On a blustery morning in December I received an email from Spotify: my personalized “wrapped” package was here. Opening the application on my phone, I scoured the splashy slideshow visualizing my listening data over the last year: my most-played songs and artists, the affective musical categories I engaged with at different moments of the day (“minimalist poignant lit” in the morning, “sorrow sentimental easygoing” in the afternoon) and a Meyer-Briggs-spoof descriptor of my “listening personality,” The Adventurer. The mass of data that Spotify has collected about me over the last year was tightly presented in an easily shareable format, revealing not only quantifiable information about my individual listening habits, but the particular persona that Spotify has chosen to describe my position as a user and music fan. The relationship between music recommendation technologies, the people who create them, and the way that music recommendation companies imagine their listeners (as exemplified by my Spotify wrapped package) is the central focus of anthropologist Nick Seaver’s Computing Taste. The book is the amalgamation of Seaver’s years of ethnographic fieldwork at various music technology companies, during which time he worked alongside and interviewed employees at various levels of corporate hierarchy—from interns, to industry conference attendees, to CEOs. Seaver notes the tricky navigation of access in his project: like the “black box” technologies that they create, Silicon Valley startups are difficult spaces to gain entry into, let alone garner details about their business decision-making processes. But Seaver’s more broad, and seemingly simple question of “[how the] people who design and build recommender systems think about music, listeners, and taste” offers an accessible starting point from which his interlocutors are able to discuss their shifting self-conception and role in the music and technology industry (17).
The book is divided into six chapters, which Seaver orders to provide a chronological development of recommendation systems and arranges around themes of information overload; listener retention; the imagined user; music and sound as information; genre and musical relationships; and data maintenance. As illustrated throughout Seaver’s book, music streaming and recommendation services are not guided by a simple algorithmic process, but rather by multiple overlapping and dynamic algorithms that are monitored and guided by human hands. Seaver shows how algorithmic logics are informed by human ideas of technological utility that organize music and sound as a form of data. Resisting the reductive distinction between human and machine agencies that are often reified within both music scholarship and industry, Computing Taste examines the complex socio-technical infrastructure of music recommendation programs and the shared discourses that have contributed to their design.
To accomplish this, Seaver calls into question the basic function of and motivation for music discovery services, putting to use the tropes and metaphors he encounters from various interviewees as theoretical jumping off points. Drawing on work from critical algorithm studies, musicology, and anthropology, Seaver complicates his interlocutors’ commitments to music discoverability as a lofty and imperative goal. In the first chapter, he interrogates the popular notion of information overload as a “commonsense problem that needed little explanation,” providing a genealogy of this conjecture from Theodor Adorno’s Introduction to the Sociology of Music through the dawn of the World Wide Web (28). Seaver posits the myth of overload as contributing to an informatic cosmology, on which the purpose and necessity of music recommendation hinges. Technical solutions, like the process of collaborative filtering described herein, not only reify the existence of information overload as a problem to be overcome, but also inform how researchers see other forms of human choice as computationally deduced. In the second chapter, Seaver charts the mutation of these ideals of music stewardship, as the commitment to making music more discoverable is supplanted by market demands of listener retention. Drawing on ethnographies of animal trapping and behavioral science writing on “captology” (the design of computers as persuasive technologies), Seaver examines how engineers try to entice people to use their services and “capture” the ears of their users (54). Seaver settles on a depiction of music recommendation softwares as pastoral infrastructures (71)—listeners are shepherded within an enclosed territory of songs and sounds, but are also cared for and tended to while there.
Perhaps Seaver’s greatest addition to the field of popular music studies is his discussion of the imagined user positions demarcated by music recommendation systems. Developers pitch their programs as having a new “post demographic” way of organizing musical data, an understanding of musical interest divorced from region and community, as well as identity factors such as gender and race (74). Throughout his research, Seaver is continually confronted with the “pyramid of listeners,” a diagram in which users are grouped into the categories of indifferents, casuals, enthusiasts, and savants (78). These categories refer not to the types of music that listeners engage with, but to the variety in their consumption patterns and their motivation to seek out new music. Conversations with his interlocutors show that categorization according to musical avidity often elides with other forms of difference, where the musical savant is imagined as young and male (reflecting the largely white, young, male, English-speaking workforce employed in these companies) and the indifferent listener is cast as female. Rethinking musical communities in terms of avidity rather than demographics offers useful terminology for thinking about listening practices but risks reinscribing musical hierarchies and snobbishness under the guise of aesthetic neutrality, a point which Seaver stresses.
Seaver goes on to discuss more algorithmic aspects of music recommendation in chapters five and six. He describes a fascination among engineers and developers with music as a signal, using machine learning processes to determine similarities and relationships between pieces of music that are not readily recognizable to the human ear. This understanding of musical relationships catalyzes an ontological shift, under which human perception is superseded by complex algorithmic pattern recognition. He goes on to describe how developers visualize spatial relationships between musical materials based on acoustic similarity, celebrating the interconnectedness of far-flung styles and regions. Scholars focused on genre will be interested in his discussion of the synonymous description of musical “clusters” (material that has been gathered together on the basis of audio analysis, rather than audience engagement) and genres. The creation of genre labels on the basis of shared acoustic information (“deep cello” (133) and “oratory”(131)) and listening context (“sleep” (131)) marks a shift in how such categories are constructed and circulated. While the relationship between identity and music recommendation is not the central focus of the book, Seaver’s engagement with musicological work in this area is evident, as he frequently returns to the co-mingling of social and machinic decisions in the differentiation between types of music. Algorithms may be able to group sounds together, but it is humans who decipher, interpret, and name such groupings, a “performative effect” that is informed by the prior knowledge, values, and vantage point of the person tasked with demarcating these boundaries (135).
Because Seaver’s text is primarily concerned with the aims and ideals of developers building music recommendation systems, his technical description of algorithmic programming is fairly minimal. While this makes Computing Taste an accessible text for readers who are interested in algorithms as they relate to music and sound, a definition of and further elaboration on the differences between machine learning and neural networks would be helpful for readers less knowledgeable about this area of study. Researchers already interested in algorithms, machine learning, and artificial intelligence will enjoy the linkages made between these hot-button topics and issues of design, labor, and taste.
Computing Taste is a stylistically rich text that elaborates on the stories, metaphors, and tropes of the music recommendation industry to draw an eclectic yet digestible theoretical framework. Seaver finishes his book with a focus on the gardening and farming metaphors that spring up throughout the music tech industry—if data is soil and the algorithm the fruit, the overseer must give up complete mastery and instead work in conjunction with the non-human forces that shape the garden’s development. Viewing my Spotify Wrapped package through this lens, I can better understand how my listening habits are both attended and tended to, informed by assumptions about the wants of listeners. Like pastoral enclosures, music recommendation systems create power “not through domination but through modulation” (152), offering a critical figuration for thinking about technology and agency. As music recommendation and other algorithmic programs play an increasingly large role in the popular music economy, Computing Taste provides an imperative account of these emergent and mutable systems, urging scholars to take seriously the corporate talk of music and tech industries as allegories for human and machine relationships.
University of California Press