From the moment I discovered the online service Pandora, I've been fascinated with it. Not with its predictive acumen, really, or with its method of distribution, but rather, with its broader technological and artistic implications. (For those unaware, it's a radio and music recommendation service that, through its algorithms, its database and the selective input of a dedicated user, “learns” the user's preferences and then streams music it deems appropriate.) It's a nifty little piece of software from the Web 2.0 generation, and it offers an novel and fairly effective method of getting one's musical fix, but it won't transform the world of music the way Wikipedia has transformed the world of information.
But its descendants? They certainly might. Its creators suggestively named the underlying technology the “Music Genome Project,” and its mission is to describe music in terms of its elementary particles. This is a noble but deceptively ambitious endeavor in reductionism, and it involves some strong underlying assumptions about the metaphysics of music; can this software, with only a few hundred genes in its vocabulary, actually develop a sufficiently expressive musical grammar? Could any mechanical software, ever? The naysayers argue against it on principle, but I think we should always hesitate to underestimate our own ingenuity. Pandora is only the very beginning, and with the exponential growth of processing power, future iterations will be progressively more sophisticated, nuanced, and automated. And beyond this project lies even more stunning theoretical possibilities; we might someday be able to create computer programs that can effectively compose music—or can at least imitate those who could—and in doing so, might also come to better understand the principles and metaprinciples of the composers themselves.
This is probably less difficult than it initially seems. Consider the relatively prolific J.S. Bach, for instance, who left fewer than two thousand works; a comprehensive analysis of his music and his methods is far from impossible, and, although it would be a daunting task by current computing standards, encoding his principles into an application which convincingly randomizes music based upon them is certainly not inconceivable. And thus, fresh manuscripts from the grandmaster of the Baroque, at our fingertips.
In the distant future, our computers will be able to “render” these projects in real-time. You might at first direct it to “Play something Wagnerian for about half an hour, with those sensational orchestral tuttis at about thirteen minutes and at about twenty-nine minutes” and then, five minutes into the session, suddenly wish to hear a tutti or a cello solo immediately, or simply wish for the work to conclude, and all that will be required to seamlessly insert these effects into the music in real-time will be a bit of fiddling with the knobs. Our computers certainly won't mind this kind of capriciousness, and they might even find it helpful. There will surely also be an evolutionary component to this kind of software; it will synchronize with a global network, check its work against what other computers have devised with similar guidelines, and continue to discover what works best.
In the very distant future, the computer might simply be able to detect your preferences through some kind of neural peripheral, and adjust itself accordingly. Just plug in, turn up, and tune out. Perhaps some of the music of the distant future won't involve sound at all—not even the illusion of it. People might well wonder what it was like not to take one's music intravenously, so to speak, and they might pity those who didn't.
In any case, whatever Pandora is, it has clearly sprouted something; I suspect this distant future isn't as distant as it seems.
