A reply to the Fed · First principles ·
We Don’t Generate Ideas. We Actualize Them.
The chairman of the Federal Reserve says super intelligence could supercharge idea generation. Ideas were never the scarce thing. Hands were. And all over the world, one person at a time, the hands are arriving.
By John Rector · about 9 minutes
Ideas wait above. Hands live below.
The firm era. The hands live inside a few firms, and only the ideas those firms choose get made.
- An idea with hands
- An idea still waiting
- A person actualizing an idea
- A person whose hands are spoken for
An idea is free to use. Making it real is not.
Nonrival · the idea
My use of it takes nothing from yours. A million people can hold the same idea at once, and it costs nothing more.
Rival · the actualization
An afternoon, a body, a signature, a risk. Whoever spends them on this idea cannot spend them on another.
Tuesday, 6:12 a.m.It arrives in the shower, the way they do: a better way to schedule a clinic’s nurses, a dye that holds on wool, a map of every pothole on your road. You did not make it; you were standing there and it came in. Then you towel off and do the arithmetic every working person has done for a century: who would I need to hire, how much would I need to raise, which company would have to say yes. The idea goes back on the shelf. It is still there.
What the chairman said
On Monday, Kevin Warsh opened the Federal Reserve’s Productivity and Jobs Task Force conference in San Francisco. It was a thoughtful speech: humble about forecasts, serious about risk, generous to the people in the room. Its economics rest on a tradition I respect. Endogenous growth theory holds that ideas are the ultimate resource. Paul Romer showed in 1990 that ideas are nonrival, so one person’s use takes nothing from anyone else’s, and their value grows with every new user.
Warsh’s question was whether super intelligence changes the picture. His answer was that it might differ from electricity and the combustion engine, which spawned decades of follow-on invention but never automated the making of ideas. Super intelligence, he said, could “supercharge idea generation.”
I agree with nearly every sentence of that speech. I disagree with the verb.
We do not generate ideas
I have argued for years that people do not have ideas; ideas have people. The line is usually credited to Carl Jung. In the Reality Equation it is not poetry. It is structure. Reality is a ratio, R = A/E: the Actual the world hands you, over the Expectation you bring. Expectation is complex, E = P + iI. Its real part is Prediction, what habit and history expect. Its imaginary part is the Ideal: the pull of a coherent idea, already present before reality arrives. Nobody sets it. Nobody generates it. No entity can touch either term of the equation, ever, by any means.
What a person can do is act. Actions leave artifacts. Artifacts become history, and history is Actual. Consciousness does not manufacture the future; it manufactures history. That is what it means to actualize an idea: to be the person through whom an idea leaves its marks on the world.
So a machine said to “generate ideas” is doing something else, something useful. What it produces at almost no cost is proposals: drafts, designs, hypotheses, plans. Those are wonderful. They are not the scarce input.
There have always been more ideas than hands. Every engineer has a drawer. Every grandmother has a business she never started.
Why hands were scarce
In 1937 Ronald Coase asked a question that sounds naive: why do firms exist at all? If markets work so well, why not hire every pair of hands by contract, task by task? His answer was that using the market has a cost. Finding people, bargaining, directing them and checking the work all take effort. When that effort costs more than directing people inside one organization, you build a firm. The firm’s boundary sits wherever organizing inside becomes as expensive as buying outside.
Put Romer and Coase side by side and the last century fits in one line. Ideas were free to use, but the hands to actualize them came bundled inside firms. So firms decided which ideas got hands. The idea in the shower had to wait for a company, a team, a fund. It had to wait for someone else’s yes.
Read that way, the well-known finding that ideas are getting harder to find may be measuring something nearby. Nick Bloom, Chad Jones and their coauthors found research effort climbing for decades while research productivity fell even faster. My reading, and it is argument rather than record: what grew harder was not finding ideas. It was actualizing them through ever-larger institutions, where each idea needs more hands, more meetings and more permission before it touches the world.
A correction, same dayIn the next section I let Warsh name the one thing no delegate can do: judgment, morality, responsibility. A delegate can already show all three. The real answer is narrower and harder. It is liability: the person who can be made to pay. I wrote that out in the follow-on, Not Moral. Not Responsible. Liable.
What super intelligence actually supercharges
It supercharges hands. A delegate drafts the contract, builds the site, reconciles the books, answers the phone, writes the code, finds the supplier and files the form. The cost Coase described, of finding, directing and checking the work, falls toward the cost of a sentence.
Then Coase’s logic runs in reverse. When organizing hands costs almost nothing, the boundary of the firm shrinks toward the smallest unit that can still do the one thing no delegate can do. Warsh named that thing himself: humans must be the deciders. Judgment, morality, responsibility. Someone has to stand behind the work, sign it, answer for it, and look a customer in the eye. That input cannot be copied or shared. It is rival in the strictest sense, and the smallest unit that carries it is one person.
That is the solopreneur. Not a freelancer waiting for a firm to pass work down, but a firm of one: the minimum an idea needs in order to become history.
Warsh also said, almost in passing, that long-run growth depends on the number of people producing new ideas “and setting them in motion.” His footnote points to the worry that slowing population growth means fewer such people. But the count that matters was never the number of humans. It was the number of humans who could get hands. For a century, firms rationed that number. They no longer do.
All over the world, this morning
Hangzhou, 9:40 a.m.A former logistics clerk runs a company of one on a floor full of other companies of one. Her delegate negotiates freight in four languages while she walks the factory floor and shakes the owner’s hand.
Lagos, 2:15 p.m.A pharmacist who always knew which medicines would run short before the rains has turned that knowing into a service clinics subscribe to. She has no staff. She has a delegate, and a phone number people trust.
Charleston, 4:50 p.m.A retired machinist sits beside an expert in a lab off Jessen Lane, turning a jig he sketched in 1994 into a file a shop three states away can cut tonight. He is sixty-eight, and this is his first company.
None of them generated anything. The idea had each of them long before. What changed is that the hands showed up.
Five predictions
“Idea generation” will give way to actualization as the question that matters.
Within a few years the serious debate about AI and growth will stop asking how many ideas machines can propose and start asking how many people can carry one into the world. Policy will follow the second question, because it is the one with a person in it.
Most new firms will be one person, and many will stay that way on purpose.
Growth will come from more firms rather than bigger ones. A firm of one that does well will add delegates, not departments, and will count its success in customers kept rather than headcount.
Where you live will matter less than whether you will stand behind it.
An idea’s chance of becoming real will depend less on being near capital and big employers and more on whether one person will put their name and face on it. The map of new firms will flatten across countries and small towns.
The official numbers will see it late.
Statistics built to count employers, payrolls and hours will undercount firms of one for years. Warsh noted that productivity has not yet broken out structurally. It will be felt on the ground long before the data says so.
The best employer becomes the first customer.
Large institutions will compete for actualizers rather than employees, buying from firms of one instead of hiring them. The advice I give my students, let your first employer be your first customer, will sound ordinary.
I have lived both eras
E2open began with two people, Paul Stern and me. On paper, its founders were about ten corporations, Toshiba, IBM, Hitachi and Solectron among them, alongside venture money led by Crosspoint, and we started in Crosspoint’s building. That is what it took, in that era, to give one idea enough hands. Today at Charleston AI I watch people walk into our incubator with an idea that has had them for years and reach real customers on a budget, with a delegate doing the work a department once did. The ideas did not get better. The hands got cheap. I wrote about the person an idea acts through in The Principle Has No Hands, and about the solopreneur’s first loyalty in The Idea’s Liaison. This is the economics underneath both.
The claim, and how it fails
- Already true
- Delegates now do much of the organizing work Coase described: drafting, building, reconciling, answering. Companies of one are multiplying in many countries, and some governments now encourage them by name.
- What has to happen
- The cost of trust has to fall with the cost of hands. Customers must believe a single person will stand behind the work, and contracts, payments and liability must fit a firm of one.
- Where I am probably wrong
- If firms of one keep starting but cannot find customers, the bottleneck has only moved from hands to attention, and whoever owns attention becomes the new boundary of the firm. Early reports from solo-founder communities already show many struggling to sell. If in ten years most firms of one live on a single platform’s terms, I was wrong about where the power went.
Tuesday, 6:12 a.m. Again.The same idea arrives in the same shower. This time you skip the arithmetic about whom to hire. You tell your delegate what you saw, and by evening there are artifacts: a page, a price, a name to call in the morning.
The idea was never yours to generate. It is yours to actualize.
Background
- Kevin Warsh, Opening Remarks, Productivity and Jobs Task Force Conference, Federal Reserve Board and Federal Reserve Bank of San Francisco, .
- Paul M. Romer, “Endogenous Technological Change,” Journal of Political Economy 98 (1990).
- Ronald H. Coase, “The Nature of the Firm,” Economica 4 (1937).
- Nicholas Bloom, Charles I. Jones, John Van Reenen and Michael Webb, “Are Ideas Getting Harder to Find?” American Economic Review 110 (2020).
- Charles I. Jones, “The End of Economic Growth? Unintended Consequences of a Declining Population,” American Economic Review 112 (2022).
- China’s AI Boom Is Fueling a Surge in One-Person Startups, Channel Insider, September 2026.
- The Reality Equation and Surprise Has an Imaginary Part.
- Do Not Negotiate the Idea and The Solopreneur Superhero.