The Extended Composer - Computers and Creativity - page 176

music-making can now be performed using a single digital device: from recording,
arrangement and production, through to networked collaboration and distribution to
listeners.
Lubart ( 2005 ) proposes a loose taxonomy of the roles that we can metaphorically
consider a computer as playing within a creative context: as a nanny , taking care of
routine tasks and freeing up our cognitive faculties for the real creative grist; as a
penpal , aiding the process of communication and collaboration; as a coach , provid-
ing exercises and collections of related knowledge through a targeted database sys-
tem; and as a colleague , working as a “synergistic hybrid human-computer system”
to explore a conceptual space in tandem. Though some of the associative elements
of the “coach” role are relevant to this discussion, we are here mostly concerned
with the latter case, in which the computer is embedded within the same creative
framework, co-operating to create a work through a succession of interactions, to
form a partnership between creator and computational system (Brown 2001 ).
The capacity for autonomy in computational systems can allow them to oper-
ate with distinct agency in the creative process, a property described by Galanter
as generativity (Galanter 2003 ). When using generative processes, the artist sets up
a system with a given set of rules. These rules are then carried out by computer,
human, or some other enacting process. 1 A purely generative work involves no sub-
sequent intervention after it has been set in motion; a work with no generative ele-
ments has no capacity for autonomous action, and so requires continual intervention
to operate.
The class of systems that we are interested in lies somewhere between those
which are purely generative and those which must be manually performed. Such a
system is interactive ; it does not produce output which is completely predictable
from an artist's input, nor does it simply follow only its internal logic. The output of
such a system follows somehow from the previous marks of the artist (and, in some
cases, the computational system itself), but its output is mediated through some
predetermined structure or ruleset. A prominent example is François Pachet's Con-
tinuator (Pachet 2003 ), which captures the performance of a user and subsequently
plays it back under some statistical transformations.
Systems capable of such creative interactions can be described as having agency .
Philosophically, agency is typically aligned with intent, goal-based planning, and
even consciousness. It is not this strong type of agency that we are attributing to
generative art systems. We have in mind a broader, anthropological description of
agency, closer to that provided by Gell ( 1998 ) in relation to art objects. Here, agency
is attributed to anything seen to have distinct causal powers.
Whenever an event is believed to happen because of an 'intention' lodged in the person or
thing which initiates the causal sequence, that is an instance of 'agency'. (Gell 1998 ,p.17)
1 For examples, see the crystal growth of Roman Kirschner's installations, Hans Haacke's Conden-
sation Cube ( 1963-65 ), or Céleste Boursier-Mougenot's Untitled ( 2010 ), in which zebra finches
are given free reign over a gallery of amplified electric guitars.