Coordination failure

Coordination failure is the breakdown that occurs when people or units possess useful local knowledge but cannot make their separate judgments and actions add up to a collective result. The book treats it not as a simple shortage of expertise or a case for putting every decision under central command, but as an information-and-coordination problem: specialization and autonomy can make a group smarter, yet only if the knowledge they produce can be shared, combined, and put to use.

Local Knowledge and the Coordination Problem

Decentralization allows people close to a task or place to draw on experience that cannot easily be summarized and passed upward. The book calls this tacit knowledge and traces the principle to arrangements as different as ancient Athenian local government and the advice given to Moses to reserve only great matters for himself. Local decision-making can encourage independence and specialization, enlarging the knowledge available to a group rather than forcing every problem through a central authority.

That strength has a corresponding weakness: information discovered in one part of a system may never reach another part where it matters. The challenge is therefore not to choose between local knowledge and collective coordination, but to preserve both—to let people acquire specific knowledge while finding ways to make it useful to the wider group. The book compares this task to Google’s use of the local knowledge embodied in millions of Web-page operators: distributed contributions become more valuable when they can be aggregated.

Intelligence Without Aggregation

The U.S. intelligence community before September 11 illustrates how a sensible division of labor can still fail. Different agencies could develop specialized expertise and distinct perspectives, benefits that might make analysis more finely grained and decisions better informed. But the system lacked an effective way to aggregate not only information but judgments. It contained decentralization without aggregation, leaving no reliable means to bring the collective insight of its analysts into a shared assessment.

Richard Shelby’s proposed remedy—a truly central intelligence agency—addressed the problem of organization by placing information under a center of authority. But centralization would also risk sacrificing the diversity, independence, and local knowledge that made the separate agencies valuable. The book’s distinction is consequential: the failure to share information did not prove that a small central group would interpret it better. A small group, however intelligent, could not simply be presumed wiser than the larger body of people who had gathered and analyzed it. The proposed alternative is aggregation without traditional top-down control.

Even relatively mechanical measures could improve the flow. Linking agency databases would allow information to move while preserving agency autonomy. Yet two years after September 11, the government still lacked a single unified watch list drawing on data across the intelligence community. The gap between information held in separate places and information available to the system as a whole made coordination failure visible in a concrete institutional form.

Hierarchy and the Loss of Information

NASA’s Mission Management Team provides a different example: information was present, but the organization’s decision process did not give it a meaningful hearing. On January 24, when Don McCormack summarized the Debris Assessment Team’s findings, he left out its supporting analysis and did not describe whether the team was divided. No member of the Mission Management Team asked a question or requested the study. The meeting’s brisk closure of discussion turned uncertainty into apparent resolution rather than an occasion for inquiry.

The underlying difficulty was not simply that the team had too much structure or too little. Its discussions were overly dependent on a leader asking a question and someone answering, yet insufficiently organized to invite every member to analyze particular issues. The Debris Assessment Team engineers had serious concerns about the foam strike and wanted images of the Orbiter’s wing before they could make an informed assessment. Their concerns did not receive a serious hearing in a deeply hierarchical NASA culture. When a group begins with a conclusion, the book warns, new information can be reinterpreted to fit it; unshared analysis then stays not merely unused but effectively unseen.

Corporate hierarchy creates a comparable barrier between local experience and organizational judgment. At Ford in the 1970s, more than fifteen managerial layers separated the chairman from a factory-floor supervisor, compared with five at Toyota. In a General Motors plant in Van Nuys, California, one assembly-line worker had spent two years welding a bracket without knowing that it held the car’s sunshade. The episode makes the costs of poor information flow tangible: workers can perform a task without understanding its purpose, while the organization loses the knowledge that might connect work on the line to the finished product.

The barriers were also procedural and personal. A chart reproduced in In Search of Excellence showed that one unnamed company’s new-product idea faced 223 possible paths through the bureaucracy before acceptance. In such a structure, each manager could become a stumbling block, and executives risked losing contact with conditions in the field. Research cited in the book likewise found that career incentives could discourage the transmission of bad news: young executives more anxious to move up communicated problem-related information less accurately, and middle managers’ upward mobility correlated with withholding information about failures such as fights or budget problems. The organization could be full of people and still deprive itself of what they knew.

Designing for Coordination

The alternative is not decentralization without connections, but an organization that lets decisions about local problems be made as close as possible to the people who understand them while ensuring that local knowledge can matter beyond its immediate setting. The book argues that the experience of workers and specialists can outweigh managerial expertise when a problem depends on knowledge acquired through practice. Pushing every problem up the hierarchy and every solution back down can interrupt precisely the information that a workable response requires.

The book notes a gap between corporate rhetoric and practice: fewer than 2 percent of American companies, by one estimate it cites, made real use of high-performance work systems. Its evidence for employee empowerment includes a puzzle-and-proofreading experiment. Two groups worked amid loud, random noise; one group had a button that could turn it off, while the other did not. The group with the button solved five times as many puzzles and made fewer proofreading errors, even though nobody pressed it. The possibility of influence over their working conditions mattered to performance. Coordination, in this account, depends not only on transmitting facts but also on giving people meaningful scope to act on what they know.

A Problem Across Scales

The book’s chapter framing places coordination failure on a broad spectrum. A chapter on decentralized agents and the El Farol problem examines how individual choices relate to a shared outcome; another turns explicitly to traffic as a failure to coordinate and considers congestion pricing. These topics extend the same structural question beyond agencies and firms: how can separate decision-makers adjust their behavior when the consequences depend on what others do?

The range of examples does not reduce coordination to central planning. Rather, it distinguishes useful decentralization from disconnected decentralization. A system can benefit from specialized judgment and local autonomy, yet remain collectively ineffective when knowledge does not cross organizational boundaries, when dissent is silenced, or when people lack a workable means to align their choices. Coordination failure names that gap between what a collection of people knows and what it can accomplish together.