Unqualified Advice

Hello and welcome to Unqualified Advice, an entertaining show for entertainment purposes.Join us as we talk about running our small businesses, what we've been learning, and how we're applying lessons from academia and real life as entrepreneurs and investors.

Unqualified Advice is a business podcast hosted by Sean Filipow and Daniel Hatke, with 57 episodes on record and a Required Pod Score of 80. PitchCentric scores this show on Booking Probability, Listen Score, and live audience signals refreshed every 24 hours.

Sean Filipow and Daniel Hatke hosts Unqualified Advice, a business show with 57 episodes published.

Our AI reads these to draft pitches. Use them as grounding for a pitch that cites a real guest and a specific topic.

• 

  Episode #24

  The Chicken Salad Index

  Sep 21, 20261h 9mS2

  Dear Show Notes Readers, This episode begins with Halloween costumes and ends with copper on the Moon. Somehow, it makes sense. Dan's kids have reached the age where putting on a costume doesn't mean pretending to be something. For a little while, you simply are that thing. A dinosaur. Elsa. Whatever you've decided to become. That gets us thinking about the costumes adults wear, the stories we tell ourselves, and the enormous difference between saying "I'm not good at this" and "I'm not good at this yet." Because almost everything worth doing requires reps. That includes making economic predictions, where we take a moment this week to admit that we whiffed—hard—on a few. We revisit the Strait of Hormuz, fertilizer, rice, inflation, energy markets, and our increasingly mediocre prediction tracker before Sean and Dan simultaneously remember that the Kelly Criterion exists. From there we discuss the Chicken Salad Index, America's looming college demographic problem, whether degrees should somehow be priced according to their economic value, and why markets are so damn difficult to centrally organize. Trade, as Sean eventually observes, finds a way. That turns out to be a useful lens for AI, too. Should the frontier labs slow down? Can they? What happens when their breakthroughs diffuse around the world? Is China inventing the future, implementing it, democratizing it—or some mixture of all three? Then Dan makes perhaps the best defense of auto racing Sean has ever heard: the racetrack is a laboratory. It's where engineers get to ask, "Can we do it?" Which naturally sends us to the Moon. Why are we going back? Water. Minerals. Science. Capability. And all of the technologies we'll have to invent along the way. That brings us back to Earth, where electrification requires staggering quantities of copper, mining requires moving staggering quantities of rock, regulations require staggering quantities of paperwork, and physics remains stubbornly unwilling to negotiate. Maybe material science changes the equation. Maybe AI helps. Maybe we learn to use less copper, substitute other materials, recycle more efficiently, or build things in ways we haven't imagined yet. We aren't good at all of that yet. So we keep taking reps. Put on the costume of someone who can figure it out. Go build something. Books Discussed Mistborn by Brandon Sanderson — the king in white, and the reason I think about a fantasy novel on the plant floor The Stormlight Archive by Brandon Sanderson — currently on book two. There are more than I said there were The Kingkiller Chronicle by Patrick Rothfuss — two brilliant books, one promised trilogy, and nineteen years of nothing Material World by Ed Conway — where I got the scale of the Atacama pit, and the second episode in a row Conway has done heavy lifting for us Tools & Platforms Mentioned Baker Hughes rig count · Bloomberg Odd Lots · Costco · Spirit Halloween Companies Discussed Ivanhoe Mines · Baker Hughes · BYD · Tesla · Porsche · GM / Chevrolet · Ford · Ram · Anthropic · OpenAI · University of Colorado Boulder · Purdue University · BloomTech Links & References The Chicken Salad Index — Dan demanded it on air, so here it is: the whole index, the modeled cost stack, and every price I have receipts for Robert Friedland on Odd Lots, "The World's Monumental Shortage of Copper" — published the day before we recorded. Go listen to it; I clearly should have Our predictions tracker — every call we've made, scored, including the ones we got wrong this week Baker Hughes rig count — published weekly since 1944, and the number the whole industry actually uses BLS Consumer Price Index, August 2026 — released two days before we recorded; the headline and core figures we argue over Purdue's "Back a Boiler" — the income share agreement I promised to look up on air Unqualified Fact-Check We said some things. Here's how we did. = Nailed it | = Close enough | = Whiffed it The headline and the core Dan said "the three point four percent headline number, if you just strip it down to core, is closer to two." August CPI, released two days before we recorded: headline +3.4% year over year — exactly right — and core at +2.4% , the lowest core reading since March 2021. "Closer to two" than to three-four, comfortably. Recalled cold, two days after the print. Heating oil up fifty-three percent Dan, off the top of his head: "Heating oil is up like fifty three percent." The CPI fuel oil index rose 52.0% over the twelve months to August 2026, the largest energy component in the report. One point off. Show-off. Oil crossed a hundred again, and it never sticks Sean said we'd gotten above $100 a barrel recently and that historically it doesn't last. Brent broke $100 on September 9 and sat around $105.82 on the 11th . And the "short-lived" read holds up: before this year, the only time in the last decade oil held above $100 was 2022 — roughly six months — and the March–May 2026 run lasted about two. Copper is the thermal metal — the one Dan asked us to check Dan flagged himself mid-sentence: "the Claude's gonna have to fact check me here, because I'm speaking out of not deep knowledge," then said copper shows up in heat sinks and piping, and that it's all through the data centers. He's right. Copper conducts heat at about 401 W/(m·K) against aluminum's 237 — roughly 1.7 times better — and it's standard in cold plates, heat pipes and vapor chambers in exactly the high-density rack cooling he's describing. Sean's follow-up is right too: metals win because heat rides on free electrons, while ceramics are stuck moving it through lattice vibrations that scatter. One asterisk, and it's a fun one — diamond is a non-metal and beats every metal at about 2,200 W/(m·K) . So if anybody's hunting for that Nobel Sean called, the exception already exists in the display case. Friedland's copper number, relayed correctly Dan: "we have to mine more copper in the next eighteen years than we have mined in the last ten thousand." That is Friedland's actual line , close to verbatim, and Dan attributed it to him properly. Worth adding what Dan didn't: it's a demand forecast , not a measurement. USGS puts all the copper ever mined at roughly 600 million tonnes against about 23 million a year today, which is nearer twenty-six years than eighteen at current output. The eighteen assumes we build what we say we're going to build. Our own scorecard Sean said "we're only sitting at about sixty percent accuracy." Checked against our own tracker: of 28 resolved predictions, 12 came true, 9 landed partially and 7 missed. Score the partials at half credit and you get 59% . He was off by one point, about his own show, from memory. Dan's response — that anything above fifty is fine on Wall Street if you size the bet right — is also correct, and it's why the Kelly criterion conversation follows. A five percent rig count Sean said "about a five percent increase in rig count in North America" per Baker Hughes, claiming his own May call. The referent matters and he was reading oil-directed rigs, which is the right instrument for the prediction he was scoring: US oil rigs ran 429 at the end of May to 450 on September 11 — +4.9% . That is his number, essentially exactly. Two other readings exist and neither is what he meant: total North America is 792 against 718 year over year, +10.3% , and North America year to date is about 1.7%, depressed by Canadian seasonality. His characterization of Baker Hughes is exactly right — weekly since 1944, and the number the industry runs on. Jackson Hole is not an FOMC meeting Sean set the table with "coming up out of the Jackson Hole FOMC summit a couple of weeks ago." Right event, right timing, wrong institution. Jackson Hole is the Kansas City Fed's Economic Policy Symposium — an academic conference with no policymaking power, attended by plenty of people who aren't on the FOMC at all — and it ran August 27–29. The actual FOMC meeting was two days after we recorded, on September 15–16. The Philippines rice record Sean said the Philippines "had the largest rice harvest since nineteen eighty seven this year." Close, and garbled in a specific way. Q2 palay came in at 4.63 million tonnes, +5.7% — the largest second quarter in a data series that begins in 1987. That's a different claim, and full-year 2026 is actually forecast below the 2023 record. The import ban detail is also off: the suspension ran through 2025 and simply expired on December 31, and in 2026 Manila deliberately chose not to re-impose it. But the thing that matters — that we called a rice crisis and got a record quarter instead — is entirely correct, which is the whole point of the segment. The reserve math Three claims in one breath, and they land differently. "Six to seven hundred million barrels released" is high — the IEA's March 11 collective action authorized 400 million , with 426 million in confirmed contributions. That's still the largest in the agency's history , so the instinct was right and the number was inflated. "Global reserves are over a billion barrels" is correct and if anything understated: IEA members hold over 1.2 billion in public stocks alone, plus China's roughly 1.3 billion. And "six months worth of cover" — the IEA obligation is 90 days , not six months, though Japan and South Korea really do sit on 200-plus. Two out of three, generously. European gas storage Sean put EU storage at "63% total capacity" against a five-year norm "closer to 80." It was about 67.8% on September 11 , and the five-year average for that date implies roughly 85% full — so a bit stale and a bit conservative on the comparison. 63% was accurate for late August. The conclusion he drew is not just right but understated: this is the lowest fill for the date in about fifteen years . Purdue, the only school Dan: "Purdue's like the only school that hasn't raised tuition in the past decade or something." It's the longest freeze — fourteen straight years, locked at the 2012-13 rate — but not the only one. Florida State is around eleven years, Wisconsin managed ten, the UNC system nine, and every public college in Indiana froze in-state tuition for 2025-26 and again for 2026-27. The "or something" is doing honest work. The Douglas firs Dan on Friedland and Steve Jobs at Reed College: "they were like, maybe we can just like plant a bunch of Douglas firs and we'll be rich." It was an apple orchard . Friedland caretook his uncle's farm south of Portland and turned it into a commune called All One Farm, where Jobs lived and worked the orchard — and the name Apple Computer came to Jobs on the drive back from that farm. Scored yellow rather than red because Dan relayed it accurately: the firs are how the story was told on the Odd Lots episode he'd just listened to, and I heard the same thing. Faithfully repeating a source's error is a different sin than inventing one — and the true version is dramatically better than either telling. Thirty-two thousand jurisdictions Dan's standing example of compliance cost is sales tax "in thirty-two thousand different jurisdictions." The number everyone uses is just over 12,000 — Vertex and Avalara have tracked it from about 10,000 in 2014 upward. He's roughly 2.5x high, which is why this is yellow and not red: the point is that the number is absurd, and twelve thousand separate tax authorities for one company to satisfy is absurd. Directionally he's dead on. The college wage premium Dan said everyone went to college "and then they made twenty percent more over their lifetimes." It's 75% . Georgetown's CEW puts median lifetime earnings at $2.8 million for a bachelor's against $1.6 million for a high school diploma — a $1.2 million gap — and BLS has the weekly premium at about 63%. Off by three or four times, and the direction of the error cuts against Dan's own argument, since the historical case for college was even stronger than he gave it credit for. BONUS — which hole in the Atacama Sean described a pit like "Everest but inverted," said it had been producing for almost a hundred years, and then caught himself on whether it was the largest by global supply — "Second." He's describing two different mines . The century-old inverted mountain is Chuquicamata , open since 1915 and the largest open pit by excavated volume. The largest copper mine by production is Escondida , which only started in 1991. The self-correction was the right instinct. BONUS — thirty thousand years of no rain Dan relayed that the Atacama "hasn't rained there in like thirty thousand years or some whoa, wicked crazy number." Nobody uses that figure. The real ones are still absurd: no rainfall recorded in about 500 years in the hyperarid core before the freak 2015 and 2017 rains, a verified 172-month dry streak at Arica, and geological evidence of aridity going back tens of millions of years. BONUS — it's Austen, and the CR-V won Dan's coding-school founder is Austen Allred, not Austin, and the ISA was 17% of income for 24 months before dropping to 14% — his 15% split the difference. And on the top three sellers: F-Series and Silverado are right, but Ram is fourth. Third is the Honda CR-V, which in the first half of 2026 outsold the F-150 specifically . Final Score: 7 green, 7 yellow, 1 red The instincts were excellent and the denominators were a disaster. The one red is a real fact wearing the wrong number; the best correction of the lot isn't even ours to own, since we inherited it faithfully from a podcast we'd just listened to; and the rig count turns out to be a win we talked ourselves out of. Chapters 00:00 - Cold Open: Trade Finds a Way 00:29 - Houston 102, Denver 54, and Conjuring Fall 03:12 - Costco Halloween, an Elsa Dress, and a T-Rex 04:08 - Peacocking: The Costume You Wear in Your Own Head 05:47 - "You're Not Good At That Yet" 06:23 - Mistborn, and the King Who Wore White in a World of Ash 08:13 - Sanderson, Rothfuss, and the Trilogy That Never Ended 09:14 - Why Endings Are the Hard Part 11:45 - "Is That What the Fed Needs to Do?" 11:58 - Level-Setting: Jackson Hole and the September FOMC 13:02 - Scoring Ourselves: The Fertilizer Call We Whiffed Hard 13:59 - Analyst Number Three and the Ships Nobody Documented 14:39 - Oil Reserves, EU Gas Storage, and Pressure at the Margins 15:34 - Headline vs. Core: What Inflation Is Actually Doing 16:21 - The Chicken Salad Index 17:35 - What's the Right Inflation Rate for a Shrinking Population? 19:59 - Inflation as a Wealth Transfer: By Class or By Age? 22:09 - CU Boulder, the Enrollment Cliff, and College as a Business 26:17 - Should a Degree Be Priced By Its Economic Value? 27:41 - Purdue, Back a Boiler, and Income Share Agreements 30:53 - The Conflation That Broke a Generation 32:13 - It's the Meme, and Memes Are Infectious 33:26 - Hayek, the Holodomor, and Why You Can't Plan From the Top 34:29 - Trade Finds a Way 36:21 - The Scorecard and the Kelly Criterion 38:10 - Rig Count, Baker Hughes, and the Oil Call That Landed 43:01 - Dario Says Stop, Elon and Sam Agree, and Dan Says Wait a Minute 44:56 - What Distillation Actually Is 46:01 - BYD's Chargers and Porsche's 800 Volts 48:43 - The Racetrack Is the Laboratory 49:48 - Why Everyone Wants the Lunar South Pole 51:47 - Friedland, Odd Lots, and the Copper Nobody Has Mined Yet 54:51 - How Expensive It Is to Play By the Rules 1:00:54 - Fifty Thousand Lawyers at the Bottom of the Sea 1:02:57 - Two of Water, One of Stone, in a Desert That Never Rains 1:03:54 - Everest, Inverted 1:05:29 - "Claude's Gonna Have to Fact-Check Me Here" 1:08:12 - Keep the Reps Up

  ListenShow Notes
• 

  Episode #23

  The Demographics of Midtopia

  Sep 14, 202646 minS2

  Notes From a Man Who Did Not Shave Dear Show Notes Readers, If demographics are destiny, what happens when some of the participants in the economy aren't actually people? That's where we start this week. Population decline has traditionally meant fewer workers, less economic activity, less surplus, and—if history is any guide—a greater chance that people start fighting over a shrinking pie. But history never had millions of AI agents doing useful work, making decisions, building businesses, and creating economic output. So we try to figure out whether that changes the equation. We talk about the Black Death and the Renaissance that came out of it. Is it actually getting rougher, or are we simply old enough to notice now? We stumble our way home to a better economic indicator than GDP: how things feel at the kitchen table. Then we ask the obvious next question. If agents really do become economically productive, who captures the surplus? Do we eventually tax agent-generated income differently from human labor? Can we use some of that productivity to give actual humans a better floor without deliberately kneecapping the technology creating it? Then we move to more practical matters. I've been moving my own agent between AI providers. So naturally we discuss sexy AI voices, revisit Her , and debate whether the ability to tell a genuinely funny joke might actually be a better test of AGI than most benchmarks. Dan thinks scientific discovery gets there first. Eventually we land on the bricklayer. Maybe AI doesn't replace the person who loves laying brick. Maybe it handles the website, follows up with customers, collects reviews, answers emails, manages the marketing, and does all the digital nonsense the bricklayer never wanted to do in the first place. That future sounds a lot more interesting than either utopia or dystopia. Midtopia, perhaps. Go build something. Cheers, Sean "You get older and you realize everyone's making it up, even the people that are in charge of it." Books Discussed Factfulness by Hans Rosling — the four income levels, and why things getting better is exactly what made families smaller Rich Dad Poor Dad by Robert Kiyosaki — invoked strictly as the patron saint of predicting the next crash, forever Influence by Robert Cialdini — the one I couldn't retrieve on air. It's the "click, whirr" material, and yes, it's exactly about inverting the laziness Shows/Films Discussed Her (2013, dir. Spike Jonze) — due for a rewatch, and it will not feel the same 30 Rock — I was reaching for a joke that turns out not to exist. See the fact-check Lost — collateral damage in a conversation about what's in the box Tools & Platforms Mentioned OpenAI Codex — the coding product, and the reason my agent has a new home this week GPT-6 Astra — OpenAI's September release Claude / Anthropic — where the agent used to live Anki — spaced repetition, aspirationally Companies Discussed Anthropic · OpenAI · Google · Meta · Amazon · BYD · Denver Botanic Gardens · The Comedy Cellar Links & References Greg Brockman on Stratechery — Ben Thompson's Sept 4 interview, the one Dan is half-remembering. Brockman: "somewhere in there, I think we're going to cross most people's AGI threshold" Gapminder's income levels — Rosling's four levels, with the actual dollar-per-day thresholds Epoch AI on revenue per employee — the numbers behind my "look at that number" argument Alliance for Lifetime Income, "Peak 65" — the boomer retirement figure I mangled, in its correct units Charlie Munger, "How to Guarantee a Life of Misery" (1986) — the source of "invert, always invert," which Munger got from Jacobi Unqualified Fact-Check We said some things. Here's how we did. = Nailed it | = Close enough | = Whiffed it The Black Death handed the peasants leverage Dan said the plague "allowed the peasants to rise up, basically." Correct, and the receipts are unusually good: the plague killed 30–40% of England's population, and within two years the Crown passed the Ordinance of Labourers (1349) and then the Statute of Labourers (1351) capping wages at pre-plague 1346 levels — a law you only write if employers are being forced to bid up pay. One honest wrinkle: real wages didn't actually jump for about thirty years, and the 1381 Peasants' Revolt was crushed. But the long arc is exactly as Dan described it. Rosling's levels, recalled under pressure Sean said Factfulness has "four or five different stages of wealth" and that America is "ninety nine percent in level four." It's four — under $2/day, $2–8, $8–32, and over $32 — and he hedged on the number anyway. Gapminder doesn't publish a US figure, but Level 4 starts at about $11,700 a year per person, which is below the US poverty line for a single adult. So essentially all of it. Both halves stand. Brockman really did say that, and Dan really did listen to it Dan recalled an interview with OpenAI's co-founder on Ben Thompson, and a line about how "for a lot of people this will be their AGI moment." It ran on Stratechery on September 4, and Brockman's version is "somewhere in there, I think we're going to cross most people's AGI threshold." Dan also added, unprompted, "of course he's out there, he has something to sell" — which is the correct amount of skepticism, since Brockman hedged too. The revenue-per-employee tell Sean said the AI labs' revenue per employee is "pretty amazing compared to even Google Meta," and that they're still losing a lot because of the capital build-out. Epoch AI (May 2026): Anthropic roughly $9M per employee, OpenAI roughly $5.5M, against Meta at $2.5M and Google at $2.1M. So two to four times over. The losses are real too, though the picture has started to split — Anthropic was projected to post its first quarterly operating profit in Q3. Invert, always invert Dan: "Do a little Charlie Munger and invert it." Correct, and Munger himself credited it to the mathematician Carl Jacobi — "Invert, always invert" — in his 1986 Harvard School commencement speech. Munger's companion version: "All I want to know is where I'm going to die, so I'll never go there." How many children died a hundred years ago Sean opened with "over fifty percent," Dan cut in with "more like a quarter," and Sean then landed on 25% by age ten and 50% by age twenty. Dan wins this one. Globally in the 1920s roughly a quarter to a third of children died before age five, so his instinct was right. In the US it was closer to 10% before age ten. And nobody's 50%-by-twenty figure survives — child mortality is front-loaded, so the gap between age ten and age twenty adds a couple of points, never twenty-five. The 50% number is real, it's just from before 1800. Four million boomers a day Sean said boomers are retiring at "four million a day roughly." The four million is right; the day is not. About 4.1 million Americans turn 65 each year through 2027 — roughly 11,400 a day. He hedged twice, which buys him the yellow. But this is the second consecutive episode where this exact number has caused trouble, so we're putting it on the wall. The Bessent BYD quote Sean relayed a quote about "the best car that money can buy is a BYD at thirty seven thousand," flagging twice that he wasn't sure it was real. It is real, and he was close. The actual line is "it is the best $70,000 car that $35,000 can buy — it is heavily subsidized," from Charlotte in early September. Two things off: the price is $35K, and the framing was an attack on Chinese industrial subsidies, not praise. Full marks for hedging. What Scarlett Johansson actually said Sean: "I believe she specifically said nobody could do that." Not quite. She declined one specific request from OpenAI in 2023, and then objected publicly when the "Sky" voice shipped sounding like hers anyway. Her ask was transparency and legislation on likeness rights — not a blanket prohibition — and she never sued. Lawyers sent letters, OpenAI pulled the voice. The "I believe" earns the yellow. BONUS — the word Sean wanted At 29:11 Sean went hunting for the midpoint between utopian and dystopian, tried "midtopian," "topian," and "topia," and then said "look that up later." The word is protopia , coined by Kevin Kelly: not a perfect world, just one that's incrementally better than yesterday. Which is, more or less, the entire thesis of this episode. BONUS — the first name Dan: "I think I listened to an interview with Brockman. What's his name? I can't remember his first name." It's Greg . President and co-founder. Final Score: 5 green, 4 yellow, 0 red A clean sheet on the reds — and I'll admit one of the two I edited out of existence before this ever went to air. Two guys with no notes, no Google, and no intention of working on Labor Day. We'll take it, though somebody should still take the boomer retirement statistic away from me. Chapters 00:00 - Cold Open: It Still Has to Start at the Bottom 00:10 - Labor Day, Denver, and a Shout-Out to a Listener 00:55 - Demographics Is Destiny — But Whose Demographics? 02:35 - When Does an Agent Stop Being a Tool? 04:43 - What Shrinking Populations Actually Break 06:53 - Flesh Bodies, GPU Bodies 09:08 - The Future Is Not as Dark as History Suggests 10:15 - The Renaissance Still Had to Start at the Bottom 10:39 - The Black Death, the Medici, and Who Picks Up the Pieces 11:49 - Where Revolutions Come From: The Second Percentile 12:50 - Why Fertility Fell: Rosling, Mortality, and Birth Control 19:05 - Doom Sells: FUD, Kiyosaki, and the Perma-Bear Trade 20:08 - When the Joneses Next Door Are Billionaires 21:58 - GDP vs. the Kitchen Table Measure 23:37 - BYD, High-Speed Rail, and What We Made Impossible to Build 24:46 - The Cure for Pain Is More Pain 25:23 - Tax the Agents 29:07 - Build Segment: Brain Surgery on a Personal Agent 31:25 - The Coworker With the Sultry Voice Agent 33:06 - Is Comedy the Real AGI Test? 38:00 - What's In the Box 39:08 - Revenue Per Employee as the Tell 40:19 - From Dirt Floors to a Bricklayer With a Marketing Agent 43:25 - Invert It: The Laziness Is Baked In 44:34 - Everyone's Making It Up

  ListenShow Notes
• 

  Episode #22

  Artificial Laziness

  Sep 7, 20261 hourS2

  Hello dear show notes readers! Dan started with a question about AI agents: how do you stop a thing that keeps writing down its own rules from going rigid? He was picturing a government. I was picturing red tape, and the policy documents I'd spent the week wrestling with. In the end, you wind up with the rule is C , and nobody remembers that C only ever followed from A and B — so when A changes, things downstream don't adapt quickly. Dan's answer: build deterministically, break the process into the smallest steps you can, and layer intelligence into only the one narrow place that actually needs judgment. It sounds a lot like training an employee, no? The middle of the episode is a boat. Ed Conway — whose Material World was my favorite book of its year — published a post the morning we recorded asking whether we've hit peak ship, and I brought it in as settled fact. Reader, I should have read to the end. The fact-checkers have notes. What survives the correction is the more interesting question anyway: what happens to a system when the curve that's been quietly fighting inflation on your behalf for forty years finally flattens out? That took us to Houthis, chokepoints, insurance actuaries, and my favorite tangent of the year — why shipping fortunes are shadowy in a way tech fortunes structurally cannot be. A Silicon Valley billionaire has one cash cow. A shipping magnate has Panama, Greece, and Angola. Then Dan announces a business. He's launching Polyiota — polyiota.com, as promised on air — consulting on AI transformation for exactly the kind of company that will never buy itself a software team. There is no such thing as ready, so you put the rivets in the airplane while it's already flying. His field report from his early efforts with his own team gives us a glimpse into the adoption headaches he hopes to help others avoid: the blinking cursor effect. The fix isn't just more intelligence. We end up somewhere I didn't see coming: not why not , but where's next . Cheap land moved technology west a century ago, and Dan's argument is that the underrated story of the next decade isn't AI at all — it's people moving again. Great Lakes. College towns. Columbia, Missouri. Then four million boomers a year start retiring, Dan does the arithmetic live and out loud with predictably mixed results, and we close on what the millennials will do with the inheritance. My guess was tattoos and beard oil. Dan, for the record, has neither. Go build something this week. Cheers, Sean Books Discussed Material World by Ed Conway — my favorite book of its year, on the raw materials that make modern life possible. If you only take one thing from it, make it how few places on Earth certain links in the silicon supply chain actually exist. Trade World by Ed Conway — the follow-up I said I was looking forward to. It came out four days after we recorded, built around bread, cloth, and cars. Tools & Platforms Mentioned Claude · Codex · Gemini · Grok · Cursor · OpenRouter · Zapier · Slack · X · YouTube · Substack Companies Discussed Anthropic · OpenAI · Google · xAI · SpaceX · Anysphere (Cursor) · Amazon · Sky News · NPR / Planet Money · Bloomberg / Odd Lots · University of Alabama · Purdue University · University of Missouri · Denver International Airport · Polyiota — Dan's new AI transformation and strategy practice Links & References Are We Approaching Peak Ship? (Ed Conway, Substack, Aug 30) — the post the middle of the episode is built on. Published the morning we recorded. Read past the setup; the ending is not where I left it. Polyiota — Dan's new practice. As promised on air. SycEval: Evaluating LLM Sycophancy (Fanous et al., Stanford, arXiv) — the study behind the "who caves when you push back" numbers. Planet Money Makes a T-Shirt (NPR, 2013) — the series I was describing, and the origin of the squirrel on my shirt. The Twitter recommendation algorithm — heavy ranker weights (GitHub) — the actual published numbers Dan was working from memory. How I AI — Claire Vo's show, correctly spelled. Peak 65: 4.1 million Americans turn 65 each year through 2027 (Alliance for Lifetime Income) — the real shape of the number I quoted. 100 Years of U.S. Consumer Spending (BLS, PDF) — where my ninety-percent claim goes to die. China dominates the shipbuilding industry (CSIS) — the correct version of the stat that flabbergasted me. Unqualified Fact-Check We said some things. Here's how we did. = Nailed it | = Close enough | = Whiffed it Material World is about six raw materials Sean hedged at "six or seven" and landed on the right one. It's six — sand, salt, iron, copper, oil, lithium — and the US subtitle says so on the cover. The new book really is called Trade World Sean: "he's coming out with a new book, which I'm excited about, called Trade World." Correct, and better timed than he knew: Trade World: The Ties That Bound Our Past and Could Unravel Our Future published September 3, 2026 — four days after we recorded. The sycophancy numbers Sean put Claude Sonnet at "about 40%" holding a correct answer under pushback, caving "about 20% of the time," with Gemini best and ChatGPT in the middle. Stanford's SycEval : Claude's non-sycophantic rate is 42.6% , its regressive sycophancy — abandoning a correct answer for a wrong one — is 18.3% , and the ranking on that metric is Gemini 9.3% , ChatGPT 14.4% , Claude 18.3% . Three numbers and an ordering, all correct. One asterisk: on overall sycophancy Gemini is the worst of the three — it changes its answer more than anyone, it just changes toward the right one. The Planet Money t-shirt, down to the squirrel Sean described the series, dated it at "ten, fifteen years ago," and read the shirt off his own chest. Published December 2, 2013 . Cotton from Mississippi, yarn spun in Indonesia, women's shirts sewn in Colombia. And Planet Money's own note on the design: "Squirrel + martini = animal spirits." He left out Bangladesh, which handled the men's shirts. That's the whole correction. China's shipbuilding, understated Sean: "last year China alone built more container ships than the US has built since World War Two combined." The real statistic is different on three axes and worse on all of them — per CSIS, China's largest state shipbuilder CSSC alone built more commercial vessels by tonnage in 2024 than the entire US industry has built since the end of WWII. One company, not a country. For scale: of 5,448 large commercial vessels on order worldwide, China holds 3,419 and the US holds three . The X algorithm weights Dan rattled off "negative two fifty, negative five hundred" for mute and block against "a like is a plus one," then said don't quote him, they're directionally accurate. He's directionally right and, if anything, too gentle. Published heavy-ranker weights: report −369 , block/mute −74 , reply +13.5 , retweet +1.0 , like +0.5 . So a block is worth −148 likes and a report −738. Two small slips: the +1 is the retweet, not the like, and bookmarks don't appear in the published weights at all. Who actually owns Cursor Sean, on Grok: "with their now, you know, owning Cursor." Right family, wrong entity. SpaceX acquired Anysphere for $60 billion in all stock on June 16, 2026 — four days after its Nasdaq debut. xAI merged into SpaceX back in February, so Grok and Cursor do live under one roof; SpaceX is just the one holding the deed. (For the record: we called this acquisition in Ep 45.) Four million boomers retiring a year Sean, citing a Fed president. The 4.1 million figure is real and it's the Peak 65 number — but it counts Americans turning 65 each year through 2027, not retiring. The retirement flow is about half that: Richmond Fed's Tom Barkin, three weeks ago, put the increase in those not in the labor force at "almost 2 million" a year over the last three years. Right number, wrong verb. The Jones Act "is not active right now" Not repealed, and not off — but Dan is closer than the flat statement suggests. A waiver that began March 17, 2026 ran 150 days, the longest suspension of Jones Act restrictions ever , and was extended again in August through November 15 . It was also narrowed on the way, from 600-plus tariff codes down to 237, mostly petroleum, LNG, and fertilizer. Deflationary in direction, as he said; estimates put the effect at a few cents a gallon. A twenty-or-thirty-million-dollar golf course Sean, hedged three ways, and every hedge earned. Alabama's golf facility was approved at $26.1 million — in February 2022 — and finished at $47.75 million . The course opened in October 2024. Nothing new surfaced the week we recorded. What Ed Conway actually argues Sean brought the peak-ship post in as a conclusion: we've hit the physical limit, the improvement has flatlined. Conway lays out those constraints faithfully — Suez's 400-metre limit, cranes that can't reach across the beam, draught rather than box count as the real ceiling — and then argues against the thesis . His words: "The main problem with this thesis, though, is that this isn't the first time it's been posited." People said the same before the Emma Maersk in 2006 at 14,770 containers; the MSC Irina now carries 24,346, sixty-five percent more. The post is a warning about betting against shipbuilders, and Sean turned it into a verdict. Nine grand to eighteen grand because of Houthis Dan's number is from the wrong crisis. At the July 2024 peak of the Red Sea disruption, Drewry's composite was $5,937 per container and the worst single lane, Shanghai–New York, hit $9,612 . The nine is right; the doubling never happened. Rates above $18,000 are real — they belong to the 2021 COVID crunch, when China–US passed $20,000 . Ninety percent of the budget went to food and housing Sean, on life a century ago. BLS has it at 65.8% in 1901 — food 42.5%, housing 23.3% — and 61.5% by 1918. To reach ninety you have to add clothing and fuel, which gets you to about 85%, and which is not what he said. The point still stands: today the same two categories are about 46%. Seven thousand boomers a day Dan, doing arithmetic out loud and flagging that his denominators were shaky. They were. Seven thousand a day is 2.55 million a year, and total US deaths from all causes at every age in 2024 were 3.07 million — of which 890,000 were 85-and-over, which is nobody's boomer. The real figure is roughly 3,000 to 4,000 a day . The cohort shrinks about 2% a year , not the ten percent we landed on. BONUS — "Grok boys" has no documented usage anywhere; Dan may simply be the origin, in which case, congratulations. Final Score: 4 green, 6 yellow, 4 red The ideas held up and the numbers took a beating. Every red is a figure imported from the right neighborhood and the wrong house — except the Conway one, which is just a man who stopped reading too early. Chapters 00:00 - Cold Open: These Are Just Employees, Dan 00:41 - How Do You Keep an Agent From Ossifying? 02:40 - This Isn't an AI Problem, It's an Org Problem 03:23 - The Rule Is C, and Nobody Remembers A and B 04:42 - Artificial Laziness and the Cost of Thinking 05:31 - I Don't Know Why I'm Paying You, Claude 06:54 - The Sycophancy Data: Who Holds a Correct Answer 07:18 - Model Vibes, Flavors, and Choosing for the Task 09:57 - Build Deterministic, Layer Intelligence Thinly 10:31 - Ed Conway and Are We Approaching Peak Ship? 12:27 - The Moore's Law of Shipping Containers 13:37 - Houthis, Chokepoints, and Just Building More Ships 14:07 - China Has Built More Ships Than We Have Since WWII 16:29 - Insurance, Actuaries, and Ships as an Asset Class 17:57 - Why Shipping Money Is Shadowy Money 19:45 - The Bouncy Ball, the T-Shirt, and the Alaskan Fish 21:27 - What We Stopped Paying For 22:40 - Making Millions of Something Is a Different Mindset 22:59 - Should We Get Dan's Brother On to Talk Amazon? 24:46 - Just Hire Me: Dan Launches Polyiota 25:01 - Riveting the Airplane While It's in the Air 27:03 - The Real Customer Is a Business Changing Hands 28:11 - Compliance Is the Differentiator 32:10 - The Blinking Cursor Problem 33:12 - Zapier, the 80/20, and Making Automation Legible 36:49 - Sizing the Engagement by the Back Office 38:24 - When Does the Model Crown Change Hands? 38:58 - Grok Boys, and Why Your Feed Is New York Coded 42:26 - Don't Linger: You Control More of Your Algorithm Than You Think 44:05 - What YouTube Knows That X Doesn't 45:34 - Good Pope, Bad Pope: Benevolence as Infrastructure 46:25 - The Denver Illuminati Detour 47:21 - When the Midwest Was the Richest Place in America 47:54 - The Jones Act Interlude 49:22 - Where's Next? Cheap Land and the Next Migration 51:13 - The Sunbelt Happened. Are the Great Lakes Next? 53:01 - The Case for College Towns 54:13 - Aggregation Theory Comes for Higher Ed 56:02 - The Silver Tsunami: Four Million a Year 56:22 - Doing the Boomer Math Live 57:28 - Tattoos and Beard Oil 58:41 - Embrace Your Actual Age

  ListenShow Notes
• 

  Episode #21

  Our Submissions to the Library of Alexandria.

  Aug 31, 202659 minS2

  The Horse Hello dear show notes readers! On this week's episode I tell Dan about something that had been haunting me. The events in the Strait of Hormuz feel eerily similar to a story captured 2,800 years ago. I had just seen Nolan's version of The Odyssey and been left with the feeling that Hormuz feels a lot like a modern-day Troy. A trade civilization built on a handful of honor rules. A horse rolled up to the gates. Somebody breaks the one rule that made the whole thing work, and everything that comes after comes through that gap. Dan's answer, which I'm still chewing on: "this sounds like post-World War Two." We came back to Troy twice more. It ends the way it always ends. Rome cleans up. That's the back half of a conversation that starts somewhere much drier. Dan opened by asking what on earth is going on with the Fed and the Treasury, and I did the responsible thing and pretended nothing was happening — thirty-year rates only back at 2007 levels, one arm of the government telling markets to speak freely and the other quietly buying bonds so they don't speak too freely. Then Dan said the quiet part out loud in the best line of the episode: Warsh says let the market speak, and Bessent says not that loudly. There are charts, an argument about whether what Bessent did counts as QE, and a detour on why a thirty-year mortgage is priced off the ten-year note — which I sent straight to the fact-checkers, because I was guessing. Then it turns into a build episode, and Dan makes an argument about agents I haven't been able to shake: you start a new conversation, and the thing knows everything until it knows nothing. It has to be born into existence every time. Which, he points out, is also us — we just happen to run about eighty years, and none of us can checkpoint our context either. That leads to his digital Library of Alexandria, my record store you can walk into on the internet, his rebuild of the whole investment process at Kabu, and the saddest sentence in modern software: all I've got is this little blinking cursor. We close on whether "artificial" is doing any work in "artificial intelligence," and whether a person is allowed to hold two contradictory ideas at once. Reader, we are. Go build something and contradict yourself. Cheers, Sean Books Discussed The Myth of America's Decline by Josef Joffe — Dan's find, on the pile, not yet read. When Reason Goes on Holiday by Neven Sesardić — arrived on Dan's doorstep the day before we recorded, still unopened. The Odyssey and The Iliad by Homer — the spine of the first half. The Republic by Plato — cited, correctly in spirit, on the permanence of nobody wanting to fix the hard thing. Also Discussed Watched / listened: The Odyssey (2026) · Tetragrammaton with Rick Rubin (the Tyler Cowen episode) · Odd Lots · Forward Guidance Tools, platforms, companies: Notion · GitHub · Discogs · Audible · Claude, Codex, Grok · Number One, Boosby, Tara, Saul and Fable (the named agents on both ends of the call) · US Treasury · Federal Reserve · Anthropic · OpenAI · xAI · Kabu Links & References Treasury doubles debt buybacks (CNBC, Aug 19) — the announcement the first act is built on. 30-year Treasury tops 5.33%, a 19-year high (CNBC) — the "levels we haven't seen since '07" number. Warsh's July 29 press conference — full transcript (Federal Reserve, PDF) — the actual "play the ball, not the referee" remarks. The typical US homeowner now stays put about 12 years (Redfin) — the number that beat both our guesses. Unqualified Fact-Check We said some things. Here's how we did. = Nailed it | = Close enough | = Whiffed it The 30-year really is back at 2007 levels Sean called it above 5.3%, at levels unseen since '07. It touched 5.33% on August 17–18 — a 19-year high, the highest since 2007. Correct to the decimal, from memory, in the first two minutes. The buyback, and the date Sean put it at "middle of the week, August nineteenth." Treasury announced it August 19 , raising the maximum operation size to at least $4 billion from $2 billion across 10–30 year maturities. One nuance: the 19th was the announcement. Buying starts September 9. Yellen did it first Treasury relaunched its buyback program in May 2024 , the first since the early 2000s. This is the load-bearing beam under Dan's "both sides are implicated," which was the whole point of the segment. Two book titles, recalled cold, both exact Dan apologized for his pronunciation twice and then got both titles and both authors right. Word for word. The Tetragrammaton episode really was last week The Tyler Cowen episode went up August 19 — four days before we recorded. "Isn't this QE executed by Treasury?" — "Yes, sir." Buybacks are funded by issuing other debt, mostly bills, so no new money enters the system — a duration swap, closer to Operation Twist. Only the Fed can create reserves. Where we're right, and it's the part that matters: it does pull long duration out of the market. "Warsh says 'Let the market speak'" Dan's line is the best six-word summary of Warsh's position anybody's produced, and it isn't a Warsh quote. He actually said: "Market participants are learning to play the ball, not the referee." Same argument, better sentence. The seven-year mortgage The one Sean sent to the fact-checkers himself. The ~7-year average life is real — it's the prepayment convention that makes mortgage-backed securities trade off the 10-year rather than the 30-year, which is exactly why Dan's pricing point is correct. But homeowner tenure has stretched to about 12 years . Right convention, right conclusion, stale number. Who came out of the horse, and who cleaned up after Not "the Spartans" — the force was Achaean broadly, Odysseus leading under Agamemnon. And Troy's destruction layer sits around 1180 BC ; Rome's founding is 753 BC , four centuries later, with Archaic Greece rebuilding first. The Troy-to-Rome line is the Aeneid 's — myth doing exactly the job we'd spent five minutes describing. Something extra in the communion wine That's Brian Muraresku's The Immortality Key — a real argument that got a real hearing, not the consensus of classicists. Sean hedged twice, which is the right amount. Revolutions start with the elites Well supported in direction (Crane Brinton's "desertion of the intellectuals," Peter Turchin's elite overproduction). The crisp 2%-envying-1% version isn't from anywhere in particular. Why universities are called universities Sean: "the repository of universal knowledge." Lovely. Wrong. From universitas magistrorum et scholarium — "the whole body of masters and scholars." Universitas meant a guild . The original university was a labor union with better robes. BONUS — "Nevin Sicerdic" is Neven Sesardić , roughly SEH-sar-ditch. And the new Congress is sworn in January 3 , not February, which costs Dan a month of his narrative-control window. Final Score: 5 green, 6 yellow, 1 red Two guys who told you in the first minutes they weren't masters of the Fed did considerably better on the Fed than on the Greeks. The universities one is going to sting for a while. Chapters 00:00 - Cold Open: Which Takes Us Back to Troy 00:22 - What's Going On With the Fed and Treasury 01:59 - Bessent Buys Bonds, and the Midterm Math 03:53 - Is Hormuz a Modern-Day Troy? 06:40 - The Odyssey on IMAX, and What Survives Translation 09:23 - The Non-Rational Thing People Are Clamoring For 12:35 - Chart Time: The Three Rates That Matter 13:58 - The 30-Year Hits Levels We Haven't Seen Since '07 14:25 - August 19: QE by Another Name? 15:36 - Why a 30-Year Mortgage Is Priced Off the 10-Year 17:50 - Warsh Says Let the Market Speak. Bessent Says Not That Loudly. 19:45 - The Myth of America's Decline 21:02 - Where Revolutions Actually Start 24:12 - Precedent Is the Real Damage 28:36 - Every New Chat Is a Person Born Into Existence 31:00 - Building the Digital Library of Alexandria 36:09 - A Record Store You Can Walk Into 40:22 - Number One, Notion, and the Schema Problem 41:58 - Mr. Claw's Gone Lazy: Going Model-Agnostic 45:56 - Why Most People Still Use It Like Google 47:26 - Tetragrammaton, Tyler Cowen, and Why Prompts Matter Again 49:45 - Rebuilding the Kabu Investment Process 52:57 - Commenting Regimes and Boosby the Code Janitor 55:19 - It's Not Artificial Intelligence, It's Intelligence 57:39 - Go Build Something and Contradict Yourself

  ListenShow Notes
• 

  Episode #20

  Roll Your Own Problems

  Aug 24, 202649 minS2

  The 7:30 PM Tuesday Problem Hello dear show notes readers! This one starts with a website. I spent a good chunk of the weekend finishing the rebuild of Howard Home Realty — Lindsay Howard's brokerage — and it went live the morning we recorded. Twenty to twenty-five hours all in, about ninety-five percent of it alongside an AI tool. Dan, being Dan, immediately turned my little side project into a tip that's going to change how I work: go extend your Claude Code transcript retention before the thing quietly auto-deletes, then mine all of it into a repeatable playbook for the next time. Audit trail first, tooling second. From there we get to the meat. Dan fired a software vendor this week and he's building the replacement himself. So we've both got a build project going, and we spend a while circling the same caution from opposite ends. It is a genuinely great moment to roll your own solutions, and you want to be very careful about rolling your own problems. Otherwise, on some Tuesday at 7:30 PM the thing breaks, it's yours, and there's nobody else to call. Then Dan asks me whether AI is a deflationary technology, and I fumble it live. I'm leaving the fumbling in, because I think I talk myself into his answer over about three minutes and that's more useful than if I'd just nodded along. He gets there in one sentence — you get more for less, period — and the carve-out he adds forty minutes later is the part I've been chewing on since: deflationary everywhere except the one place all the money is currently going. The back half turns to a soft jobs print, a read on the new Fed chair with the politics stripped out, and Dan asking the question of the episode: what is actually inside the manufacturing number? If Meta is out there training electricians and plumbers to build data centers, is manufacturing coming back in a shape our statistics structurally cannot see? He handed that one to Claude on the record, and the answer is in the fact-check below. I'll spoil this much — his instinct was right and his follow-up guess was not, which is a very Dan way to go two-for-one. We land somewhere I keep circling back to, which is that a thing can be the best outcome in the long run and genuinely painful in the short one, and that being honest about is the best way to face it. Cheers, Sean Go out, build something this week. Tools & Platforms Mentioned Claude / Claude Code — the ninety-five percent of the Howard Home Realty build Cowork — including the mobile release that obsoleted a handful of tools I'd built to do the same job Claude Tag — Dan's been using it in one of his businesses, and it's the jumping-off point for the lock-in conversation Claude Fable — token appetite, orchestration, and hashtag skill issue Codex — Dan's on-the-go workhorse, and also the thing building UIs nobody asked for GitHub Copilot — my underwhelming work experience with it Buzz — Jack Dorsey's agent-native group chat workgroove.ai — work governance and auditability across humans and agents HQ — shared-context layer across coding agents Squarespace — the before picture Google Search Console / Google Analytics / Google Cloud — everything Squarespace was quietly doing for me IDX / MLS — the real-estate listing plumbing that was broken on the old site Epic Systems — Dan's example of why healthcare AI is a slower story X / Twitter bookmarks — Dan's over-stuffed information surface, and the raw material for his ingestion engine Companies Discussed Anthropic · OpenAI · Microsoft · Meta · Block · Google · Squarespace · Epic Systems · Goldman Sachs · Morgan Stanley · workgroove.ai · HQ · Howard Home Realty Links & References Howard Home Realty — Lindsay Howard's brokerage, live as of the morning we recorded. This is the cross-link I promised her on air. The before-and-after case study — on our own hosts page, if you want to see what twenty-five hours actually bought. Meta: America's Workforce Academy — the electrician-and-plumber training program Dan was describing, announced two months before we recorded. BLS Employment Situation — the July report, released two days before this conversation, and the reason Dan opened with "bad job numbers." FOMC press conference transcript, July 29, 2026 — Warsh in his own words, including the AI-capex-is-raising-prices line that undercuts our read of him. Claude Code data usage and retention — the cleanupPeriodDays setting behind Dan's tip. Default is 30 days; zero disables cleanup entirely. Introducing Claude Tag — what Dan's been testing. Jack Dorsey's Buzz — the Slack-killer with agents as first-class citizens. workgroove.ai and HQ — the two tools Dan named for making human-and-agent work legible. Philippine Statistics Authority, April 2026 labor force survey — where my eleven percent went to die. Why K-12 enrollment is declining — the demographic story under the elementary school closures. Unqualified Fact-Check We said some things. Here's how we did. = Nailed it | = Close enough | = Whiffed it The jobs print really was bad — worse than Dan said, actually Dan opened the macro half with "bad job numbers, unexpectedly low." He undersold it. The July employment report, released August 7 — two days before we sat down — showed nonfarm payrolls fell by 23,000 , against a consensus expecting somewhere around +83,000 to +95,000 . That's a miss of more than a hundred thousand jobs, and it wasn't a slow month, it was an outright decline. May and June got revised down by a combined 103,000 . Unemployment ticked down to 4.1%, but only because people left the labor force rather than because anyone was hiring. Full marks, and then some. "We added two hundred ninety five thousand jobs" Dan, reaching for the print that busted the recession narrative. No month in 2025 or 2026 came in at 295,000. The closest real narrative-buster was December 2024 at +256,000 ; the best 2026 candidate is March at +178,000 , which the House Budget Committee chair described as "triple expectations." The phenomenon Dan is describing absolutely happened — the hive mind kept calling for weak prints and kept getting embarrassed. The specific number is a composite of a memory. Philippine jobs are up eleven percent quarter over quarter Me, on something I'd seen that morning and explicitly flagged I hadn't read closely — which turns out to have been the right instinct. Philippine Statistics Authority numbers show employment going from 47.94 million in January to 48.89 million in April 2026 , which is about two percent , not eleven. Year over year it's roughly half a percent , and the less flattering details are that unemployment actually rose from 4.1% to 4.7% and underemployment worsened. But there's a real kernel here: the BPO and IT-BPM sector grew about four percent in 2025 to roughly 1.9 million workers , which genuinely does violate the "AI guts outsourcing" story. Right narrative, invented number, partial credit for hedging. Warsh is riding a deflationary AI wave Dan: "He's pro AI productivity. We've got a huge deflationary force on the horizon… things are kind of moving in his direction." That was true of the Warsh who hadn't been confirmed yet — the "structurally disinflationary" framing dates to around December 2025, before he had the job . As Chair he's said close to the opposite, twice, on the record. Senate testimony July 15 , asked whether AI would raise measured prices over the next twelve months: "I suspect it will." Press conference July 29 : "The business capex boom, for example, is driving up prices of memory and logic chips and associated AI infrastructure." He has held rates at 3.50–3.75% at both of his meetings and took three dissents in July — Hammack, Kashkari and Logan, all of them hawkish, all wanting a hike. And after the bad print, markets knocked down September hike odds without ever pricing a cut above roughly one percent. Where Dan is right: the cagey-and-not-communicating narrative is real and well documented, from CNBC's "more opaque Fed" to the Washington Post's "the new Fed chair won't tell you what he thinks" to Bloomberg's "left investors baffled." It's the deflation read that's a year out of date. Meta is training electricians and plumbers Dan nailed this one down to the framing — that it's Meta building its own jobs program, and that a company that has been engineering its own servers for years saying this out loud "says a lot." It's called America's Workforce Academy , announced June 8, 2026 : $115 million in year one, a free five-week program covering tuition, housing and a daily stipend, NCCER credentials, and a job guarantee for graduates . Electricians, plumbers, welders, fiber techs, data-center techs. Pilots in Baton Rouge, Columbus, Indianapolis and Houston , explicitly tied to Meta's roughly $600 billion US data-center buildout through 2028. Meta calls it the largest private-sector skilled-trades commitment with a job guarantee in American history. Dan also wasn't wrong that this isn't unique — BlackRock announced a comparable ~$100M trades investment in March. The trades don't show up in the manufacturing number — the question Dan handed to Claude on air This was the best question of the episode: is "manufacturing coming back in a different way" invisible in the official number because the work is landing in the trades? Yes, and it's definitional rather than arguable. Manufacturing is NAICS 31–33 — establishments that physically fabricate things, which is where a chip fab or a server plant lands. Electrical contractors are NAICS 238210 and plumbing and HVAC contractors are 238220 , both sitting inside Construction, NAICS 23 . Once the data center is actually running, its own staff fall under NAICS 518210 , inside the Information sector. So a crew wiring a hyperscale campus can never appear in manufacturing payrolls, no matter how much renaissance language gets wrapped around the groundbreaking. The AI buildout is split across three separate BLS supersectors, and only the physical hardware — chips, servers, transformers, if they're actually made here — counts as manufacturing. Dan's instinct: exactly right. The two thousand electricians don't all go to zero Dan's follow-up guess, which I endorsed on air, was that a good number of the construction crew sticks around permanently. The evidence runs the other way and it isn't close. Meta's Hyperion campus in Louisiana peaks near 7,500 construction workers and settles at roughly 1,000 permanent jobs — about seven and a half to one. The Hamm Institute puts construction at 0.7–2.0 workers per megawatt against 1–2 permanent staff per megawatt at hyperscale. Crews demobilize after twelve to thirty-six months and move to the next site; that's the whole business model of construction trades. In Dan's defense, the "2,000 electricians" figure is a perfectly fair ballpark for a gigascale build — electrical is typically the largest single trade on these jobs. It's the stickiness that doesn't survive contact. College enrollment is plummeting — U of A down twenty percent, maybe Delaware Me, and this is a near-miss with a real source underneath it. A Tucson.com piece published August 8 — literally the day before we recorded — reported University of Arizona first-year enrollment projected down about 23% , roughly 5,800 incoming freshmen against 7,500 the prior fall. So the number was real and about twelve hours old. But it's the freshman class, not the school: UA's total enrollment was down about 2.7% , and graduate enrollment actually rose. Delaware I invented from vibes — reporting there describes enrollment recovering . And nationally, enrollment rose about one percent in fall 2025 , up 187,000 students, with first-year enrollment up too. Real number, wrong noun, and the national trend is pointed the other way. Elementary schools are closing because there aren't enough kids Both of us, and this holds up completely. The US fertility rate hit an all-time low of 1.599 births per woman in 2024 , down from 2.1 in 2007. Public K-12 enrollment fell from 50.8 million in fall 2019 to 49.4 million in 2024 , and NCES projects it below 47 million by 2030 . On the ground in 2025–26: San Jose Unified closed five elementary schools after a 20% enrollment drop, Cedar Rapids closed five , Kyrene in Arizona closed four while sitting on capacity for 20,000 students with about 12,000 enrolled, and Fort Bend ISD in Texas consolidated seven. Not anecdote — pattern. "Buzz, just released by Jack Dorsey" Dan. Dorsey did announce it, on July 21, 2026 , but it's built by Block — his company — rather than being a personal side project. It's Apache-2.0 licensed, and the agent-native architecture Dan described is real: agents are first-class channel members with their own keypairs, over the Nostr protocol, model-agnostic by design. The "all agent ready on the back end" part is honest about intent and generous about status — it's pre-1.0 , desktop client at v0.4.22, mobile incomplete, workflow approvals still being built. Coverage is explicit that you shouldn't move your team over yet. Which, to be fair, is roughly what Dan said next. Final Score: 4 green, 4 yellow, 2 red Dan sweeps the trades round and drops the Fed round; I found a genuinely fresh number and stapled it to the wrong noun. The manufacturing question was worth the whole segment. Chapters 00:00 – Cold Open: Rolling Your Own Problems 00:23 – A Client Site, Live That Morning 03:59 – Dan's Tip: Stop Letting Your Transcripts Auto-Delete 08:06 – What Squarespace Does For You That You'll Now Do Yourself 12:33 – So Would You Call It a Deflationary Technology? 16:32 – The Agent Layer Becomes the Operating System 17:57 – When a Vendor Won't Tell You It's Broken 20:10 – A Tech-Support Agent for a Client Who Isn't Technical 24:58 – Advice for the Cohort Booing AI at Graduation 27:56 – Okay, Bug Brothers, You Got the Job 30:21 – Your Own Personal Chief of Staff 36:20 – Fable, Token Budgets, and Hashtag Skill Issue 38:43 – The Jobs Print Nobody Wanted 40:11 – What's Actually Inside the Manufacturing Number? 45:38 – Best in the Long Run, Painful in the Short 46:54 – The Airplane's in the Air, I'm Still Putting the Rivets In

  ListenShow Notes

Based on semantic analysis of episode topics and host coverage, this show is a strong guest fit for executives in:

Industry fit is computed by PitchCentric using vector embeddings of the show's episode catalog.