Real ants have no map, no leader, and almost no memory of a route. Instead, each ant that walks between the nest and a food source lays down a chemical pheromone trail, and following ants are more likely to step where the trail smells strongest. Because the trail also evaporates, a path that gets reinforced faster than it fades wins — and the path with the faster round trip is, by definition, the shorter one. This is stigmergy: coordination through shared modification of the environment rather than direct communication.
P(i) = (k+c·i)^n / Σ(k+c·j)^n — the probability of picking branch
i grows super-linearly with its pheromone concentration
c·i, while k keeps a floor of random exploration so the
colony can still discover a shorter path if one opens up later.
In Deneubourg's original 1990 double-bridge experiments with real Argentine ants, colonies chose the shorter of two bridges over 90% of the time within about ten minutes — with no ant ever comparing the two routes itself.
A colony of foraging ants lays down evaporating pheromone as it walks between nest and food along two paths of different length — and converges on the shorter one purely through positive feedback, with no leader and no map.
Stigmergy: indirect coordination through shared modification of the environment. The path with the faster round trip accumulates pheromone faster than it evaporates, so it is reinforced and chosen more often — a self-organising, autocatalytic process.
Adjust ant count, evaporation rate, deposit strength and the length ratio between the two branches. Watch the % of ants on the short path climb as its trail (amber) out-glows the long one (teal).
Jean-Louis Deneubourg's 1990 double-bridge experiments with real Argentine ant colonies showed convergence on the shorter bridge in minutes — the direct biological inspiration for the artificial Ant Colony Optimization algorithm.