20th century

The 20th century, as depicted in the book, was shaped by a recurring organizational question: how can a large group preserve the knowledge and judgment dispersed among its members while bringing them together to solve problems? Its examples range from corporate hierarchies and U.S. intelligence to a worldwide scientific collaboration and a celebrated hedge fund. Across these cases, the central distinction is not simply between centralization and decentralization: specialization can create valuable local knowledge, but without ways to share and aggregate judgments, that knowledge may never become collective understanding.

Corporate Structure and Collective Decision-Making

For much of the twentieth century, successful corporations were understood as vertically integrated, hierarchical, and centralized. A company was expected to control much of its supply chain; layers of managers dealt with problems at their level and passed more consequential questions upward; and ultimate authority often rested with a small group or a single CEO. General Motors exemplified the apparent qualification: its Buick, Chevrolet, and Cadillac divisions operated day to day much like independent businesses, yet headquarters retained the authority to make the major decisions shaping corporate strategy and organization. At the other extreme, Henry Ford’s company owned even the iron ore and sand used in its cars.

That managerial structure coexisted with a long-running rhetoric of worker empowerment and teamwork. In the second decade of the century, major companies experimented with profit-sharing and worker voting rights. In the 1930s, the human relations movement associated with Elton Mayo argued that workers became happier and more productive when management listened to them, though the book notes that the studies, in retrospect, seem to show the effect of higher pay. By the 1950s, committee meetings and teamwork had become prominent corporate ideals. William H. Whyte’s critique in The Organization Man captured the tension: companies praised groups and middle managers, even as authority remained concentrated above them.

The book distinguishes genuine collective judgment from consensus-seeking. Consensus is not necessary to draw on a crowd’s knowledge, and the effort to reach it can suppress disagreement in favor of cautious, lowest-common-denominator outcomes. Familiar groups may exchange only familiar ideas rather than invite provocative debate. The committee-made camel, in this account, is not the inevitable result of collective decision-making; it is the product of a process that mistakes agreement for good judgment.

Hierarchy, Information, and Authority

The costs of rigid hierarchy were visible both in corporate structures and in ordinary work. In the 1970s, more than fifteen managerial layers separated Ford’s chairman from a factory-floor supervisor, while Toyota had five. At a General Motors plant in Van Nuys, a worker had spent two years welding a bracket without being told that it held the car’s sunshade. When workers repeatedly failed to install it, a supervisor showed them the costly repair required at the end of the line. The episode makes concrete what hierarchy could obscure: people might perform a narrow task without receiving the information that would help them understand its purpose.

Information also had to travel through formal channels before a new idea could be accepted. A chart reproduced in Thomas J. Peters and Robert H. Waterman’s In Search of Excellence showed 223 possible paths through one company’s bureaucracy for a new product proposal. With so many layers between executives and workers, senior managers could struggle to know whether their picture of the company matched conditions in the field. Career incentives could worsen the problem: a study of young executives found that those most anxious to advance communicated problem-related information less accurately, while research on middle managers linked upward mobility with withholding news of failures such as fights and budget problems.

The book’s proposed alternative is not employee participation as a slogan or a financial stake alone, but a broader distribution of real decision-making power. Stock options given widely to employees were associated with gains in productivity, profits, and market returns, but most U.S. corporations concentrated their options among a small group of executives. Alfred Sloan’s approach at General Motors instead required managers to make a case for proposals and resisted judging an idea by its advocate’s status. Jack Welch’s effort to make General Electric a “boundaryless corporation” reduced divisions between business units and levels of management and used Work-Out sessions in which managers faced public criticism from employees. Welch did not consistently welcome opposing views in every major decision, but the reforms helped GE flourish in contrast to many older industrial corporations.

Decentralization and Its Limits

Decentralization draws strength from independence, specialization, and knowledge rooted in particular places or experiences. The book invokes Friedrich Hayek’s concept of tacit knowledge: useful understanding that cannot readily be summarized or transferred because it is tied to a specific job, place, or experience. People close to a problem may be better positioned to solve it than distant managers. Yet decentralization has a corresponding weakness: information discovered in one part of a system may never reach the rest. The organizational challenge is to preserve local specificity while making the resulting knowledge collectively useful.

The contrast between the intelligence community and a more integrated information system clarifies that distinction. Before September 11, the intelligence agencies’ division of labor offered specialization and diverse analytic perspectives. But the agencies lacked a mechanism to aggregate judgments as well as information. The book argues that a central authority might improve sharing while sacrificing the diversity, local knowledge, and independence that decentralization provides. Linking agency databases, for example, could improve information flow while preserving autonomy; remarkably, two years after September 11, the government still lacked a unified watch list drawing on data across the intelligence community. The diagnosis is not that decentralization itself fails, but that decentralization without aggregation leaves the system without an effective way to combine what its parts know.

Distributed Work in Science and at Sea

The book sets the value of decentralized contributions against cases where collective work produced results that no single expert could readily supply. In May 1968, the U.S. submarine Scorpion disappeared while returning to Newport News from duty in the North Atlantic. The Navy knew its last reported location but not what had happened or how far it had traveled after its last radio contact. The search began across an area twenty miles wide and thousands of feet deep. Rather than rely on a handful of submarine and ocean-current experts, naval officer John Craven assembled people with varied expertise, including mathematicians, submarine specialists, and salvage men.

Craven asked each participant to estimate the likelihood of different scenarios, with wagers and bottles of Chivas Regal as prizes. The bets addressed separate elements of the submarine’s final journey, including why it ran into trouble, its speed as it descended, and the steepness of its descent. No individual estimate located the submarine. Craven used Bayes’s theorem to combine the guesses into a composite picture of how the Scorpion died and an estimate of its final location. The episode presents aggregation as the step that turns specialized, partial judgments into a useful whole.

A similar principle operated in the World Health Organization’s international SARS research effort. In early February 2003, China reported 305 cases of severe respiratory disease in Guangdong Province since November 2002, including five deaths; tests had ruled out influenza viruses. Reports followed from Hanoi and Hong Kong, and by early March the illness was recognized as a new disease. The WHO issued a global warning and activated surveillance, while also organizing eleven laboratories in countries around the world to investigate the cause.

Beginning March 17, the laboratories held daily teleconferences, shared analyses and results online, and exchanged virus samples. They could examine the same samples simultaneously, testing possibilities and eliminating candidates found in some patients but not others. On March 21, researchers at Hong Kong University had isolated a likely candidate, while scientists at the U.S. Centers for Disease Control separately identified a coronavirus. Labs in Germany, the Netherlands, and Hong Kong sequenced it; in early April, infected monkeys in a Netherlands laboratory developed SARS. By April 16, only a month after the collaboration began, the laboratories were confident that the coronavirus caused the disease.

The discovery could not be attributed to one scientist or laboratory. Cynthia Goldsmith, an electron microscopist at the Centers for Disease Control and Prevention in Atlanta, first spotted the coronavirus, but weeks of work across the network established that it caused SARS. The process of ruling out other viruses also narrowed the field. The WHO orchestrated the network, but did not direct each laboratory’s work: the participants agreed to share relevant data and speak every morning, then worked out how to divide the investigation. In the book’s account, collective discovery depended on coordinated exchange without a single authority controlling every contribution.

Expertise and the Danger of Concentration

The account of Long-Term Capital Management supplies a counterpoint to the successes of distributed judgment. Started in 1994 by John Meriwether, the hedge fund brought together computer-modeling specialists and prominent financial theorists Myron Scholes and Robert Merton. Investors faced a $10 million minimum and managers took 25 percent of annual profits, yet demand grew after the fund posted strong returns for four consecutive years. Its concentration of expertise and confidence did not protect it when circumstances changed.

In August 1998, Russia defaulted on its debt amid investor unease following the collapse of Asian economies. A flight to quality made risky assets difficult to sell, leaving the fund with billions of dollars in rapidly declining holdings. It lost $4.5 billion in a couple of months, and efforts to sell assets pushed prices down further, creating losses for Wall Street banks. In September, a consortium of thirteen banks bailed out the fund. Set beside the submarine search and the SARS laboratories, the episode warns against treating a small, celebrated group of experts as a substitute for robust ways of testing judgments against changing information.

The Century’s Organizational Problem

Across its corporate, governmental, scientific, and financial examples, the book presents the twentieth century not as a simple march from hierarchy toward decentralization, but as a series of attempts to reconcile local knowledge with system-wide intelligence. Hierarchy can impede information and silence dissent; decentralization can leave useful discoveries isolated. Collective performance depends on structures that let people develop independent judgments and specialized knowledge while creating reliable means to share, test, and combine them.