THE MAZE
Can a colony discover the shortest path?
A procedural maze. Nest on one side, food somewhere else. No ant knows where the food is, no ant knows the maze, no map is provided.
Can a colony discover the shortest path?
71×41 procedural maze with loops, one nest, one unlimited food source, 350 ants.
Ants sense only neighbouring cells. Searching ants lay a home trail; ants carrying food lay a stronger food trail that fades with trip length. Other ants follow trails probabilistically. Trails decay every tick.
The colony's typical food→nest trip reaches at least 80% efficiency against the breadth-first shortest path.
Theoretical shortest path computed by breadth-first search (a global algorithm no ant has access to).
Grid movement, simplified pheromone model. Shows that local reinforcement can approximate a shortest path; does not model real ant biology.
Can thousands of local decisions approximate a global solution?