This blog post comes out of a discussion with Ritwik Banerji about the ‘hidden’ role of ethnography in the work involved in creating new experimental systems for music improvisation. Ritwik put it to me that “it seems that a lot of work … involves a kind of ‘implied ethnography’ – that is, it’s clear that the author/designer has lots of personal experience with the domain they’re designing for, and yet the technical documentation of such systems makes scant mention of it.” This was a welcome invitation to reflect on my past practice since I had once been a student of social anthropology and am now, as an associate professor 25 years on, re-engaging with ethnography as a methodology. Have I been implicitly using ethnography all along, and could/should this component have been more explicit in the presentation of my work in an academic context? I will begin with some scene setting.
It is 2010, and I am working with fellow members of the improvising collective, Not Applicable, Sam Britton (computer), Lothar Ohlmeier (bass clarinet), and Tom Arthurs (trumpet), to create an improvised performance using what I tended then to call “Live Algorithms” (Blackwell, 2009) (the performance was later documented in the record “Long Division” on the Not Applicable label). We had been invited by a producer friend at the North Sea Jazz Festival to create this experimental content for a one-off show.
This was philosophically motivated work, at the heart of which was the idea of the software doing things “autonomously,” making meaningful actions in an improvised context without someone directly operating it. Achieving this involved, in part, the smart use of complex evolved systems and, in part, our own informed aesthetic manipulation of the resulting system (a description of the system and design process can be found in Bown, 2011).
The former, the system development, through systematic experimentation with algorithms, aspired to simulate all of the richness of expression and interaction that occurs between two improvisers. My practice is grounded far more in artificial life than artificial intelligence, and I liked to think of my musical systems as more amoeba-like or insect-like than simulating the smarts of any recent human composer. Of course, this richness of expression was not achieved in any complete sense. Instead, it resulted in a curious digital object that gave some sense being a complex system that musical improvisers be interested in playing with, through its blend of unpredictability and responsiveness.
In a paper with academic collaborators Alice Eldridge and Jon McCormack (Bown et al. 2009), we discussed how these types of software behaviours needed a new category: they were object-like (not humans or animals, or “artificial intelligences,” artifacts under our direct authorial control), but they also felt a bit lifelike. We felt they deserved some attribution of agency beyond those of Latourian objects (books and doors and the like that still acted in social realms but not with the wiggly zeal of the musical machines we had constructed). The term we gave them was “behavioural objects.” We also distinguished between performative and memetic agency, the former being agency enacted in the moment of performance, where the software operates in the absence of a human controller, the latter being agency in the actor-network theory sense, enacted over the much longer period of creative development leading up to a performance, where the software is one of many materials in a sociotechnical system.
I also mentioned above the additional work of the artist in aesthetic manipulation bolted onto these imperfect, messy complex systems. Philosophically, like other live algorithms artists of the time (Blackwell, 2009), I saw myself as attempting to offload some part of the creative and aesthetic decision-making to the system itself, to edge oneself out of the process, leaving the system with this “performative agency,” but at the same time taking charge of the result, as the composer of this work-as-system. Studying the creative work of working with algorithms in this way became an increasing focus in design, using creative reflective practice to document the interplay between developing an algorithm and refining the artistic outcomes that use that algorithm.
Coming back to Ritwik’s question, I am led to revisit the plethora of methodologies I have encountered in over 10 years of working in this field. In a book chapter published last year (Bown, 2021), I traced how music and AI work had transitioned from more of a computational philosophy perspective (what would it mean to make a computer that could create music? And how do we do it?) to a design perspective (how do we make usable tools to support people making music?), and is now in transition to more seriously embracing a sociological perspective, i.e., understanding creative tools not only in a creative practice niche or design problem but in terms of how they are adopted and may transform creative practice in communities of practice. This latest transition is well underway but is not mature; the competing concerns of these various “inter”-disciplines have not been jostled into a neat order. Note that there is a longstanding tradition of anthropological perspectives in creative, technological music practice, such as in the work of Simon Waters (Waters, 2007), but that there is, separate to this, a gradual awakening coming from the direction of more computer-science-oriented creative work.
Methodological standpoints don’t just fuse together or super-impose in a body of work like this; they are in conflict for discursive space (words on a page), time (research time), and attention (researcher awareness — what we have the time to read and understand, and what debates we participate in). It is not straightforward to do justice to the multiple competing concerns of creating algorithms, experimenting artistically, understanding and solving design problems, and simultaneously grappling with underlying cultural issues. We have to pick a methodological focus.
Critical to considering the relationship between this work and ethnography, then, is what space is given to doing ethnography. My work and that of my colleagues making creative music algorithms may have been ethnographically informed and involved thinking about people and their cultures of creative practice, but it was not carried out in an ethnographic way. Time wasn’t spent assuming an ethnographic stance nor analysing the role of class, socioeconomic background, gender, race, and ethnicity. The work was embedded in a cultural milieu of electroacoustic free improvisation, particularly based in London’s experimental music scene, and didn’t devote time to looking outside of it. Indeed, Ritwik’s own work brought quite a different perspective to this live algorithms community by centering questions of ethnography and through the original idea of using music technology experimentation to shine a light on human behaviour.
An Artist-Developer’s Stance
My live algorithms work, like others, took an artist-developer’s stance, and in the early 2010s, I was perhaps typical of artist-developers exploring new creative technologies in that I didn’t yet associate with a design perspective either. This changed gradually over the 2010s as design thinking rose in prominence, and personally for me finding myself situated in a design faculty from 2011. The user-centred designer is very different from the creative artist making things with technology. Artists position themselves directly in what they are doing, whereas designers seek greater objectivity. To be sure, both overlap in their concerns, and in fact the artist-programmer is often a hybrid innovator: making software for their own immediate use, and then, in addition, potentially as tools for third parties. One might broadly sketch out a spectrum, then, and extend it to include the ethnographer. It might look something like this:
|assume creative direction and authorship of work, engage with design issues, seeks objectivity by being honest in reflection. The artist-programmer may also study their creations, both via the tools of the practice-based researcher, and of computer science (quantitative and qualitative tests of system behaviour).
|assumes a user-centred perspective informed by ethnography, but still has to make something, thus assuming creative license. As above, the artist may be a designer, putting themselves as the first, guinea pig user, but there is a tension here with a principle of good design: to emphasise with users who are different to oneself.
|the primary goal is to understand human cultural behaviour as it emerges in particular spatiotemporal contexts, not complicated by ulterior goals such as to build things. This may inform design but mustn’t be confused with the work of design.
These descriptions directly consider how the artist-programmer and designer overlap, as the former seeks to understand software’s use and potential in other people’s hands as well as their own. They may still position themselves at the centre of the software use via the use of practice-led research: reflection on their own creative practice with technology. This perhaps plays into Barry, Born, and Weszkalnys’ (2008) notion of a subordination-service relationship between methodologies: design in service of creative practice. Over recent years I’ve taken to flexibly situating myself across creative practice and design, diversifying the methods I use in order to address different research questions. This has been most pronounced in a recent collaborative project where a collaborative team moves between practice-based and design research processes. But in this early work, my role as an artist, doing that key aesthetic manipulation, and putting my name on the work, marked a clear boundary from the work of design. During this time, I have been guided by the clarity of a paper by my Ph.D. peer Marcus Pearce (Pearce et al. 2002), which reminds researchers to prise apart their methodological practices based on their project aims.
The overlap between designer and ethnographer too occupies an entire academic cohort following thinkers like Lucy Suchman and Paul Dourish (Suchman, 1987; Dourish, 2007). Ethnography is considered part of a designer’s toolkit, although the ethnographic inclinations of designers vary considerably according to the trade-offs found in the list above. If we trust in this spectrum, we may claim that the artist-programmer has too much of a stretch to reach the world of the ethnographer; methodologically, there is too much to take into account.
But developing a three-way interaction between reflective creative practice, design and ethnography should be explored more thoroughly. Ritwik’s own work uses systems developed as an artist-technologist as a probe to explore human behaviour. In conjunction with asking “how can I make the system better” like many of us, he adds, “what does the system show us about the world” (Banerji, 2012, 2019). In my own work, this was indirectly manifest, and perhaps this is how I can best answer Ritwik’s question in the personal and “genred” relationships that formed with specific instrumental musicians as I conducted this work. The project peaked in its collaboration with clarinetist François Houle, who deeply studied the system’s behaviours.
But above all, it was a shared understanding of free-improvised electroacoustic music that was essential to the system’s working. An anecdote helps frame this ethnographically. In 2012 I was demoing some other work of mine on the ABC science show Catalyst. We had wrapped up filming, and I told the presenter about my system for live improvisation. He got excited and asked to jam with it with his guitar, but it became rapidly apparent that the experimental bleeps and drones my system started making were only jarring noise to him. He was expecting something more like an accompaniment system, and I was reminded that free improvised music is a narrow scene, impenetrable to many. Recognising our Bourdieuian boundaries, we politely agreed to look no further.
Banerji, R. (2012). Maxine’s Turing Test: A Player-Program as Co-Ethnographer of Socio-Aesthetic Interaction in Improvised Music. Proceedings of the Artificial Intelligence and Interactive Digital Entertainment Conference (AIIDE ’12), 2-7.
Banerji, R. (2019). Feeling Like an Agent. Array: the Journal of the International Computer Music Association, 4-7.
Barry, A., Born, G., & Weszkalnys, G. (2008). Logics of interdisciplinarity. Economy and Society, 37(1), 20-49.
Blackwell, T. (2009). Live algorithms. In M. Boden, M. D’Inverno, & J. McCormack (Eds.), Computational Creativity: An Interdisciplinary Approach (pp. 1-3). Dagstuhl, Germany: Schloss Dagstuhl-Leibniz-Zentrum für Informatik.
Bown, O., Eldridge, A., & McCormack, J. (2009). Understanding interaction in contemporary digital music: From instruments to behavioural objects. Organised Sound, 14(2), 188-196.
Bown, O. (2021). Sociocultural and design perspectives on AI-based music production: Why do we make music and what changes if AI makes it for us? In E. R. Miranda (Ed.), Handbook of Artificial Intelligence for Music (pp. 1-20). Cham, Switzerland: Springer.
Dourish, P. (2007). Responsibilities and implications: Further thoughts on ethnography and design. Proceedings of the 2007 Conference on Designing for User eXperiences, 2-16.
Pearce, M., Meredith, D., & Wiggins, G. (2002). Motivations and methodologies for automation of the compositional process. Musicae Scientiae, 6(2), 119-147.
Suchman, L. A. (1987). Plans and situated actions: The problem of human-machine communication. Cambridge, England: Cambridge University Press.
Waters, S. (2007). Performance Ecosystems: Ecological approaches to musical interaction. Proceedings of the Electroacoustic Music Studies Network Conference, 1-20.