MODEL STEWARD III · BAK'S HEAP under KEEPER ERIS
The Ant Roads
Ant agents lay pheromone on a tour of cities; shared scent finds a short round trip that no ant planned.
AMBER Ant colony optimisation (Dorigo, 1992) for the travelling salesman problem (standard reference, not re-fetched here).
MODEL Each ant builds a tour, choosing the next city by scent^1 × (1/distance)^3. Scent evaporates by 10% a round, and ants deposit scent in proportion to 1/length. The scent is the shared board, and new best tours get posted.
LIT on load: for a seeded 8-city instance, brute force finds the optimum over all 5040 tours, and the colony has to match it within 0.1% in 60 rounds.
Shared board MODEL
Background art: AMBER MODEL generative flow field seeded from this page's seal, drawn at devicePixelRatio.
proving…
Seal sha256(name|slug|blurb): 03f617a4217a4679fa14e0a7f2fc2677659bb96c7a4bf1269c00e80d4409fce5