← Back to Paradoxes

Schelling's Segregation Model

When Mild Preferences Create Extreme Patterns

Even when individuals prefer diversity, the collective outcome is extreme segregation.

In 1971, economist Thomas Schelling (later Nobel laureate) discovered a troubling truth about social dynamics: even when every person in a neighborhood would be perfectly happy with diverse neighbors, the neighborhood can still become completely segregated. The paradox? Nobody wanted this outcome.

Imagine residents who only ask: "Do at least 30% of my neighbors look like me?" This mild preference—compatible with 70% different neighbors—inevitably produces neighborhoods that are 90%+ homogeneous. The model reveals how individual tolerance and collective segregation can coexist.

Watch Segregation Emerge

Interactive Simulation
Speed:

Parameters

Tolerance Threshold 30%
Agents are happy if at least 30% of neighbors are similar. They would accept up to 70% different neighbors.
Population Density 85%
Grid Size 50×50

Live Statistics

Step 0
Unhappy Agents 0
Happiness Rate 100%
Segregation Index 0%
Moves This Step 0
Group A
Group B
Empty
Equilibrium Reached! All agents are satisfied.

Segregation Index Over Time

Happiness Rate Over Time

The Mechanism

How Agents Decide

similar_neighbors / total_neighbors ≥ tolerance
An agent counts neighbors of the same type and compares to threshold.
If unhappy → move to random empty cell
Unhappy agents relocate, but their departure may make others unhappy.
Cascade effect → emergent segregation
Each move changes multiple agents' satisfaction, creating chain reactions.

Why Does This Happen?

The Key Insight: Local Choices, Global Consequences

When a few agents at the boundary become unhappy and move, they change the neighborhood composition for everyone nearby. This triggers more moves, creating a cascade effect. The boundaries between groups become sharper and sharper until equilibrium—but by then, the damage is done.

The model demonstrates a profound truth about emergent behavior: the collective pattern is not a simple sum of individual preferences. A population where everyone wants 30% similar neighbors doesn't stabilize at 30% segregation—it overshoots dramatically.

Mathematically, this is because the system has multiple stable equilibria. Random mixing is unstable—any small perturbation grows. Complete segregation is stable—once achieved, everyone is happy. The system rolls downhill toward segregation even when no one intended it.

Try These Experiments

🔬
The Tipping Point

Set tolerance to 25%, then slowly increase it. Watch how the segregation index jumps dramatically somewhere between 30-40%. This is the critical threshold where integration becomes unstable.

📊
Density Matters

Compare high density (90%) vs low density (60%). Lower density means more empty spaces—more room to escape—leading to faster but sometimes less extreme segregation.

⚖️
The Paradox Zone

Set tolerance to exactly 50%. Agents would be happy with half-and-half neighbors—perfect integration! Yet watch as the system still segregates. Why? Because once a few agents cluster by chance, the cascade begins.

Real-World Implications

🏘️
Residential Segregation

Studies of American cities show segregation patterns consistent with Schelling dynamics. Individual prejudice explains some but not all of it—the model shows how mild preferences amplify into extreme outcomes.

🏫
School Choice

When families have school choice, even small preferences for "schools like us" can produce highly segregated school systems—a Schelling effect in education policy.

💬
Online Echo Chambers

Social media "unfollowing" based on mild disagreement can create extreme polarization—a digital Schelling segregation where like-minded users cluster.

🧪
Policy Design

The model suggests that fighting segregation requires understanding tipping points. Small interventions at critical moments may prevent cascades that larger interventions can't reverse.

The Deeper Lesson

Schelling's model is a canonical example of emergence—complex global patterns arising from simple local rules. It warns us that we cannot judge a system's outcomes by looking at individual intentions. Even a world of tolerant people can produce intolerant-looking outcomes.

This insight extends far beyond segregation: financial markets, traffic patterns, ecosystem dynamics, and social movements all exhibit emergent behaviors that surprise and sometimes dismay the individuals who create them. The whole is not just different from the sum of its parts—it can be its opposite.

In a world of individually tolerant agents, collective intolerance emerges unbidden.