||This paper presents a survey of current computational techniques for machine musicianship: composition, improvisation, and performance. It then presents an experiment testing the musical difference between one of the techniques and a similar, 'naïve', approach, to quantify the difference. My hypothesis is that there should be a quantifiable difference between a melodic sequence generated using an event tree versus a melodic sequence generated using 'direct output' when deriving the sequence from the orbit of a of a non-linear dynamic system. I employ 4 different non-linear dynamic systems with chaotic orbits as the basis of the experiment, which are used to generate control and experiment melodies. The musical output of each run is then in 4 dimensions: absolute pitch frequency, average pitch change, average rhythm change, and key/scale position.