How the patterns emerge

The background of this site isn't a video or random animation - it's a cellular automaton on a hexagonal grid: a fixed rule decides, generation by generation, which cell lives next. Depending on the rule, the long-run behaviour looks very different - and falls into a small number of classes.

None of this is new: Stephen Wolfram showed in his "A New Kind of Science" project that the long-term behaviour of cellular automata can be sorted, roughly, into a handful of classes - from rigid/periodic to chaotic. For the hexagonal grid here (code: src/analysis/ruleClassifier.py) the same idea applies, with four classes:

Extinct - the population dies out completely. Periodic - the automaton eventually loops exactly, the same state recurs. Explosive - the cells fill practically the whole grid. Chaotic - neither happens, the pattern keeps moving unpredictably. Measured, not guessed: each rule is actually run until one of the four conditions is met.

Two measured examples

Both run live here, with the same starting cell (grid centre) and grid size that ruleClassifier.py actually used to classify them - not a re-staged example.

Most of the rules that actually show up in the videos and the player are chaotic in this sense - they don't die out, don't repeat exactly, and don't overrun the screen. That middle ground is what ends up sounding most like music.