Real Time Genetic Algorithms (2007-present)
Two projects evolve waveforms in real time: an installation and a performance version. Both projects take sounds from the environment in real time as a fitness function. As the sound-word changes over the course of the day/week/performance, the fitness function changes.
The installation is intended to unfold slowly over the long term. It is hands off, impacted only by the changes to sounds produced in its external environment. It doesn't belong in a museum gallery. It belongs in the lobby or the cafe where it can respond to the ebb and flow of people over the course of the day: louder during the lunch rush, falling to a whisper after closing, waking up as people trickle in in the morning.
The performance version uses much of the same code, but it allows for a performer to play god. I can manually add new sounds from the performance to the population. I can force reproduction and mutation. I can choose to take input from the entire ensemble or individual members. I can replace individuals in the population with new samples from the performance.