Astronomers who simulate galaxies do not always get the same result, even when they start from identical conditions. New research from Leiden University shows that this is not a flaw, but a consequence of how galaxies behave—and how they are modeled.