The Apparatus as Dialogic Partner

In the previous sections we have examined the various forms of interactions between the human performer and other improvisers, as well as the apparatus. The discussion now shifts to the perspective of the apparatus. How is the

apparatus able to listen and speak? How may it interact in a dialogic way?

3.6.1 The Problem of Computational Perception

The most basic condition for dialogue is the ability to hear and listen. In other words, to perceive someone else speaking. To consider the programmable computer capable of a subject-subject relation suggests that it should have that potential for being self-aware, independent and able to respond just as a human. Cybernetics (von Foerster, 2003) describes this functionality in living beings as

autopoietic: a self-regulating system which is capable of computing its own

organisation. The autopoietic system is open to its environment, but organisationally closed. Obviously, this model does not straight-forwardly apply to computers.72 _{However a computer may approximate such a systemic behaviour} and then also become able to interact dialogically with an algorithmic performer. As illustrated in Fig. 7 (meso level interaction), the communication between algorithmic performer and apparatus is asymmetric in relation to the improvising group. While the performer is listening to the musical improvised activity, she interacts with the apparatus through program code. On the opposite side, the

72_{The ongoing development of artificial intelligence and robotics may eventually close this gap.}

improvising performers perceive both, the visual-embodied coding as well as the sonic activity of the apparatus as the contributions of the algorithmic performer to the improvisation. I have termed this configuration the 'mediation triangle' for how the communication works in indirect ways between group and apparatus, mediated by the performer – as a kind of 'human filter'.

3.6.2 The Problem of Re flexivity and Action (Responsibility)

The second challenge for the dialogic apparatus is the condition of reflexivity and responsibility. These are similarly important elements of dialogic interaction. It is the ability to perceive oneself 'through' the other, or rather: in relation to the other. Reflexivity is based on perceptual and interpretive abilities at the same time. It requires an awareness of oneself as well as the perception of the other. This reflexivity lies at the basis for every act of meaning making (to interpret an utterance in relation to oneself). Furthermore, without reflexivity it may be impossible to decide when to contribute to an improvisation (kairos). Following this standpoint it would not be possible for the apparatus to participate in an improvised dialogue in the way a human performer would respond. Again this suggests the performer as a necessary mediator between the apparatus and the performance environment in order to create an 'algorithmic' self or identity of the apparatus. The concrete affordances of a particular performance situation in regards to the interactivity and agency of the improvisers are difficult to imagine outside of the immediate event. Following Kanellopoulos (2011: 122):

the process of improvisation creates an attitude of consciousness and a mode of being in the world where art and life are united through a deep sense of responsibility to sounds and musicians.

I propose that a reflexivity or responsibility of the apparatus (toward the improvising group) can only be developed adequately in collaboration with a programming performer (who senses deeply) in the situation of performance. The programming performer then interacts with the apparatus to create its algorithmic responsibility in the live. In other words: the performer holds various responsibilities through the different modes of interaction: That of the functionary- performer (her own) and that of the programmer-performer (of the apparatus). Algorithmic kairos, or the ability of the apparatus to act reflexively and self- responsibly, does not exist prior to performance, but needs to be constructed and

refined during performance – driven by the programming interaction of the performer.

Beyond this direct intersubjective action there is the linguistic dimension in which an algorithmic kairos may well contribute to a dialogic apparatus interaction. On the syntactic level of coding automatic suggestions for code words may be proposed by the apparatus to the typing performer. This can propose unexpected directions in the programming – a form of auto-completion. This function is significant for live coding efficiency during performance. In my practice this has facilitated a quicker development of live coded sound (reducing the afore mentioned delay through typing) and in this way made the immersion in the activity of programming (as flow) easier.

3.6.3 The Problem of Nontriviality (Alterity)

Live coding is a non trivial activity anyhow, as in many cases (like sound synthesis), the relation between code and result is necessarily nonobvious – the resulting surprises or frictions with intuition or convention are a large part of its benefit (equally for learning, art, and science) (Rohrhuber, in Blackwell et al., 2014: 142).

Following Rohrhuber, the notion of the non-obvious or nontrivial relationship between action and result is not only a benefit of live coding practice, but in this context also a central pre-condition for dialogic interaction. If creativity is understood as 'the ability to come up with ideas or artefacts that are new,

surprising, and valuable' (Boden, 2004: 1, original emphasis), non-trivial apparatus

behaviour may facilitate exactly such creative thinking for the performer – or the group. Care needs to be taken for the implementation however. In some cases of Dialogic Coding practice I experienced the exact opposite situation: the very predictable apparatus behaviour constrained my personal creative thinking process. This became particularly evident in the use of repeatedly playing patterns. Part of live coding techniques is the approach of creating players which 'just play' based on given rules, often endlessly because they will be changed and updated continuously – so there is no need to explicitly start and stop them. The problematic consequence of this approach is that the (silent) musical and compositional thinking needs to be performed alongside the endlessly repeating (audible) sound patterns. The repetition of these sounds may draw a performer's attention so as to interfere with the silent articulation of an imagined new sonic idea. More exactly, it distracts the attention needed for completing the translation

processes for articulating this prototypical idea into apparatus language (code). As a consequence, the influence of hearing the repeating loop slows down or even completely blocks the imagination and thought process. It becomes hard to transform the imagined sonic idea into concrete sound.

As this example reveals, nontrivial behaviour may only support the development of new ideas in the process of improvisation when it appears in a non-obtrusive form. At the same time, if the behaviour is strongly predictable it similarly influences the performer's action and thinking process.

3.7. Summary

In this chapter I have outlined the interactions of the performer during improvised group performance reaching out to all agents actively involved. These interactions with the different agents were categorized into four levels. On each level I have identified particular forms of listening, speaking or acting as forms of dialogic performance. Witnessed by a co-present audience, the relationships not only unfold in the close temporal proximity of a live performance event in the improvising group, but they also extend through time and space to past utterances in various symbolic forms: sound, music, code, gesture, rituals. More importantly for this research are the interactions taking place between the performer and the apparatus. I have described how this interaction takes place and how to develop it into a dialogic relationship. This describes the properties of the apparatus in Dialogic Coding practice: It needs to be able to 'listen', 'reflect' in some ways and thus functions in nontrivial ways. These conditions can be created through programming but need to be implemented by the programming-performer carefully and in response to the situation. The viewpoint of the performer and the various roles she performs in the pursuit of creating this dialogic agent will be the focus of the next chapter.