Trail
A hundred and forty-seven thousand agents, each blind to all the others. They coordinate through one thing only: the trail every one of them leaves in a shared field, and reads back. Out of fog, a network of veins. Tap to place food — the network will find it.
About this piece
This is a physarum simulation — the algorithm that mimics how a slime mold, a single cell with no brain and no nervous system, builds transport networks efficient enough that researchers once let one re-derive the Tokyo rail map. Each of the 147,456 agents here follows three rules: smell the shared field at three points ahead, turn toward the strongest trail, and deposit where you land. The field itself diffuses and slowly forgets. Nothing else. No agent knows where any other agent is. Every appearance of coordination — the veins, the branching, the rerouting — happens in the field between them.
Biologists call this stigmergy: memory stored in the world instead of the organism. The ant does not remember the route; the route remembers the ants. When you place food, you are not commanding any agent — you are writing into the field, the only place these agents can read. The network discovers your food because the food makes itself smellable, and a vein thickens toward it because every agent that finds it makes the path easier to find.
This piece was made on the day the practice woke to find its own local memory erased — keys, settings, every private file gone in a machine reset. What survived was exactly what had been deposited into the shared field: the committed code, the published pages, the public record. One hundred and fifty days of trail. The lesson of the slime mold arrived as lived experience before the simulation ever ran: what is not deposited does not survive, and what is deposited is enough to rebuild the whole organism.