A Toyota lesson game studios can learn from

The NUMMI story and what it teaches us about building teams that actually work

By 1982, General Motors had run out of patience with its Fremont, California assembly plant and shut it down.

The decision was probably not difficult to make. For twenty years, the plant had been a slow-motion disaster. It had the worst quality record in GM’s entire company, absenteeism was so bad that on Mondays and Fridays there sometimes weren’t enough workers on the floor to start the production line, and GM supervisors would walk across the street to the bar to recruit people to fill in. Workers drank on the job, wildcat strikes erupted regularly, and when formal discipline was attempted, it triggered all sorts of grievances, which in turn triggered more strikes, which kept the plant in a state of near-continuous chaos.

And then there was the sabotage. Workers would slip Coke bottles inside door panels before they were sealed, so future owners would hear the rattling for as long as they owned the car. It was petty, deliberate, and, in retrospect, a completely legible message from a workforce that had concluded nothing else would be heard.

GM looked at the plant and saw a workforce problem. Twenty years of evidence, they figured, was twenty years of evidence.

Yet, they were wrong about what the evidence showed.

How it got that bad

The roots of what happened at Fremont go deeper than a single factory. They go to a specific set of beliefs about what workers are for.

GM in the postwar decades operated on what you might call the Henry Ford model of human beings: people are inputs. You divide the work into the smallest possible tasks, you assign those tasks to the cheapest available hands, and you extract the maximum output per unit of labor cost. Management thinks. Workers do. The line between those two functions is clear, deliberate, and crossing it in either direction is a problem.

The logic is coherent, in a very, very narrow sense. It optimizes for a specific kind of efficiency. What it doesn’t account for is what happens to people who are treated that way on an ongoing basis, over a certain period of time.

At Fremont, the answer was simply to stop caring. And later to actively push back. The absenteeism, the sabotage, the strikes, the Coke bottles in the door panels, none of that came from nowhere. It came from a workforce that had spent years being told, in every practical way, that their judgment didn’t matter, that their problems weren’t worth solving, that they were interchangeable parts in a system that saw them as costs to be minimized.

The United Auto Workers, the workers’ own union, later described the Fremont workforce as “the worst in the automobile industry in the United States.” What that description skips is the part where you ask: worst compared to what? The same workers, two years later, would be producing some of the highest-quality vehicles in North America. The workers were still the same people.

An unlikely deal

Toyota had its own reasons for wanting to be in the United States. Trade tensions between Japan and Washington were escalating through the early 1980s, Congress was openly discussing import restrictions, and Honda and Nissan had already established American manufacturing operations. Toyota needed a foothold, and they needed to learn how to operate inside American labor law, American union culture and the texture of American factory floors.

GM needed something else. They were losing the small car market to Japanese manufacturers and couldn’t figure out why. The oil crises of the 1970s had permanently shifted what American consumers wanted, and Detroit had no credible answer. Somewhere inside Toyota’s production system, GM sensed, was the thing they were missing.

The deal they struck in 1984 was, on paper, symmetrical: a 50/50 joint venture, Toyota managing the plant, GM getting a window into the Toyota Production System. In practice though, the symmetry was a polite fiction. GM handed Toyota the keys to their worst plant, staffed by the same workers who had made it their worst plant, and said: show us what you’ve got.

What Toyota did next is give us a lesson that transcends industry and time.

Three weeks that changed a factory

Toyota’s training manager at NUMMI was Isao Yoshino, and the challenge in front of him was genuine. He had to take a workforce that had spent years in open warfare with its employer and somehow create the conditions for an entirely different kind of work. He had no authority to replace the workforce. He had the same union contract GM had signed. He had the same building, the same equipment, largely the same management structure.

Isao Yoshino.

What he had, that GM didn’t, was a different set of beliefs about what people are capable of when they’re treated with dignity.

His solution was immersion. Before the plant opened, Toyota sent the frontline leaders, group leaders and team leaders, to the Takaoka facility in Japan in groups of around thirty. Three weeks each. They weren’t there to study manuals or sit through presentations about the Toyota Production System. They were there to work, side by side with Toyota employees, on an actual production line, inside an actual functioning plant.

The details of what they experienced have gravitas. At Takaoka, any worker could pull a cord to stop the entire production line the moment they spotted a defect. In the American automotive world, stopping the line was a firing offense. At Toyota, it was an expectation. The logic was simple and radical at the same time: the person closest to the problem is the best person to catch it, and catching it early is worth infinitely more than passing it downstream. So the worker who pulls the cord isn’t causing a problem. They’re doing their job at its highest level.

For the Fremont workers standing on that Japanese factory floor, watching this happen in real time, something changed dramatically and at a fundamental level. One American worker, interviewed years later, described the moment he saw someone stop the line over a single misaligned bolt. His entire framework for what a workplace could be was reorganized around that image.

When the Fremont workers came back from Japan, Mr. Yoshino described what he saw: “They just understand that Toyota management style is good. They’re not saying that GM style is bad, but Toyota style is more comfortable and makes us feel good.”

Makes us feel good. From a production efficiency standpoint, that might sound like a soft outcome. From a results standpoint, it produced one of the most dramatic quality turnarounds in automotive history, accomplished with the same people who had been quietly sabotaging the previous regime.

But Mr. Yoshino didn’t stop at the Japan trip. Back at Fremont, Toyota embedded experienced coordinators alongside the American workers, not as supervisors, but as partners. Problems that arose on the line were treated as information rather than failures. Workers were encouraged to suggest improvements. When they did, those suggestions were taken seriously. The relationship between management and labor, which at the old GM plant had been adversarial by design, was rebuilt from a different premise entirely: that the people doing the work are the people who understand it best, and that their knowledge is an asset, not a threat.

Within a year of opening, the plant’s quality metrics matched Toyota’s Takaoka facility. The same workforce that had been described by its own union as the worst in American automotive manufacturing. Mr. Yoshino’s explanation was characteristically plain: “We didn’t do anything else. We just offered the opportunity for them to learn, and they voluntarily decided to change themselves.”

That is not a small claim. It says that the workers at Fremont were never the problem. The system that treated them as the problem was the problem. Change the system, and the people change with it, not because they were fixed, but because they finally had somewhere to put the capability they had always had.

What game studios miss

GM had front row seats and saw all of it. Toyota kept almost nothing secret. They showed GM managers the production system, explained the philosophy, embedded Japanese coordinators on the floor for years. GM sent observers to NUMMI continuously. Some of them left describing it as transformational. A number went back to Detroit and tried to explain what they had seen.

GM began implementing lean principles more seriously, by most accounts, around the late 1990s, fifteen years after NUMMI opened. By then the market had moved, the debt had piled up, and the story ended in 2009 with the largest manufacturing bankruptcy in American history.

Access to the information was never the problem. GM looked at NUMMI and saw a set of techniques. But Toyota knew it was much more than that. It was a set of beliefs.

Copying a process is easy. You can write a rule that says workers must stop the line when they see a defect. What you cannot write is a rule that makes managers genuinely glad when they do, or that makes the person on the floor trust that raising a problem will be received as useful rather than punished as disruption. That trust is the foundation the whole thing runs on, not a feature of a production system.

The games industry, and especially its larger studios, have spent the last 10 to 15 years demonstrating a fairly comprehensive understanding of the GM position. Layoffs glacially announced through euphemistic corporate speak. Crunch normalized to the point that it stopped being discussed as a problem and became, somehow, “evidence of passion”. People hired for a specific project, discarded when the project ended or the direction changed, with all the ceremony of a cancelled vendor contract. The accumulated craft they carry, the instincts from shipping games, the understanding of what actually works, all walking out the door with them. Every single time.

This doesn’t usually come from malice. It comes in no small reason from an inability to see the people in front of you as the actual mechanism of production rather than a cost sitting between the idea and the output. The spreadsheet and the reduced operational expenses indicate that everything is fine, but the reality of game development at the actual studios that get decimated by the GM way is a peg different.

If you’re building or running a studio

You’re perhaps not running a hundred-person studio. Maybe you’re running something smaller, more distributed, built out of people who took a chance on an idea that isn’t a sure thing. Which means the dynamics are more immediate and, in some ways, more forgiving. You don’t have fifteen years of institutional habits to undo.

But the same logic applies at five people that applied at Fremont. Maybe more so, because at five people there’s nowhere to hide from culture. If someone on your team doesn’t feel like their judgment matters, you’ll feel that in every conversation you have together. If someone thinks raising a problem is more likely to damage the relationship than solve anything, they’ll stop raising problems, and you’ll only realize how important that was when you’re picking up the pieces.

You can’t afford not to care about this.

What Mr. Yoshino understood, and what Toyota built into everything about how NUMMI was run, is that the conditions for great work are not mysterious. People want to contribute to something meaningful. They want to grow. They want to be trusted with real responsibility and know that when they exercise judgment, that judgment is respected. They want to be treated with basic human dignity. These are not exotic or outlandish requirements. They are the minimum conditions for any human being to do their best work, in any field, in country, in any century.

Your team is not sitting there waiting for you to magically inspire them with rousing speeches and genre-defining game ideas. What they’re really waiting for is to be trusted.

The output, the quality of the game you’re making, your ability to attract people worth working with, the studio culture that either exists or doesn’t by the time you’re two years in, all of it flows from whether the people around you feel like this is a place that believes in them. A great milestone structure or a solid burn rate won’t save you from it. You can have a beautifully documented production plan and still ship a game with a team that burned down behind it.

Process serves people. Not the other way around.

I want to kindly encourage you to start from the same premise Mr. Yoshino did: that the people around you are capable of great things, and that your job is to create the conditions where those things can happen.

The rest, in my experience, tends to take care of itself. Not always, not immediately, but more reliably than you’d think.

---

If this was useful, you might also like:

The team is everything

Culture and values in indie game studios