The Bill for Not Looking

There were days on my glass leg at Lansing Delta Township when the glass would not stay put. The robot laid the urethane bead, swung the glass toward the body, and somewhere in that short trip the bond let go and the glass slid off the edge. The line stopped, and people started cutting glass back out of cars that already had it in them. Was it the urethane? The glass? The robot? The operator?

One of those four suspects belonged to Dale, the pipefitter who fed urethane to every robot cell on that leg. On the bad days he appeared without being sent for and worked the problem as though his own feed were the likeliest culprit. Then it got solved, and nobody could say which part of the answer was his.

I have written about Dale before. The plant’s records logged the hours the line was down, never the thousands it wasn’t, or the man who made them. Dale never saw a record of his own contribution. There wasn’t one to see.

In late September 2026, New York City showed what happens when that blindness scales to a quarter of a million people.

What the settlement requires

The city announced a $131.5 million settlement with DoorDash over wages owed to its delivery workers. About 260,000 drivers are to be contacted about repayment. DoorDash said the errors were not intentional and admitted it had gotten it wrong.

I believe them about the intent. That is exactly the point.

The settlement requires DoorDash to hand the city its worker-pay data for three years. And the city, with outside nonprofits, is building software that lets delivery workers record their own data and use it to dispute their pay. Samuel Levine, the commissioner who led the investigation, called the problem an “asymmetry” between the company that collects everything and the workers who suspect they are being shorted.

That is Dale’s problem, stated by a regulator.

The company held the whole record. The worker held none of it. When a worker cannot see the ledger, errors need not be intentional to be permanent. They run quietly until someone from outside forces the books open.

So here is the argument. Every operation keeps a ledger of what its people contribute, whether it means to or not. If the company refuses to build that ledger honestly, someone else will build it — a regulator, a court, a city hall — and they will build it to their specifications, not yours. Legibility is not optional. The only choice is who designs it.

That’s the argument.

The phrase that left the boardroom

The mayor framed the settlement with a phrase he borrowed from DoorDash’s own chief executive, who had talked about entrepreneurs acting like a “greedy algorithm.” In computer science the term describes a procedure that takes the best available choice at each step without regard for what comes after. The mayor turned it into a charge.

I am not writing to cheer a mayor or to boo one. I have spent too much of this year explaining that my work is not a socialist project to hand anyone that argument for free. The Wall Street Journal notes that politicians on the left and the right are watching how this plays. That is the part that should frighten a board.

The diagnosis has crossed the aisle. When a CEO’s own words become a campaign line, the selection problem I write about has left the boardroom and entered the ballot box. Boards that select for the shortest time horizon are, in effect, installing greedy algorithms with pulses. The public has noticed. And the public’s instrument is not a better CEO. It is enforcement.

Why the enforcer’s ledger costs more

Nothing in that settlement was unaffordable to build from the inside. Showing a driver how their pay was calculated is not a moonshot. Giving a worker a view of their own contribution is a reporting feature. The company had the data. It chose not to make it visible to the people it described.

Now the ledger exists anyway. It has a nine-figure price attached, three years of outside monitoring, and a city-built app whose purpose is to generate disputes. That is the enforcer’s version of worker visibility: adversarial, retrospective, and designed around the question of what you owe, not what you could gain.

A builder would have designed the same visibility for the opposite reason.

The ledger I kept

I should admit that I kept a ledger too.

At Lansing Grand River, and later at Lansing Delta Township, every team of five to seven had an operator certification matrix, one of the tools we took from the Toyota Production System. Down one side, the team members. Across the top, the jobs. In each cell, a circle cut into four quarters, the kind consultants call a Harvey ball.[1] The first quarter was filled in when the team member knew the job: had read and understood the standard work. The second, when he could do it to quality. The third, when he could do it at quality and at speed. The fourth, when he could train someone else.

It was a real record of capability. Rudimentary, but real. A glance at the board told you who knew what.

Here is what we used it for. We tracked training. And when we had too many line stops, or a big defect, we went back to the matrix to see whether the team leader had done his job, and we used it to discipline the team member. Sometimes the team leader.

In other words, we read our own ledger the way the city now reads DoorDash’s. After the fact. To assign blame.

It never once became a reason to recognize anyone. A team member who taught himself the jobs in the team next door was worth a great deal to quality. He could see a problem coming across the boundary before it became a defect. Nothing encouraged him. There was no column for it, no pay for it, no line anywhere that said thank you. The matrix could see what he had done. We had simply decided that seeing was for punishing.

Dale would never have shown up on it at