Right now, middle managers are being told the same thing in a dozen different dialects of corporate bullshit (pardon my French). Coinbase calls it “no pure managers.” Block says there is “no need for a permanent middle management layer.” Meta, Amazon, Google, Citi, and others are flattening layers, widening spans of control, pushing managers back into individual-contributor work, and dressing the whole thing up in the language of AI, speed, efficiency, and “player-coaches.” Same message: fewer people in the middle, more reports per manager, more hands-on work, and a vague promise that AI will somehow make the math work.
But this is not some bold new management breakthrough. For roughly fifty years, American companies have gone through recurring spasms of middle-manager demonization. Every decade or so, the same idea comes back wearing a new costume: cut the layers, speed things up, get rid of bureaucracy. Then companies overload the survivors, watch the coordination break, and eventually rediscover that strategy does not execute itself, people do not coach themselves, and work does not magically align because someone deleted a box from the org chart.
Before we pretend AI has revealed some brilliant new way to run companies, let’s look at the last fifty years of brilliant new ways that mostly turned into the same old mess.
LATE 1970S TO EARLY 1980S: JAPANESE COMPETITION AND THE FIRST PERMISSION SLIP
The modern attack on middle management did not begin with AI, agile, or reengineering. It started when American companies began comparing themselves unfavorably with Japanese manufacturers. Toyota and other Japanese competitors looked faster, leaner, and more disciplined, while U.S. corporations looked bloated, slow, and layered. William Ouchi’s Theory Z argued that American firms could learn from Japanese-style management, especially around consensus, long-term employment, and more holistic worker involvement. The implicit critique was that American command-and-control bureaucracy had grown too thick.
This wave was not yet a full-scale war on middle managers. The 1981–82 recession hit manufacturing and blue-collar workers hardest. But it created the cultural permission to ask whether all those layers of American management were really necessary.
LATE 1980S TO EARLY 1990S: DELAYERING, REENGINEERING, AND THE WHITE-COLLAR RECESSION
This was the first major purge. Michael Hammer’s 1990 Harvard Business Review article, “Reengineering Work: Don’t Automate, Obliterate,” and the 1993 book Reengineering the Corporation argued that companies should redesign work from scratch rather than automate old processes. Middle managers were cast as relics of industrial-era handoffs, and information technology was supposed to let work flow without all those people in the middle.
Jack Welch at GE became the most visible executioner of the model. Between 1981 and 1985, GE’s workforce fell from 411,000 to 299,000, and Welch compressed the hierarchy by eliminating management levels and widening spans of control. IBM also cut heavily during this era, shedding more than 100,000 employees between 1986 and 1994.
The macro data backed up the idea that this was a different kind of downturn. The 1990–91 recession was described as a “white-collar” or “middle management” recession because supervisory and nonproduction workers were hit unusually hard. Permanent staff reductions rose from 111,000 in 1989 to 316,000 in 1990 and 556,000 in 1991.
The reversal was predictable. By the mid-1990s, companies that had reengineered aggressively began struggling with coordination failures, institutional knowledge gaps, and weak execution. The cuts saved money, but the work of coordinating, translating, coaching, and integrating did not disappear.