The Fractional Republic

The problem

America mints elite aspirants far faster than elite roles. About four in ten recent college graduates work in jobs that don't require their degree — a number the New York Fed has tracked for years. Nearly sixty thousand PhDs are conferred annually into a market of a few thousand professorships. Law, consulting, media, and management run the same narrowing pyramid. The historian Peter Turchin calls the pattern elite overproduction: a society that produces surplus claimants to status produces frustrated counter-elites, and frustrated counter-elites are how stable countries stop being stable. Whatever one thinks of the grand theory, the underemployment number needs no theory at all.

The two classic responses are both wrong. Produce fewer aspirants — pull up the ladders — and you betray the covenant. Manufacture fake sinecures and you get a mandarinate. The third option is to change the geometry: multiply the podiums by dividing them.

The premise: AI is a leveler, so far

The early empirical record on workplace AI points one consistent direction. In the largest field study to date, AI assistance raised novice customer-agent productivity by roughly a third while barely moving the experts (Brynjolfsson, Li & Raymond). In controlled writing tasks, AI cut the gap between weaker and stronger professionals roughly in half (Noy & Zhang, Science). Among consultants, the bottom half gained twice what the top half did (Dell'Acqua et al.). The pattern: AI lends every worker the best colleague's brain. Expertise — the thing licenses, salaries, and status monopolies were built to ration — is becoming abundant.

What stays scarce is everything else: accountability, judgment, presence, hours, and above all slots. So the slots are what policy must multiply.

The proposal

1. Tiered licenses for the AI era. Every licensed profession — law, medicine, accounting, architecture, engineering — opens limited-scope practitioner tiers built around AI decision support with audit trails. The precedent already carries American health care: nurse practitioners, a few hundred thousand strong, now anchor primary care with outcomes at parity within their scope. Dental therapists and prescribing pharmacists extended the pattern. Utah's regulatory sandbox for legal services shows the method: test scopes, measure outcomes, then license what works. The license monopolies made sense when expertise was scarce. The machines are ending the scarcity; the licenses should admit it.

1. The fractional default. The federal government — the nation's largest white-collar employer — makes every position job-shareable and part-time-eligible by default. The statute already exists (the Part-Time Career Employment Act of 1978) and has been ignored for two generations; enforce and modernize it, and require federal contractors to report fractional availability. One AI-amplified senior role becomes two or three senior roles — held by people who also build, teach, care, or farm with the rest of the week.

1. Pro-rata parity. Part-time work receives pro-rata equal pay, benefits, and advancement — the reform the Netherlands legislated a generation ago, which made part-time careers first-class and Dutch workers among the most satisfied measured. Fractional cannot mean second-class, or nobody ambitious will take the fraction.

1. Repair the status of the trades. Double registered apprenticeships — America runs roughly 600,000; Germany, with a quarter our population, runs over a million. Strip degree requirements from public jobs that don't need them, as several states began doing in the 2020s. And let the American Year (№ 005) carry trade tracks, so the service covenant attaches its status to tools as readily as titles.

The background — and where this is going

Status is a ratio: aspirants over podiums. Elite overproduction is what happens when the numerator grows for fifty years while the denominator stands still. AI now lets the podiums divide without the sound shrinking — a senior role at three days a week is still a senior role, and the professional it frees for the other two days is a neighbor, a coach, a builder, a parent. A society of part-time titans and full-time citizens is more connected than one of exhausted winners and credentialed spectators — and much harder to destabilize.

Say the quiet part plainly: if machines are about to do more of the work, only two questions matter — who owns the machines, and who gets the hours. Proposal № 001 answers ownership: everyone, through the Fund. This proposal answers hours: everyone, through slots that divide. Capability abundance owned narrowly and worked by few is how republics curdle. We'd rather multiply the podiums.

Nothing here compels anyone. Full-time ambition remains fully available; the door gets wider, not the room smaller.

The honest objections

"The leveling evidence is early and task-specific." Correct, and we publish it as such. The effect could invert as frontier work concentrates. That is precisely why this catalog pairs hours with ownership: if AI levels, the licenses should admit it; if AI concentrates, the Fund (№ 001) catches the gains anyway. The citizen is hedged in both worlds.

"Tiered licenses mean second-class service for the poor." The nurse-practitioner record says otherwise within scope — and the honest comparison for the underserved is not the full-price expert, who was never coming, but nobody. Access at audited scope beats absence at any price.

"Fractional senior roles are a coordination fantasy." Job-sharing has decades of quiet precedent, and AI is itself the coordination technology — the shared context lives in the tools now. We set defaults and parity rules; markets decide where fractions make sense.

"This sounds like rationing ambition." It is the opposite: ambition currently rations itself against a fixed number of chairs. We are adding chairs. Nobody is sent anywhere, capped at anything, or docked for working five days if five days is the life they want.

Sources

• Federal Reserve Bank of New York, underemployment of recent college graduates (newyorkfed.org); NSF Survey of Earned Doctorates
• Turchin, End Times: Elites, Counter-Elites, and the Path of Political Disintegration (2023)
• Brynjolfsson, Li & Raymond, "Generative AI at Work" (NBER); Noy & Zhang, Science (2023); Dell'Acqua et al., "Navigating the Jagged Technological Frontier" (2023)
• AANP nurse practitioner data (aanp.org); Utah Office of Legal Services Innovation; U.S. DOL registered apprenticeship data (apprenticeship.gov); OECD, part-time employment, Netherlands