C. Code - The Computational Engine of Meaning
The Code dimension analyses the thinking part of a digital work: the algorithms, data structures and decision-making routines that turn a musician’s idea into sound, picture and movement. Together with the other six dimensions it creates a multimodal ecology in which sense, action, representation, sound, embodiment and computation constantly influence one another.
Key Analytical Concerns
Four coupled concerns structure the analysis. • Reasoning Presence examines how the system’s perceived “mind” makes choices and how intelligible those choices are (examples include neural-net score generators, weighted probability tables and echo/obliterate functions). • Graphical Behaviour & On-Screen Imagery asks how algorithmic parameters are rendered visually and read as musical instructions, with artefacts such as animated polygons, colour-coded noteheads, scrolling scores and dynamic UI elements. • Algorithmic Indifference / Ignorance explores when and why the system refuses to respond and its affective impact. • Conditional Networks & Structural Randomisation studies the layered conditionals (e.g. probability, timing, sensor data) that govern material flow; typical artefacts are Max/MSP-TouchDesigner patches, brain-computer-interface loops and “Time-Out” hand-sign overrides.
Corpus analysis surfaces several recurring motifs: • Reasoning Presence & Ambiguous Intent shows performers treating the system as a quasi-sentient partner, prompting a constant “why did it do that?” mindset. • Visualisation as Semiotic Code reveals that colour palettes, animation speed and shape become loaded symbols that directly shape expectations and decisions. • Algorithmic Indifference & Anxiety highlights that threshold-based silence or ignored input can generate performance anxiety, foregrounding non-human agency. • Conditional Networks & Structured Randomness demonstrate how layered probabilistic and sensor-driven conditionals fuse fixed directives with chance, sharpening improvisational acuity and producing a dynamic dramaturgy.
Together these insights suggest that code can function as a performative mind, that visualising its state can bridge the gap between numbers and musical intent and that curated randomness and technical limits can be integral to the artistic outcome.
Sub-field analysis
These are the main themes that emerged from the DigiScore corpus analysis. They strongly indicate that the Code dimension operates as the engine that converts abstract compositional intent into behaving agents; its design decisions shape the material constraints (what can be done) and the interpretive affordances (what can be read). Each sub-field is introduced with a short description followed by the key themes that emerged from the corpus. Themes are numbered C1-C21 for ease of cross-reference.
A. Reasoning Presence & Ambiguous Intent
Performers repeatedly characterized the system as a quasi-sentient agent that “listens”, “decides”, occasionally “ignores”. In one example, the algorithms blended probabilistic weighting with extensive habit models, for example, derived from 502 days of rehearsal recording. In this example, the system’s responses were informed both by stochastic processes and by long-term performance patterns offering a sense (or perception) of familiarity. A similar approach was used in the Jess+ project and the Solaris score. This hybrid behaviour forced musicians into a meta-cognitive mode of listening, prompting the constant question “why did it do that?”, or “did it do that because of its training?” and thereby expanding their interpretive strategies far beyond conventional cue reading.
Key themes. • C1 - Hybrid stochastic habit reasoning: Output is shaped by both random processes and statistical patterns harvested from, say, the performer’s own history. • C2 - Interpretive uncertainty: Ambiguous decisions provoke on the spot explanations, expanding the performer’s interpretive toolkit. • C3 - Agency attribution: Musicians begin to treat the code as an intentional partner rather than a neutral tool.
B. Visualisation as Semiotic Code
In many scores, colour-coded noteheads, dynamic polygons and scrolling scores encoded core sonic parameters (e.g. amplitude, panning and FM index). Performers described navigating these visual “maps” of gesture and timbre, and how they treat the visual layer not as decoration but as an explanatory interface that directly shaped expectations and decision-making process. Consequently, design choices like colour palettes and animation speed become semantically loaded, steering performers toward particular interpretive and perceptual pathways.
Key themes. • C4 - Visual semiosis: Design choices (hue, size, motion) become semantic wrappers that translated raw numbers into perceptible musical meaning. • C5 - Predictive visual scaffolding: Real time visual feedback guided performers’ expectations and decision making, reducing cognitive load. • C6 - Aesthetic steering: Colour and animation can subtly biased performers toward particular interpretive pathways.
C. Algorithmic Indifference & Anxiety
Several systems employed threshold-based suppression of interactivity, deliberately ignoring a performer, sometimes with outright silence. Experiencing this selective “ignoring” did provoke performance anxiety in some, yet it also foregrounded the non-human agency embedded in the algorithm, prompting musicians to reconsider established control hierarchies and their own position within the performative network.
Key themes. • C7 - Selective gating: Threshold based suppression created moments of silence or visual dimming that acted as negative feedback. • C8 - Anxiety as a reflective catalyst: Being “ignored” pushed musicians to renegotiate control hierarchies and reflect on their role in the network. • C9 - Agency asymmetry: The imbalance between performer initiative and system response became a productive tension rather than a technical flaw.
D. Conditional Networks & Structured Randomness
Multiple layers of conditionals, ranging from Max/MSP randomisation and weighted probability distributions to “Time Out” hand-sign overrides, interweaved fixed directives with stochastic elements in multiple systems. This interplay generated a dynamic dramaturgy in which the performer attempted to remain perpetually alert, thereby sharpening improvisational acuity and fostering a heightened sense of co-creation between human and code.
Key themes. • C10 - Layered conditionals: IF/THEN rules combined stochastic weighting, sensor data and temporal constraints into a single decision pipeline. • C11 - Curated randomness: Structured probability offered surprise while preserving a sense of control for the performer. • C12 - Real time negotiation: The performer became a co-creator who continuously adapted to emergent algorithmic choices.
E. Composer as System Designer
Many musicians felt that their role shifted from author of a fixed score to “designer of chance”. Tools such as Decibel ScorePlayer, AutoConductor and bespoke Max patches become creative partners, expanding compositional practice to encompass software engineering, probability design and system architecture. The boundary between composition and system design was felt to be deliberately porous.
Key themes. • C13 - Design as composition: Writing code was integral to the compositional process; the score becomes a rule set rather than a static object. • C14 - Porous boundaries: The line between musical notation and software implementation was deliberately blurred. • C15 - Collaborative agency: Composer, code and performer formed a creative network.
F. Learning from Data / Machine-Learning Habits
In several scores, deep-learning models were trained on extensive performance archives in order to extract habitual gesture patterns, while speech-analysis AI uncovered hidden structural cues. Additionally, data-driven generation produced personalised material that could reveal previously unknown performer traits, enriching both analytic insight and creative output through a feedback loop of habit detection and generation.
Key themes. • C16 - Habit detection: Machine learning models created a statistical portrait of the performer’s gesture vocabulary. • C17 - Personalised generation: Algorithms generated material tailored to the individual’s habitual repertoire. • C18 - Feedback loop: Generated material informed subsequent performances, creating a closed learning cycle.
G. Pragmatic Constraints & Technical Failure
Practical limits such as fixed sampling intervals (e.g. a 10-second sample window), patch crashes and latency issues imposed concrete constraints on the performance that had an impact on their musicking. These real-world limitations shaped artistic decisions, often prompting a simplification of gestures and drew attention to the materiality of code itself as a performative constraint that must be negotiated alongside artistic intent.
Key themes. • C19 - Materiality of code: Latency, buffer sizes and crash behaviours were audible/visible properties that influence artistic choices. • C20 - Graceful degradation: Designers embedded fallback behaviours (e.g. fade outs) to preserve musical continuity. • C21 - Constraint driven creativity: Restrictions inspired novel gestural vocabularies and compositional strategies.
Takeaways
These themes repeatedly surfaced across many of the diverse works surveyed and together outline that the code can be as a co-creative, semi-transparent agent whose logic, opacity and visualisation shape aspect of the entire performative ecosystem. Emerging from this analysis process we could claim that:
- Code is a performative mind. Algorithms are experienced as intentional agents, prompting performers to ask why and to attribute meaning to statistical events.
- Visualising code-state bridges the gap. Colour, size, motion and animation act as semantic wrappers that translate raw numbers into understandable musical cues.
- Randomness is curated. Structured probabilistic networks give both surprise and a sense of control; the performer’s role becomes a real-time negotiator of chance.
- Composer becomes system architect. The act of writing a score now includes writing the rules that will govern the piece, aligning compositional and software design practices.
- Pre-recorded material gains agency via code. Fixed audio samples are re-animated through conditional DSP, turning “dead” media into active partners.
- Technical constraints are musical constraints. Sampling intervals, latency and patch stability shape the pragmatic aesthetics of the work.