The Expensive Hour

First principles · Baumol, 1966

The Expensive Hour

AI is the largest productivity event ever aimed at thinking. The interesting question is not what gets cheap. It is who gets the raise.

John Rector· · 6 min read· Artificial Intelligence

Tuesday · 7:41 a.m.

You hand your daughter across the half-door to her teacher, and the envelope waiting in the cubby is the tuition increase — the third in three years. That evening you buy a sixty-five-inch television for less than one month of the increase. It is a better television than last year’s model, and it costs less than last year’s model. Nobody at the daycare got worse at their job. Nobody at the television factory loves your daughter.

In the mid-1960s, two economists — William Baumol and William Bowen — were hired to figure out why the performing arts were always broke. Theaters, orchestras, opera, dance: all of them permanently in deficit, decade after decade, no matter who managed them. Baumol and Bowen expected to find mismanagement. They found arithmetic.

The violinist’s wage is not set in the concert hall. It is set everywhere else — by what her time could earn out in the fast part of the economy. When factories and offices get more productive every year, they can pay more every year, and the concert hall has to keep up or lose the violinist to them. So the quartet gets a raise it never earned on stage, and the ticket price carries it to you.

Baumol called it the cost disease, which was a failure of naming, because nothing is broken. It is simply what an economy does with an hour that refuses to compress: it reprices it. Whatever cannot be sped up becomes relatively more expensive, forever, while everything that can be sped up becomes cheaper, forever. You have lived inside this law your whole life. The television, the flight, the long-distance call collapsed in price. College, childcare, the hospital, the haircut, the live show climbed. One lane compounds. The other stands still. One labor market connects them — and the connection is the whole machine.

Figure 01

The shape of unbalanced growth

The fast lane — productivity compounds

Output per hour rises every year, so wages can rise while prices fall.

  • Televisions, phones, flights
  • Software, search, drafting
  • Anything a machine can repeat

The slow lane — the hour is the product

Output per hour is fixed by the hour itself, yet wages must keep pace — so prices rise.

  • The quartet, the classroom
  • The bedside, the job site
  • Anything that requires being there
The mechanism of Baumol and Bowen’s unbalanced growth, drawn as shape rather than data. The lane assignments are the argument of this piece; the mechanism itself is the economic record.

The turnAim the law at the present

AI is the largest productivity event ever pointed at cognitive work — at drafting, research, analysis, correspondence, code, the entire paper interior of every profession. Nearly everyone is asking the cheap question: what falls to zero? Baumol teaches the other question, the one with the money in it: when one lane speeds up, who gets the raise?

For sixty years the line between the fast lane and the slow lane ran between industries — the factory on one side, the concert hall on the other. That line has now moved inside each job, inside each afternoon. The research memo is the fast lane; the hour in the room is the slow one. The draft compresses toward nothing; being there still takes sixty minutes.

The hours that cannot compound — presence, witness, custody, the signature with a body behind it — are about to collect the largest relative raise in economic history.

And note what the law does not require: improvement. The raise arrives unearned. The nurse of 2030 will be no better at bedsides than the nurse of 2020; her hour gets dearer anyway, because everything around her hour got fast. When it happens, people will call it inflation, or gouging, or a crisis of care. It is none of those. It is the arithmetic of a string quartet, finally running at machine speed.

March 2029 · 4:12 p.m.

The invoice has two lines. The first covers everything you used to think you were paying for — the research, the modeling, the memo — and it is priced like electricity: four dollars and change, metered. The second line reads counsel, in person, two hours, and it is most of the bill. You pay it without complaint. The first line told you what the machine concluded. The second line was a person putting her name, her license, and her afternoon between you and the consequences.

PredictionsWhat the repricing looks like

  1. The blended hourly rate dies.

    Professions have always hidden the compressible hour and the human hour inside one average price. An average cannot survive one of its two inputs going to zero. Invoices will split — machine lines priced like utilities, human lines priced like scarcity — and any firm that keeps averaging will be undercut on the first line and underpaid on the second.

  2. “In person” becomes a printed price tier.

    The default version of nearly every service becomes the absorbed version — handled machine to machine, no human hour consumed. A human present becomes a named surcharge, listed the way “expedited” is listed today. The first time you see it on a rate card it will feel cold. Within a decade it will feel honest.

  3. The raise lands on the people everyone told to learn to code.

    The electrician under the house, the nurse at the bed, the teacher in the room, the caregiver at the door — their hour cannot be compressed, so their wage chases the AI-amplified benchmark upward without their workday changing shape. The career advice of the 2010s inverts almost exactly.

  4. Counterfeit presence becomes the signature fraud — and verification becomes an industry.

    When the premium is on a human being there, the money is in faking a human being there: avatars, recordings, “live” sessions one operator supervises twelve at a time. Expect certificates of presence — proof that a particular person spent a particular hour in a particular room — sold the way audits are sold now.

  5. The political argument of the 2030s is presence inequality.

    The affordable version of everything — school, medicine, counsel, care — becomes the absorbed version, and it will be good. The human-hour version becomes the expensive one. “Who still gets a person” replaces “why is college so expensive” as the complaint of the decade, and it is a harder complaint, because this time too the arithmetic has no villain.

Where I standLiving on the fast side of the line

I live on the fast side of this line on purpose. My first thought each morning is to delegate the whole outcome to my personal AI — I wrote about that reflex in The First Thought — and almost everything I produce in a day now costs me almost nothing. This site publishes daily. The drafts arrive finished. What is left on my calendar is only rooms: the classes I teach in person, the tables I sit at. And I have watched what happened to the price of those hours, mine and everyone else’s. Nobody pays me for documents anymore. They pay for the hour with me in it. I argued in The Last Line Is You that when the machine takes every other step, the queue moves to you. This is the same truth wearing its other uniform: the queue is where the waiting pools, and the expensive hour is where the money does.

The ledgerWhat this claim rests on

Already true
A half-century of relative prices is on the record: manufactured goods fell while services built from person-hours — education, care, medicine, live performance — climbed, exactly as unbalanced growth predicts. And every professional invoice that already lists “consultation” apart from “preparation” has drawn the line this piece runs on.
What has to happen
AI has to act as a complement: the people wielding it must earn more, so that the benchmark wage the slow lane competes against keeps rising. And labor has to stay mobile enough to transmit that benchmark from the fast lane to the slow one. Both have held through two centuries of mechanization. Both are assumptions, not laws.
Where I am probably wrong
The mechanism has a reverse gear. If AI replaces more than it amplifies — if displaced analysts flood into presence work faster than demand for presence grows — then the supply of human hours swells and the expensive hour gets cheap instead. In that world I have the sign backwards and this piece ages badly. The instrument to watch is wages in care and the trades: if they stall while cognitive output explodes, believe the stall, not me.

Go back to the cubby. The tuition letter was never a malfunction; it was a forecast that arrived a few decades early. Four players, forty minutes. The hour with your daughter in it was never going to get faster, and so it was never going to get cheaper. We are walking into an economy where almost every other hour collapses toward free, and the law that priced the quartet will do what it has always done, at a speed it has never done it.

The expensive hour is the one with a person still in it. It always was. Now it will say so on the bill.

Background

Author: John Rector

John Rector is a Charleston-based entrepreneur, author, and AI strategist. He co-founded E2open, the supply-chain software company acquired for $2.1 billion in 2025, and in 2026 opened Charleston AI, a 3,000-square-foot lab that helps people and organizations understand and use artificial intelligence. He is the creator of The Reality Equation — a lecture series, book, and curriculum exploring attention, prediction, and how reality is experienced — and the author of more than two dozen books. He writes and speaks widely on artificial intelligence, attention, and the future of human work.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Discover more from John Rector

Subscribe now to keep reading and get access to the full archive.

Continue reading