Schelling Segregation Model: The Emergence of Spatial Order

The Emergence of Spatial Order from Individual Preference — Schelling (1971) / Beinhocker Ch. 5

How Does Segregation Emerge?

Place two types of agents on a grid — say red and teal. Give each agent a modest preference: "I'd like at least 30% of my neighbors to be like me." That's not prejudice — it means being perfectly happy as a minority. Yet when unhappy agents relocate to random empty cells, the population spontaneously sorts itself into stark homogeneous clusters. Neighborhoods that no one explicitly chose emerge as if drawn by an invisible hand. This is Thomas Schelling's celebrated insight: individual tolerance does not aggregate into collective integration. Mild micro-motives produce extreme macro-behavior — a hallmark of complex adaptive systems.

Schelling's Paradox

With a threshold of just 30% (agents are happy being a minority), the simulation produces neighborhoods that are 70-80% homogeneous. The segregation index — the average fraction of same-type neighbors — far exceeds what any individual demands. No one wants segregation, yet everyone gets it. This is emergence in its purest form: a macro pattern that cannot be predicted from individual rules alone.

Grid View

Type A

Type B

Empty

Segregation Index Over Time

Average fraction of same-type neighbors. Starts near 0.5 (random); rises as agents sort themselves. Higher = more segregated.

Happiness Rate Over Time

Fraction of agents satisfied with their neighborhood. Rises to 1.0 as system reaches equilibrium.

Moves Per Tick

Number of agents that relocated this tick. Falls toward zero as system stabilizes.

Interface Density Over Time

Fraction of neighbor-pairs that are of different types. Falls as segregation increases. In a random grid with 50/50 ratio, expect ~0.50.

Insights from Schelling's Model

> Micro-macro disconnect: A 30% threshold means agents are happy as minorities, yet neighborhoods become 70-80% homogeneous.
> Tipping dynamics: Once a few agents move, they change the composition of two neighborhoods, potentially triggering cascades.
> Empty cells matter: More vacancy means easier movement and faster (but sometimes different) segregation patterns.
> Asymmetric groups: When one group is smaller, its members face more pressure to move, creating distinct spatial patterns.
> Policy relevance: Schelling showed that integration policies must address systemic dynamics, not just individual attitudes.
> Beinhocker's context: Schelling's model is a canonical example of emergence — complex macro patterns arising from simple micro rules with no central coordination.