In May of 2008, Patrick Flanagan learned a hard lesson about computer music: no one cares about your algorithms. After months of research, coding, and practice, Patrick debuted his implementation of a music learning and generation algorithm called Factor Oracle. This algorithm allowed his computer to extrude new music based on statistical patterns in bebop solos, Stevie Wonder songs, and Slayer guitar riffs. The audience was curious, maybe even impressed, but mostly confused. The connection between the knob-twiddling on stage and the barrage of notes issuing from the speakers was opaque; more than that, the logic behind the whole performance was too complex. Patrick decided that better algorithms alone wouldn’t save live performances of music driven by artificially intelligent musical agents; the agents needed to be embodied.