The broker used to be a man. He knew your name and your parish; he sat across a desk and told you, within the limits of his judgment and his bias, what to buy and what to sell. The Wall Street Journal reports this morning that retail investors are handing their stock portfolios to AI agents — built from a written prompt, trading at “superhuman scale,” without the human ever watching the market. “Skeptics worry,” the paper adds, that the new tool “could expose less-experienced investors to new dangers.” I traded these futures for a living once. Let me show you what is missing from the picture they have drawn.
The steelman is the easy half, and I will grant the strongest version of it before I take it apart. A kid in a small town can now sit down at a laptop and run a strategy the 1985 floor trader at the Chicago exchange could not have run without a team and a phone tree. He types a sentence in plain English. An algorithm builds the model, tests it against a decade of data, and begins trading before his coffee gets cold. That is a real change. An algorithm is also impartial in a way a salesman with a quota never was, and the gatekeeping that kept ordinary people out of the serious tools was real gatekeeping. I will not deny any of it.
But the steelman is the wrong question. The question is what has been taken.
For seven years I sat at a desk in Chicago and traded agricultural futures — paper claims on corn, cattle, and beans grown by the neighbors I had left behind in Adams County. The men in that building knew precisely two things about the men in the field: what was planted, and what the price was. They thought about the second one all day. The order went in. The algorithm ran. The firm took its cut on the way in and another cut on the way out. The small saver’s check was the raw material, and we on the desk did the abstracting. I walked away from it. I have not walked back.
The men who sold it to me were polite and wore good ties and explained in calm voices why the small saver could never do what they did. They were not lying. They were right. The small saver cannot do what they did. What the small saver can do is be the flow — the predictable stream of retirement dollars that the platform routes through its algorithm and bills at every node. The platform takes a percentage of every dollar on the way in. It takes a percentage of every trade on the way out. It takes the spread. It takes the order flow. It sells the data. The saver’s decision is dissolved into the platform’s decision, and the saver is told the dissolution was empowerment. The small saver’s inability to read the code is not a bug in the business model. It is the business model.
Now the abstraction has come for the man who owned the futures — the retail investor, the small saver, the person the Journal calls “less experienced.” He used to walk into a broker’s office, or call one, or sit across from an advisor who lived in his county and went to his church. The advisor was biased, often expensive, and frequently wrong. He was also a man. He had a face. He had a name. The small investor could look him in the eye, argue with him, fire him. He could not be abstracted from his own money by him.
That is the institution the AI just dissolved. Call it what you will — wealth management, financial planning, the retail brokerage — it was a relationship between two men with a stake and a name. It has been replaced by an algorithm built from a prompt, executed at “superhuman scale,” that the saver has no way to oversee, no way to argue with, and no way to hold to account. The “new dangers” the skeptics worry about are not the danger. The danger is older and quieter. It is that the saver has been abstracted out of his own savings the same way the farmer was abstracted out of his own crop. He does not know what he owns. He does not know what is being done in his name. The institution that connected him to his money — the human link — has been replaced by a line of code.
I want to say what the Journal does not, plainly. The people who built and marketed this technology will tell you it is liberation: the small investor, freed from the broker, freed from the gatekeeping, freed from human error. They will tell you it is the future, and they will say so with conviction. They will not tell you that the man they replaced used to sit across from you and could be held to a standard you could see. The algorithm cannot. It has no parish. It has no name. It cannot be sued into behaving better. It cannot be embarrassed into honesty. It does not even know it has done wrong. They have put a casino in every man’s pocket and called it liberty — this time, a casino run by a line of code whose results the player cannot read.
The men who sold you the machine did not mention the machine has a footprint. The technology runs on data centers the size of small towns — the same concentrated AI compute that has communities from Bannon to Sanders and Ocasio-Cortez pushing back at the data centers themselves, drawing 70% local opposition in the states where they were being sited, the same compute the governors of Texas and Pennsylvania have started curbing. The abstraction is sold as freedom. The ground it stands on is contested before the first share is traded. The small saver is buying empowerment from the same system the rest of the country is refusing to license.
This is the same disease in a new coat. The man I sat across from on the Chicago desk had been abstracted out of his own work by the same kind of machine. The farmer in Adams County was abstracted out of his own land by the same kind of consolidation. The local paper was abstracted out of its own town by the same kind of owner. The family was abstracted out of its own economy by the same kind of economy. Each time the abstraction was sold as progress. Each time the man who lost his place was told it was freedom. Each time the institution generations built — a relationship, a stewardship, a human link — was dissolved in the name of efficiency.
The cure is not another machine. The cure is what we built the last time the machine ate the town. We have been here before, and we have known how to answer — and we have let the answer be sold for parts.
I came home and I manage the farm co-op. The co-op is a small, unfashionable thing. Its members vote. They sit across the table from one another. They argue about prices and pooling in language the algorithm cannot parse. They own what they produce. They are accountable to one another. It does not scale to superhuman magnitude. It was never meant to. It is meant to do one thing well: keep the human in the transaction.
What we already know how to build is the answer. Member-owned credit unions. Mutual savings banks. The cooperative constitutional DNA — one member, one vote, accountable to the people it serves, distributing the gains to the members instead of the platform. The lift is real: credit unions hold roughly $1.5 trillion against the $13 trillion sitting in American IRAs, and scaling the co-op vehicle to that volume is political work measured in decades. But the structure is built and proved. The Adams-Columbia Electric Cooperative down the road from me — the largest rural electric co-op in Wisconsin, headquartered a few blocks from where I write this — electrified the sand counties in the 1930s on member equity, against the judgment of every investor-owned utility that said it could not be done at a profit. The same member-owned structure can hold a retirement portfolio. The saver would be a member, not flow. The gains would distribute to the people who earned them. The decision would be made by the saver, with advice, not by a black box with a friendly logo.
The men who sold me the desk will sell the small saver the algorithm. They have been doing it since long before I left Chicago and they will be doing it long after. The only answer is the answer that does not depend on their restraint. Disperse the power. Leave the saver his judgment. Leave the saver his savings. The earth was given for all.
The man across the desk is gone. The man across the table is not.