Organizational decentralization

Organizational decentralization is the distribution of decision-making authority toward people close to the problems at hand. In The Wisdom of Crowds, it is valuable because it lets organizations use specialized experience and local knowledge, engage workers, and coordinate without relying on constant orders—but it succeeds only when information can travel and people remain on the same team. The central question is not whether to choose decentralization over organization, but how to give local judgment a route into collective action without suppressing the independence that makes it useful.

Local Knowledge and the Problem of Aggregation

The case for decentralization begins with knowledge that cannot readily be detached from the place, job, or experience in which it arises. Friedrich Hayek’s idea of tacit knowledge helps explain why a manager removed from a problem may be less able to solve it than the person who encounters it directly. Organizations therefore should not assume that every difficulty must be sent up a hierarchy and every answer sent back down. For many local problems, specialization and proximity can matter more than managerial expertise.

That advantage carries a structural risk. Decentralized people and units may discover valuable facts without those facts reaching anyone else who needs them. The challenge is to preserve local knowledge while making it collectively useful: independence and specialization must coexist with channels for coordination and aggregation. The same design problem appears in a market, a corporation, or an intelligence agency. A system that distributes work but cannot combine what its parts know has decentralization without the means to act as a whole.

The book’s account of collective judgment clarifies why both sides matter. A wise group needs diverse, independent opinions, decentralized access to knowledge, and some mechanism that aggregates individual judgments. Aggregation is not a demand that every participant surrender autonomy to a central decision-maker. It is what allows separately acquired information to contribute to a shared result. Conversely, local knowledge alone cannot guarantee a good collective decision if it remains isolated or if participants’ judgments are shaped by one another rather than by what they know.

Authority, Engagement, and Coordination

Decentralization can change how people perform even when they do not use the authority they are given. In one experiment, people worked on puzzles and proofreading while loud, random noises recurred. One group had a button that could turn the sound off; the other group had no such control. No one pressed the button, yet the group that had it solved five times as many puzzles and made fewer proofreading errors. The mere knowledge that they could affect their working conditions was associated with a striking performance difference.

At Toyota, decentralization is also a way to coordinate production. The Toyota Production System trains frontline workers in a wider range of skills and gives them an understanding of the production process as a whole. Rather than isolating each worker at a narrow station on a classic assembly line, the system organizes workers into teams that take responsibility for their production process. Any worker can pull a cord to stop the line when something needs fixing. Like the button in the experiment, the cord is rarely pulled; its availability is itself part of the system. Workers can respond to problems where they arise, reducing the need for constant supervision and allowing managers to concentrate elsewhere.

The argument is not that every decision should be delegated. Firing decisions, for example, remain an area in which corporations delegate little authority. But a corporation faces different kinds of problems, and coordination and cooperation problems can often be handled through decentralized solutions. Where the knowledge needed to solve a problem is in the heads of the workers dealing with it, giving them authority is more than a gesture: it places decision power where relevant knowledge resides.

Hierarchy, Information, and Group Failure

Formal hierarchy can prevent local information from moving upward, even when the people closest to a problem recognize that something is wrong. In the account of NASA’s Mission Management Team and the debris assessment engineers, the engineers had serious concerns about the foam strike and said they needed images of the Orbiter’s wing to make an informed analysis. Their concerns did not receive a serious hearing. The team’s discussion was too structured around one person asking questions and others answering, yet not structured enough to ensure that members were asked to analyze particular issues. The result was a narrow supply of information and analysis, while a conclusion already reached shaped how later information was interpreted.

The same organizational weakness appears in the description of large corporations. At Ford, more than fifteen layers of managers stood between the chairman and a factory-floor supervisor; at Toyota, there were five. In one General Motors plant in Van Nuys, California, a worker had been welding a bracket for two years without being told that it held the car’s sunshade. When a supervisor showed two workers what happened if the bracket was missed—the carpet had to be torn out so the bracket could be welded in later—one was surprised to learn what the part did. The example makes the cost of distance concrete: workers can perform a task without receiving information that would help them understand its place in the whole process.

Bureaucratic distance also obstructs ideas and obscures conditions on the ground. A chart reproduced in In Search of Excellence showed 223 different paths through an unnamed company’s bureaucracy that a new-product idea might have to traverse before approval. With so many layers between executives and workers, top management could struggle to tell whether its picture of the corporation matched reality. Incentives made the flow still less reliable: a 1962 study found that young executives more anxious to advance communicated problem-related information less accurately, while another study of middle managers linked upward mobility with withholding news of things that had gone wrong. Hierarchy can therefore suppress not just local decisions but also the information on which higher-level decisions depend.

The Boundary: Shared Goals and Corporate Strategy

Decentralization inside an organization is not the same as fragmentation. The benefits of competition are real, but internal rivalries can undermine the reason for having a company at all, eroding economies of scale and increasing the cost of monitoring behavior. The divisions at Enron operated as separate islands, with their own senior executives and even their own information-technology systems. Some divisions could not communicate with one another; the company also spent millions of dollars on redundant technology. When units compete for internal resources rather than in a genuine outside market, apparent efficiency can be an illusion. Decentralization works only if those making local decisions still understand that they belong to the same enterprise.

Nor does decentralization mean that a corporation can dispense with collective decision-making at the level of strategy. Companies may acknowledge the value of pushing responsibility downward while still leaving major cognitive problems to one person: the chief executive. Decisions about new products, factories, demand forecasts, prices, and mergers are problems that may be well suited to collective judgment, even within a relatively small group. The book notes the paradox of the 1990s: as businesses talked more about decentralization and bottom-up mechanisms, CEOs were increasingly treated as superheroes and corporate saviors. That belief placed the answer to organizational success in the person at the top, even as the case for distributed knowledge pointed elsewhere.

The resulting principle is conditional rather than absolute. Local decisions should be made as close to the problem as practical; workers should have room to use what they know and improve the operations they understand. At the same time, organizations need ways for their separate judgments and discoveries to circulate, be considered, and contribute to common decisions. Without local authority, the system discards knowledge and engagement. Without communication and shared purpose, decentralization leaves knowledge stranded in disconnected units. Its promise lies in holding independence and coordination together.