The Enclave That Already Exists — Summary
Renee made the map in her first month on the job. She overlaid MARTA’s coverage area on a map of the county boundaries surrounding Atlanta. The result was so legible that she has shown it to every class of planning students she has spoken to in the twelve years since. The map shows a transit network serving Fulton County, DeKalb County, and the city of Atlanta. It shows Cobb County, directly to the northwest, where the network stops at the county line. Cobb County has about 800,000 people. It refused MARTA, explicitly, in 1971, when the regional transit authority was being assembled. It has reaffirmed the refusal several times since.
Regional transit in 1971 was being built to connect the labor pool of a metropolitan area to the locations of its employment. The transit network was, functionally, a labor market infrastructure project. Cobb County’s refusal was a refusal to participate in that infrastructure. What gets refused when a wealthy enclave refuses regional transit is not a bus line. It is connectivity itself. The refusal ensures that the people who would most benefit from the connection remain outside the enclave’s economic geography. This is the template: exit from shared infrastructure, justified in neutral administrative language, with a sorting effect on who can access the exiting community that benefits those already inside it.
The exit-voice dynamic operates in built infrastructure with a specific logic that amplifies over time. When a wealthy community exits from a shared public system — whether transit, schools, water, or any other infrastructure — two things happen. The exiting community stops consuming the shared system, which reduces its political constituency for maintaining and investing in that system. And the exiting community builds or buys a private alternative, available to those who can afford it and unavailable to those who cannot. Over time, the shared system loses investment because the people with the most political capital have exited it. As the shared system degrades, the case for remaining weakens, which incentivizes further exit among those who can afford it. The cycle is self-reinforcing and has no internal correction mechanism. It requires a political intervention, which is increasingly difficult to organize because the people with the most political capital are the people who have already exited.
The private infrastructure of the American enclave is not primarily dramatic. It is mundane. The neighborhood where the sidewalks are maintained because the HOA requires it and the adjacent neighborhood where they are not. The school where the PTA budget supplements the public allocation and the school where it does not. The park that is well-lit and the park that is not. Each of these is a small exit. Together they constitute a parallel infrastructure, built from private capital, available to the people inside it, invisible as a system to those looking at any single piece.
What automation accelerates is not the enclave itself, which predates automation by decades. It accelerates the mechanism by changing who holds the automation dividend. When automation displaces workers, the productivity gain flows somewhere. In the current configuration, the displacement is broad enough across sectors that the surplus labor supply remains large, and the productivity gain flows to capital, which is already concentrated in the people and institutions for whom exit from shared systems is easiest. Capital accumulation and exit capacity are correlated. As the automation dividend concentrates, it funds further exit infrastructure, which further degrades the shared systems, which further widens the gap between what the exited experience and what the remainder experiences.
The enclave logic operates at every scale where capital concentration is sufficient to fund exit. The walled compound in Lagos. The private city in Manila. The gated development in São Paulo. The special economic zone in Bangladesh that operates under different governance, different infrastructure, different rules than the surrounding territory. These are not the same political or historical phenomenon. They are the same economic logic applied in different contexts, producing a built environment that is physically recognizable across contexts that are otherwise completely different.
Renee is presenting to a regional planning commission about a transit extension that would connect a job-rich suburb to the existing MARTA network for the first time in the system’s history. The suburb’s representatives are in the room. They are polite. She has given this presentation four times in eight years. Each time the response has been the same: acknowledgment of the analysis, interest in the concept, concern about cost and local control. The map is on the screen behind her. MARTA’s coverage area and the county boundaries. Everyone in the room has seen it before. Some of them have been in the room for all four presentations. She is no longer presenting to change the outcome of the vote she knows is coming. She is presenting because the map needs to be in the room, on the record, so that the decision being made is made with full knowledge of what it means. The people in the room know what the map means. They are not confused. The question is whether knowing is the same as deciding.