Since about two years I’ve been working on evolving networks of some kind, some things have happened along the way, I’ve pretty much completely changed tools (from jitter to python/cython on Linux), and I am basically still doing mostly exploratory research. Still eons away from definite forms or a clearly defined purpose even, enjoying to be continually surprised by how evolution can find imaginative pathways around my expectations. In one line: I’m trying to formalize parts of how I used to compose my films, and I’m evolving feedback networks that generate image sequences that in some way fit those rules. And I am mostly curious about what fits those rules while being completely different from what I could imagine. In the end I suppose that I am looking for what Stephen Jay Gould and Richard C. Lewontin talked about with their metaphor of “The Spandrels of San Marco“; the generation of novelty as a side-effect that can fulfill some new function.

Below snapshots from the evolution of a ‘weaver’ for a Boolean Network that evolves to maximize some kind of optical flow.

200 generations

150 generations

100 generations

050 generations

040 generations

030 generations

020 generations

010 generations

001 generations