Page 61 - Composing Processes and Artistic Agency

50 The topography of composing work

and selecting musical material. Their five main characteristics are: (1) being
based on rules, (2) consisting of non-arbitrary and (3) calculated operations
that follow a fixed sequence and, (4) in a finite number of calculation steps, (5)
produce a result (see Vempala 2014: 38). Gerhard Nierhaus (2012: 4f.) specifies
different types of algorithms used in the process of composing, such as stochastic
models, generative grammars, recursive transition networks, chaos and self-
similarity models, genetic models, cellular automatons and neural networks.
If we look at their operative effect, we can define other groups, such as algo-
rithms with memory functions, random functions and exclusionary functions.
Using algorithms in connection with digital computers produces a sort of
“assisted composing”. Algorithms deliver impulses for composition and are
therefore epistemically relevant. Elisabeth Harnik, an improvisation musician
and composer, says in an interview (in Nierhaus 2012: 28) that “certain proce-
dures and sets of rules” act like a “counterpart” and “fundamental stimulus”
for her. Furthermore, the use of algorithms increases formalisation, under-
stood as the possible ways of abstracting work processes. It also raises the
level of rationalisation (closely linked to formalisation), understood as the
conscious ordering, controlling and accelerating of work processes.
However, in practical terms, there are limited possibilities of formalisation and
rationalisation. In close cooperation with other composers, Gerhard Nierhaus
(2012, 2015) has tried to formalise composition decisions in an experimental
setting. In a dialogue format, eight selected composers who generally do not
use algorithms to compose articulated their structural ideas for generating
and working on musical material. A program was subsequently developed to
implement the ideas. The composers were confronted with the results by being
given a limited amount of algorithmically generated musical material, which
they commented on and evaluated. Further programming steps and discussion
of the results followed. The aim of the experiment was not to replace the
composers’ artistic intuition by an individually tailored program, but to shift
the focus of music analysis onto the level of intuitive evaluation. The study
shows that even when composers have at their disposal an elaborate set of
formalised rules in the shape of algorithms, they do not necessarily accept the
results it generates, yet nor do they think them trivial.
Algorithms are immaterial tools, which – like musical instruments – can
only be used meaningfully with effective artistic practical knowing. Many of
the fundamental problems of creating still remain, such as having to confront
imponderability, semantic openness, potential alternative choices and con-
tingence during the decision-making process. The deployment of artificial
intelligence is consequently limited. If, hypothetically speaking, the composers’
intelligence in music composition was entirely transparent to them, would
they be in a position to compose faster and better? It is doubtful. The social
and cultural complexity of the creative process cannot be eliminated because
composing is embedded in concrete social contexts, which precede the action
of composing and pre-structure it. Composers are neither completely con-
scious of these contexts, nor can they clearly grasp them. The contexts remain