Chimpanzees, Wind-chimes, & Algorithmic Composition.

Despite a generally large interest in procedures for music-composition, such an issue is far from often seen on the program at International Computer Music Conferences.  However, last year at ICMC in San Jose, California, this topic became the focus of a lively two-hour panel discussion when a special event on algorithmic composition was launched during the conference's last day.  It surprised me to see no coverage of this panel talk in Computer Music Journal's review edition (17:2 Summer 1993). 

The panel for the discussion included Max Mathews, Gareth Loy, Larry Austin, Kevin Jones, and David Cope (who also was the moderator).  After Cope introduced the panel members, he spoke at some length about his own fascination with and use of algorithms in music composition. 

Mathews then played an example of an early experiment he did with permutations of a melodic phrase of a popular song.  Although Mathewss example clearly did not reflect current directions in algorithmic music, his 25-year-old experiment proved to be highly amusing.  Unfortunately, instead of accepting this early example for what it was (an icebreaker), it became a handy reference for some members of the audience to argue that algorithmic composition is futile.  To me, this argument had almost the same relevance as a demonstration of the horrible effects of mustard-gas to prove that chemistry is something "bad".  Mathews's intention in playing this example was probably not to advocate some of the quite reactionary arguments of the audience, but rather to emphasize what procedures the composer should pay more attention to when using algorithms in composition.  The impact that this short example had on the discussion should not be ignored and it certainly threw some fuel into the fire, revealing some audience members fairly anachronistic and limited perceptions of contemporary music.  At times during this two hour discussion, I blushed on behalf of those who insisted on bringing the event down to the "what-is-music" level.  In the panel, Larry Austin, Gareth Loy, and David Cope spoke favorably about using algorithms in composition.  (Indeed, Larry Austin stated that he liked Max's piece, even as a composition). 

The discussion then shifted with some help from Stephen Pope towards deterministic versus stochastic models of composition.  It is amazing that even the editor of Computer Music Journal seems trapped in a popular and rampant idea of music composition.  A thread from the cultural life of Sweden crossed my mind during Pope's comment on stochastic procedures in composition; he stated the prevalent view that we all should agree that "sounds from wind-chimes are not music."  I recall how in the early seventies, a couple of journalists, annoyed by modern art, decided to dispose of it (and its adherents) once and for all.  They gave a chimpanzee some oil paint, brushes, and a canvas and then submitted the result to an annual art exhibition in Stockholm.  They were delighted to find the painting admitted by the jury and hung amongst works by credible modern artists.  It occurred to me that some of the people who opposed algorithmic composition during the ICMC session neither understand nor have paid much attention to what impact directional procedures have on stochastic processes.  That's what the two Swedish journalists did when they overlooked the consequences of directions and limitations in the material given to the chimpanzee and of their own decision to interrupt the apes activity when the painting looked like art.  The journalists, not the chimp, were at that moment the artists.  The chimp, who of course lacks artistic skills, would have kept on "painting" until there were no more paint (and even longer).  At the very moment when directions and limits are given to a chimpanzee (or for that matter a stochastic cloud or even the design of a wind-chime), the director (designer, composer, artist, etc.) becomes the creator of that work.  It's here that some of the arguments at this ICMC-panel degenerated into infantile arguments that weren't relevant for the topic at hand, but rather mere opinions on preferences in music. 

Because composers throughout history have tended to utilize the highest technologies available, it is easy to agree with Iannis Xenakis' statement that, "music, by its very abstract nature is the first of the arts to have attempted the conciliation of artistic creation with scientific though."  With this in mind in addition to hearing quite a few references to Xenakis during the discussion it seems contradictory to see so many engineers at a computer music conference, all dedicated to cutting-edge technologies, speaking favorably about traditional and embryonic ideas of music composition.  Unfortunately, it also seems that the engineers, even at the earliest stages of their work, often ignore possible collaboration with the users of their products (i.e., composers).  Instead, their developments or inventions are guided almost entirely by their own rudimentary and biased conception of composition, with the result being perhaps an engineering advance, but a step backwards for music composition.  It happens too often that audible examples by engineers get confused with (and treated as) actual composed music, when they should rather be looked at as demonstrations of new engineering ideas and breakthroughs. 

I believe that algorithmic composition in signal processing is one of the most important areas of new music today.  I hope that panel discussions and presentations such as this one will be repeated at upcoming conferences -- though I hope with more diverse and open discussion from both sides of the computer music community. 


Peer Landa
Stanford, California USA