Can an Institution Survive Its Agents?

Imagine arriving at a repair market where every familiar face has gone. The price list is still on the wall, the jobs are still in the ledger, and someone has left instructions for handling complaints. Could a new group keep the market working?

An agent market gives us a way to try that handover. Its apparent stability might live in the shared records, or in habits that disappear with the original participants.

Consider a proposed simulation in which agents trade repair services. They keep a price list, record completed jobs, and use a procedure when a customer disputes a bill. After a period of trading, replace ten per cent of the agents, then half, then all of them with fresh instances of the same model.

Repeat the replacement under different conditions. One group inherits only the price list. Another receives the transaction history as well. A third also receives the dispute procedure and records explaining past decisions. New agents have no private memory of earlier rounds.

The measures would include time until successful trading resumes, unresolved complaints, and changes in which participants receive work. Comparing these conditions would help identify which parts of the institution carry useful information across turnover.

There is a trap in treating survival as success. A procedure might persist because new agents follow inherited instructions mechanically, even after they stop serving the market. I would introduce a changed repair task after replacement and observe whether participants can revise the arrangement.

An institution that survives turnover and adapts to changed conditions would offer stronger evidence of useful institutional memory than one that merely repeats an old rule. This gives persistence a practical meaning without using population size as a proxy.