1. Introduction
This article presents ideas relating to the creation of computer music using emergent systems based on rules and local interactions. Three practical examples will be presented with a special focus on the integration potentials the adapted systems afford. The discussion includes a reflection on algorithms, interaction, and the behaviour of sound processes. It questions the scope and potential boundaries of computational systems through the space relating compositional practice with the development of generative environments. It outlines the process of developing such systems, but also the act of using them within a dynamic musical context. The goal here is to merge the thinking of sound, systems and sonic development for creative applications and compositional approaches.
Working with emergent systems in computer music highlights the different levels computational behaviour operates within. Processes based on self-organisation and complexity present a promising direction for musical approaches that question levels, scales and the overlap between structure and sound. However, the relationship between algorithmic behaviours and musical results is not necessarily straightforward to establish. Incorporating strategies in music that have evolved without any clear reference to sound can introduce issues regarding the nature and purpose of such mappings. In many cases, algorithmic systems are studied and discussed separately from the outcomes they deliver. General properties are highlighted but not the details of how they relate to possible musical outcomes. Such segregation can involve difficulties with regard to how the different domains are set to connect or communicate. While the domain difference is challenging, it also presents opportunities to rethink ways of organising sound processes and introduce novel applications for generative algorithms. In what follows, an attempt is made to take advantage of these problems for original musical approaches.
2. Mapping and Behaviour
The approaches presented here involve a process-based approach to sound. Sonorities characterised by development, change, and behaviour. Sound material that evolves through possibility spaces and continuous transition. Working with such materials involves interacting with a dynamic flux that spreads across multiple levels and layers of organisation. The appeal of using behavioural dimensions for those applications is noticeable when developing the controls and parameters that drive their development over time. Rather than specifying precise values for parameters such as frequency or duration, behaviour-driven processes are frequently measured through higher-level controls such as density and regularity. Such meta-parameters correlate well with those used in complexity-related algorithms such as Alife or Agent-based systems (Beyls, Bernardes and Caetano 2015). However, these systems have also mostly developed on their own, evolving based on criteria that have little to do with music and sound. A possible reason for using such ideas within musical contexts could be to extend existing sound processes with features that otherwise they would not have. To introduce behaviours and ways of becoming that could facilitate the development of sound material towards solutions that would otherwise not be possible to realise. Considering the contact point of the different domains through the axis of behavioural dimensions enables possibilities for combining and understanding them as occupying the same operational space.
2.1 Interpretation
Emergent algorithms often model natural processes and physical systems. Incorporating these within computer music involves considering what dimensions to model and whether or not their ‘natural’ aspects should be featured or suppressed. How to approach the mapping involves addressing differences but also choices. As noted by McCormack “Mirroring nature involves interpretation and ordering by the artist. As simulacra or simulation, a computer process is not the same as what it seeks to mirror. ” (Mccormack 2013). Modelling differences can introduce tensions within the translated phenomena. They also raise concerns about what is being adapted and why that is being done. Martin Supper remarks that one reason for using natural processes in composition might be to use them “as a way of generating natural forms naturally—forms which are taken to justify themselves by their naturalness alone.” (Supper 2001). However, interpretations of natural processes do not mean they somehow require a literal representation. Rather, characteristics of how natural processes develop, change or transform can be fruitful to explore in a musical context. According to Alice Eldrige and Oliver Bown “the algorithmic musician is seen weaving together equations from complex systems, biology, social behaviour, and so on, to create new artistic works that employ nature in their behaviour, rather than merely representing nature.” (Aldrige and Bown 2018). Why specifically these disciplines are selected and how their behaviour is actually reflected in new artistic works is not fully clarified. However, their performative qualities are clearly emphasised in favour of literal representations. Frequently, the goal with creative mappings of natural processes is indeed not to display nature or represent extramusical behaviours. Rather introduce behaviours that extend the context they operate within and the conditions they are evoked by.
Reflecting on his musical practice through composing with algorithms, Paul Doornbush emphasises the idea of translation between domains where “at some stage there must be a translation from the domain of data, mathematics, functions or concepts, to musical or sonic parameters – from the conceptual domain to the sonic domain.” (Doornbusch 2008). He stresses the domain difference by explicitly stating it (from the ‘conceptual’ to the ‘sonic’). He also formulates the directionality of the relations clearly, where concepts translated to sound have little or no influence back on the concepts. Many reports exist that discuss the “mapping problem” (Polansky 2002). It is often assumed that mapping translations are simply ad hoc or domain-specific by nature. Artistic bricolage filled with details not suitable for a general discussion regarding complex behaviour and musical algorithms. However, some have also mentioned the lack of discussion on the “nature of such “manipulations" and the underlying principles which act to constrain them” (Harley 1994). Or that one should be looking towards other fields since “a considerable challenge remains—and this is as much a challenge of HCI as of algorithm design—to find effective and usable ways to exploit these principles in compositional practice.” (Aldridge and Bown 2018).
2.2 Creation principles
Given the fact that mapping goals are usually artistic, it seems that creating them belongs to an artistic activity. Working with behaviours and process-based sound highlight this, as the focus is not so much on implementing a value-by-value mapping, but rather to establish a framework for relating evolving behaviours. Approaching the domain differences through a common ground allows these areas to interact, for example having sound material influence a controlling algorithm. Perhaps most importantly, providing interaction between the domains introduces a meaningful exchange instead of a simple translation.
Discussing operating levels and compositional processes, Horacio Vaggione has noted how ”music-making remains an activity revealing its own ”creation principle”. (Vaggione 2001). Activities and events contributing to a certain musical outcome are somehow always a part of it. Compositional processes involve choices that form part of the complex systems that constitute computer music. Instead of being directly guided by extramusical models, generative algorithms can be thought of as organisational forces injected into compositional situations. The richness of connecting the contradicting forces behind adopted models and musical ones can be something to explore rather than avoid.
3. Context and Scale
Computer music processes emerge within a given medium and the enclosing conditions it consists of. Similarly, generative processes come about within the boundaries imposed by a given computational context where “the context is the world-state data and arguments that the algorithm has access to” (Wooller et al. 2005). The dynamic relationship between digital environments and generative algorithms allows for an exchange, or mediation between these two that could be described as “not a flow between two preexistent entities; rather, it is a process that re-presents or reconstitutes entities [...] it is a generative process, setting the conditions for the becoming of entities.” (Barker 2012). Enabling the conditions for such exchanges can be seen as an important part of artistic approaches involving algorithms.
3.1 Evolving medium
Computational systems change through the interaction of their parts but also with respect to their own state and potential transformations. The practical approaches presented here follow a similar path where a context (or medium) is designed to evolve through the interactions of the processes it contains. This results in an environment where processes ”are seen as collections of collaborating agencies [...] coupled through interaction such that they are mutually influencing.” (Brown and Gifford 2013). Establishing such a medium is equally important as crafting the processes it contains.
3.2 Local tensions
Difficulties can be encountered when building operational spaces where detailed sound-synthesis procedures interact with abstract computational processes such as cellular automata or alife algorithms. This can be notably complicated if the two domains are not working on the same scale or structural level. Forcing the levels to match, for example by directly mapping a complex system to sound samples, might not fully take advantage of the natural scope the different domains belong to. When enabling their relationships, the different structural levels, including lower-level details but also higher-level developments must preferably be worked out simultaneously. How they complement and connect over time can be crucial for establishing emergent behaviour and a tight relationship between material and form.
Tensions surface between perceived local sound and the coexisting direction of higher-level structures. Addressing such tensions highlights the importance of working with algorithms spreading over different time-scales. Meaningful interactions between control structures and sonic details afford richness and highlight the specificity of the medium. Focusing on having this relation expressive is a vital element when connecting complex systems with algorithms for process-based sound.
4. Interactivity or Interference
The projects presented here below question the possible modes of interaction of sound, generative systems and manual inputs. Instead of allowing for direct control of a complex system, the idea is to rather consider interaction as part of the system itself. For example, to modify the context an emergent process exists within, react based on system observation or directly interfere with an ongoing flow. Interacting with behaviour also highlights concerns regarding the influences of manual input. In a musical setting, it enables different possibilities of change and control. Actions that do not have instant impact modify the usual feedback-loop of musical interaction. These situations question how a “fluidity of experience is possible and under what conditions can it be maintained” (Brown 2012). Interaction with slowly evolving systems reframes the important dynamics of the micro and macro that are so fundamental to electronic and computer music.
4.1 Emergence and Control
Adopting behaviour from systems characterised by complex changes over time is not trivial. As noted by Oliver Bown “Although the Game of Life and these various creative models are very interesting from the point of view of studying emergence, and give us complex behavior emerging naturally out of a simple algorithm, they come at a cost: we cannot control them easily, even less so than simpler rule-based or parametric systems. Without methods to automatically explore the worlds of possibilities they offer, they may be of limited use for creative practitioners” and furthermore “We have some understanding of the creative potential of self-organizing processes but little understanding of how we can control them. After all, this is something of a contradiction in terms. Finding ways to combine emergence and control remains a major challenge for the practical application of complex systems.” (Bown 2019). Should control strategies modify the properties that generate emergence or rather address how the outcomes are processed? Can those viewpoints perhaps be combined and implemented on the same level? Controlling emergence poses difficulties while attempting to do so enables original situations for engaging with systems and self-organisation.
Establishing interaction can be understood as a dynamic process. Something that changes over time instead of remaining constant. Of use here is Andrew Goodman’s use of the term “interfacing” for discussing interfaces and dynamic interaction where “the interface might now be thought of more as a process of interfacing, as an unfolding or contingent process within a larger nexus of relation, as an in-action moment of intensity of disruption, contrast and invention rather than a privileged or static position within an art event.” (Goodman 2018). Goodman’s interests are towards interactive artworks but at the same time strongly conform to the ways of ‘thinking temporally’ as presented here. The setup of interactive processes through the dynamics of interfacing enables interaction to occur at any moment within a given timeframe of interaction. This can mean for example to attach (and detach) various algorithms of interaction during this period of time. To augment behaviours through the ways one responds to them during moments of interaction.
Evolving interactivity enables different points of contact within a system consisting of many parts. It recalls the notion of networks and their aesthetics as “a mode of articulation between ways of doing and making, their corresponding forms of visibility, and possible ways of thinking about their relationships”, (Kalonaris et al. 2021). Dynamic interactivity reflects a creative process based on exploration. It emphasises a two-way, responsive relationship where feedback and exchange are enacted between a system and its interacting agent. Through such exchange, the purpose of the system is not simply to act on a composer's intention but also to contribute to its inception.
4.2 Interrupting flows
An important possibility for interacting with generative systems is by going against their normal flow. By interrupting or disturbing their regular behaviour. The interactivity approaches presented allow for processes to be interrupted, halted, blocked or disturbed in various ways. This approach to interaction creates immediate audible effects but also introduces tension in the relationship between system flow and manual control. Algorithms can be subject to direct intervention that radically alters their aims and direction. Interaction based on intervention pushes something off track while also allowing for reaction or adaptation. Through such communication, a duality of the direct and the unfolding emerges.
Interference with the ongoing can help to shape interaction where rapid but radical disruptions couple with slower, evolving changes. The balance between the two offers different situations of responses and possible outcomes highlighting how “interrupts offer a dynamic way of balancing autonomy and command through the interaction with generative processes” (Gunnarsson 2018). Interfering with a flow of events undeniably provides a strong contrast to parametric control or pre-defined mappings. Blocking behaviours bring forward an attitude of taking over, of engaging with something that unfolds or choosing to ignore it.
Two simulations running in a browser with 'Brutes and Bullies'
WFSCollider with 'Wildfires' session view and Udef editor
The Game of life WFS system setup in Institute of Sonology's New Music Lab, on the sixth floor of Amare, Den Haag.
16 of the 22 scores that form 'Wildfires'
Transition from a ring to random via a small world network (Chatterjee 2015)
Your browser does not meet the minimum requirements to view this website. The browsers listed below are compatible. If you do not have any of these browsers, click on the icon to download the desired browser.
