Live Algorithms

A Live Algorithm takes part in improvised, collaborative performance, sharing the same modes of communication and expression as its partners. Autonomous rather than automated or controlled, the device enjoys the same constraints and freedoms as its human associates. Ideally the machine improviser conforms to the established practice of a medium, capable of imitating and developing shared gestures, behaviours and ideas. Furthermore, the live algorithm would be expected, at times, to contribute novelty and surprise, to experiment and take risks, and to assume leadership. Other performers experience the live algorithm as if it were a human, with a sense of validity and belief. This idealised concept of a live algorithm was articulated by the Live Algorithms for Music (LAM) research network, building in the work of a small number of researchers who have pioneered autonomous music systems. LAM is concerned with a particular type of live algorithm, those that are capable of performing free musical improvisations with other musician. (It is understood that the improvisations may be vocal but do not involve words.) This musical domain is ostensibly an easier route for the study of performance algorithms since bodily presence, movement, facial expressions etc. have only secondary importance, and the prohibition of language, as is common in this domain, apparently liberates the system from vast semantic problems. The aim, it is stressed, is not to reproduce the capacity of a human performer. Instead, we wish to explore a mechanical, algorithmic world, expanding our own potential for music making. Although designing a live algorithm with the ability to imitate and develop shared ideas is already a formidable undertaking, the additional requirement of innovation is an even harder research challenge. Without the capacity to innovate, however, the audience and the performers would lose the belief that the live algorithm was truly engaged with the performance and not merely accompanying it. We suggest that it is the ability to innovate that distinguishes autonomy from automation and randomness.