The Minimum Viable Agent World

Picture three settlements in a small trading game. One produces energy, another repairs tools, and the third stores records. Agents exchange resources, send messages, and form groups. A trade has consequences tomorrow because ownership and balances persist between rounds.

The designer supplies the basic mechanics: what can be owned, transferred, remembered, or destroyed. Agents decide how to organise within them. They might keep a shared price list, appoint a record keeper, or agree on a way to settle disputed trades.

I would begin with a task that makes exchange useful, such as maintaining a common transport link. The proposed experiment would then introduce a shortage in one settlement. That disturbance would test whether existing agreements help the population adapt or become obstacles that participants work around.

Authority could be distributed in several ways. One settlement might control all records; each might govern its own trades; or several groups might share responsibility. The last arrangement illustrates polycentric governance: multiple centres of authority operating alongside one another, sometimes with overlapping responsibilities. Ostrom’s work provides a starting point for comparing such arrangements.

The measures would include successful exchanges, unresolved disputes, dependence on intermediaries, and the survival of shared rules after a shortage or participant replacement. Repeated runs with different starting conditions would help distinguish a recurring pattern from a memorable story produced once.

This is where the civilisation question sits within Games. An institution needs some continuity: records, expectations, or procedures that remain useful beyond the immediate encounter. A small persistent world would let us examine that continuity before describing a much larger artificial society.

Further reading

Elinor Ostrom, “Beyond Markets and States: Polycentric Governance of Complex Economic Systems” (2010).