Where Are the Humans?

Dan: Meta's closed, just-works personal agent launched September 8 and hit #1 on Apple's US App Store within two weeks.

Dan: The block landed the weekend of September 20 — officially over agent identification and credentials, and in practice over a $68 billion ad business that depends on humans browsing sponsored listings.

Dan: Undersold — Walmart signed on as a Muse shopping partner at Meta Connect on September 23, four days before we said Walmart could be more AI-native than Amazon.

Dan: Meta jumped 11% on Monday, September 21, on a Wells Fargo target raise from $640 to $796 and Muse sitting at #1 in the App Store.

Sean: Since April 15, 2026, card-paying sellers have ad costs deducted from proceeds by default, with a net-30 invoice as the alternative — the lost card points are real, but no wire is required.

Dan: Hedged on air — YouTube is right at about 10% of Alphabet's $402.8 billion in 2025 revenue, but search and other is a little over half, not 90%.

Sean: Founded in 2004, but its breakout was pandemic-era, capped by PepsiCo's $550 million stake in 2022 — right about the rise, wrong about the birth.

Dan: John Philip Sousa, whose 1906 essay The Menace of Mechanical Music warned recordings would kill amateur music-making — and who made plenty of records anyway.

Sean: August 2026 manufacturing average hourly earnings rose 3.77% year over year against 3.09% for all private workers — a gap of exactly 0.68 points.

Sean: Timing right, with an all-time high in April 2009 and mostly negative readings 2010–2013, but the cause was composition (layoffs take the lowest-paid first), as Sean's own SMI page says.

Sean: Manufacturing added 23,000 jobs in a year while the five build-out industries gained 64,000 and the rest lost 41,000 — 4.6 jobs moved for every net job added.

Sean: The JOLTS manufacturing job-openings rate rose from 3.3% to 4.4% between July 2025 and July 2026, up about a third.

Dan: The August 2026 jobs report put unemployment at 4.1% — close enough for a guy who admitted he hadn't looked.

Sean: Medium-voltage switchgear quotes 52–80 weeks in 2026 and the 15–38kV gear data centers need runs 80–104 weeks, or about 18–24 months.

Sean: Right direction and right humility, but the evidence is stingier — Meta's Hyperion settles near 1 in 7.5 and Stargate runs from 1 in 64 to about 1 in 6.

Dan: On a lifecycle basis, IPCC medians put solar PV at about 41–48 gCO2e/kWh against about 12 for nuclear.

Dan: Right book, right chapter and right term (social arbitrage), but Camillo's documented footwear trade was Crocs and UGG — a Deckers brand, not Skechers.

Dan: Sheel Mohnot's bot ranks on 3-, 6- and 12-month returns, not 30/60/90 days, and reports +347% since May 2023 against +162% for the Mag 7 — wrong lookbacks, right result.

In a world disintermediated by personal agents, Meta is the better ten-year hold than Google. Agents don't click search ads, so search revenue erodes, while human attention concentrates in entertainment products where Meta's ads convert.

Walmart becomes the agent-friendly retailer and Amazon the agent-hostile one. Because Walmart's economics depend far less on ad revenue, it can open its catalog to personal agents, and sellers shift optimization effort toward Walmart as a result.

Manufacturing wage growth pulls further ahead of the private-sector average, from +0.68 percentage points today to above +1.0. It stays positive for a sustained period as the tariff-driven redistribution of manufacturing labor wears off and the churn that has been suppressing wage pressure stops.

“I truly believe we're at an inflection point and this thing is gonna go up above one." (The mechanism, ~48:57): "The tariff impacts and the tariff network redistribution is eventually gonna wear off… and I think we're gonna see an inflection in various American manufacturing wage sectors." (The horizon, ~54:01): "If you go watch these numbers over the coming twelve to eighteen months, you're gonna see this continue to play out." (The failure mode, ~53:11): "Overall degradation in demand. Somehow Oracle goes bankrupt next week." (Dan's addition): "Rates are gonna go up.”

Over the next decade the comfortable upper middle class sees the weakest earnings and net-worth trajectory of any cohort, as high earnings built on credentialed white-collar work stop compounding. Trades, manufacturing and other atom-economy "doers" move up, and the resulting status anxiety becomes a political force.

“My pet theory is that the upper middle class is the one that's gonna get walloped over the next decade… their expenses are too high because they have high earnings. And I'm questioning their ability to have high earnings going forward the way history has foretold." (The political half): "It's the two percent where revolutions come from. I think it's the upper middle class that is where the angst is… the near to power who are watching themselves and their cohort backslide." (The other side of the trade): "The tradespeople, the manufacturing, the doers, the people in the physical and atoms world, they're gonna come up.”

Two-sided. Sean: public perception of AI-assisted work settles on a distinction between things made *by* AI (slop, judged harshly) and things made *with* AI (human intent visible, judged as craft). Daniel: the distinction dissolves instead. Like the phonograph, AI use "becomes water," watermarking fails because non-US models won't comply, and nobody asks who held the pen.

Walmart will successfully position itself as the pro-US-small-business alternative to Amazon's China-factory-direct model, leveraging brick-mortar / data integration to drive omnichannel adoption.

Resolution: The growth half landed cleanly: Walmart's Q1 FY2027 results (reported May 21, 2026) showed US eCommerce +26% and advertising +36–37%, driven by store-fulfilled delivery/pickup (omnichannel), Walmart Connect, and a Marketplace that crossed ~200,000 active sellers. But the strategic-framing half ran the *opposite* way: rather than positioning as the pro-US-small-business alternative to Amazon's China-direct model, Walmart aggressively recruited Chinese sellers — Marketplace Pulse found China-based merchants reached 34% of active sellers, with nearly 60% of 2025 new sellers China-based — while Amazon's late-2024 "Haul" pushed Amazon further toward the same China-direct model. Measurable omnichannel/marketplace growth came true; the differentiating pro-US-SMB positioning did not (evidence runs counter) → Partially. Sources: Walmart earnings (SEC), Marketplace Pulse, Modern Retail, CNBC (2026-06-05).

A meaningful share of the construction-phase trades workforce on a large data center stays on permanently after commissioning — the headcount does not collapse to near-zero once the facility is running.

Resolution: Flagged 🔴 in the Ep 54 fact-check and the evidence has only hardened since. **Stargate (1 GW): ~6,400 construction workers → 100–1,000 permanent staff — 1.6% to 15.6%**, and the permanent roles are facility technicians and engineers, a different skill set rather than a retained construction crew. Hyperscale permanent staffing runs **0.2–0.3 people per MW** (against 2.0–3.0/MW at small 1–5MW sites), so a 300MW campus supports roughly 60–90 permanent people. Brookings (published 4 May 2026, updated 10 Aug 2026) found a county's first large data center raised data-processing employment **+56% over a decade** and telecoms +43% — but that is **only ~100–200 jobs**, with **wages unchanged** and home prices up just 2–5%. **The steelman the hosts didn't make, and it holds.** In Virginia, Texas, Ohio and Arizona the buildout is continuous enough that trades **chain from project to project for years**, so trades employment persists *regionally* even though almost nobody stays on any single site. If the claim is "the crew that builds it stays to run it," it is false. If it is "the work stays in the region," it is defensible — and Meta's training program guarantees *jobs*, not jobs at one site, which would satisfy the spirit while failing the letter. Score both readings as the entry originally asked. Sources: Brookings, Stargate reporting, Hamm Institute per-MW staffing. (2026-08-29)

Consumers hand purchasing authority to AI agents before businesses do, because the cost of an agent's mistake scales with the order size; B2B's first purchase stays human even where reorders get automated.

The agent interface becomes the operating system — the primary layer through which people touch every other tool — and it must be a *horizontal* layer that survives changes of model provider and harness, not a vertical stack per vendor.