randomness in Nature does not exist then telepathy is possible.”
(Figurska, Stańczyk, Kulesza, 2008, p. 185)
Humans are so predictable it's pathetic. In the banal world of music, “music which is indeterminate as regards to performance” risks contamination from the endless well of disappointment which is the human condition. The indeterminacy supposedly gained by the utilization of human choice in this type of music may ultimately become as fixed as “determinate” music, due to the increasing ability to actively model human behaviour.
To take a simplistic example, Karlheinz Stockhausen's Klavierstück XI consists of a set of nineteen isolated musical fragments spread out on a large sheet of paper. The instructions state that a player let their eye wander at random over the page, performing the fragment on which their eye settles, at a speed, dynamic and touch from a selection of options, then letting their eye wonder randomly until it settles upon another fragment, playing it, and repeating this process until one of the fragments has been played three times (Maconie, 1976, p. 100). However, research into eye-tracking has shown that the way in which people scan a page is not random at all and, in fact, is highly predictable.15 Given the limited amount of possible progressions through Stockhausen's work, if an eye-tracking analysis was carried out on how performers' eyes scan over the piece, using a large enough group of people, and this was then combined with a sophisticated enough statistical model16, it would be possible to calculate the order in which people would perform the work to a high degree of statistical certainty. The score could then be essentially “fixed” and made determinate. These modern applications of combinatorics and eye-tracking mean that the agency of the performer is severely undermined due to the ability to use computers to predict their performance of this open score. With the right amount of data and the correct modelling tools, it is a possibility that many indeterminate works lose their political edge and become simply articulations of human predictability, rather than of rational agency. Or, imagine a piece of experimodern music in which, at a particular moment in time, the score asks the performer to make choice from one of nine options, each given equal value in the score. Here the performer functions as a random number generator, despite the well documented inability for humans to successfully create random strings of numbers. Currently, an algorithm is able to predict any human's next “random” number in a series after the enumeration of just seven numbers, with a 45% success rate.17 (Schulz, Schmalbach, Brugger, Witt, 2012). The same research also presented an algorithm that could identify an individual's pattern of random numbers over those of others. Interestingly, the same reasons that prevent humans from being good random number generators and make their strings of numbers so recognizable: “cycling, seriation and repetition” (Schulz, Schmalbach, Brugger, Witt, 2012, p. 1) are the same pattern-based techniques that form the bulk of conventional musical composition techniques, exactly the type of order that the intervention of a human agent is designed to dismantle in the moment of performance.
Similarly, the decision to play at a specific point in a Cageian time-bracket is essentially a procedure of random-number generation within bounds, and thus, equally predictable. The increasing predictability of human beings implies that much of Cage's indeterminate output could (and will be) made determinate, given the right data and modelling18. In many ways, the use of performers as active agents in the shaping of indeterminate music makes it more predictable and, in fact, builds in the very structures its construction was meant to avoid: “cycling, seriation and repetition”.19
One is reminded of Cage's comments on the “problems” of improvisation: “Improvisation is something that I want to avoid. Most people who improvise slip back into their likes and dislikes, and their memory, and they don’t arrive at any revelation that they’re unaware of” (Feisst, n.d. p. 3 - quoting Turner, 1992). Big Data-style analysis removes any revelations. Whilst some of Cage's indeterminate pieces may ultimately be made determinate, improvisation itself, as a discipline, is also susceptible to the modelling of the behaviour of its performers. The analysis of keyboard usage in Beethoven's piano sonatas , seen later in this publication, shows that key presses across the keyboard tend towards a normal distribution, underlining the predictability of human interaction with an instrument. This moment of performer/instrument connection is the point at which the instrument imposes its own topological and haptic priorities upon the player who, in the moment of improvisation activates a behaviourally-conditioned psyche that collapses all possibilities to near zero – a singularity of maximum predictability; not a moment of freedom, but a statistical peak in a computer-cluster made sounding. Whilst, much determinate music thrives on the predictability of its performers, indeterminacy and improvisation may suffer from an ontological crisis following the encroachment of the Petabyte Age into contemporary music practice.
Big Data could mean the end of the idea of indeterminate music, as humans become ever more predictable and its ontological space collapses, simultaneously undermining much of the rhetoric around the politically empowering aspects of itself and improvisatory practices. In-the-moment musical choice becomes the compulsive Tourette-ic glossolalia of a pornographically and pop-culturally conditioned desire that is as predictable as Big Data makes it easy to predict. But, then again, how aesthetically interesting is it to model utopia, anyway?
Outside of Big Data, the construction of ever more sophisticated algorithms for writing Contemporary Music applies pressure from the aesthetic side of things. Recently, Iamus, a computer cluster at the University Of Málaga has been able to produce increasingly convincing pieces of contemporary composition in 8 minutes and with no human intervention.20 The success of Iamus is a testament not to the sophistication of its evolution-based algorithm, but to the predictability of most contemporary composition21.
Just as Big Data and the modelling of behaviour is a challenge to human agency, interventionless algorithmic composition is a challenge to the unpredictability of compositional aesthetics. However, projects like Iamus also point towards a brighter compositional future. As the algorithms become more sophisticated and able to more accurately stylistically replicate the works of humans, they become invaluable not in their ability to produce innovative compositions, but in their ability to stylistically copy, and then produce multiple versions in massive numbers. Once widespread enough, and especially once these algorithms have anonymously (and embarrassingly) won a few composition prizes, they will act as a destabilizing force that readjusts the balance of supply and demand in the compositional market place. The consequence of this will be that, just as other commodities saw massive drops in their value as their production became automated, the mass-production 22 of the type of schlocky, uninteresting (yet well-crafted) festival music that these computer clusters are best at re-creating, will rapidly and irreparably drive down the worth of such music composed by humans, flushing it out of the system and aesthetically improving the eco-system of New Music. This will, in turn, spur composers on to new, undreamt-of creative spaces, which a computer would be unable to predict; although it would always be there, gradually chewing up the value of works as it subsumes them into an ever-expanding stylistic repertoire....
The composition of our desire and sexuality becomes the way of circumventing Big Data's subsumption of human agency, reconfiguring our psychosexual nexus on an individual level to renew the possibilities of indeterminacy and improvisation, as behavioural predictability encroaches on their very efficacy and threatens them with non-functionality.