University of Sussex, Brighton, UK
How might we go about the analysis of musical works (algorithmic compositions, or more fashionably, generative pieces) which are designed to vary with each run, even without any online interactive facility? This paper shall consider some of the issues and make a cautious start at providing some analytic methodology, which I hope that in the context of a conference, we might profitably discuss! I mean to tackle, not necessarily large-scale essentially deterministic and ambient works (like much of Brian Eno’s earlier work pre-Koan or Jem Finer’s LongPlayer, founded on isorhythmic layers) but programs which can vary their output each time they are started. These works might be represented by those in ‘The Algorithmic Stream’ or rand()% streaming algorithmic radio stations, Karlheinz Essl’s Lexicon Sonate (even in its non-interactive version), my own (over-enthusiastically dubbed) Infinite-Length Pieces (Collins 2002) or indeed, algorithmic artworks in the form of programs produced for various computer music systems from PD to CLM to the SuperCollider demo examples.