Introduction
Spotify welcomes the growing interest in streaming media but is concerned about information it received regarding methods used by the group of researchers responsible for this project. This information suggests that the research group systematically violated Spotify’s Terms of Use by attempting to artificially increase plays, among other actions, and to manipulate Spotify’s services with the help of scripts or other automated processes. Spotify determines that the group of researchers was aware that such actions explicitly violate its Terms of Use and aimed to mask this violation by technical means. In light of the above you are hereby asked to confirm by 26th of May 2017, in written form, that you have received this note and that the group of researchers has ended such actions that are in violation of Spotify’s Terms of Use, and that it does not intend to take up such actions again in the future. Note that in this context, violation against the Terms of Use may imply responsibility for possible damages resulting from this violation.
On May 19, 2017, legal counsel of the Swedish music streaming service Spotify contacted the authors of this book via email to inquire about the methods used in a then-ongoing research project, titled “Streaming Heritage: Following Files in Digital Music Distribution.” During the autumn of 2013, this project had received over $1 million1 in grant funding from the Swedish Research Council for investigating the challenges and consequences of streaming services, such as Spotify, for the heritage sector.2 A guiding question for the project was how people’s practices and approaches toward cultural forms such as songs, books, or films—practices including the production, expression, and exchange of those cultural forms—are transformed under the shift from commodity ownership to commodified experience.3
While research within this project soon went beyond the initial focus on cultural heritage, the project itself had been explicit and open about its critical approach from the very start. In fact, various forms of public activism were part of the project’s interventionist research design, for which it had received—in a fierce national competition—the second highest of the council’s assessment grades. Blog posts, newspaper articles, and interviews on both television and radio, not to mention the 2013 project application, had outlined the project’s strategy as one based on digital methods, including the use of bots and the dissemination of self-produced sounds via Spotify.4 The project opted to study Spotify using the same tools that organize online information.5 More importantly, Spotify’s apparent lack of interest in data sharing necessitated an approach that would go beyond interviews, direct observation, and other standard methods of media industries research.6 While the project team members met and engaged with several Spotify employees over the years, listening to company spin and “industrial self-theorizing” felt inadequate.7 To follow a formal procedure of gatekeeper introduction or even to seek official endorsement by Spotify would have limited and biased this research.
Complementing front-end inquiries (such as interviewing) with experimental back-end studies of digital media infrastructure, metadata generation, and aggregation practices, the project instead aimed to initiate public debate about the often subtly changing standards, values, and politics of cultural dissemination online. Given this aim, the question is not why Spotify responded to this research in May 2017, but why had this not happened earlier and why only by mailing a formal complaint. We will return to these questions at the end of this book, which is based on the very research that Spotify aimed to stop.
Back in 2013, Spotify was still widely seen as pioneering solutions to issues of concern to the music industry. Its à la carte, on-demand music streaming had arguably created a sustainable revenue model for artists. Spotify seemed to have stopped the downward spiral in earnings and income that had plagued physically distributed music recordings since 2002, and it had delimited informal practices of media circulation, effectively advancing the fight against unauthorized (or pirate) forms of file sharing.8 Despite the Spotify hype of those years, however, critical questions also kept coming up. What was the substance of the service that the company claimed to offer, given that its founders themselves had little of value to offer? Spotify was neither providing a cultural good nor offering valuable contacts in the field of cultural production. Initially, the company could not afford licensing for the music or the bandwidth to distribute it. With the launch of its first beta version in 2007, Spotify largely built both on the peer-to-peer (P2P) technology already in wide use within Sweden’s file-sharing community and on a vast, unauthorized music catalog that had been shared by members of this community.9 How were artists to be remunerated? Spotify’s business model never benefited all musicians in the same manner but rather appeared—and still appears—highly skewed toward major stars and record labels, establishing a winner-takes-all market familiar from the traditional media industries.10 How then is this streaming market different from traditional media markets? And what distinguishes Spotify from a regular media company?
The history of our encounters with Spotify has developed along these questions. It began with a brief conversation with Spotify’s then head of marketing, Sophia Bendz, in Stockholm in the fall of 2012. Asked if Spotify would not qualify as a regular media company—given its declared business interest of providing content to audiences while selling those audiences to advertisers—Bendz rushed to praise Spotify’s achievements as a tech company. A follow-up meeting with a company executive in Spotify’s headquarters (before we received the grant funding) led to mutual expressions of respect but little more. At this point, Spotify had already been made aware of our research interest in working with the company’s data. Over the following years, we met individually with engineers, marketers, data professionals, and academics related to the company. But conversations often fell apart as soon as Spotify’s tech identity was questioned. At the same time, the company’s attempts to add television content made it obvious that Spotify increasingly operated like a traditional American media enterprise. Between 2013 and 2017, the company’s equity rounds, market and debt capitalization, and changing board of directors tied its corporate strategies more and more to US-based financial interests. At the end of a decade-long process of financialization, Spotify is neither particularly Swedish nor solely about music.
In this respect, Spotify’s transformation resembles Facebook’s. “Digital content curators fit quite well within the established parameters of media organizations,” Philip Napoli and Robyn Caplan observe in their aptly entitled article “Why Media Companies Insist They’re Not Media Companies, Why They’re Wrong, and Why It Matters.”11 Although Spotify, of course, also offers a technical solution for music distribution, the aggressive discursive framing of Spotify’s operation as being primarily technological has tended to obscure its long-term entrepreneurial, financial, and culture-changing strategies. Such “politics of standards” often have regulatory consequences.12 Since 2015, for instance, Spotify has implemented a plan—and the technology—to generate data based on its music streaming that allow for the study of human behavior at scale. The company acts, in other words, not only as a music provider but also as a private data broker. It has openly promoted its massive collection of contextual data as a service to marketers and, in November 2016, even launched a global outdoor ad campaign in fourteen different countries, with ads jokingly showcasing aggregate data sets: “Dear 3,749 people who streamed ‘It’s the End of the World as We Know It’ the day of the Brexit vote, hang in there.”13
Figure 0.1
Spotify London subway advertisement in November 2016. Photograph by the authors.
The ad implied that Spotify knows you at scale—it knows what you listened to and what it meant to you. But what is the commodity being sold here? And how does this service relate to the European Union’s new General Data Protection Regulation and its provisions on profiling?14
At the time of writing (early 2018), Spotify’s desktop interaction design looks very different from how it did in 2008 or 2012. Once, user interaction was organized around tracks, search options, and community-activating features, such as self-made playlists. Today, Spotify’s interaction design reorganizes music consumption around behaviors, feelings, and moods, which are channeled through curated playlists and motivational messages that change several times a day. A large number of these playlists are created by third-party playlist services such as Filtr, Topsify, or Digster, owned by Sony, Warner, and Universal, respectively—the three major record labels that, in turn, own stakes in Spotify. In 2016, evidence emerged that Spotify had begun filling some of their mood-oriented playlists with fake artists, illustrating the company’s interest to use playlists as a device for framing and measuring user behavior.15
The present situation—where music has become data, and data has become contextual material for user profiling at scale—invites us to pause and reflect about the way that songs, books, or films are now typically made accessible. Internet critics and journalists have documented that social media companies such as Facebook use tools to microtarget their users based on emotional states.16 Similar claims have been made about Spotify by Stanford psychologist Michal Kosinski, who developed a model for behavioral prediction now used by Cambridge Analytica, a firm notorious for “psyops” electoral manipulation in support of Brexit and the Trump campaign.17 As Kosinski and others argue in a paper entitled “The Song Is You: Preferences for Musical Attribute Dimensions Reflect Personality,” industries should abandon the traditional order of knowledge that organizes music according to genre and style and, instead, strategically exploit the link between music choices and personality traits.18
Access and Interventionism
This book and the project on which it is based developed out of a concern with the tendencies of mainstream work either to overlook and suppress such difficult ethical, political, and ideological issues of digital media, or to prejudge and scandalize these issues. We observed an increasing divide between political and public debates about the social responsibilities of digital media companies, such as Facebook and Google, and the social research that aspired to cover the digital. Fields such as digital sociology foreground the computational dimension of social inquiry as well as social life, locating research largely in a “field of devices.”19 Yet political events and debates involve more heterogeneous constellations of actors and also underline a widely recognized “need to inform public debate about online business practice,” as researchers and “journalists have long done in the offline world.”20 In responding to this need for greater transparency, we felt that the adoption of existing tool kits from media industries research or the digital humanities to be similarly insufficient. The obvious power of digital media companies to establish their own order of knowledge through sponsored and applied research or, as seen in our own case, through coercive legal practices suggested that using qualitative social science methods to “study up”—to research the powerful—would not yield satisfactory results in this case.21 While media companies have always aimed to control research access and output, sociologist Noortje Marres rightly points to the “highly troubling relations of dependency” between today’s platforms and researchers, especially where the latter simply “sign up to the terms of use stipulated by digital industries,” which Marres identifies as a major ethical “problem of complicity with the organizations that serve as our sources of information.”22 Why make Spotify the object of studying up in the first place, given the company’s soft brand image and origins in Sweden’s illicit file-sharing community, a company that allegedly still remains so close to its users?
An established but still useful idea we adopted is to focus on the process that allowed Spotify to grow from micro- to macrosize, rather than on the powerful global organization itself. As Michel Callon and Bruno Latour have suggested, size alone is insufficient for substantially distinguishing between institutions or organizations on the one hand and individuals or groups on the other. “There are of course macro-actors and micro-actors,” Callon and Latour explain, “but the difference between them is brought about by power relations and the constructions of networks that will elude analysis if we presume a priori that macro-actors are bigger than or superior to micro-actors.”23 Accordingly, the task is not for the researcher (microactor) to study the corporation (macroactor) but to trace the manifold transactions that allowed an initially opportunistic and semilegal project to turn into the allegedly ultimate solution for music listening. Again, at the beginning, Spotify founders Daniel Ek and Martin Lorentzon themselves had little of value to offer, apart from the advertising tech firms they had worked with. It was through a long chain of associations and by enlisting bodies, materials, discourses, techniques, feelings, and laws that this faceless Swedish group of engineers and digital advertising technologists became the voice of digital culture.
Following the process by which Spotify created lasting asymmetries between itself and us, in terms of capital as much as cultural power, throughout this book, we will come to open not only one but many of the black boxes of music streaming. In the terminology of science and technology studies (STS) from which this metaphor is imported, actors grow with the number of relations that can be put into black boxes, that is, be made invisible by their own success. A black box contains that which no longer needs to be reconsidered, those things whose contents have become a matter of indifference.24 Think of streaming, for instance, as the solution to illicit downloading. Or consider the way that Spotify’s infrastructure, including its reliance on P2P networks (until 2014) and user devices, has been black-boxed. Gideon Kunda, in his seminal study of a computer company, wondered about the way in which this company understood culture as “something to be engineered,” in the sense of a “mechanism” of normative control extending well beyond the corporation itself.25 Macroactors’ power largely relies on their ability to keep such boxes firmly closed and to make us believe that it could not be otherwise.
This book traces, on various levels, the process of Spotify’s becoming. It partly enters its black boxes technically, by employing digital methods such as bots, or small software scripts. Over the last decade, such digital methods have been developed and widely deployed to perform social science and humanist inquiries on big or midsize datasets.26 Part of the research also relies on interviews, document analysis, archival work, and nonparticipant forms of observation. Overall, the research design has been strongly influenced by autoethnographic and self-reflexive forms of fieldwork, as they are common in social anthropology and ethnology.
A reason for this is the difficulty usually encountered when industries are subjected to studying up. Over the years, much discussion has surrounded the problem of access that besets social and cultural studies of media companies and the way they organize processes of content production and distribution. Informed consent, taken as an iron rule in clinical research, seldom applies fully in critical studies of digital industries, for which experts advise that researchers “move beyond overly simplistic ideas about informed consent.”27 In research on media organizations, researchers often embrace perceptions of difference rather than ideals of objectivity through immersion.28 As organization scholar Barbara Czarniawska notes, “an observer can never know better than an actor; a stranger cannot say more about any culture than a native, but observers and strangers can see different things than actors and natives can.” An idea of “symmetric fieldwork” that does not strive to be “nice to the natives” is both fairer and more adequate in an industry context, because it “allows one-self to be problematized in turn—at a certain cost to the researcher, of course.”29
In drawing from these methodological insights and experiences, this book subscribes to an idea of interventionism that aims to change, or at least shed light on, an emerging order of knowledge. Such an intervention is not an activist call to arms but rather an effort to generate questions, by means of our own probing and partly performative research practice, about the conditions of transparency in digital culture. A famous model for this kind of approach is Harold Garfinkel’s “breaching experiments,” which aimed to break the mundane, routine, implicit, tacit, and taken-for-granted social rules in a given setting. Being interventions into the normal stream of daily life as participants experienced it, these experiments consisted of low-level disruptions of ordinary scenes (such as taking items from a stranger’s basket in a grocery store) that aimed to make the familiar appear strange. Although they were a series of (public) demonstrations rather than formal experiments, these tests were productive in showing how the social order is maintained and how the trust arrangements that stabilize it can be analyzed.30 More specifically, we understand Garfinkel’s demonstrations as a model for how the social order governing our access to the field could be turned against itself. What if instead of adopting the conventional interview routine, with its clearly demarcated roles of insider and outsider, we were to have an ethnographer observe each conversation in order to understand how corporate spin and the power relationships of studying up play out in our own research practice? What if we were to complement interviews, document analysis, and overt observation of the front end of Spotify with experimental, covert access to its back end in order to compare and verify the information given and to confront interlocutors with the results? How about making following files a research-guiding metaphor, in accordance with economic and social anthropology’s interest in the social life and cultural biography of things, as if it were possible to follow the transformation of audio files into streamed experiences in the simple way a postman would follow the route of a parcel?31 How about distributing sounds and music via Spotify in order to gain firsthand experience of its aggregators while also programming bots, or software scripts, that would listen to this music, thus short-circuiting the system? And how to design this project in such a way that it could unveil and challenge ethical standards of social research that take overt and covert as simple binaries, instead of acknowledging the “continuum of deception” that digital industries are themselves actively part of?32
The title of this book, Spotify Teardown, signals our intent to break from corporate protocols by taking up these questions. Borrowing the notion of teardown from reverse engineering processes, we aim to disassemble the way Spotify’s product is commonly conceptualized. Our teardown, however, is meant in an imaginative rather than purely technical sense. We primarily mean to perform teardowns—publicly, through the interventions that this book documents and itself forms part of—as a form of creative engagement with our research object. In this sense, “interventionist methods” are also “inventive methods,” that is, methods or means by which “the social world is not only investigated but may also be engaged.” As Celia Lury and Nina Wakeford have pointed out, inventiveness is not intrinsic to methods but rather emerges in relation to the purposes to which they are put.33 An actor is strong insofar as he or she is able to intervene. Thus, by examining the associations used by actors such as Spotify to grow to macrosize with “the same daring as the actors who make them,” internet research may, we hope, recapture part of the strength it seems to have lost.34
Platforms and Companies
One reason why some research currently seems to lack determination in confronting the critical ethical, political, and ideological issues of digital media is that research is also regularly enlisted by macroactors. It is not accidental that Facebook, Google, Microsoft, Intel, and Spotify all invest in social research. Technology drives both the news and the academic agenda. Organizational environments, with their built-in instrumentality, also might predispose us to think about methods in empiricist terms, as an essentially technical affair to be taken up like a hammer from a toolbox. This book does not offer models, nor does it present findings that are easily operationalized. It follows Eszter Hargittai and Christian Sandvig’s call for a more “positive conceptualization of method as a creative act,” revealing the “messy details” of what we were doing, including failures and open questions.35 The turn to method also had the pragmatic function of opening an interdisciplinary space for collaboration between social anthropology, media studies, economic history, musicology, and ethnology—the disciplines that form our various organizational and intellectual homes.
This is not to say that the digital methods presented in this book could not be reused or repurposed. Some of them were inspired by those developed by Richard Rogers and the Digital Methods Initiative in Amsterdam and thus have already been “repurposed or built on top of the dominant devices of the medium.”36 Here, Spotify’s “system is the method,” as Klaus Bruhn Jensen puts it, insofar as usage of the service produces data that are already “documented in and of the system” and, “with a little help from network administrators and service providers,” can be used as an empirical base for research.37 This is largely how we collaborated with a team of programmers at Humlab, Umeå University’s digital humanities lab, while also consistently building on more than one empirical base.
In highlighting the performative, playful, ad hoc character that governs parts of this research, we do not mean to avoid laying bare the project’s conceptual premises. While this book, throughout its four main research chapters and the interventions that punctuate its main narrative, visits different sites to answer different questions, its conceptualization started from an engagement with a basic concern of what we were actually studying. Given our critical take on Spotify and our call for more public debate around issues of transparency, how were we to ascribe agency and to whom? As a research group conducting experiments on Google’s advertising privacy settings once noted, “We like to assign blame where it is due. However, doing so is often difficult.”38 In the case of our project, the difficulty largely related to the way that Google, Facebook, and Spotify had previously been conceptualized: as platforms. In what ways does the notion of a platform relate to that of a corporate actor or firm, and is it productive for analyzing the relationship between the two?
In December 2017, the European Court of Justice decided that Uber, widely understood to be a platform that merely connected unlicensed private car drivers to passengers, was in fact operating like a conventional taxi firm. Uber claimed that it had only acted as an intermediary offering information rather than transportation services. The court found sufficient evidence to the contrary and rejected this claim, with no possibility of appeal, creating a regulatory precedent for “looking at what tech companies actually do, not how they do it,” as one journalist put it. “For legal purposes, if it quacks like a duck and waddles like a duck, it’s a duck, not an alien robot from outer space.”39 As we will elaborate in later chapters, Spotify, Facebook, Twitter, and YouTube can indeed be regarded as media companies that sell audiences to advertisers, implying legal responsibility for the content disseminated through their interfaces. At the same time, one also has to acknowledge the differences between ducks. In the digital sphere, some of them come in flocks, or systems, and develop a competitive dynamic rather different from conventional ducks. Some live and die alone.40
Figure 0.2
Spotify once described itself as a platform to manage one’s music—but should it still be seen as one? Advertisement for the “Discover Weekly” playlist in the New York City subway in 2016. Photograph by the authors.
The notion of platform has been declared a paradigm of internet research by some,41 but it has done very little to shed light on the differences between digital companies and other varieties; to distinguish the manifold relations between services and providers, supply chain and partner networks, and algorithmically personalized catalogs and global strategies; and to develop a means to measure the relevance or irrelevance of all these actors. Organizationally, the center of any platform is hard to pin down. This, perhaps, is not surprising given that this concept was first launched by the digital industries themselves.
In the computer industry, the term platform has been used since the mid-1990s, when Microsoft began describing Windows as a platform. After circulating first within management and organization studies, the term entered media scholarship simultaneously with the rise of the notion of Web 2.0.42 Partly in response to Netscape founder Marc Andreessen’s description of Facebook and Amazon as platforms, Ian Bogost and Nick Montfort launched the notion of platform studies in 2007 at the Digital Arts and Cultures Conference. They further elaborated on the concept in their book Racing the Beam: The Atari Video Computer System (2009) and on their joint blog, http://
In this current use, platform most often implies a techno-economic view of services such as Facebook or Spotify. In a computational sense, Spotify is a platform to build applications on. At the same time, a platform using the social web is also a business and therefore should be described in terms of business models and pricing structures, according to this view. Abstracting a conceptual model from Facebook, Apple, Amazon, and Google, for instance, Benjamin H. Bratton defines platform as a “standards-based technical-economic system,” while José van Dijck sees it as a “techno-cultural construct” within some “socioeconomic structure.”45 Similarly, Anne Helmond adopts a “techno-economic outlook” on Facebook in order to examine how its “technical architecture” links to an “economic model.”46 Whatever the wording, an interplay between computation and economics is seen as foundational for platforms, usually mediated through the figure of the user, whose apparent unpredictability introduces ideas of generativity into platform theory. Put simply, in order to investigate platforms, platform scholars need to triangulate by relating user participation, computing technology, and economics in one way or another.
Information technology companies and some internet scholars keep the platform terminology because it comes with two major promises. The first is the promise of social web platforms as representing some sort of “ontological distinctiveness,”47 as being new and technologically unprecedented. There was never anything like Spotify before the advent of the social web. The second promise is that platforms epitomize a new form of market, or what the economics literature has dubbed a multisided market. Multisidedness means that a market is not ordered hierarchically—as in a shop where a salesperson may coerce you into buying an unwanted pair of shoes—but rather contains strong network elements, or “positive indirect network externalities.”48 In this view, a platform is less like a market and more like a network that brings two or more different sides together, like a shopping mall that links shops to customers. There is, in any case, some kind of positive interdependence or externality between groups of actors that are served by an intermediary, the platform. Think of Spotify Free, for example, where the platform is said to positively affect supply and demand by providing more listeners with more music while decreasing fixed and marginal costs. Advertisers are meant to benefit from the presence of many consumers, while consumers benefit from the information provided by many advertisers.49
When it comes to the task of identifying Spotify’s corporate responsibilities, strategies, organizing principles, and the markets it creates or folds into, this techno-economic notion of the platform has limited value. Strategically launched by the digital industries, it connotes egalitarian ideas of participation and sociality. It also implies that some businesses are intrinsically multisided: after all, in plain language, a platform is a “raised level surface,” with several sides.50 Corporate stakeholders appear as just another type of user in this setting, as if Facebook, Google, and Spotify were indeed community services that only “helped change,” as noted in the case of Apple, the “economic and legal conditions for music production and distribution.”51 Furthermore, centering accounts of Spotify or Facebook on platforms downplays historical trajectories of market behavior that had multisidedness as a dominant feature long before the advent of digital infrastructures—think of television or newspapers.52 The widespread idea of the social web as being an emergent phenomenon, rather than a strategically or centrally planned one, is contingent on the notion of the generative internet, that is, subscribing to some version of the theory that a network of adaptable machines without centralized control would result in open and socially beneficial innovation processes.53 Yet, there is also evidence that suggests interaction over the internet to be both strategic and centralized, such as political economic analyses of Apple, Microsoft, Google, Amazon, and Facebook as global media corporations.54
Most importantly, however, the platform in its current use is a concept that offers a flat rendering of digital markets. Not all digital markets take on the character of a platform, and not all behavior on platforms is transactional. Explaining the web as organized around platforms often focuses attention on monopoly actors, such as Google or Facebook, and their competitors, rather than acknowledging how platforms relate to other actors. There is certainly value in studying even unsuccessful tech companies.55 Markets also never appear in isolation but are always codependent on other markets. As Karine Nahon has argued, being inside the box of platforms means focusing too much on one kind of power and one perspective of the problem, instead of looking at the more complex ecosystems with different forms of power working at different levels.56 Organizations need both an inside and an outside (and the inside/outside binary) to stabilize, and their study needs to cut across these different domains. Finally, the platform view tends to ignore so-called negative network externalities or side effects, like an enthusiastic freeway user attempting to ignore the risk of crowding.57 It is precisely those negative externalities or unexpected consequences of the social web that today are most widely discussed: surveillance and private data brokerage, precarious creativity, platform capitalism, and the unregulated global proliferation of intermediaries.58
For these and many other reasons,59 using the term platform comes with a great many caveats. Since it has been widely adopted, it will also be used in this book. Using the term, however, does not mean to stick to an essentialist idea of what platforms are as companies. To the contrary, we use the term here merely as a shorthand to evoke the “action nets” involved in making Spotify grow and prosper, that is, in accordance with the processual orientation of this research overall, the emphasis is on organizing rather than organizations. To think of Spotify and other digital macroactors as action nets means to grasp this quality of companies as temporary entanglements of unlike yet related actors.60
About This Book
While it is not the task of this book to review or remedy existing gaps in platform theorizing, we have to qualify and provide nuance to our use of the term—and our understanding of our research field—against this background of critique. In order to not render overdetermined the partly experimental or ethnographical work that follows, a few guiding signposts here may suffice. First, as already stated, an overall aim is to de-essentialize both the idea of the platform and that of the corporate organization, emphasizing instead the changing, interspersed, multidirectional character of Spotify. We strive to avoid an “ontologization of an epistemology,” that is, the notion that in order to understand digital culture, there is no alternative to the platform view (and the digital methods devised for studying platforms).61 Consequently, the book begins with the question “Where Is Spotify?,” opting to approach a taken-for-granted object through the history of its venture capital funding rounds.
Second, despite a surge of research interest in digital intermediaries, as well as Spotify’s self-definition as intermediary, we prefer to treat Spotify as a “mediator” in Latour’s sense, that is, as an actor that transforms, translates, and modifies the meaning of the elements it is supposed to carry.62 Chapter 2, “When Do Files Become Music?,” paints a vivid picture of the many mediating actors and infrastructures involved in music streaming today. In a less technical wording, which resonates with the traditions of qualitative social and humanist inquiry to which our idea of access is indebted, one may call Spotify a producer. In chapter 3, “How Does Spotify Package Music?,” the productive forms of refashioning files for various constituencies is laid out in detail. Chapter 4, “What Is the Value of Free?,” sets out to explore the market dynamics of Spotify. In doing so, it also sheds light on the relation between markets, variously theorized in previous literature as intertype competition, platformization, or platform systems.63
Instead of presenting knowledge as a set of prepackaged findings reducible to any one single concept, this book deliberately delves into the extensive, irreducible fabric of contingencies that marks the lived realities of our own knowledge production. Borrowing from Allaine Cerwonka and Liisa H. Malkki, we are ready to “improvise theory”—to recreate the centrality of surprise encounters with the often-undisciplined realities of our various research fields.64 While most of the mutual mentoring and productive misunderstandings among ourselves or between us and our informants—human and digital—have been edited out for reasons of readability, our book is still coauthored, interspersing research chapters with experimental interventions. The interventions are shorter texts that, in one way or another, interfere with Spotify and/or established research methods. They can be read independently, but they are also thematically linked to discussions in previous chapters.
Different approaches have informed and inspired the interventions in this book. Consequently, the interventions can be presented in two different ways. First, according to the methodological approaches they test and introduce: 1) following the controversies of a social media campaign; 2) establishing a record label for research purposes; 3) intercepting network traffic by way of packet sniffers; 4) conducting a reflexive analysis of methodological decision making; 5) building a digital application as part of an arts performance; and 6) engaging in web scraping of job listings. Second, the interventions could be presented according to the topics they concern: 1) the local political significance of tech corporations; 2) the power dynamics and financial logics of music distribution; 3) the infrastructural entanglements of streaming services; 4) the perils of collecting data; 5) online advertisement logics and the phenomenon of ad blockers; 6) and Spotify’s recruitment history. Our section on methodological decision making (Intervention: Too Much Data) critically reflects on the process of working with experimental digital methods and discusses data-driven knowledge production in relation to qualitative research. In general, however, the interventionist chapters are meant to serve as experimental suggestions—and practical “how-tos”—for future research. They frequently ask more questions than they provide answers. At the same time, they also offer concrete examples of how alternative and inventive research can be done. Therefore, one way to think about our interventions is to read them as provocation pieces meant to inspire research, without taking on a prescriptive character.
Importantly, this book itself has turned into an intervention, due to Spotify’s legal response quoted in the opening of this introduction. In the conclusion, we will follow up on how the narrative of this unplanned rapport between us and our object of study unfolded while we were in the very process of writing up the research. Back in May 2017, upon receiving Spotify’s request for information, we immediately assured Spotify, in written form, about the motivation of our research and the fact that all activities that potentially could be understood as a violation of Spotify’s terms of service had stopped. We never collected any private user data, nor did we reveal company secrets or damage the behavior or interests of subjects. Yet, bypassing our suggestion of a joint conversation about findings and methods, Spotify then moved on to contact the Swedish Research Council directly to suggest cutting the funding for the project. Given that such a move by a major platform appears to be unprecedented, we will dedicate some final reflections to the relationship between ethical and legal frameworks for research and to the ethics and legal frameworks that govern corporate behavior. While Spotify did not follow through with legal action, the questions raised in the process, and the public debate that accompanied it, are of interest for future research in this field.