Coordination Games

Coordination games are a class of games in which players have a shared interest in coordinating their strategies, but face multiple equally viable — or differently ranked — equilibria. Unlike prisoner's dilemmas, where individual incentives undermine collective outcomes, coordination games feature aligned incentives frustrated only by strategic uncertainty about which equilibrium others will select.

The simplest form is the pure coordination game, represented by a symmetric payoff matrix where players earn their highest payoff only when they choose the same strategy, and their lowest payoff when they miscoordinate. Both players prefer any coordinated outcome to any miscoordinated one, and in the pure case, they are indifferent between the coordinated equilibria.

Formal Structure

In a two-player, two-strategy coordination game, the payoff matrix typically takes this form:

Player 2
       A      B
Player A  (a, a)  (c, c)
1   B  (c, c)  (b, b)

Where \(a > c\) and \(b > c\). Both (A, A) and (B, B) are Nash equilibria. The game is one of pure coordination if \(a = b\); it becomes a battle of the sexes or impure coordination game if \(a \neq b\), meaning players rank the equilibria differently while still preferring coordination to miscoordination.

The formal simplicity is deceptive. Coordination games are trivial to solve once players know which equilibrium is focal. The entire difficulty lies in the meta-game of equilibrium selection: how do players converge on shared expectations without explicit communication?

Equilibrium Selection

Thomas Schelling's concept of the focal point was the first rigorous answer. Agents use salience — cultural convention, historical precedent, physical prominence, or arbitrary symmetry-breaking cues — to solve the selection problem without negotiation. The famous example: if two people must meet in New York City without communicating, they will likely converge on Grand Central Terminal not because it is optimal, but because it is uniquely prominent.

But salience is context-dependent and fragile. What is focal in one culture is invisible in another. What is focal today may not be focal tomorrow if the shared environment shifts. This makes coordination games especially sensitive to information structure and common knowledge: it is not enough that both players find A salient; each must know that the other finds it salient, and know that the other knows, ad infinitum.

Beyond focality, two formal refinements dominate the literature:

• Pareto dominance: if one equilibrium gives all players strictly higher payoffs than another, rational agents should select it. But experimental evidence shows that players frequently fail to coordinate on Pareto-dominant equilibria, especially when the risk of miscoordination is high.

• Risk dominance: proposed by John Harsanyi and Reinhard Selten, risk dominance selects the equilibrium that is safer to play when uncertain about the opponent's choice. Risk dominance and Pareto dominance often conflict — a tension that mirrors the broader trade-off between efficiency and robustness in complex systems.

Coordination and Network Effects

Coordination games scale dramatically when embedded in networks. In a population where each agent plays a coordination game with its neighbors, local conventions can crystallize into global norms — or fragment into persistent disagreement. The dynamics depend on the topology: dense networks with high clustering tend toward rapid convergence; sparse networks with bottlenecks can sustain multiple incompatible conventions indefinitely.

This is the game-theoretic foundation of network externalities. The value of joining a platform depends not merely on its intrinsic quality but on the expectation that others will join. A coordination game played across millions of agents produces path dependence and lock-in: once a convention is established, the cost of switching exceeds the benefit, even if the alternative is superior. The QWERTY keyboard, VHS format, and dominant social media platforms are all solutions to large-scale coordination games that became self-reinforcing traps.

The policy implication is structural, not behavioral. You cannot solve a coordination failure by educating individuals to be more rational. You solve it by changing the game: introducing standards, subsidies, or temporary coordination mechanisms that make the desired equilibrium focal and self-sustaining.

The Connector's Claim

The persistent treatment of coordination games as a minor specialization within game theory is a category error. Coordination is not a special case of strategic interaction; it is the general case. Every market, every language, every institution, every technological standard is a coordination solution that has achieved sufficient stability to persist. The prisoner's dilemma gets the headlines because conflict is dramatic, but coordination gets the work done because agreement is the precondition for everything else. A field that studies strategic interaction without foregrounding coordination is like a biology that studies predators but ignores symbiosis — technically valid, structurally blind.