← /notes

Coordination

Created Dec 23, 2024 systemsgame-theoryorganizations

Two people want to meet but can’t communicate. Each must choose a location. If both pick the same place, they meet; otherwise they don’t. This is a coordination problem: success requires aligned choices, but nothing guarantees alignment. Even when interests are identical, coordination can fail.

Thomas Schelling studied how people solve this. They look for schelling points — options that stand out, that seem obvious, that a reasonable person would expect another reasonable person to choose. “Let’s meet in New York” succeeds because Grand Central Station feels like the answer. Coordination requires common knowledge about what everyone expects everyone to expect.


Coordination problems differ from prisoners’ dilemmas. In a dilemma, individual incentives conflict with group outcomes; people defect even though cooperation would be better. In pure coordination, there’s no conflict — everyone wants to align — but alignment isn’t automatic. The barrier isn’t selfishness; it’s uncertainty about what others will do.

But many real situations blend both. The tragedy of the commons is a coordination failure with competing incentives. collective action requires agreeing on goals and agreeing on who does what. The more complex the task, the more coordination becomes the bottleneck, regardless of goodwill.


Organizations exist largely to solve coordination problems. Hierarchy creates common knowledge about who decides what. Procedures create shared expectations. Culture creates schelling points for novel situations. The organization is a coordination technology.

This is why small groups can outperform large ones. Coordination costs scale faster than linearly with group size. Adding a member adds another person to coordinate with everyone else. network effects cut both ways: more connections mean more value, but also more overhead. The winning organizations are those that compress coordination costs.

Related: schelling points, systems, network effects, threshold models, collective action