Schelling’s Segregation Model, explained.
Schelling's segregation model shows how individual preferences about nearby neighbors can generate substantial separation at the population level.
Why it happens
The final pattern need not be anyone's explicit goal. Each move changes another person's neighborhood, creating a chain of responses. Local decisions and global outcomes operate at different scales.
Schelling explored how individual rules can generate aggregate patterns. Even a preference that does not demand complete separation can produce clustering through repeated moves.
Read the result
Vary the tolerance threshold, then resample the initial arrangement. Compare local satisfaction with overall separation. The pattern illustrates a mechanism under the specified moving rule, not a complete explanation of real segregation.
A worked example
A mixed neighborhood changes gradually
People initially live in a mixed arrangement but prefer at least some neighbors of their own type.
One person moves to satisfy that preference, changing the neighborhood composition for others and prompting further moves.
The resulting clusters can be stronger than the segregation any single person requested.
OPTIONAL DEEPER DETAILGo deeper: inside the model
Inside this model
A bounded 16 × 16 grid begins with equal groups and 20% vacancies. An agent is unhappy if the share of like neighbours among occupied adjacent cells is below its threshold. In shuffled order, unhappy agents move to a random vacancy; new sites need not immediately satisfy them.
Where this idea is useful
A practical use
Use the model to distinguish individual intentions from collective outcomes in neighbourhood or organisational clustering.
A common misconception
“A segregated outcome proves everyone wanted complete segregation.”
A collective pattern can emerge from weaker local preferences. Real outcomes can also involve policies, constraints, discrimination, and history omitted by this model.
What this explanation leaves out
- This toy model omits discrimination, wealth, housing supply, policy and history. It cannot explain real segregation on its own. Agents with no occupied neighbours are treated as satisfied.
Does a low preference threshold guarantee integration?
No. Starting conditions, available vacancies, moving rules, and path dependence can affect the final arrangement. Compare multiple runs before treating one outcome as representative.
Could many reasonable local choices combine into a collective outcome nobody intended?
Associated thinkers
Further reading
Explore the original research or the teaching reference behind this experiment.