Skip to main content
computer-science economics political-science

​
Focal point

​
Description

A focal point is the option that uncommunicating parties converge on because it is salient to all of them, and each knows the others find it salient too. Thomas Schelling introduced the idea in The Strategy of Conflict (1960): asked to meet a stranger in New York City on a given day with no way to communicate, most people name the same prominent place and time (in his informal sample, noon at Grand Central). Nothing makes noon or Grand Central a better rendezvous in payoff terms — they are simply the obvious ones, and their obviousness is common knowledge, which is exactly what lets strangers coordinate. The load-bearing element is the recursion, not the salience alone. A feature that is prominent to me does no coordinating work unless I expect it to be prominent to you, and expect you to expect it prominent to me. That mutual expectation is what selects one option out of many that are otherwise equally good. Schelling’s phrase for the underlying capacity: people “can often concert their intentions or expectations with others if each knows that the other is trying to do the same.” The diagnostic question — of the equally-valid options, which one is obvious to everyone, and does everyone know it’s obvious? — turns “how will they coordinate without talking?” into a search for the salient default. It fires wherever coordination must happen without communication or before a protocol exists: strangers meeting, standards converging, belligerents settling on a boundary, distributed nodes breaking a tie.

​
Aliases

“Schelling point” is the eponym, after Thomas Schelling; “coordination point” names the same thing by its function. The name borrows the image from optics and geometry, where a focal point is where parallel rays converge — here it is where independent expectations converge. The two senses share the deep structure: many separate lines drawn toward a single distinguished point. (The optics sense is pure polysemy — a different concept wearing the same word — not a mis-application of this one.)

​
Triggers

User-initiated: User asks how parties will coordinate without communicating, or notes that everyone “just knows” which option to pick. Vocabulary cues: “focal point,” “Schelling point,” “the obvious choice,” “we’ll both know to…,” “default,” “coordinate without talking.” Agent-initiated: Agent notices a coordination problem with several equally-good solutions and no communication channel, and asks which option is salient enough to be common knowledge. Candidate inference: “there are many valid choices here — is there one obvious enough that everyone expects everyone to pick it?” Situation-shape signals: Coordination without a protocol; tie-breaking that must be predictable to independent parties; de-facto standards forming without a committee; tacit boundaries in conflict; defaults that everyone converges on not because they are best but because they are obvious.

​
Exclusions

  • One option is simply better — when a coordination option dominates on payoff, parties choose it for its value, not its salience; the focal-point mechanism (salience selecting among payoff-equivalent options) is not doing any work.
  • An explicit coordination protocol is available — when parties can pass messages and agree (a vote, a contract, a standards committee), convergence comes from the protocol, not from shared salience; that is consensus, not a focal point.
  • The salience is private, not common knowledge — a feature obvious to me but that I have no reason to think others perceive cannot select a coordination point. The focal point requires the recursive expectation that all parties are drawn to the same salient option.
  • There is only one equilibrium — a focal point is a selector among multiple coordination points. When the game has a unique solution there is nothing to select, and the concept category-mismatches with equilibrium, the balance-point primitive itself.

​
Structure

Internal structure of focal-point: a table of its component slots and the concepts that fill them. = a set of roughly-equal options + a feature that makes one salient + common knowledge of that salience. The first two are inert on their own: many equal options with no salient feature leaves the parties guessing, and a salient feature that only I perceive cannot coordinate anyone. The concept is the conjunction, and the third role — the recursive mutual expectation of salience — is what does the actual selecting.

​
Relationships

Relationship neighborhood of focal-point: a graph of the concepts it connects to and the concepts it is a part of.
  • consensus — protocol versus salience. Consensus converges by passing messages under a rule; a focal point converges with no messages at all. The focal point is what remains available when communication is impossible.
  • equilibrium — a focal point presupposes multiple equilibria and picks which one the players land on. Equilibrium is the steady-state primitive; focal-point is the salience-based selector among several of them. (With a unique equilibrium there is nothing to select — see Exclusions.)
  • commitment-device — Schelling’s paired strategic moves. A public, irreversible commitment can manufacture a focal point by making one option unmistakably the obvious one; a pre-existing focal point gives a commitment something to converge on.
  • seeding — analogous small-cause-large-effect shape at different timescales: seeding fixes an emergent structure over developmental time; a focal point fixes a coordination outcome in a single simultaneous choice. Both are cases where a low-cost detail disproportionately determines a large outcome.

​
Examples

Schelling, T. C. (1960). The Strategy of Conflict. Harvard University Press. · economics

Schelling asked people where and when they would go to meet a stranger in New York City, given only that the stranger was trying to meet them on the same day and that they could not communicate in advance. Despite the enormous number of possible times and places, answers clustered sharply: most named noon, and a majority named Grand Central Station. No time or place is a better rendezvous on its own merits; they win because they are the prominent, obvious choices — and, crucially, because each person expects the other to find them obvious too.Inference: The coordinating force is not the salience of the landmark but the common knowledge of that salience. The diagnostic move when parties must coordinate without communicating: don’t ask “what is the best option,” ask “which option is obvious to everyone, and does everyone know it’s obvious?” The answer is often unremarkable on the merits and unbeatable on coordination.

Schelling, T. C. (1960). The Strategy of Conflict — tacit bargaining and limited war. · political-science

Adversaries who cannot openly negotiate — belligerents in a limited war, rivals avoiding escalation — still coordinate on where to stop, by converging on salient limits. A river, a coastline, a round-numbered parallel of latitude, or a qualitative line (“no nuclear weapons,” “no attacks across this border”) becomes the tacit boundary because it is conspicuous to both sides and each expects the other to recognize it. “There is stability at the river,” Schelling observed, “and perhaps nowhere else”: a boundary halfway across an open plain has no purchase, but a natural feature does.Inference: Salient discontinuities do coordination work that smooth gradients cannot. When two parties must settle on a limit with no way to bargain over it, the limit snaps to whatever qualitative marker both can see and both know the other can see — which is why “somewhere reasonable in the middle” is unstable while “at the river” holds. Look for the conspicuous line, not the optimal one.
SchellingCoin is a decentralized-oracle design: many independent reporters each submit a value (say, the ETH/USD price), and the protocol rewards those whose report lands near the median of all reports. No reporter can communicate or collude, yet each is paid for matching the others. The only value every honest reporter can expect every other honest reporter to also aim at is the true one — so the truth becomes the point they converge on. Buterin’s formulation: the truth is the most powerful Schelling point.Inference: A focal point can be engineered into a mechanism, not merely observed in behavior. If you pay independent agents for agreeing, and the one feature all of them can jointly expect to target is the correct answer, honest reporting becomes the coordinating equilibrium. The design imports Schelling’s stranger-meeting logic into a trustless protocol: replace the shared cultural salience of “noon at Grand Central” with the shared epistemic salience of “the real number.”