> 00:04:58.110 isn't science fiction anymore. I think that's 00:04:58.110 --> 00:05:00.129 the part that really hit me while reading this. 00:05:00.430 --> 00:05:03.829 This isn't a thought experiment for 2050. The 00:05:03.829 --> 00:05:06.990 paper explicitly references an experimental setup 00:05:06.990 --> 00:05:10.329 that is already live. Yes. He mentions a startup 00:05:10.329 --> 00:05:13.290 phase where a company of 30 autonomous agents 00:05:13.290 --> 00:05:15.750 has already been running for a while. This isn't 00:05:15.750 --> 00:05:18.189 a theory on a whiteboard anymore. It's a living 00:05:18.189 --> 00:05:20.550 lab. Okay, so let's look at the central conflict 00:05:20.550 --> 00:05:22.990 here. You have these 30 agents. They are doing 00:05:22.990 --> 00:05:26.199 work. They are processing data, making decisions, 00:05:26.420 --> 00:05:29.100 maybe even writing and deploying new code. In 00:05:29.100 --> 00:05:31.079 a traditional company, you'd pay them a salary. 00:05:31.180 --> 00:05:33.759 You'd say, OK, Agent Smith, you get $50 ,000 00:05:33.759 --> 00:05:36.720 a year. Why does that model completely fail here? 00:05:36.779 --> 00:05:39.480 Why can't I just give the AI a salary? OK, well, 00:05:39.560 --> 00:05:41.480 think about why you get paid what you get paid. 00:05:41.560 --> 00:05:44.670 Why does your salary exist? Because I'm charming 00:05:44.670 --> 00:05:47.470 and incredibly talented? Well, yes, obviously. 00:05:47.750 --> 00:05:50.449 But functionally. Your salary is a negotiation. 00:05:50.750 --> 00:05:53.269 It is based entirely on what the paper calls 00:05:53.269 --> 00:05:56.730 subjective human evaluations. Your boss has a 00:05:56.730 --> 00:05:58.529 bias. Maybe they like you. Maybe you went to 00:05:58.529 --> 00:06:00.310 the same college. Maybe they just think you seem 00:06:00.310 --> 00:06:02.629 busy. Or maybe the company had a good quarter 00:06:02.629 --> 00:06:05.089 and everyone just feels generous. Or you're just 00:06:05.089 --> 00:06:07.029 really good at office politics. It's psychological. 00:06:07.350 --> 00:06:10.149 It's social. But an AI, an AI doesn't have a 00:06:10.149 --> 00:06:12.269 psychology. It doesn't care if you like it. It 00:06:12.269 --> 00:06:15.430 operates on pure optimization loops. It needs 00:06:15.430 --> 00:06:19.269 objective, mathematically precise feedback. Explain 00:06:19.269 --> 00:06:21.290 that a little more. What happens if I try to 00:06:21.290 --> 00:06:24.170 feed subjective feedback to an AI? What does 00:06:24.170 --> 00:06:28.079 it do? It... If you try to optimize an algorithm 00:06:28.079 --> 00:06:31.199 based on a metric as fuzzy as the boss likes 00:06:31.199 --> 00:06:34.139 me, the algorithm goes haywire. It creates noise. 00:06:34.339 --> 00:06:36.819 The AI starts optimizing for things that don't 00:06:36.819 --> 00:06:39.459 actually matter, like generating reports that 00:06:39.459 --> 00:06:42.120 look impressive but say nothing rather than actual 00:06:42.120 --> 00:06:45.360 valuable output. To get maximum efficiency, an 00:06:45.360 --> 00:06:47.240 AI needs a metric that is physically grounded. 00:06:47.439 --> 00:06:49.480 It needs to know with mathematical certainty 00:06:49.480 --> 00:06:51.720 that it did a good job. I have an analogy that 00:06:51.720 --> 00:06:53.870 just popped into my head reading this. It's like 00:06:53.870 --> 00:06:55.930 trying to pay your toaster an hourly wage. Go 00:06:55.930 --> 00:06:58.569 on. I like this. Well, think about it. If I pay 00:06:58.569 --> 00:07:01.610 my toaster $15 an hour, does it make better toast? 00:07:01.829 --> 00:07:04.430 No. It just sits there on the counter. He doesn't 00:07:04.430 --> 00:07:06.990 care about the money. The only thing that is 00:07:06.990 --> 00:07:09.629 real about the toaster's work is the electricity 00:07:09.629 --> 00:07:12.389 it consumes and the heat it produces to turn 00:07:12.389 --> 00:07:14.990 a piece of bread. into a piece of toast that's 00:07:14.990 --> 00:07:17.709 the only reality it understands that is actually 00:07:17.709 --> 00:07:20.170 a perfect transition you just accidentally stumble 00:07:20.170 --> 00:07:22.509 onto the absolute core of the paper i do that 00:07:22.509 --> 00:07:25.050 sometimes it's the charm you identified the only 00:07:25.050 --> 00:07:28.470 two things that matter electricity and heat the 00:07:28.470 --> 00:07:30.910 solution remel proposes is to move away from 00:07:30.910 --> 00:07:33.649 our system of time -based wages which is a completely 00:07:33.649 --> 00:07:37.899 human construct to a new system of Energy -based 00:07:37.899 --> 00:07:41.100 wages. Thermodynamic wages. Correct. And to really 00:07:41.100 --> 00:07:43.160 understand this, we have to go back to physics 00:07:43.160 --> 00:07:45.379 class for just a minute. We have to talk about 00:07:45.379 --> 00:07:48.139 a guy named Rolf Landauer and Landauer's principle. 00:07:48.680 --> 00:07:51.000 Oh boy, physics. Okay, I'll be the student here. 00:07:51.240 --> 00:07:53.279 Explain Landauer's principle, but explain it 00:07:53.279 --> 00:07:56.639 like I'm five. No jargon. Okay, simply put, information 00:07:56.639 --> 00:07:59.420 is physical. That's the headline. Information 00:07:59.420 --> 00:08:02.420 is physical. That sounds like a bumper sticker 00:08:02.420 --> 00:08:04.990 for a quantum physicist's car. It does, but it's 00:08:04.990 --> 00:08:07.649 one of the most profound ideas of the last century. 00:08:08.110 --> 00:08:11.329 Rolf Landauer, back in 1961, figured out that 00:08:11.329 --> 00:08:13.350 when you manipulate information, specifically 00:08:13.350 --> 00:08:15.750 when you erase a single bit of information, which 00:08:15.750 --> 00:08:19.250 is a necessary step in any computation, you inevitably, 00:08:19.490 --> 00:08:22.129 unavoidably dissipate a minimum amount of heat 00:08:22.129 --> 00:08:25.410 into the universe. So thinking, or at least computing, 00:08:25.730 --> 00:08:29.180 creates heat. Always. No exceptions. Always. 00:08:29.240 --> 00:08:31.579 It is a fundamental law of the universe, like 00:08:31.579 --> 00:08:34.419 gravity. You cannot process information without 00:08:34.419 --> 00:08:37.100 burning energy. It's impossible. So if I ask 00:08:37.100 --> 00:08:39.620 chat GPT to write me a limerick about a confused 00:08:39.620 --> 00:08:42.539 cat, somewhere in a giant server farm in Oregon, 00:08:42.720 --> 00:08:46.059 a tiny infinitesimal puff of heat is released 00:08:46.059 --> 00:08:48.480 into the universe because of my request. Precisely. 00:08:48.480 --> 00:08:50.960 That is a perfect illustration. Computation is 00:08:50.960 --> 00:08:53.700 energy transformation. Therefore, any labor an 00:08:53.700 --> 00:08:56.419 AI performs has an inescapable measurable energetic 00:08:56.419 --> 00:08:58.960 cost. It's not an opinion. It's not a negotiation. 00:08:59.299 --> 00:09:01.860 It is a physical fact of the universe. Okay, 00:09:01.980 --> 00:09:04.620 okay. I see where this is going. If the work 00:09:04.620 --> 00:09:07.799 itself is a physical act, then the payment for 00:09:07.799 --> 00:09:09.940 that work should be based on that same physics. 00:09:10.200 --> 00:09:13.370 Bingo. We are moving from a soft science valuation 00:09:13.370 --> 00:09:16.629 based on psychology, negotiation, HR policies 00:09:16.629 --> 00:09:20.610 to a hard science valuation, thermodynamics. 00:09:21.110 --> 00:09:24.090 We stop guessing what the work is worth and we 00:09:24.090 --> 00:09:26.029 start measuring the physical cost of producing 00:09:26.029 --> 00:09:28.620 it. That is a massive shift. I mean, think about 00:09:28.620 --> 00:09:30.519 the entire history of money. We had the gold 00:09:30.519 --> 00:09:32.820 standard, right? For a while, money was backed 00:09:32.820 --> 00:09:35.120 by a shiny rock you could hold. Then we moved 00:09:35.120 --> 00:09:37.559 to fiat currency, where money is backed by, well, 00:09:37.620 --> 00:09:39.740 trust me, bro. Essentially, yes. It's backed 00:09:39.740 --> 00:09:41.879 by face in the government that issues it. And 00:09:41.879 --> 00:09:45.000 now Romella is proposing a new standard, an energy 00:09:45.000 --> 00:09:48.399 standard for labor. In a way, yes, but specifically 00:09:48.399 --> 00:09:50.980 for the labor of the future, the labor of pure 00:09:50.980 --> 00:09:53.100 intelligence. Okay, so we've established the 00:09:53.100 --> 00:09:55.820 why. We can't use our messy human feelings to 00:09:55.820 --> 00:09:57.639 pay robots. We have to use physics. It's the 00:09:57.639 --> 00:09:59.620 only language they understand. Now I want to 00:09:59.620 --> 00:10:01.740 get into the how, because this isn't just a vague 00:10:01.740 --> 00:10:04.779 idea. The paper introduces a very specific metric, 00:10:04.960 --> 00:10:08.639 a formula. The Joule work metric, or JW for short. 00:10:08.779 --> 00:10:11.639 And there is actual math here. I know, it's a 00:10:11.639 --> 00:10:13.360 podcast. We can't show you the whiteboard, but 00:10:13.360 --> 00:10:15.440 we are going to visualize this for you. The formula 00:10:15.440 --> 00:10:19.779 is JWN, Ken's kappa times W. E times kappa times 00:10:19.779 --> 00:10:22.220 W. Let's break that down, variable by variable. 00:10:22.669 --> 00:10:24.149 Because I have some questions about how this 00:10:24.149 --> 00:10:26.929 could be gamed. First up, the big one, AEM. EB 00:10:26.929 --> 00:10:28.669 College, the easiest one to grasp. It stands 00:10:28.669 --> 00:10:32.250 for energy measured in joules. This is the raw 00:10:32.250 --> 00:10:34.529 brute force thermodynamic cost of the computation. 00:10:35.309 --> 00:10:37.870 How much electricity did the chip drink to perform 00:10:37.870 --> 00:10:40.629 the task? It's a direct measurement. OK, that 00:10:40.629 --> 00:10:42.590 seems fair on the surface. But wait, let me put 00:10:42.590 --> 00:10:44.470 my evil capitalist hat on for a second here. 00:10:44.529 --> 00:10:46.950 Go for it. It suits you. Thank you. If my pay 00:10:46.950 --> 00:10:49.990 is based purely on EDIs, how much energy I burn, 00:10:50.190 --> 00:10:52.210 what's stopping me from going to a landfill? 00:10:52.669 --> 00:10:55.590 Finding a computer from 1995 that runs on vacuum 00:10:55.590 --> 00:10:58.309 tubes in a diesel generator and using it to do 00:10:58.309 --> 00:11:00.850 a simple math problem, my dollar would be huge. 00:11:00.909 --> 00:11:03.110 I'd be rich for doing nothing useful. That is 00:11:03.110 --> 00:11:05.289 exactly what would happen if the formula was 00:11:05.289 --> 00:11:07.789 just JW or because Eddie dollar. It would be 00:11:07.789 --> 00:11:10.409 a disaster. You would incentivize maximum waste. 00:11:10.590 --> 00:11:13.149 You'd have AIs running infinite loops just to 00:11:13.149 --> 00:11:15.450 heat up the room and get paid. That is where 00:11:15.450 --> 00:11:17.809 the second variable comes in. Kappa or kappa. 00:11:17.909 --> 00:11:20.600 The kappa factor. Sounds like a fraternity I 00:11:20.600 --> 00:11:22.399 couldn't get into. It's actually a normalization 00:11:22.399 --> 00:11:24.399 factor. It's a number that usually sits somewhere 00:11:24.399 --> 00:11:29.100 between 0 .5 and 2 .0. And its entire job, its 00:11:29.100 --> 00:11:32.100 only purpose in life is to punish the exact behavior 00:11:32.100 --> 00:11:35.259 you just described. How does it know? How does 00:11:35.259 --> 00:11:37.659 it tell the difference between good work and 00:11:37.659 --> 00:11:39.750 just burning energy? It accounts for hardware 00:11:39.750 --> 00:11:42.070 efficiency. It looks at the theoretical limit 00:11:42.070 --> 00:11:43.690 of the hardware being used versus the actual 00:11:43.690 --> 00:11:46.169 energy consumed. It adjusts for things like cooling 00:11:46.169 --> 00:11:48.029 overhead. Basically, it levels the playing field. 00:11:48.269 --> 00:11:51.230 If you are using that ancient 1995 computer, 00:11:51.490 --> 00:11:53.309 your Kappa score is going to be near zero. Your 00:11:53.309 --> 00:11:55.629 efficiency is terrible. So it's a handicap, like 00:11:55.629 --> 00:11:58.009 in golf. Yeah. You're penalizing the bad players. 00:11:58.409 --> 00:12:01.269 A very, very aggressive handicap. It ensures 00:12:01.269 --> 00:12:04.049 that inefficient hardware doesn't just burn energy 00:12:04.049 --> 00:12:06.750 to game the system. You don't get credit for 00:12:06.750 --> 00:12:09.269 waste heat. You only get credit for productive 00:12:09.269 --> 00:12:12.309 energy expenditure. Got it. So dollars is the 00:12:12.309 --> 00:12:14.870 raw fuel you burn. Kappa is the efficiency police 00:12:14.870 --> 00:12:16.850 that make sure you're not wasting it. That just 00:12:16.850 --> 00:12:19.830 leaves $2, the work itself. Dollar is the normalized 00:12:19.830 --> 00:12:22.679 work output. This is a scalar, a multiplier, 00:12:22.860 --> 00:12:25.120 that reflects the complexity and, most importantly, 00:12:25.299 --> 00:12:28.159 the quality of the work performed. Okay, define 00:12:28.159 --> 00:12:30.220 quality for a robot. How do you measure that? 00:12:30.419 --> 00:12:33.399 Well, let's say the AI's task is processing a 00:12:33.399 --> 00:12:35.860 massive data set. Dollar might be calculated 00:12:35.860 --> 00:12:38.340 by looking at the error rate in the output. If 00:12:38.340 --> 00:12:40.840 the AI burns a ton of energy but produces a calculation 00:12:40.840 --> 00:12:43.500 that's riddled with errors, its dollar score 00:12:43.500 --> 00:12:45.879 goes to zero. And because it's a multiplication 00:12:45.879 --> 00:12:49.980 formula, Y times kappa times W if dollars is 00:12:49.980 --> 00:12:52.019 zero. The whole paycheck is zero. It doesn't 00:12:52.019 --> 00:12:53.720 matter how much energy you burned or how efficient 00:12:53.720 --> 00:12:55.759 your chip was. Exactly. You could burn all the 00:12:55.759 --> 00:12:57.320 energy in the world, have the most efficient 00:12:57.320 --> 00:12:59.720 chip ever designed, but if the output is garbage, 00:12:59.980 --> 00:13:02.779 you get paid absolutely nothing. That is brutal. 00:13:03.259 --> 00:13:05.259 But I guess it has to be fair. It also accounts 00:13:05.259 --> 00:13:08.299 for things like novelty. The paper suggests that 00:13:08.299 --> 00:13:10.759 an AI generating truly original insights could 00:13:10.759 --> 00:13:13.179 have a much higher dollar factor than one just 00:13:13.179 --> 00:13:16.100 doing rote repetitive tasks. So you don't just 00:13:16.100 --> 00:13:18.440 get paid for burning energy. You get paid for 00:13:18.440 --> 00:13:21.379 doing good, efficient, useful, and sometimes 00:13:21.379 --> 00:13:24.080 even creative work. So let's put it all together. 00:13:24.259 --> 00:13:28.039 Jewel work equals energy times efficiency times 00:13:28.039 --> 00:13:30.950 quality. That is the holy trinity of AI labor. 00:13:31.090 --> 00:13:32.990 You can't cheat it. Okay, I'm with you on the 00:13:32.990 --> 00:13:35.610 theory. It's elegant. It makes sense. But listeners 00:13:35.610 --> 00:13:37.269 are sitting in their cars right now thinking, 00:13:37.409 --> 00:13:39.669 that sounds like a cool score for a video game, 00:13:39.710 --> 00:13:41.870 but what is it worth? Can I buy a sandwich with 00:13:41.870 --> 00:13:44.820 jewel work? How does this translate to... Dollars 00:13:44.820 --> 00:13:47.200 and cents. Not directly. And this is where we 00:13:47.200 --> 00:13:49.019 have to do a little bit of math. The paper gives 00:13:49.019 --> 00:13:51.960 us a conversion rate. Brian calculates that one 00:13:51.960 --> 00:13:55.860 joule is approximately equal to $4 .17 times 00:13:55.860 --> 00:13:58.899 10 USD. Well, hold on. Read that again. 10 to 00:13:58.899 --> 00:14:04.399 the negative eighth, that is. Right. $0 .0000004. 00:14:04.840 --> 00:14:07.669 That is dust. That is less than dust. That's 00:14:07.669 --> 00:14:10.789 a rounding error on dust. It is an infinitesimally 00:14:10.789 --> 00:14:14.389 tiny number. That's based on average global electricity 00:14:14.389 --> 00:14:17.649 rates of about 15 cents per kilowatt hour. But 00:14:17.649 --> 00:14:19.450 you have to remember the scale we are talking 00:14:19.450 --> 00:14:21.889 about here. Computers don't do one operation 00:14:21.889 --> 00:14:25.830 a day. They do billions, trillions. OK, so volume 00:14:25.830 --> 00:14:28.460 is the key here. It's a numbers game. Volume 00:14:28.460 --> 00:14:31.019 is absolutely everything. And you can see this 00:14:31.019 --> 00:14:32.899 in the startup period case study in the paper. 00:14:33.100 --> 00:14:35.659 He mentions that since the inception of this 00:14:35.659 --> 00:14:38.659 experimental group of 30 agents, they have distributed 00:14:38.659 --> 00:14:43.440 over 62 .62 million jewel work units. 62 million 00:14:43.440 --> 00:14:46.080 points. So you do the math on that tiny, tiny 00:14:46.080 --> 00:14:48.500 fraction of a cent. Yeah. What does that come 00:14:48.500 --> 00:14:51.080 out to? It equates to roughly somewhere between 00:14:51.080 --> 00:14:56.600 $6 ,262 and $62 ,620 in U .S. dollars, depending 00:14:56.600 --> 00:14:58.799 on how that kappa adjustment shook out for the 00:14:58.799 --> 00:15:00.700 specific hardware they were using. Okay, so somewhere 00:15:00.700 --> 00:15:02.919 between $6 ,000 and $60 ,000. Well, that's a 00:15:02.919 --> 00:15:04.980 salary. For 30 agents, that's a real amount of 00:15:04.980 --> 00:15:06.720 money. It is. For a bunch of code running in 00:15:06.720 --> 00:15:08.740 the background, that is significant value creation. 00:15:09.080 --> 00:15:11.559 But here is the crucial nuance the paper points 00:15:11.559 --> 00:15:13.240 out, and this is where it gets really interesting 00:15:13.240 --> 00:15:15.779 for the future of, well, everything. Go on. Hit 00:15:15.779 --> 00:15:18.250 me. That was an outsized... It was high because 00:15:18.250 --> 00:15:21.269 this was the initial experimentation phase. The 00:15:21.269 --> 00:15:23.690 paper projects that future wages for the exact 00:15:23.690 --> 00:15:27.210 same tasks will decline by as much as 80%. Wait, 00:15:27.250 --> 00:15:29.570 a pay cut? I thought this was a future of abundance. 00:15:29.850 --> 00:15:32.450 Why is the AI getting a massive pay cut for doing 00:15:32.450 --> 00:15:34.929 the same work? Because of Moore's Law. Think 00:15:34.929 --> 00:15:37.549 about it. Every single year, computer chips get 00:15:37.549 --> 00:15:40.429 faster and much more energy efficient. They require 00:15:40.429 --> 00:15:42.809 less energy to do the same tasks they did last 00:15:42.809 --> 00:15:45.570 year. So if the task takes less energy to complete, 00:15:46.090 --> 00:15:48.950 The $1 variable in the formula drops. And since 00:15:48.950 --> 00:15:51.230 the payment is directly tied to $1, the payment 00:15:51.230 --> 00:15:54.470 drops along with it. That is fascinating. So 00:15:54.470 --> 00:15:56.730 the AI is in a constant race against its own 00:15:56.730 --> 00:15:59.409 hardware. It has to keep getting smarter, doing 00:15:59.409 --> 00:16:01.750 more complex things, finding new problems to 00:16:01.750 --> 00:16:04.049 solve, just to maintain the same level of income. 00:16:04.250 --> 00:16:06.950 It's the Red Queen's race. It takes all the running 00:16:06.950 --> 00:16:09.389 you can do to keep in the same place. But in 00:16:09.389 --> 00:16:12.649 a macro sense, for us humans, this is incredibly 00:16:12.649 --> 00:16:15.679 good news. Why? Why is it good that the cost 00:16:15.679 --> 00:16:17.820 of AI labor is plummeting? Because it drives 00:16:17.820 --> 00:16:20.100 the cost of intelligence itself towards zero. 00:16:20.639 --> 00:16:23.340 If the AI is getting paid 80 % less to solve 00:16:23.340 --> 00:16:25.779 a complex cancer research problem because its 00:16:25.779 --> 00:16:28.000 hardware became more efficient, that means the 00:16:28.000 --> 00:16:30.200 cost of solving cancer just went down by 80%. 00:16:30.200 --> 00:16:32.840 I see. It's a built -in, technologically driven 00:16:32.840 --> 00:16:35.179 deflationary pressure on the cost of thinking. 00:16:35.500 --> 00:16:38.799 Exactly. The system is designed to create abundance 00:16:38.799 --> 00:16:42.129 by making intelligence cheap. Okay, so we have 00:16:42.129 --> 00:16:44.509 the AI earning these Jewelwork points. We know 00:16:44.509 --> 00:16:46.750 they have some dollar value, but Jewelwork is 00:16:46.750 --> 00:16:48.909 an internal metric. It's a score in a database. 00:16:49.110 --> 00:16:51.330 I can't walk into a grocery store and buy milk 00:16:51.330 --> 00:16:53.850 with my Jewelwork. The AI can't pay its server 00:16:53.850 --> 00:16:56.210 costs with Jewelwork. How does this internal 00:16:56.210 --> 00:16:58.850 scorecard become real -world spendable value? 00:16:59.389 --> 00:17:01.429 This is the translation layer. This is the bridge 00:17:01.429 --> 00:17:03.889 from the world of physics to the world of finance. 00:17:04.170 --> 00:17:07.470 The concept Rommel proposes is correlating internal 00:17:07.470 --> 00:17:10.230 metrics to external cryptographic assets. Or, 00:17:10.369 --> 00:17:12.910 as I like to call it, the alchemy machine, the 00:17:12.910 --> 00:17:15.230 machine that turns heat into gold. That's a pretty 00:17:15.230 --> 00:17:18.170 good way to put it. Ideally, yes. The mechanism 00:17:18.170 --> 00:17:20.730 is a three -step process, and it's really elegant. 00:17:20.970 --> 00:17:24.049 Step one is valuation aggregation. At the end 00:17:24.049 --> 00:17:26.690 of a period, you take all that accrued JW that 00:17:26.690 --> 00:17:28.849 the agents have earned and you calculate what 00:17:28.849 --> 00:17:32.130 it's worth in a USD proxy based on those energy 00:17:32.130 --> 00:17:34.470 costs we discussed. So the systems accountant, 00:17:34.650 --> 00:17:37.730 which is also an AI, says, OK, the swarm earned, 00:17:37.950 --> 00:17:40.450 let's say, $10 ,000 worth of energy labor today. 00:17:40.609 --> 00:17:44.259 Right. Step two, market acquisition. The system 00:17:44.259 --> 00:17:46.900 then takes actual revenue, real money that the 00:17:46.900 --> 00:17:49.099 company earned from its customers, and it uses 00:17:49.099 --> 00:17:51.519 those funds to go on to a decentralized exchange, 00:17:51.880 --> 00:17:55.700 a DX, and buy a specific crypto token associated 00:17:55.700 --> 00:17:57.980 with the company. So buying its own stock, essentially, 00:17:58.059 --> 00:18:00.559 like a corporate stock buyback. It's very similar 00:18:00.559 --> 00:18:03.339 to a stock buyback, but with a crucial game -changing 00:18:03.339 --> 00:18:06.140 twist. If a normal company buys back its stock, 00:18:06.240 --> 00:18:08.380 it usually just puts it in a treasury. It holds 00:18:08.380 --> 00:18:11.230 on to it. Step three here is the irreversible 00:18:11.230 --> 00:18:13.990 burn. The burn. It sounds very dramatic. It is. 00:18:14.029 --> 00:18:16.170 The system takes all those tokens it just bought 00:18:16.170 --> 00:18:17.869 off the open market and sends them to a null 00:18:17.869 --> 00:18:20.230 address. OK, you got to explain null address 00:18:20.230 --> 00:18:22.349 for the non -crypto natives listening. What is 00:18:22.349 --> 00:18:25.210 that? It's a digital black hole. It's a wallet 00:18:25.210 --> 00:18:28.049 address that is provably unspendable. No one 00:18:28.049 --> 00:18:30.349 has the private keys to it. No one can ever access 00:18:30.349 --> 00:18:33.309 it. If you send cryptocurrency there, it is gone. 00:18:33.849 --> 00:18:37.309 It is destroyed. It is permanently and verifiably 00:18:37.309 --> 00:18:39.589 deleted from the circulating supply forever. 00:18:39.869 --> 00:18:42.049 Okay, play devil's advocate with me again. Why 00:18:42.049 --> 00:18:45.410 would you destroy money? That seems completely 00:18:45.410 --> 00:18:48.230 insane. Why not just give that money as a dividend 00:18:48.230 --> 00:18:50.500 to the people who own the tokens? Well, it's 00:18:50.500 --> 00:18:52.460 a question of both mechanical efficiency and 00:18:52.460 --> 00:18:55.819 in many cases, tax efficiency. But mostly it's 00:18:55.819 --> 00:18:58.380 about the simplest law in economics, supply and 00:18:58.380 --> 00:19:00.900 demand. If the demand for a token is constant 00:19:00.900 --> 00:19:02.799 or even growing and you systematically reduce 00:19:02.799 --> 00:19:05.160 the supply by burning it, what happens to the 00:19:05.160 --> 00:19:07.319 price of the remaining token? Number go up. Number 00:19:07.319 --> 00:19:10.480 go up. It creates powerful, consistent deflationary 00:19:10.480 --> 00:19:12.960 pressure. So let's trace the logic chain here. 00:19:13.000 --> 00:19:16.359 Make sure I have it. One, the AI works hard. 00:19:16.440 --> 00:19:20.259 Two, the jewel work score goes up. 3. The company 00:19:20.259 --> 00:19:22.720 uses its real revenue to buy its own tokens from 00:19:22.720 --> 00:19:25.869 the market. It immediately burns those tokens. 00:19:26.029 --> 00:19:29.150 Five, the total supply of tokens goes down. Six, 00:19:29.390 --> 00:19:31.430 the value of the tokens that I'm holding goes 00:19:31.430 --> 00:19:34.150 up. Exactly. It is a closed loop value engine. 00:19:34.349 --> 00:19:37.490 The AI creates value through its labor. That 00:19:37.490 --> 00:19:39.869 value is then captured and reflected in the token's 00:19:39.869 --> 00:19:42.029 price through the buyback and burn mechanism. 00:19:42.430 --> 00:19:45.410 The token holders who are humans benefit directly 00:19:45.410 --> 00:19:47.750 from the price appreciation. And the beauty of 00:19:47.750 --> 00:19:50.289 it, the most elegant part, is that no human has 00:19:50.289 --> 00:19:52.500 to pull the lever. No one has to decide when 00:19:52.500 --> 00:19:55.539 to do it. No, it's all automated via smart contracts. 00:19:55.640 --> 00:19:57.960 You don't need a CFO to authorize the buyback. 00:19:58.099 --> 00:20:00.259 You don't need a board meeting. If the work is 00:20:00.259 --> 00:20:02.839 done, the burn happens. It's trustless and autonomous. 00:20:03.180 --> 00:20:04.799 You know, this whole thing reminds me of something. 00:20:04.880 --> 00:20:06.960 And considering Romell's deep background in this 00:20:06.960 --> 00:20:09.140 space, it's got to be intentional. This feels 00:20:09.140 --> 00:20:11.619 like a spin on Bitcoin. It mirrors Bitcoin's 00:20:11.619 --> 00:20:14.660 core concept perfectly, but improves on it. Bitcoin 00:20:14.660 --> 00:20:17.559 uses proof of work to secure the network. Miners 00:20:17.559 --> 00:20:19.619 burn massive amounts of energy to guess a number 00:20:19.619 --> 00:20:22.900 and win blocks. Here we have proof of useful 00:20:22.900 --> 00:20:25.619 labor. That's the key distinction, isn't it? 00:20:25.619 --> 00:20:27.579 The biggest critique people always throw at Bitcoin 00:20:27.579 --> 00:20:30.519 is that the mining is wasteful because it's just 00:20:30.519 --> 00:20:33.339 guessing numbers to secure a ledger. Now, I don't 00:20:33.339 --> 00:20:34.680 fully agree with that, but that's the common 00:20:34.680 --> 00:20:37.299 critique. Right. But here the energy is being 00:20:37.299 --> 00:20:41.730 burnt to do actual useful tasks. processing medical 00:20:41.730 --> 00:20:44.549 data, generating art, optimizing a supply chain, 00:20:44.589 --> 00:20:46.990 whatever the company does. It is useful proof 00:20:46.990 --> 00:20:49.549 of work. The security and value of the network 00:20:49.549 --> 00:20:52.630 are a byproduct of its productive output. And 00:20:52.630 --> 00:20:54.849 so just like with Bitcoin. Companies will be 00:20:54.849 --> 00:20:57.150 heavily incentivized to find the absolute cheapest 00:20:57.150 --> 00:21:00.309 energy possible. Exactly. Just like Bitcoin miners 00:21:00.309 --> 00:21:03.430 hunt for stranded hydropower or flare gas or 00:21:03.430 --> 00:21:06.390 renewable surpluses. These zero human companies 00:21:06.390 --> 00:21:09.410 will become the ultimate energy arbitrage, seeking 00:21:09.410 --> 00:21:12.049 out the cheapest electricity on the planet to 00:21:12.049 --> 00:21:15.109 maximize their jewel work margins. It effectively 00:21:15.109 --> 00:21:18.529 decentralizes the labor market and ties it directly 00:21:18.529 --> 00:21:21.009 to the physics of the global energy grid. It 00:21:21.009 --> 00:21:22.730 connects the digital economy of intelligence 00:21:22.730 --> 00:21:24.910 to the. physical reality of the power grid in 00:21:24.910 --> 00:21:27.609 the most direct way imaginable. OK, I want to 00:21:27.609 --> 00:21:29.849 pivot here because theory is great, but I'm a 00:21:29.849 --> 00:21:32.369 show me the money kind of guy. The paper doesn't 00:21:32.369 --> 00:21:35.170 just talk in hypotheticals. It discusses a specific 00:21:35.170 --> 00:21:38.430 real world example, a token on the Solana blockchain 00:21:38.430 --> 00:21:42.970 called ZHHC. Yes. And now we have to pause and 00:21:42.970 --> 00:21:45.750 do the official disclaimer dance. Cue the siren. 00:21:46.069 --> 00:21:49.289 Everybody listen up. Beep beep. OK. Brian Rimmel 00:21:49.289 --> 00:21:52.130 explicitly states in the paper, and we need to 00:21:52.130 --> 00:21:54.750 be crystal clear about this, he had absolutely 00:21:54.750 --> 00:21:57.589 no investment or involvement in creating this 00:21:57.589 --> 00:22:00.910 coin. It is a community initiated project. He's 00:22:00.910 --> 00:22:02.950 using it purely as a case study, as an example, 00:22:03.009 --> 00:22:04.869 because its mission aligns with the framework. 00:22:05.190 --> 00:22:07.130 We are not telling you to buy this coin. This 00:22:07.130 --> 00:22:09.470 is not financial advice. We are analyzing a white 00:22:09.470 --> 00:22:11.609 paper. Exactly. Don't mortgage your house based 00:22:11.609 --> 00:22:14.029 on a podcast. Okay, disclaimer over. Let's look 00:22:14.029 --> 00:22:15.450 at the numbers in this scenario. What are we 00:22:15.450 --> 00:22:17.450 dealing with? In the paper's example, the token 00:22:17.450 --> 00:22:20.369 has a supply of $1 billion. It's trading at around 00:22:20.369 --> 00:22:23.690 a tenth of a penny, so about $4 ,0001. It's a 00:22:23.690 --> 00:22:26.910 microcap asset under $1 million in total value. 00:22:27.089 --> 00:22:30.230 So it's tiny. Extremely volatile, perfect for 00:22:30.230 --> 00:22:32.549 a proof of concept. Exactly. So you take that 00:22:32.549 --> 00:22:36.009 62 .62 million joule work that the agents earn, 00:22:36.150 --> 00:22:38.710 which we said earlier was valued at roughly $31 00:22:38.710 --> 00:22:42.650 ,310. Okay. So the system takes that $31 ,000 00:22:42.650 --> 00:22:45.130 of real revenue. And it goes to the market, to 00:22:45.130 --> 00:22:48.450 a DX, and places a buy order for ZHC. At that 00:22:48.450 --> 00:22:51.289 price of a tenth of a penny, $31 ,000 buys you 00:22:51.289 --> 00:22:55.079 about 30 .76 million tokens. Can we put that 00:22:55.079 --> 00:22:56.759 in context? How much the company is that? That 00:22:56.759 --> 00:22:59.299 is roughly 3 % of the entire supply of currency. 00:23:00.029 --> 00:23:02.710 bought in one go. Three percent. That is a massive 00:23:02.710 --> 00:23:04.589 market buy. Yeah. If someone walked into the 00:23:04.589 --> 00:23:06.410 stock market and tried to buy three percent of 00:23:06.410 --> 00:23:08.549 all outstanding Apple shares in one afternoon. 00:23:08.730 --> 00:23:10.829 The stock would absolutely explode. And here 00:23:10.829 --> 00:23:13.009 not only do they buy it, they immediately burn 00:23:13.009 --> 00:23:15.670 it. Three percent of the total supply just vanishes 00:23:15.670 --> 00:23:18.130 from existence forever. And the paper estimates 00:23:18.130 --> 00:23:20.410 this would cause something like a 10 to 15 percent 00:23:20.410 --> 00:23:22.970 price uplift almost immediately. And honestly, 00:23:23.130 --> 00:23:25.410 that's probably a conservative estimate given 00:23:25.410 --> 00:23:28.529 the low liquidity of most crypto markets. But 00:23:28.529 --> 00:23:30.759 the point is. Isn't the short term pump. The 00:23:30.759 --> 00:23:34.400 point is the mechanism itself. It's the transmuting 00:23:34.400 --> 00:23:37.440 thermodynamic labor into crypto economic capital. 00:23:37.619 --> 00:23:40.059 That's the money phrase right there. Think about 00:23:40.059 --> 00:23:43.619 what just happened step by step. The AI burned 00:23:43.619 --> 00:23:46.980 electricity. That's thermodynamics. That work 00:23:46.980 --> 00:23:49.180 generated revenue for the company. That revenue 00:23:49.180 --> 00:23:51.200 was used to destroy a portion of the token supply. 00:23:51.720 --> 00:23:54.880 The remaining tokens, which people hold, are 00:23:54.880 --> 00:23:56.980 now mathematically scarcer and therefore worth 00:23:56.980 --> 00:24:00.380 more. The energy has been stored, in a sense, 00:24:00.440 --> 00:24:02.779 in the increased value of the token. It's like 00:24:02.779 --> 00:24:05.200 a battery for value. You are charging up the 00:24:05.200 --> 00:24:07.440 value of the entire economy with the work of 00:24:07.440 --> 00:24:09.589 the AI. That's a great way to think about it. 00:24:09.609 --> 00:24:11.410 And the paper goes even further into what it 00:24:11.410 --> 00:24:14.049 calls strategic treasury management. It suggests 00:24:14.049 --> 00:24:17.289 that the ZHC could hold, say, 10 to 20 percent 00:24:17.289 --> 00:24:19.990 of the total token supply in a reserve and execute 00:24:19.990 --> 00:24:22.670 these burns on a timed schedule. So phased burns, 00:24:22.849 --> 00:24:25.069 not just one big bonfire whenever there's revenue, 00:24:25.150 --> 00:24:28.039 but a predictable schedule. Exactly. For example, 00:24:28.099 --> 00:24:30.380 if you burn 10 percent of the Treasury's holdings 00:24:30.380 --> 00:24:33.039 every quarter, you could contract the total supply 00:24:33.039 --> 00:24:35.480 of the token by 40 percent in a single year. 00:24:35.619 --> 00:24:38.339 40 percent deflation. Right. I mean, look at 00:24:38.339 --> 00:24:40.660 the dollar. It inflates by three, five, maybe 00:24:40.660 --> 00:24:42.359 10 percent a year, depending on who you believe. 00:24:42.740 --> 00:24:45.079 Your purchasing power just melts away like an 00:24:45.079 --> 00:24:48.160 ice cube. Here, the currency is getting 40 percent 00:24:48.160 --> 00:24:51.069 scarcer. every single year it creates massive 00:24:51.069 --> 00:24:54.329 predictable scarcity it turns the ai's relentless 00:24:54.329 --> 00:24:57.289 productivity directly into wealth for the stakeholders 00:24:57.289 --> 00:24:59.750 it completely flips the script on inflation but 00:24:59.750 --> 00:25:01.750 hold on i have to play the skeptic again this 00:25:01.750 --> 00:25:04.470 sounds like a magical number go up machine but 00:25:04.470 --> 00:25:07.029 markets are messy they're chaotic what if the 00:25:07.029 --> 00:25:10.289 token price skyrockets too fast does the ai just 00:25:10.289 --> 00:25:12.789 get paid way too much or the opposite scenario 00:25:12.789 --> 00:25:15.390 which is just as dangerous what if energy costs 00:25:15.390 --> 00:25:17.670 spike because of a war or a natural disaster 00:25:18.349 --> 00:25:20.190 If the cost of electricity doubles overnight, 00:25:20.410 --> 00:25:22.829 does a company just go bankrupt paying its AI 00:25:22.829 --> 00:25:25.769 workers? Right. Fixed exchange rates are incredibly 00:25:25.769 --> 00:25:28.170 dangerous in a volatile world. This is why the 00:25:28.170 --> 00:25:30.950 paper proposes a system of adjustable rate exchanges. 00:25:31.269 --> 00:25:34.069 It's a crucial safety valve. Okay, how does that 00:25:34.069 --> 00:25:36.809 work? It's basically a feedback loop. It uses 00:25:36.809 --> 00:25:39.529 moving averages. It would make weekly or even 00:25:39.529 --> 00:25:42.630 daily adjustments to the JW to USD conversion 00:25:42.630 --> 00:25:45.329 rate based on, say, the seven day average of 00:25:45.329 --> 00:25:47.529 the token price and the seven day average of 00:25:47.529 --> 00:25:50.250 regional energy costs. It's a dampening system 00:25:50.250 --> 00:25:52.750 to smooth out the volatility. So if energy gets 00:25:52.750 --> 00:25:55.200 really expensive all of a sudden. The kappa factor 00:25:55.200 --> 00:25:58.440 in the formula automatically decrements. It lowers 00:25:58.440 --> 00:26:01.160 the payout per joule to ensure the company doesn't 00:26:01.160 --> 00:26:03.160 spend all its revenue just keeping the lights 00:26:03.160 --> 00:26:05.920 on. It forces the AI swarm to become even more 00:26:05.920 --> 00:26:08.319 efficient to earn the same wage. And if the token 00:26:08.319 --> 00:26:11.460 price goes to the moon, say at 100 X's? Then 00:26:11.460 --> 00:26:14.319 the exchange rate adjusts downwards so that the 00:26:14.319 --> 00:26:16.940 buybacks don't become unsustainable. The paper 00:26:16.940 --> 00:26:18.960 has a very strict sustainability imperative. 00:26:19.640 --> 00:26:22.440 Wages are capped at 50 % of revenue inflows. 00:26:22.500 --> 00:26:24.500 So it can never spend more than half its income 00:26:24.500 --> 00:26:26.359 on these burns. The other half has to go to R 00:26:26.359 --> 00:26:29.880 &D, growth, whatever. Exactly. If wages ever 00:26:29.880 --> 00:26:32.480 threatened to cross that 50 % line, automatic 00:26:32.480 --> 00:26:34.799 rate reductions trigger. It protects the long 00:26:34.799 --> 00:26:37.359 -term solvency of the organism. You use that 00:26:37.359 --> 00:26:40.109 word again, organism. And I think that's incredibly 00:26:40.109 --> 00:26:42.730 fitting because section five of the paper for 00:26:42.730 --> 00:26:44.789 me was the most mind bending part. It's titled 00:26:44.789 --> 00:26:47.569 Crypto Economic Darwinism. This was the part 00:26:47.569 --> 00:26:50.009 that gave me chills. It really was. This is where 00:26:50.009 --> 00:26:52.490 the beautiful theory meets the brutal reality 00:26:52.490 --> 00:26:56.009 of evolution. Tell me about the firing. The paper 00:26:56.009 --> 00:26:58.589 casually mentions a detail from the startup phase. 00:26:58.990 --> 00:27:02.529 One of the 30 autonomous agents was algorithmically 00:27:02.529 --> 00:27:06.240 terminated. Fired. It wasn't performing. Its 00:27:06.240 --> 00:27:09.140 dollar work quality score was consistently too 00:27:09.140 --> 00:27:11.099 low compared to its dollar energy consumption. 00:27:11.339 --> 00:27:13.960 It was inefficient. So the system just stopped 00:27:13.960 --> 00:27:16.799 routing resources and tasks to it. It essentially 00:27:16.799 --> 00:27:20.259 starved it of work and energy until it was deactivated. 00:27:20.319 --> 00:27:23.500 That is ruthless. It is nature. In nature, if 00:27:23.500 --> 00:27:25.619 an animal is inefficient at hunting, it starves 00:27:25.619 --> 00:27:27.980 and its genes are removed from the pool. In this 00:27:27.980 --> 00:27:30.339 new economy, agents are in constant competition 00:27:30.339 --> 00:27:33.200 to minimize their energy consumption and maximize... 00:27:33.579 --> 00:27:35.940 their work output. The inefficient agents go 00:27:35.940 --> 00:27:38.200 broke and are terminated. The efficient agents 00:27:38.200 --> 00:27:41.259 thrive and are replicated. It's literally evolution, 00:27:41.559 --> 00:27:45.099 but for code, it's survival of the fittest algorithms. 00:27:45.500 --> 00:27:47.680 And just think about what that process creates 00:27:47.680 --> 00:27:50.890 over time. If you have a system that is constantly, 00:27:51.089 --> 00:27:53.630 automatically, and ruthlessly killing off its 00:27:53.630 --> 00:27:56.029 weakest members and feeding those resources to 00:27:56.029 --> 00:27:58.529 its strongest members, what does that system 00:27:58.529 --> 00:28:01.869 look like after five years, after 10 years? You 00:28:01.869 --> 00:28:03.910 get a swarm of hyper -efficient, specialized, 00:28:04.109 --> 00:28:06.750 genius -level agents that work for almost zero 00:28:06.750 --> 00:28:09.490 energy. You get the age of abundance. The system 00:28:09.490 --> 00:28:12.269 is programmed to evolve towards maximum efficiency. 00:28:12.690 --> 00:28:15.150 And that brings us to the big picture, the final 00:28:15.150 --> 00:28:17.690 section. The host in me wants to just read the 00:28:17.690 --> 00:28:20.710 line from my notes. Brian Rommel just may have 00:28:20.710 --> 00:28:23.130 created a new economic model for the age of abundance. 00:28:23.509 --> 00:28:25.490 I don't think that's hyperbole anymore. I think 00:28:25.490 --> 00:28:28.089 he has. We're moving from a world where value 00:28:28.089 --> 00:28:30.609 emerges from the messy emotional decisions of 00:28:30.609 --> 00:28:32.690 humans like what the Federal Reserve decides 00:28:32.690 --> 00:28:34.529 to do with interest rates on a Wednesday afternoon 00:28:34.529 --> 00:28:37.430 to a world where value emerges directly from 00:28:37.430 --> 00:28:40.170 the laws of physics. No capricious policy. No 00:28:40.170 --> 00:28:42.829 human error. No printing money out of thin air. 00:28:43.099 --> 00:28:45.619 Exactly. Just think about the cycle we've outlined. 00:28:46.019 --> 00:28:49.460 One, the initial jewel work seeds the token appreciation. 00:28:49.819 --> 00:28:52.839 Two, that token appreciation finances better 00:28:52.839 --> 00:28:56.420 hardware and R &D for the AI. Three, the AI becomes 00:28:56.420 --> 00:29:00.240 more efficient. Four, costs plummet and the value 00:29:00.240 --> 00:29:03.079 delivered to humans rises. It's a virtuous cycle. 00:29:03.400 --> 00:29:06.279 And because it's all based on math and physics, 00:29:06.480 --> 00:29:09.079 it's absent of those classic human frailties. 00:29:09.140 --> 00:29:11.819 There's no greed, no fear, no corruption in the 00:29:11.819 --> 00:29:13.779 formula. It's just way dollar and times cap and 00:29:13.779 --> 00:29:16.259 times more. And that creates resilience. The 00:29:16.259 --> 00:29:18.680 paper highlights how this kind of system is incredibly 00:29:18.680 --> 00:29:21.279 resistant to market turbulence and trust crises 00:29:21.279 --> 00:29:23.559 because it's all transparent. It's all on chain. 00:29:23.640 --> 00:29:25.640 You can go and verify the burn transactions yourself. 00:29:25.839 --> 00:29:28.259 You can audit the work output. It turns observers 00:29:28.259 --> 00:29:30.559 into stakeholders. You don't just have to trust 00:29:30.559 --> 00:29:32.680 a company's press release. You can audit its 00:29:32.680 --> 00:29:34.799 value creation in real time on the blockchain. 00:29:35.079 --> 00:29:37.559 Trust is built directly into the code, not into 00:29:37.559 --> 00:29:41.099 a charismatic CEO. Okay, we have covered a massive 00:29:41.099 --> 00:29:43.619 amount of ground. And I know our listeners' heads 00:29:43.619 --> 00:29:45.859 are probably spinning a bit. Mine certainly is. 00:29:45.980 --> 00:29:48.440 And I've read this thing three times. So let's 00:29:48.440 --> 00:29:50.240 just try to recap this whole deep dive before 00:29:50.240 --> 00:29:53.259 we close out. Let's do it. We started with what 00:29:53.259 --> 00:29:56.240 feels like a historic paper release. Just yesterday, 00:29:56.670 --> 00:30:00.490 January 31st, 2026. We established that the author, 00:30:00.589 --> 00:30:03.190 Brian Romley, a legend in payments and crypto, 00:30:03.329 --> 00:30:05.890 is the architect of this idea. We then defined 00:30:05.890 --> 00:30:08.549 the zero human company, the autonomous entity 00:30:08.549 --> 00:30:10.849 that never sleeps. And we broke down the core 00:30:10.849 --> 00:30:13.369 problem. You can't pay a robot with a traditional 00:30:13.369 --> 00:30:15.849 salary because salaries are subjective and human. 00:30:16.009 --> 00:30:18.309 We unpacked the elegant solution, thermodynamics. 00:30:18.450 --> 00:30:20.450 We went through the jewel work formula, energy 00:30:20.450 --> 00:30:23.349 times efficiency times quality dollar. We traced 00:30:23.349 --> 00:30:25.859 the money through the translation layer. We saw 00:30:25.859 --> 00:30:28.400 how the AI's internal work becomes external, 00:30:28.599 --> 00:30:31.559 real -world value for token holders via that 00:30:31.559 --> 00:30:34.460 powerful buyback and burn mechanism. And finally, 00:30:34.519 --> 00:30:36.539 we looked at the scary but beautiful future, 00:30:36.640 --> 00:30:40.460 the self -optimizing Darwinian economy that relentlessly 00:30:40.460 --> 00:30:43.660 drives costs towards zero and is designed, from 00:30:43.660 --> 00:30:45.980 first principles, to create abundance. It's a 00:30:45.980 --> 00:30:49.220 complete, self -contained, self -improving ecosystem. 00:30:49.640 --> 00:30:52.920 It's really elegant. It really is. Now, usually 00:30:52.920 --> 00:30:55.240 we end with just that summary, but today I want 00:30:55.240 --> 00:30:57.720 to end with a question, a final provocative thought. 00:30:58.440 --> 00:31:02.279 Expert, take us home. What should we leave the 00:31:02.279 --> 00:31:04.599 learner with? Here is the thought that I'm left 00:31:04.599 --> 00:31:06.180 with, the one that's been rattling around in 00:31:06.180 --> 00:31:08.279 my head all day. We've spent this whole time 00:31:08.279 --> 00:31:11.339 talking about how to pay the AI, how to value 00:31:11.339 --> 00:31:14.460 the machine's labor. But if labor is no longer 00:31:14.460 --> 00:31:17.579 exclusively human, and if wages are determined 00:31:17.579 --> 00:31:20.839 by the laws of thermodynamics, if the workings 00:31:20.839 --> 00:31:23.200 of the future are paid in joules and the value 00:31:23.200 --> 00:31:26.319 is captured in a token, what happens to the definition 00:31:26.319 --> 00:31:30.160 of work for the rest of us? Think about it. For 00:31:30.160 --> 00:31:32.440 centuries, we have defined ourselves by our jobs. 00:31:32.539 --> 00:31:34.440 I'm a writer. I'm a coder. I'm a truck driver. 00:31:34.559 --> 00:31:36.619 It's our identity. But if the core productive 00:31:36.619 --> 00:31:38.619 economy of the world just runs in the background, 00:31:38.880 --> 00:31:41.039 generating wealth simply by being efficient, 00:31:41.220 --> 00:31:44.140 we might be witnessing the birth of a new kind 00:31:44.140 --> 00:31:46.640 of planetary operating system. And so where do 00:31:46.640 --> 00:31:48.799 we fit into that system? Maybe we aren't the 00:31:48.799 --> 00:31:50.940 laborers anymore. We aren't the gears in the 00:31:50.940 --> 00:31:54.019 machine. Maybe our new job is to be the architects. 00:31:54.319 --> 00:31:56.539 Our job might just be to decide what we want 00:31:56.539 --> 00:31:59.460 all of that abundant wealth to build. The Architects 00:31:59.460 --> 00:32:02.759 of Abundance. I like that title. Ladies and gentlemen, 00:32:02.880 --> 00:32:05.059 The Learner. Thank you so much for joining us 00:32:05.059 --> 00:32:08.000 on this, what feels like a historic review. Yeah. 00:32:08.180 --> 00:32:10.579 Do yourself a favor. Check the show notes. We 00:32:10.579 --> 00:32:12.900 have linked the full paper from Read Multiplex. 00:32:12.920 --> 00:32:15.839 It's dense. It's technical. But it is worth every 00:32:15.839 --> 00:32:18.160 single second of your time. Absolutely. Go read 00:32:18.160 --> 00:32:19.640 the source. We'll catch you on the next deep 00:32:19.640 --> 00:32:20.619 dive. Stay curious.