Schelling's Model of Segregation

How Individual Preferences Lead to Collective Patterns

In 1971, economist Thomas Schelling proposed a simple yet profound model that revolutionized our understanding of segregation. His key insight? Even when individuals have only a mild preference for similar neighbors—not outright hostility toward others—the collective outcome can be extreme segregation.

This was a breakthrough in complexity science: it showed that individual intentions don't always match collective outcomes. People might be tolerant and open-minded at the individual level, yet still produce highly segregated communities at the social level.

The Segregation Puzzle

A Surprising Observation

Cities across the world exhibit striking patterns of residential segregation by race, ethnicity, income, and other characteristics. The natural question is: why?

Before Schelling's work, many assumed that high levels of segregation must result from:

• Strong prejudice or discrimination
• Explicit policies forcing separation
• Deep-seated intolerance between groups

Schelling's Question

Schelling asked a different question: Could segregation emerge even without strong prejudice? Could it arise from mild preferences alone?

His model revealed something counterintuitive: yes. Even if people are happy living in mixed neighborhoods as long as they're not in the extreme minority, the system can still segregate completely.

The Model: A City on a Grid

Setting Up the World

Schelling's model is beautifully simple. Imagine a city represented as a checkerboard grid. Each square can be:

• Empty: No one lives there
• Occupied by Group A: Represented by one color (e.g., blue)
• Occupied by Group B: Represented by another color (e.g., red)

What Matters: Neighbors

Each agent (person) cares about their immediate neighbors—typically the 8 surrounding squares (like a chess king's movement). An agent's "neighborhood" consists of these adjacent occupied squares.

The Key Parameter: Similarity Threshold

Each agent has a preference for how many of their neighbors should be similar to them. This is captured by a similarity threshold:

Simple Rules, Complex Outcomes

The Algorithm

The model follows these simple steps:

1. Initialize: Randomly place agents on the grid, leaving some spaces empty
2. Check satisfaction: For each agent, count the fraction of similar neighbors
3. Move if unhappy: If the fraction is below their threshold, the agent is unhappy and moves to a random empty space
4. Repeat: Continue until no agents want to move (equilibrium)

What This Means

Notice what the model does NOT include:

• No one actively dislikes the other group
• No one prevents others from moving in
• No explicit coordination or planning
• No external enforcement of segregation

Agents simply want "enough" similar neighbors—they don't require all-similar neighborhoods. Yet watch what happens...

Interactive Simulation

Run the Model Yourself

Now let's run the full simulation. Start with a random mixed distribution and watch how the city evolves over time. You can control the key parameters and see how they affect the outcome.

Emergent Segregation Patterns

What You Should Observe

When you run the simulation with even a modest threshold (like 30-40%), you'll typically see:

• Cluster formation: Agents of the same type group together
• Boundary sharpening: Mixed boundaries gradually disappear
• Large homogeneous regions: Neighborhoods become dominated by one group
• Stable equilibrium: Eventually, almost everyone is satisfied, and movement stops

The Paradox in Action

Here's the striking result: with a threshold of just 30%, meaning "I'm happy as long as 30% of my neighbors are like me," the final state often has neighborhoods that are 70-90% homogeneous!

Why Does This Happen?

The mechanism is a feedback loop:

1. Some agents happen to be in areas with fewer similar neighbors and move
2. Their departure makes remaining different-group neighbors now unsatisfied
3. These neighbors also move, creating growing clusters
4. Clusters attract more similar agents, making them grow larger
5. The process amplifies until clear boundaries form

The Critical Role of Thresholds

Exploring Different Thresholds

The similarity threshold is the critical parameter. Let's explore how different values affect outcomes:

The Tipping Point

There's often a critical threshold around 30-40% where the system tips from mixed to highly segregated. Below this, some integration persists. Above it, segregation is nearly inevitable.

Real-World Implications

What the Model Tells Us

Schelling's model has profound implications for understanding social phenomena:

1. Individual vs. Collective Intentions

The model shows that we cannot infer individual preferences from collective patterns. A highly segregated city doesn't necessarily mean its residents are highly intolerant. Moderate preferences can produce extreme outcomes.

2. The Difficulty of Integration

Even if everyone would be happy in a mixed neighborhood, the system naturally drifts toward segregation. Achieving integration requires either:

• Very low thresholds (high tolerance for being in the minority)
• Active intervention to maintain mixing
• Constraints that prevent free movement

3. The Role of Empty Spaces

The model shows that having available empty spaces (housing mobility) can actually increase segregation! When people can easily move, they're more likely to seek out similar neighbors. Limited mobility can paradoxically lead to more integration.

4. Feedback Loops and Path Dependence

Small initial imbalances get amplified through feedback. Where you end up depends strongly on where you started. This is why historical patterns of segregation can be so persistent.

Applications Beyond Housing

The Schelling model applies to many domains beyond residential segregation:

• Workplace diversity: Self-segregation in offices and organizations
• School segregation: Sorting by race, income, or academic achievement
• Social networks: Clustering in friendships and online communities
• Economic sorting: Income-based neighborhood segregation
• Political polarization: Geographic sorting by political beliefs

Extensions and Variants

Asymmetric Preferences

What if one group has different thresholds than another? For example, if Group A needs 40% similar neighbors but Group B only needs 20%?

Multiple Groups

The model can be extended to more than two groups. With three or more groups, patterns become more complex but segregation still tends to emerge.

Different Neighborhood Definitions

Instead of the 8 immediate neighbors, we could define neighborhoods differently:

• Larger radius (24 neighbors in a 5x5 area)
• Distance-weighted (closer neighbors matter more)
• Different shaped neighborhoods

Wealth and Housing Prices

More sophisticated versions include:

• Housing prices that increase in desirable areas
• Wealth constraints limiting where agents can move
• Preference for neighborhood quality, not just similarity

Conclusion

The Power of Agent-Based Models

Schelling's segregation model is a masterclass in agent-based modeling and complexity science. From simple rules—agents moving to satisfy mild preferences—emerges a complex, unexpected pattern: high segregation.

Key Lessons

• Emergence matters: Macro patterns don't necessarily reflect micro intentions
• Individual rationality ≠ collective optimality: Everyone acting on reasonable preferences can produce outcomes no one intended
• Small differences amplify: Modest individual preferences can create dramatic collective outcomes
• Feedback loops are powerful: Small initial clustering triggers cascading segregation
• Path dependence is real: History matters; small early differences can determine final states

Beyond Segregation

The principles demonstrated by Schelling's model apply far beyond residential segregation:

• Any system where agents have preferences about their neighbors
• Any process involving local interactions and global patterns
• Any phenomenon where feedback loops amplify small differences

From opinion dynamics to ecosystem ecology, from financial markets to cultural evolution—Schelling's insights about how simple individual rules create complex collective patterns continue to illuminate our understanding of complex systems.