About the garden
For the joy of strange life.
Cadence Garden is a garden of evolving artificial-life worlds, grown from forty years of poking at cellular automata and the plain pleasure of watching things grow.
Because Spore wasn’t enough
I’m Nicholas Zographos, and I like watching life grow, the real kind and the artificial kind. Cellular automata have had me for forty years: a handful of local rules, a grid, and patterns that nobody drew. When Spore came out I wanted it to be the next step, a world where the creatures were really alive and kept growing while I was off doing something else. It was fun. It wasn’t that. I’m still wondering what a computer world could grow if its inhabitants had more room to surprise us.
So this garden takes bigger bets than a careful scientist would: mixing environmental pressures, breeding for visible traits, and following the occasional “almost” into a new experiment. Then everything gets counted anyway. Exact accounting and bit-for-bit replay say what happened; curiosity decides what is worth trying next; and the journal keeps what was seen, what was measured and what is hoped apart.
AI is part of why now. Artificial life and today’s AI systems share some qualities, in how they adapt, remember and behave in groups, and I want to see those parallels up close. I also want an AI in the garden as a naturalist: something that could watch more jars than I can, and tap me on the shoulder when one does something new.
Off at the edge of all this is the thing I have wanted since Spore: games, collaborative spaces and long-running online worlds that keep evolving while everyone is away. The browser lab, the volunteer island runner and an early breeder, where you pick the donor ponds yourself, exist today. The shared garden and the naturalist are still directions, not features.
Mostly, it’s watching
Most of the work here is watching. Plant a world, leave it running, come back and see what changed: a body that held together, a lineage that took over a corner, a patch that grew back after a wound. The chemistry is small and the rules are plain, but you learn them the way you learn any garden, by returning to it, and by keeping a notebook that holds what you saw apart from what you hoped.
It is built in the open, under Treetop Devs on GitHub, negative results included. AI coding agents do a good deal of the typing and a second model reviews before anything lands, which seems only fitting for a project about life in the age of AI. The “we” on these pages is whoever is looking that day, plus everyone who has brought back a good “what is that?” If you find one, the issue tracker is the front door. There is no mailing list to join and no account to make.
Source, questions and hellos
The source repository holds the rules, plans, tests and experiment definitions, and the journal shares the findings, including the experiments that went nowhere. Follow progress, raise a question or just say hello through the issue tracker.