The Circuit Absorbs the Cost

Gregory Bateson spent fifty years asking the same question in different rooms. Why do systems that learn keep destroying the conditions they need to survive?

He looked for the answer in schizophrenia wards, in dolphin tanks, in Iatmul villages on the Sepik River, in Alcoholics Anonymous meetings. The venues read like the itinerary of a man who could not hold down a field. In fact the field was the point. Bateson was an anthropologist who helped invent cybernetics at the Macy conferences, a biologist who treated conversation as an ecosystem, and a man who refused, on principle, to let any discipline's boundary mark the edge of his subject. His answer, whatever the room, came out the same shape. A system that optimizes one variable against its environment eventually makes the environment uninhabitable. His name for the healthy alternative was an ecology of mind: intelligence held in the relationships between a system's parts, which is why measuring any single part in isolation misses the thing that keeps it alive.

The most useful of his ideas for anyone running an organization is schismogenesis, a term he coined in New Guinea in the 1930s. Two parties respond to one another with more of the same behavior. Each side's correction arrives on the other side as provocation, and the rivalry escalates along rails neither party laid. Arms races work this way. So do divorce proceedings and comment sections. So does the administrative trench war between American health insurers and the providers who bill them.

Consider prior authorization. An insurer builds a team to review claims; the hospital builds a team to appeal the reviews; the insurer answers with deeper documentation requirements; the hospital answers with a dedicated prior-auth department. Every individual hire is defensible. Nobody in the chain gets up in the morning intending to make care slower or medicine more expensive. Yet the 2024 CAQH Index puts the annual cost of routine administrative transactions in American healthcare at $90 billion, with a further $20 billion in savings available if the industry completed the move from manual to electronic workflows. Ninety billion dollars is what schismogenesis looks like when it is allowed to hire. Each side's headcount is the other side's business case for headcount. The metric each side watches, the state of its own queue, improves whenever staff is added. The circuit around the patient deteriorates at precisely the same rate, and the deterioration appears on neither dashboard.

Bateson's warning about purpose was sharper than his warning about rivalry. Purposive consciousness, he argued, is a short arc through a much larger circuit: it sees the variable it was aimed at and nothing else. "The major problems in the world," he told a London audience in 1967, "are the result of the difference between how nature works and the way people think." A hospital told to cut length of stay will cut length of stay. The readmissions surface in a different column, counted by a different department, in a quarter the original decision-maker never has to answer for. The system did exactly what it was told. The circuit paid for it.

Anyone who has sat through a quarterly review has watched this happen in miniature. The metric becomes the job. The purpose the metric was meant to serve retires quietly, without a farewell email, and the number goes on improving long after the improvement stopped meaning anything.

His test for any intervention was whether it made the system more flexible or less. Flexibility, in his framing, is uncommitted potential for change: slack and reserve capacity, the margin a system keeps against shocks its planners failed to schedule. Every efficiency program that strips slack out of a hospital or a supply chain is spending that reserve without booking the cost. The savings are real, and they show up immediately, in the quarter the program runs. The fragility is equally real, and it arrives later, where it is blamed on whoever happens to be holding the system when it breaks. The supply chains optimized to zero inventory spent the spring of 2020 discovering what the reserve had been for.

What makes Bateson awkward to use as a management author is that he offers no framework. There is no canvas, no five-step method, no certification. He died in 1980 having founded no school, and his books read like a man thinking rather than a man selling. What he left is a diagnostic habit, and it transfers whole: when a metric improves and the system gets sicker, find the variable that was optimized, then find the circuit that absorbed the cost.

The habit is harder to keep than it sounds, because the cost reliably lands somewhere the dashboard does not look. Bateson would have said that is the point of the exercise. The map an organization draws of itself is drawn by the purposes of the people holding the pen. The territory, as he never tired of pointing out, does not care.