evolutionary search

Evolutionary search is what I reach for when a problem is too tangled to solve head on: describe what good looks like, then let a population of designs breed, mutate, and compete until something clever falls out. It's optimization with the humility built in, because you're admitting you don't know the answer, only how to recognize it.

On the Cappelletti Pavilion it handed us two lattice topologies we'd never have drawn by hand, both stable, one noticeably lighter.

Fig. 6 · Four design qualities against a fitness function. Three carry a measure and are scored; the fourth has none and leaves the search.

The catch is the fitness function, and writing one is harder than it sounds. Good has to become a number before the search will believe it, so every quality you cannot count quietly drops out of the competition. Lighter is easy. Legible is not.