The sight of Bill Boeing was a familiar one on the factory floor. His office was in the building next to the converted boatyard where workers lathed the wood, sewed the fabric wings, and fixed the control wires of the Boeing Model C airplane. there is no authority except facts. facts are obtained by accurate observation read a plaque affixed outside the door. And what could need closer observation than the process of his aircraft being built? One day in 1916, Boeing spotted an imperfectly cut wing rib, dropped it to the floor, and slowly stomped it to bits. “I, for one, will close up shop rather than send out work of this kind,” he declared.

When David Calhoun, the soon-to-be-lame-duck CEO of the company Boeing founded, made a rare appearance on the shop floor in Seattle one day this past January, circumstances were decidedly different. Firmly a member of the CEO class, schooled at the knee of General Electric’s Jack Welch, Calhoun had not strolled over from next door but flown some 2,300 miles from Boeing’s headquarters in Arlington, Virginia. And he was not there to observe slipshod work before it found its way into the air—it already had. A few weeks earlier, the door of a Boeing 737 had fallen out mid-flight. In the days following his visit, Calhoun’s office admitted that it still didn’t know quite what had gone wrong, because it didn’t know how the plane had been put together in the first place. The door’s restraining bolts had either been screwed in wrong, or not at all. Boeing couldn’t say, because, as it told astonished regulators, the company had “no records of the work being performed.”

The two scenes tell us the peculiar story of a plane maker that, over 25 years, slowly but very deliberately extracted itself from the business of making planes. For nearly 40 years the company built the 737 fuselage itself in the same plant that turned out its B-29 and B-52 bombers. In 2005 it sold this facility to a private-investment firm, keeping the axle grease at arm’s length and notionally shifting risk, capital costs, and labor woes off its books onto its “supplier.” Offloading, Boeing called it. Meanwhile the tail, landing gear, flight controls, and other essentials were outsourced to factories around the world owned by others, and shipped to Boeing for final assembly, turning the company that created the Jet Age into something akin to a glorified gluer-together of precast model-airplane kits. Boeing’s latest screwups vividly dramatize a point often missed in laments of America’s manufacturing decline: that when global economic forces carried off some U.S. manufacturers for good, even the ones that stuck around lost interest in actually making stuff.

The past 30 years may well be remembered as a dark age of U.S. manufacturing. Boeing’s decline illustrates everything that went wrong to bring us here. Fortunately, it also offers a lesson in how to get back out.

In Bill Boeing’s day, the word manufactory had cachet. You could bank at the Manufacturers Trust. Philadelphia socialites golfed at The Manufacturers’ Club. Plans for the newly consecrated Harvard Business School called for a working factory on campus. The business heroes of the day—Ford, Edison, Firestone—had risen from the shop floor.

There, they had pioneered an entirely new way of making things. The American System of production—featuring interchangeable parts, specialized machine tools, moving assembly lines—was a huge leap beyond European methods of craft production. And it produced lopsided margins of victory for the likes of Ford, GM, and Boeing. To coordinate these complex new systems, two new occupations arose: the industrial engineer, who spoke the language of the shop floor, and the professional financial manager, who spoke the language of accounting.

[Charlie Warzel: Flying is weird right now]

At first the engineers held sway. In a 1930 article for Aviation News, a Boeing engineer explained how the company’s inspectors “continually supervise the fabrication of the many thousands of parts entering into the assemblage of a single plane.” Philip Johnson, an engineer, succeeded Bill Boeing as CEO; he then passed the company to yet another engineer, Clairmont Egtvedt, who not only managed production of the B-17 bomber from the executive suite, but personally helped design it.

After the Second World War, America enjoyed three decades of dominance by sticking with methods it had used to win it. At the same time, a successor was developing, largely unnoticed, amid the scarcities of defeated Japan. The upstart auto executive Eiji Toyoda had visited Ford’s works and found that, however much he admired the systems, they couldn’t be replicated in Japan. He couldn’t afford, for instance, the hundreds of machine tools specialized to punch out exactly one part at the touch of a button. Although his employees would have to make do with a few general-purpose stamping presses, he gave these skilled workers immense freedom to find the most efficient way to run them. The end result turned out to be radical: Costs fell and errors dropped in a renewable cycle of improvement, or kaizen.

What emerged was a different conception of the corporation. If the managerial bureaucrats in the other departments were to earn their keep, they needed a thorough understanding of the shop floor, or gemba (roughly “place of making value”). The so-called Gemba Walk required their routine presence at each step until they could comprehend the assembly of the whole. Otherwise they risked becoming muda—waste.

When the wave of Japanese competition finally crashed on corporate America, those best equipped to understand it—the engineers—were no longer in charge. American boardrooms had been handed over to the finance people. And they were hypnotized by the new doctrine of shareholder value, which provided a rationale for their ascendance but little incentive for pursuing long-term improvements or sustainable approaches to cost control. Their pay packages rewarded short-term spikes in stock price. There were lots of ways to produce those.

Which brings us to the hinge point of 1990, when a trio of MIT researchers published The Machine That Changed the World, which both named the Japanese system—“lean production”—and urged corporate America to learn from it. Just then, the Japanese economy crashed, easing the pressure on U.S. firms. In the years that followed, American manufacturers instead doubled down on outsourcing, offshoring, and financial engineering. This round of wounds was self-inflicted. Already infused with a stench of decay, manufacturing was written off as yesterday’s activity.

At GE, which produced three of Boeing’s last four CEOs, manufacturing came to be seen as “grunt work,” as the former GE executive David Cote recently told Fortune’s Shawn Tully. Motorola—founded as Galvin Manufacturing and famed for its religious focus on quality—lost its lead in mobile-phone making after it leaned into software and services. Intel’s bunny-suited fab workers were the face of high-tech manufacturing prowess until the company ceded hardware leadership to Asian rivals. “Having once pioneered the development of this extraordinary technology,” the current Intel CEO, Pat Gelsinger, wrote recently, “we now find ourselves at the mercy of the most fragile global supply chain in the world.”

Phil Condit, the talented engineer who had overseen design of the hugely successful 777, was atop Boeing when I visited the company in late 2000. He was no stranger to the shop floor. Traversing Boeing’s Everett plant in a golf cart, he pointed out the horizontal tail fin stretching above us. Hard to believe it was larger than the 737’s wing, he marveled. Waiting back in his office—still located on the bank of the Duwamish River but greatly swollen by the recent merger with McDonnell Douglas—was a different sort of glee. “Wow! Double wow!” his mother had emailed him, referring to Boeing’s closing stock price that day. And, it would soon emerge, he wanted to get some distance from what he described to the Puget Sound Business Journal as “how-do-you-design-an-airplane stuff.” The next year, he moved Boeing’s headquarters to Chicago, pulling the top brass away from the shop floor just as the company was embarking on a radically new approach to airplane assembly.

Its newest plane, the 787 Dreamliner, would not be an in-house production. Instead Boeing would farm out the designing and building to a network of “partner” companies—each effectively its own mini-Boeing with its own supply chain to manage. “It used to be you’d have some Boeing people develop the blueprints, then march over and say, ‘Hey, would you build this for me?’” Richard Safran, an analyst at Seaport Research Partners and a former aerospace engineer, told me. “Now, instead, you’re asking them to design it, to integrate it, to do the R&D.”

The allures of this “capital light” approach were many: troublesome unions, costly machine shops, and development budgets would all become someone else’s problem. Key financial metrics would instantly improve as costs shifted to other firms’ balance sheets. With its emphasis on less, the approach bore a superficial resemblance to lean production. But where lean production pushed know-how back onto the shop floor, this pushed the shop floor and its know-how out the door altogether.

Beyond that were the problems that a Boeing engineer, L. J. Hart-Smith, had foreseen in a prescient white paper that he presented at a 2001 Boeing technical symposium. With outsourcing came the possibility that parts wouldn’t fit together correctly on arrival. “In order to minimize these potential problems,” Hart-Smith warned, “it is necessary for the prime contractor to provide on-site quality, supplier-management, and sometimes technical support. If this is not done, the performance of the prime manufacturer can never exceed the capabilities of the least proficient of the suppliers.”

Boeing didn’t listen. Wall Street dismissed Hart-Smith’s paper as a “rant,” and Boeing put each supplier in charge of its own quality control. When those controls failed, Boeing had to bear the cost of fixing flawed components. Most troubling was the dangerous feedback loop Hart-Smith foresaw. Accounting-wise, those fixes, which in reality are the costs of outsourcing, would instead appear as overhead—creating the impression that in-house work was expensive and furthering the rationale for offloading even more of the manufacturing process.

In the short term, this all worked wonders on Boeing’s balance sheet: Its stock rose more than 600 percent from 2010 to 2019. Then the true folly of this approach made its inevitable appearance when two strikingly similar crashes caused by faulty software on Boeing planes killed a total of 346 people.

[James Surowiecki: What’s gone wrong at Boeing]

Today, if you stand along the Seattle waterfront long enough, sooner or later you’ll catch sight of a train headed south carrying the distinctive shape of a Boeing 737. Though it’s colored a metallic green and missing its tail—clearly not the finished product—it’s the kind of thing you point to and say, Look kids, a Boeing plane’s on that train! Not so. The logomark on the side spells it out: Spirit AeroSystems of Wichita, Kansas, has built this fuselage, which isn’t coming from Boeing. It’s going to Boeing.

A plane is a complex system in which the malfunction of one piece can produce catastrophic failure of the whole. Assembly must be tightly choreographed. But now—especially with Boeing continually trying to wring costs from its suppliers—there were many more chances for errors to creep in. And when FAA investigators finally toured the premises of Spirit AeroSystems—maker of the blown-out door as well as the fuselage it was supposed to fit in—they did not find a tight operation. They found one door seal being lubricated with Dawn liquid dish soap and cleaned with a wet cheesecloth, and another checked with a hotel-room key card.

A dark age doesn’t descend all at once. The process of emerging from one also takes time. It must begin with a recognition that something has been lost. Boeing’s fall just might have provided that rush of clarity. You could be from the 12th century and still know that soap and cheesecloth aren’t for making flying machines. Boeing’s chief financial officer recently admitted that the company got “a little too far ahead of itself on the topic of outsourcing.” It is in talks to reacquire Spirit AeroSystems and is already making the composite wings of its next-gen plane, the 777X, in-house at a new, billion-dollar complex outside Seattle. “Aerospace Executives Finally Rediscover the Shop Floor,” Aviation Week declared on the cover of a recent issue.

As for the rest of corporate America, one of the strongest signals may be coming from the company Boeing has striven so hard to emulate: GE. Under operations-minded boss Larry Culp, the company is finally—only 40 or so years late—pushing itself through a crash course in lean manufacturing. It is belatedly yielding to the reality that workers on the gemba are far better at figuring out more efficient ways of making things than remote bureaucrats with spreadsheet abstractions.

In the crucial field of semiconductors, meanwhile, Intel has recognized that Moore’s Law (the doubling of computing power roughly every 18 months) flows not from above but from manufacturing advances it once dominated. It has undertaken a “death march,” in the words of CEO Pat Gelsinger, to regain its lost edge on the foundry floor. The CHIPS Act has put a powerful political wind at his back. Green and other incentives are powering a broader, truly seismic surge in spending on new U.S. factories, now going up at three times their normal rate. No other country is experiencing such a buildout.

Add all the capacity you want. It won’t reverse the country’s long decline as a manufacturing superpower if corporate America keeps gurgling its sad, tired story about the impossibility of making things on these shores anymore. It’s a story that helped pour a whole lot of wealth into the executive pockets peddling it. But a half century of self-inflicted damage is enough. The doors have fallen off, and it’s plain for all to see: The story was barely bolted together.