Each creature carries a heritable phenotype trait (body size, mapped 0–1 and shown as sphere size/color) plotted on a fitness landscape — a colored terrain whose height encodes reproductive fitness. Every generation, creatures near fitness peaks are more likely to reproduce; offspring inherit the parent's trait plus a random mutation.
R = h² · S
S = selection differential (mean fitness parents − mean population)
h² = narrow-sense heritability (≈ constant here)
- Selection pressure — how sharply fitness favors the terrain peak; higher values kill off low-fitness individuals faster.
- Mutation rate — standard deviation of the random trait change passed to offspring each generation.
- Population size — number of instanced creatures simulated; larger populations reduce genetic drift.
- Play / Step gen — auto-advance generations continuously, or step one generation at a time.
This is the core logic behind real breeding programs and Darwinian selection experiments (e.g. Lenski's E. coli long-term evolution experiment).