Something of a curio: I used Rainwaves this year in the Contextual Studies module as my algorithmic composition, but it was actually done back in 1992 when I was a student myself (at City University, studying Music Information Technology under Jim Grant and Simon Emmerson).
The piece was actually written in Lightspeed C (on the original Mac LC) for my programming module. The finished software outputs four channels of MIDI notes, each generated by a separate equation. However, each of the equations was of a similar type, a variant of the “hailstone equation” I found when reading Clifford Pickover’s book Computers, Pattern, Chaos and Beauty. The main characteristic of an equation of this type is that it initially outputs a haphazard sequence of values which then settle down into a rhythmic cycle.
So: work out a way to map the outputs onto MIDI note numbers; fiddle with the starting values to get some artistically satisfying results; record some sections into a sequencer (Opcode Vision); add an outro. I only had one keyboard at the time (a Korg Wavestation) and so just took the stereo output from that directly into a hired-in DAT machine. Finished.
Interesting. I didn’t do any manipulation of note values at all, and yet it all sounds very musical despite the fact the “melodies” are being generated by equations playing themselves out. Yes, there was a selection process and, yes, there was an element of quantization going on in terms of the choice of notes (insofar as the numbers output from the equation had to be mapped onto the 0-127 range permitted under the MIDI protocol). But otherwise I didn’t manipulate any of the variables or play a single note to make it artificially seem musical: it just is.
22 years old and still fresh as a daisy!