Block 11, Article 1 — They Already Knew
They already knew.
That is the thing to understand about the people building bunkers in New Zealand, reserving berths on private spacecraft, and quietly acquiring land in the Southern Hemisphere. They are not pessimists or cranks. They are the people with the best view of the system, acting on what they see. The compounds being built in New Zealand and the Southern Hemisphere, the seed vaults, the former Navy SEALs on standby to be summoned by prearranged signal — these are not expressions of ideology. They are investment decisions made by people who ran the models and did not like the output.
Douglas Rushkoff was paid to consult with a group of them. He expected a conference on technology and society. What he got, in a desert compound with no cell service, was a single question asked in multiple variations by multiple very wealthy people: How do we maintain authority over our security guards after the event? The food supply was accounted for. The perimeter was accounted for. The only unsolved variable was loyalty. One proposal involved combination locks on the provisions, keyed only to the billionaire’s biometrics. Another involved disciplinary collars — shock devices worn by the security staff as a condition of survival. The answer Rushkoff offered — that the way to secure a guard’s loyalty after the event is to treat him like a person before it — was rejected. Too dependent on the social fabric the compound was designed to outlast.
They are investing in exits. That is information. What they are exiting from — and when they expect to need to leave — is the question this block answers.
What do they know?
California has always been where America’s economic future arrives first — the dot-com boom, the gig economy, the smartphone era each showed up in the Bay Area years before reaching the rest of the country. What is happening to California’s workforce right now is not a regional story. It is a preview.
The state’s unemployment rate stood at 5.3 percent in April 2026 against a national rate of 4.3 percent — above 5 percent for nineteen consecutive months running. There are 1.9 unemployed workers per job opening in California, against 1.1 nationally. Thirty percent of unemployed Californians have been searching for at least six months. The information sector — tech and Hollywood combined — has shrunk 17 percent since mid-2022.
Through the first five months of 2026, U.S. tech companies announced 123,653 job cuts — a 66 percent increase over the same period in 2025, making tech the single largest job-cutting sector in the economy. Of workers laid off between January and April, 47.9 percent of the cuts were attributed directly to reduced need for human labor from AI and workflow automation. Meta has eliminated roughly 33,000 positions since 2022; its April 2026 round of 8,000 cuts was the first framed explicitly as headcount traded for compute, not cyclical adjustment.
The generational signal is the sharpest data point in the picture. Stanford HAI’s 2026 AI Index found that software developers aged 22 to 25 — entry-level, exactly the population the Class of 2026 is trying to join — saw employment fall nearly 20 percent since 2024, even as headcount for older, more experienced developers kept growing. The compression is not hitting the workforce evenly. It is hitting the entry point first, which means it is also shrinking the pipeline of experienced workers five years out. The median time for a laid-off tech worker to find a new role has stretched from 3.2 months in 2024 to 4.7 months in 2026.
California Federation of Labor Unions president Lorena Gonzalez named what the state’s response has produced so far: an executive order with a 180-day study clock and no enforceable rights. “Catastrophic job loss from AI is not inevitable,” she said. “It’s a political choice.”
The people running the displacement are not hiding it, either. Dario Amodei, running Anthropic, said AI would eliminate half of entry-level white-collar jobs within one to five years. Mustafa Suleyman, running Microsoft AI, told the Financial Times in February 2026 that AI would achieve human-level performance on most professional tasks — accounting, legal work, marketing, project management — within twelve to eighteen months. The World Economic Forum found 41 percent of employers worldwide planned to reduce headcount through AI automation by 2030. Ford’s CEO put it plainly: AI will cut in half the number of white-collar jobs in the United States.
These are the people running the displacement. On the record. Naming a schedule shorter than their own earlier estimates.
Salesforce cut 4,000 customer support roles and attributed it directly to agentic AI. HP announced up to 6,000 cuts by 2028. Duolingo announced it would no longer use human contractors for any work AI can handle. Snap cut sixteen percent of its entire staff because AI was writing more than sixty-five percent of its code.
Michelah described it this way, to a New York Times moderator: “It’s like a desert. There’s nothing really there. You can be out there, but you’re not being hydrated.” She is in her twenties. She did everything she was told.
The credential no longer unlocks the market.
The Class of 2026 is approximately 2.2 million people, one of the largest graduating classes in American history, and statistically among the least likely to find the work their degrees were supposed to open. The unemployment rate for recent college graduates stood at 5.7 percent in the first quarter of 2026 — a multi-year high, running above the national rate for the fifth consecutive year. More than four in ten were already underemployed, working jobs that do not require the degree they just earned.
They are carrying the bill regardless. The average borrower in the Class of 2026 graduated with $35,000 to $43,000 in student loan debt — monthly payments beginning now, due for the next decade or two, whether or not the degree produces the employment it was supposed to produce. The Randstad Workmonitor survey found 76 percent of employers predict at least half of all entry-level roles will disappear within five years. One in three graduates say their college prepared them to use AI in the workplace.
The institutions most exposed to what follows are not the Harvards. Harvard has an endowment. The institutions most exposed are the land grant colleges — the regional public universities, the HBCUs, the first-generation institutions built specifically for the population now running the calculation. They survived the defunding of the 1980s by shifting to a tuition model. The tuition model requires enrollment. Enrollment requires students who believe the credential is worth the debt. When that belief fails rationally — when the entry-level market is visibly contracting — the enrollment collapse follows.
The land grant college opened in 1862 to educate the farmer’s son. Expanded in 1944 to educate the veteran. Defunded in 1980 to shift costs to the student. It closes in 2030 because the student did the math. The original commons commitment ends not with a repeal but with a rational individual decision made forty million times.
The commons paid twice.
The jobs being eliminated now were already replacements. The fisherman’s grandson became a paralegal. The miner’s daughter became a data analyst. The logging town’s children left for the cities and took the jobs the service economy was hiring for — entry-level, credential-dependent, screen-based. Those jobs are the ones going first.
AI did not emerge from private ingenuity untouched by public investment. DARPA funded the foundational research from the 1960s forward and built ARPANET — the direct ancestor of the internet. The National Science Foundation, the Department of Energy, and the National Institutes of Health funded the machine learning breakthroughs. Public universities trained every researcher who built the models. The training data was scraped from the accumulated written output of human civilization — libraries, universities, public archives, the labor of generations. The public paid at every stage. The public was compensated at none of them.
AI displacement is framed as disruption — innovation, creative destruction, capitalism working exactly as designed. Who paid for the foundation the disruption is built on, and who is already arranging an exit from the consequences of deploying it, were not in that frame.
In June 2026, Bernie Sanders introduced the American AI Sovereign Wealth Fund Act. The legislation would require the largest AI companies to transfer fifty percent of their stock to a federal sovereign wealth fund — giving the public a direct ownership stake in the infrastructure it paid to build. Sanders cited Norway and Alaska as proof the principle is not ideological: a social democracy and a Republican-governed state reached the same conclusion about infrastructure built on public foundations. The committee that would hold the hearings is the same room Block 8 documented. The pattern is identical.
That claim is Block 12’s argument. Here, it is a marker — a flag on the map showing where the bill is addressed, if the room ever becomes capable of addressing it.
What the exit strategy reveals.
The desert meeting Rushkoff attended was not an eccentricity. It was a planning session. The people in that room — the owners of the extraction apparatus, the beneficiaries of the royalty rates and the tax provisions and the captured committees documented across the prior ten blocks — ran the models and did not like the output. They are taking the profit extracted from public land and public infrastructure and investing it in private exits from the consequences of that extraction. That is the revealed preference of the people who built the system. It is the clearest statement of what the system was for.
What they know — and what this series has been documenting — is that the commons was depleted in a specific sequence, by specific mechanisms, over a specific span of time. The depletion was not secret. The Ogallala measurements were public. The fishery data was public. The climate models were public. The topsoil surveys were public. The extraction continued anyway, because the time horizon of the quarterly earnings call ends before the aquifer does, and because the people making the extraction decisions had already begun pricing in personal exits from the consequences.
The three crises arriving together — the displacement wave, the debt spiral, and the natural capital column that does not recover — are not three separate failures. They are the same failure, arriving in different registers, on a known schedule, in a room that is structurally incapable of responding. The next three articles document each one in turn.
The people buying land in New Zealand already know how the accounting comes out.
The structural argument behind the AI displacement mechanism lives in Essay 12 of The Narrow Gate.
The record is public. Look it up:
| Stanford HAI, 2026 AI Index (developer employment data) | hai.stanford.edu/ai-index/2026 |
| Challenger, Gray & Christmas, job cuts reports | challengergray.com/blog/category/job-cuts-report |
| Federal Reserve Bank of New York, Labor Market for Recent College Graduates | newyorkfed.org/research/college-labor-market |
Ask an AI assistant: “What does Stanford HAI’s AI Index and California’s 2026 unemployment data show about AI’s effect on entry-level tech employment, and which job categories are most exposed?”
— — —
Steve Sagnotti
is a serious amateur photographer, writer, and technologist based in Oregon. With his camera he tries to capture common images not often seen, leading to common questions not often asked.
© 2026 Steve Sagnotti
— — —
Sources
1. California’s unemployment rate held at 5.3% through April 2026 (EDD, California’s April Unemployment Rate Remains Steady at 5.3%); U.S. technology companies announced 123,653 job cuts between January and May 2026, per Challenger, Gray & Christmas (Los Angeles Times, Cisco to lay off more than 400 workers in California, June 2026). Stanford HAI’s 2026 AI Index confirms software developer employment ages 22–25 fell nearly 20% since 2024 (Stanford HAI, 2026 AI Index Report — Economy chapter).
2. California Federation of Labor Unions, AFL-CIO, press release, May 21, 2026: California Labor President Lorena Gonzalez Responds to Governor’s AI Executive Order.
3. Rushkoff’s own retellings consistently describe “ex-Navy SEALs” / contracted security, not active-duty personnel (NPR, In ‘Survival of the Richest,’ author Douglas Rushkoff examines the escape plans of the tech elite, September 6, 2022).
4 Rushkoff has repeated this detail across interviews and excerpts (Rushkoff Archive, Catastrophe is Just the Figure; Next Big Idea Club podcast, How the Mega-Rich Plan to Outsmart Doomsday).
5. Sen. Bernie Sanders introduced the bill June 18, 2026: one-time 50% stock tax on AI companies with $200M+ annual AI revenue, ~$7 trillion projected fund size, ~$1,000/year projected dividend per American, seven-member independent commission (Sanders.senate.gov, NEWS: Sanders Introduces Legislation to Create $7 Trillion AI Sovereign Wealth Fund).
Leave a Reply