This week’s Futures tale is When the music ends by Philip Ball. Remarkably, given Phil’s long-standing association with Nature, this is the first Futures story he’s penned for us — and it’s very nice to welcome him to the foild. Phil kindly took some time out to explain what inspired his story. You can find out more about Phil’s actiovites at his website.
Writing When the music stops
There’s a long history of attempts to create music by some kind of automated or algorithmic procedure, going back at least to Mozart’s dice-determined assembly of musical fragments (Musikalisches Würfelspiel). The advent of modern computers obviously broadened the possibilities considerably, but aside from compositions tightly constrained by prevailing harmonic and stylistic rules (for example, Baroque chorales — K. Ebcioglu J. Logic Programming 8, 145-185; 1990), the results have tended to be unconvincing: lacking in inspiration at best, bland and formulaic pastiche at worst (J. A. Biles https://www.it.rit.edu/~jab/GenJam94/Paper.html and L Spector & A. Alpern in Proceedings of the Twelfth National conference on Artificial Intelligence, AAAI-94, pp.3–8. AAAI Press/MIT Press; 1994). Computers can come up with endless variety, but it is hard to identify reliable criteria for automating the process of deciding which variants are worth pursuing and developing. In short, computers are even worse judges of their own work than we are of ours.
And this is where I felt things stood, until in 2012 I came across Iamus — a computer developed by researchers at the University of Malaga in Spain that composes in a modern, often atonal classical style. It uses an evolutionary algorithm that takes a small musical fragment, itself generated by the computer, and mutates it according to various rules so as to unfold a fully fledged composition. Or rather, endless variations of them. Somehow the Malaga team, led by computer scientist Francisco Vico, has found a scheme that enables Iamus’s work to evolve towards forms that sound not only plausible but interesting and fulfilling. Opinions about Iamus’s oeuvre vary, as they do about any kind of music. But the results — some of them chamber works, others full orchestral scores, delivered straight from the computer as playable manuscript without human intervention — have been deemed good enough to be recorded by the London Symphony Orchestra for release as a CD. Personally, I quite like it, in moderation. The LSO’s chairman Lennox Mackenzie offered cautious encouragement: the music “went nowhere”, he said, but “it does have something.”
After I had corresponded with Francisco for some time, he invited me to Malaga in July to speak at a short meeting on the future of computer music. Having written a book about the cognition of music, I suggested that it should be possible in principle to identify some of the musical structures that excite human emotion and to build these into the selection criteria used by systems like Iamus for composing. I said at the conference that I’d be persuaded that the age of automated composition had really arrived when I hear computer-generated music that makes me cry. And I would not be surprised to see that happen.
On the trip home I thought over the implications. If we really can understand in an objective way why people respond emotionally to music — and this does seem to be possible to some degree — then there is no obvious reason why “emotionality” shouldn’t be an ingredient of computer music, to be adjusted by a dial just as we might alter the tempo or instrumentation. But what if we solve that problem too well? Francisco is sure that systems like Iamus can take us into areas of musical space that we haven’t even discovered yet. What if one of them is a hyper-emotional domain beyond anything previously composed?
So I sketched out my thoughts for this story on the plane back to London. Bach’s music has always been capable of bringing tears to my eyes, but it was shortly after my return that I stumbled across Purcell’s ‘When I Am Laid in Earth’ from his opera Dido and Aeneas, which still leaves me helpless. Imagine this emotional intensity cranked up tenfold — or a hundredfold! — by algorithmic means. What would it do to us? And yet the thought that this music should ever be banished forever from our ears is no less unbearable.
After that, the story wrote itself.
I should probably add that I don’t seriously think what Francisco and his colleagues are doing is going to hasten the end of civilization. It seems most likely that Iamus and its progeny will be valuable aids to composers – as well as perhaps giving us a superior class of muzak. The story is really about this extraordinary capacity music has to elicit an emotional response, and how helpless we are to resist its poignant charms. And that is something, of course, to celebrate, not to fear.