Coordination problems

Coordination problems are situations in which people must make their actions fit together while anticipating that others are trying to do the same. They range from deciding where friends will meet or how pedestrians will share a sidewalk to organizing companies and enabling buyers and sellers to trade. Unlike questions with a definitive answer, coordination problems often have no clear optimum; their solutions emerge from interdependence, shared expectations, and institutions that let dispersed knowledge become useful to a group.

Definition and Scope

A coordination problem differs from a question that can be answered by identifying a fact, such as how many printers a company will sell. It arises when each person’s choice both affects and depends on what other people choose. A driver in heavy traffic, a subway rider choosing a seat, or a group deciding where to eat cannot settle the matter by considering only a private view of the best option: the outcome depends on how others act as well.

The same structure appears in economic life. Buyers and sellers have to find one another and trade at a fair price; companies must organize their operations; workers and resources must be matched to tasks; and factories must decide what to produce. Such questions are not simply about discovering the right answer independently. Each participant has to form expectations about the choices of others, and those expectations shape the choices that follow.

Interdependence in Everyday Life

Pedestrian movement offers a visible example of coordination without a central director. People adjust their pace, direction, and stride in response to one another. A New Yorker may slow slightly in the “simple pass” to avoid a collision, while pedestrians may form platoons at crosswalks to protect themselves from traffic. These small adjustments allow a moving crowd to achieve a measure of order without anyone assigning each person a route.

That kind of coordination is distinct from solving a prediction problem, such as forecasting the result of a football game. In prediction, people try to identify what will happen; in coordination, they must act with an eye to what others believe should happen and how those beliefs will affect their behavior. The solutions are often good rather than uniquely optimal, and they may depend on institutions, norms, and history—elements that shape collective behavior even as people’s behavior reshapes them.

Local Knowledge and Collective Organization

Coordination depends in part on knowledge held by people close to a particular task or place. Such knowledge may be difficult to summarize or pass along because it comes from specific experience, yet it can be essential to finding a workable solution. Decentralization allows people to specialize and use this local knowledge while continuing to coordinate their actions. Its weakness is that useful information discovered in one part of a system may never reach the rest.

The challenge, then, is not simply to choose between central control and dispersed decision-making. A group needs ways to let people retain specialized, local knowledge while making that knowledge collectively useful. The intelligence community illustrates the distinction: agencies had specialized information and analysis, but lacked a means of aggregating judgments across them. Linking information is not enough if the organization cannot also bring separate assessments together; centralizing authority may improve control while sacrificing the diversity, independence, and local knowledge that decentralized work can provide.

Rules, Incentives, and Failure

Coordination can be imposed through authority or coercion: an army marching in formation and workers on an assembly line can be made to move together by formal direction. But many interactions among private citizens in a liberal society are not governed closely enough for top-down instruction to resolve them. They call instead for arrangements in which people voluntarily make their actions fit together, and there is no guarantee that independent choices will produce a good result.

The rules governing decisions can change whether a group learns from its members or follows a misleading signal. In an experiment using two urns—one containing twice as many light marbles as dark ones, the other twice as many dark as light—participants guessed which urn had been chosen. When students were rewarded for the group’s accuracy rather than their individual correct answers, they paid more attention to their own imperfect evidence and less to earlier guesses. The group became more accurate, demonstrating that the structure of incentives can affect how well a collective decision uses information.

Organizational hierarchy can produce the opposite effect. At one NASA meeting about damage to the Orbiter, the team’s discussion was shaped by a senior leader’s questions and answers, while members were not systematically invited to examine specific issues or voice their assessments. Engineers’ concerns and requests for images did not receive a serious hearing. More generally, a rigid hierarchy can obstruct information: in one corporation, a new product idea faced 223 possible paths through the bureaucracy before it could be accepted. Coordination therefore requires more than getting people to work within the same system; it also depends on whether the system allows relevant knowledge and dissent to circulate.

Markets and the Coordination Problem

Decision markets offer one proposed means of aggregating dispersed judgments. FutureMAP, an intelligence initiative funded by the Defense Advanced Research Projects Agency, included a plan for small internal markets in which analysts from different agencies could trade contracts based on expectations about possible events, including events in the Middle East. A proposed public-facing Policy Analysis Market would have applied a similar idea to forecasting. These markets were presented as a possible way to gather judgments across bureaucratic boundaries, where ordinary channels could leave information divided and political pressures could shape what analysts felt able to conclude.

Across these examples, the central difficulty is to make interdependent action possible without losing the knowledge distributed among individuals. Authority can coordinate, but may suppress local judgment; decentralization can preserve specialization, but may leave insights isolated. Coordination problems endure because neither independent choice nor central direction automatically solves the task. Groups must find institutions and practices that help people anticipate one another, share relevant information, and turn separate decisions into workable collective outcomes.