The andon cord went a lot in his zone. That was the first thing you noticed, and if you did not know what you were looking at, it looked like a problem. Pulls were high. Stops were almost nonexistent. Anyone who has run an assembly floor just sat up, because that combination is the whole thing. High pulls mean the team is catching problems — they see it, they say it, they are not hiding anything to protect a number. Few stops mean they are solving it inside the takt time, before the line ever goes down. Pulls without stops is not a quiet team. It is a team that is awake and competent at the same time, which is the rarest condition in manufacturing and cannot be ordered, bought, or installed. His standard work sheets were perfect. Training and certification records complete, every one. Almost no defects out of his zone. His name is Ramon Hernandez and he was the best team leader I ever had. I ran the Trim Shop at Lansing Grand River for General Motors. Ramon took care of his team. His team took care of the product. That sentence is not sentimental; it is a causal chain, and it ran in that order, and it never ran in the other order. He had a training instinct I have never seen matched. Put a new team member in front of him and inside the first hour Ramon had read the hands, the posture, the hesitation pattern — and calibrated his coaching to that specific person. Not to a curriculum. To that person. And he carried an encyclopedic mental model of every bottleneck the trim line had ever produced, which meant that during a launch, when GM was bleeding time against a hard date, Ramon found the constraint before the engineers finished their first meeting. He solved problems that did not have names yet, because the vehicle had not existed long enough to generate a problem history. None of that came from a syllabus. We paid for it Here is the part that matters, and it is the part everyone gets wrong about Grand River. We made him. Deliberately. On purpose. With money. A billion-dollar plant. Body shop, paint shop, a novel general assembly. Tom LaSorda’s plant, the first one he built in a unionized shop. I was one of the first shift leaders hired, and production was a full year out, and a few of us looked at each other and said: there is no plant here. What are we going to do for a year? Invest in people. That was the answer. That was the design. So we built a little assembly line out of wood. Little wooden cars, with andon pull cords, standardized work, team leaders, quality stations — the whole architecture in miniature. Ken Knight’s idea, one of LaSorda’s apostles. We put out a brochure to Local 652 saying come work at the new plant, and people came in droves, and we could afford to be selective because we had the volume. You cannot interview a floor worker off a résumé. So we watched them build wooden cars. That plant went on to JD Power Gold. Now find that year in General Motors’ accounts. You cannot. The building capitalized over thirty-nine years. The tooling over seven. The presses, the conveyors, the paint booths — every one of them on the balance sheet, depreciating politely, visible to any analyst who cared to look. The year we spent making Ramon Hernandez was expensed in the period incurred. Gone by the next 10-K. A billion dollars of that plant is still on the books. The single best thing in it never was. Loss to the floor Then Pilot came for him. We needed a Pilot team member to work the next car, and Ramon was the obvious choice, and I let him go. He went on to become one of the best pilot team leaders GM had — he brought in two vehicles after the first CTS, with results. Loss to the floor. Ascension for him. I felt it every shift. My numbers got worse. Measurably, immediately, for months. And GM’s asset got better by exactly the same event. There is no system anywhere on earth that connected those two facts. My P&L recorded the loss. Nothing recorded the deposit. If you had audited my shop that quarter you would have found a supervisor whose performance declined and no explanation, because the explanation was that I had transferred a capital asset out of my cost center and there was no line to transfer it on. That’s the argument. Not that companies are cheap. That the instrument only has one column, and it says loss. Producing from inventory An oil company that pumps a reservoir and books the cash as earnings is not more profitable than its neighbors. It is producing from inventory. Every accounting regime knows this. Under ASC 932 an extractive firm cannot simply report the revenue — it must charge depletion against a recorded reserve and publish a standardized measure of what remains in the ground. That regime exists for one reason: a company that is earning and a company that is selling what it has left look identical on an income statement. Right up until they don’t. Formation is the only reserve in the modern economy that can be pumped to zero with no depletion charge, no reserve disclosure, and no line item anywhere that would reveal it happening. Sixty people report directly to Jensen Huang. This gets offered as proof that AI has changed the architecture of leadership. Ask one question about those sixty: where were they formed? Not one of them at Nvidia. They arrived formed — decades of deposits made somewhere else, at somebody else’s expense, by people whose names are not in the article. Nvidia did not build a structure that produces sixty executives. It built one that consumes sixty. Grand River made Ramon. Somebody else got him. That is not a scandal — that is what should happen, that is what ascension is. The scandal is that no ledger anywhere records that we made him, so the next plant manager who considers spending a year on wooden cars has no evidence to point at, and will not, and cannot be blamed for it. Nobody in this story is a fool Gary Becker settled this in 1964, and his answer is worse than greed. Specific human capital is useful only to this firm, so this firm will fund it. General human capital — the training instinct, the bottleneck model, the discernment — travels. Ramon carried his to Pilot and would have carried it out the door to Toyota if he had wanted to. Which is exactly why no rational firm funds it. You cannot capture what walks out on Friday. Becker’s resolution was that the worker funds it, by accepting a low wage during the forming years. That is what an apprentice wage was. That is what the guild was for. We abolished the apprentice wage. Correctly — in most of its historical forms it was a mechanism of exploitation. And we replaced it with nothing. General formation now has no funder at all. Not the firm, which cannot capture it. Not the worker, who can no longer pay. Not the state, which funds schooling, and everyone who has stood on a floor knows schooling is a different animal. LaSorda funded it anyway. He was right, and it worked, and he could not prove it, and the proof is that nobody has done it since at that scale in this country. That is why the Zunft existed. The guild is not a nostalgia object. It is the institution that collectivizes a cost no single actor can rationally bear. Take it away and you do not get a freer market in formation. You get no market in formation, because the good has no buyer. Then what We already ran this experiment to completion. Tool and die. Machinists. Industrial maintenance. The apprenticeship pipelines came apart in the late 1980s for exactly the reasons above, and for twenty years it worked beautifully — because the reserve was full. Every firm that stopped forming looked smarter than the ones that didn’t. The numbers agreed. The numbers were measuring a withdrawal. The reserve ran dry around 2010. I ran a plant in Ohio and could not hire a maintenance technician at any price. Not a competitive price. Any price. The men who can do that work are old, there is nobody behind them, and the reason is a decision made around 1988 that appeared on no balance sheet, in no earnings call, in no case study anywhere. It is among the most consequential things that happened to American manufacturing in my working life and it is not written down. That is not a forecast. That was my Tuesday, every Tuesday. What a ledger would do Suppose Ramon had an account. Not a résumé — a résumé is a claim. An account. A lifetime formation ledger, attested, with deposits recorded when they were made: the year on wooden cars, the year he learned to read a new man’s hands in the first hour, the launch where he found the constraint before the engineers. Verified by someone with no stake in the answer, firewalled from whoever did the training. Depreciating on a half-life, because judgment decays and pretending otherwise gives you a credential instead of a capability. Then when Pilot takes him, it is a transfer, not a loss. My cost center is credited. The deposit has a name and a date and a maker. And the next man who wants to spend a year on wooden cars has something to point at. None of this is exotic. We do it for oil. We do it for buildings. Under ASC 842 we finally learned to put a leased asset on the balance sheet — something we do not own, whose value we consume over time — because pretending it wasn’t there was distorting every statement in the market. The template has been sitting there since 2016. We just never pointed it at a person. Ramon Hernandez is retired. Everything he knew about reading a man’s hands in the first hour went home with him, and not one line of it was ever written down, anywhere, by anyone. Grand River paid for all of it and could not tell you what it cost or what it bought. Book the depletion charge. Everything else follows from that one line. Ask your CFO what the company spent last quarter on titanium, on litigation, on cloud storage. You will have a number in four minutes. Then ask what it spent forming a supervisor. What did they tell you? सर्वे ज
