Pushing music: People’s continued will to archive versus Spotify’s will to make them explore

Introduction

With more than 30 million songs available in most music streaming services, how do people choose what music to listen to? Do they explore the vast library of available music or stick with familiar tunes? Such questions go beyond the individual user to also include the economics and politics of streaming services, where certain user-patterns might be more desirable for the stakeholders delivering the service. This article uses Spotify as a case and explicates action potentials for saving and exploring music. I argue that Spotify is designed to push users towards exploring music, whereas users still remain attached to their personal music history. Granted, Spotify enables people to explore music and to maintain music. The database of available music, recommendation algorithms and social features for discovering music through peers provide users with opportunities to explore music. Features for saving albums and artists and for creating playlists give people means to maintain collections of ‘their music’. Yet, I propose that Spotify is designed to make actions for exploring music more perceivable. Unlike features for saving music, features for exploring move beyond being mere features and become autonomous actors in people’s music consumption. This skew towards exploring comes with a cost: streaming music leaves no permanent trace unless the listener takes further actions.

Music collections are firmly tied to personal meanings and life narratives (DeNora, 2000; Giles et al., 2007; Kibby, 2009; Van der Hoeven, 2018). Music experiences often co-occur with a certain time and place, and may later acquire an affective power as musical indices, evoking memories of the past at the same time taking on new layers of meaning (Turino, 1999). Yet, musical memories are subject to the trajectories of life (taking on additional meanings) and are intricately interwoven between personal and collective memory (Van Dijck, 2006). We may hence expect users of music streaming services to remain attached to their musical preferences and history, organizing music into repositories, even if these are not physical or based on individual ownership, and even if these repositories are at the risk of disappearance if the service closes down.

Whereas the database of Spotify is an ordered and flexible archive, the ‘database available in the cloud breaks the dialectic of subjectivity and objectivity, providing a system of ordering that is produced by someone else’ (Marshall, 2014: 68). We thus need to separate between the archive of the service provider and the individual’s need for her or his own archive. As Featherstone (2006) proposes, ‘the will to archive is a powerful impulse in contemporary culture’ (p. 595), the volume of digital information makes the will to archive ever more pressing. With a reference to Simmel, Featherstone argues that we have everything, but possess nothing, and we yearn for a framework through which we can make sense of the volume and speed of cultural products (Featherstone, 2000: 162–163). Whereas ‘archive’ in terms of a subscription-based access to (and not ownership of) music may appear as an ill-fitted notion, we can still expect the ‘will to archive’ to be present among Spotify users.

I have outlined an argument where the need and will to retain music in streaming services is explained as a way to make sense of the vast volume of content, and as a way to safeguard the personal meaning of music. This line of reasoning rests upon what Hesmondhalgh (2008) identifies and criticizes as the dominant conception of music, emotion and identity, where music is conceived as a ‘positive resource for active self-making’ (Hesmondhalgh, 2008: 330). Although people may indeed mobilize music in positive ways, Hesmondhalgh points to how an adequate conception of music and identity needs to enquire how such goals of self-realization are linked with power. The self as defined and constituted partly by the (cultural) products we consume align well with neo-liberalist conceptions of the self as an individual enterprise. Hesmondhalgh acknowledges that music matters for people, yet he reminds us to examine musical consumption and how music matters, or comes to matter, as imbued with the values of self-realization.

The objective of this article is to investigate exploring and archiving music from the point of view of Spotify as a service provider on one hand and people on the other hand. This implies paying attention to what happens in the space between the technology and the user. For this purpose, the notion of affordance is a pertinent analytical lens. Gibson’s (1979) theory of affordances has recurrently been proposed to enable a third way between technological determinism and social constructionism (Hutchby, 2001; Leonardi, 2011; Nagy and Neff, 2015; Schrock, 2015) by acknowledging the technical capacities of an object as well as human perceptions of how that object could be used. An analysis of affordances requires investigating music streaming services both as matter and as human practice. I hence raise the following research questions:

I address RQ1 by extracting key points from the extant literature in an analysis of Spotify as a music streaming service. This discussion is followed by an analysis of qualitative data from Spotify users on their needs and experiences with exploring and archiving music (RQ2).

Theoretical framework and related research

Gibson (1979) defines affordance as what an object offers an animal or a human being. Affordances are not the object’s physical properties, and they must always be considered relative to the acting being. A seat or a bench looks sit-on-able if this object is perceived as horizontal, flat, rigid and knee-high relative to a perceiver. Hence, the sit-on-able affordance of a bench is neither an objective nor a subjective property: ‘It is equally a fact of the environment and a fact of behaviour’ (Gibson, 1979: 179). Gibson argues that what we perceive when we look at objects are primarily their affordances. For example, when we look at a bench, we perceive it as sit-on-able, more than its separate physical qualities. The question in terms of Spotify is consequently, what do we perceive when we look at this service? As affordances are constituted in relationships between people and the materiality of the technology, this question cannot be addressed by examining human practice alone, or by examining materiality alone. It is an elusive concept; affordances relate to matter or the objective features and functions, yet affordances and objective features are not the same.

Affordances point to the action-potentials of an object, yet in order to examine and understand affordances, it is essential to start with the matter. My emphasis on matter as crucial for understanding affordance is an important positioning statement when compared with how affordance is typically appropriated in communication studies. Dreading the label ‘technological determinists’, communication scholars neglect the character of affordances as always relational, always both object and perceiver, and end up with an appropriation of the term that refers to how humans choose to use technology (Nagy and Neff, 2015). Given that ‘an affordance is a relationship between the properties of an object and the capabilities of the agent that determine just how the object could possibly be used’ (Norman, 2013: 11), I delineate the properties of the software (materiality) and how humans perceive and act upon these properties.

What does Spotify ‘want’?

The objective of the article is to examine exploring and archiving of music. The affordances of interest are concomitantly formulated as ‘explorable’ and ‘archivable’. Action-potentials for browsing music (material), connecting with friends (material), recommendation algorithms (machine agency), friend feeds (machine agency) and editorial playlists (material) point towards explorable as an affordance. In contrast, the opportunities to create playlists (material), save albums and artists in ‘Your Music’ (material) and play local music files stored on the computer (material) point towards archivable as an affordance. Spotify enables users to retrieve their recently-played music, yet this list merely features the recent listening history, and it is not a full, chronological account. This list of recently played music is hence transient, and is therefore not included as a machine agency feature pointing towards archivable.

If affordance concerns the action-potentials of an object, machine agency seems to interfere or push certain action-potentials over others. The machine agency of Spotify pushes people towards exploring and discovering music, whereas the personal and mnemonic meaning of music is supported via material features that ultimately rely on human action. Figure 1 responds to the first research question of how Spotify facilitates discovering, collecting and archiving music, yet it is still incomplete without the required specifics of how humans act (the second research question).

OPEN IN VIEWER

Figure 1.

The materiality and machine agency of Spotify are more strongly oriented towards explorable as an affordance than archivable.

Method

In order to examine how people express needs for exploring and archiving music, I rely on 23 focus-group interviews with 107 users (aged 18–57, average age 28) of Spotify. These interviews were conducted between 2010 and 2013, and hence before Spotify introduced ‘Your Music’ and before the introduction of personalized recommendations such as ‘Discover Weekly’ and ‘Your Daily Mix’. This study is also limited in terms of studying a relatively homogeneous group; most participants were relatively young, urban and expressed an above-average interest in music.

Participants were recruited at the pop/rock festival Øya in Oslo, Norway. The interviews were scheduled after the festival and conducted face-to-face. In the interviews, participants were asked about their use of Spotify, what music means to them and how they explore and discover music. The focus-group interviews were recorded, transcribed verbatim and then coded and analysed using Nvivo 11.

The analysis followed a two-step procedure. I first examined how each participant talked about exploring music and listening to music they are already familiar with, and grouped the participants into three categories based on whether they tend to use Spotify primarily for exploring music, for a combined pattern of exploring and listening to music they are already familiar with or primarily for listening to music they are already familiar with. Next, I conducted a thematic analysis and generated codes and categories from the data. A thematic analysis allows for key themes to be determined prior to the analysis, yet the specific codes and categories emerge through close reading and coding the data (Ezzy, 2002). Participants are anonymised in the below analysis, and referred to using fictive names.

How people express needs for exploring and archiving music

Table 1 presents an overview of the participants, grouped according to their listening patterns when using Spotify.

OPEN IN VIEWER

Table 1.

Three types of streaming listening patterns among participants.

Listening pattern	Example quotes
Explore: 31 participants Average age: 30 15 women 16 men	‘It’s cool to just come over something. To get lost in related artists, and to find new stuff, like related artists of related artists. I’ve found great music that I didn’t know I was looking for’. (Christopher, 30) ‘You’re being served with new music, and that’s great [. . .]. It opens up for new stuff that you weren’t aware of’. (Elisabeth, 27)
Explore and same old songs:56 participants Average age: 27 28 women 28 men	‘I listen a lot to music I already know. And I use related artists on Spotify. I like it, because I can check them out a bit. If I don’t like it, I move on’. (Anders, 23) ‘I make playlists, I have different lists. I have one that I’ve entitled a specific year. And then when a new year comes, only the best ones make it to the new list. Because I kind of miss how it used to be, like that album used to be that holiday. I miss that, and I want it to be like that’. (Malin, 25)
Same old songs:13 participants Average age: 26 6 women 7 men	‘I don’t listen to new music that much. I’ve returned to what I used to listen to. I know of the music before I play it’. (Andreas, 29) ‘I feel like a fossil or dinosaur. I’m not very oriented towards checking out new music. I like the bands I’ve always liked’. (Solveig, 30)
Un-coded: 7 participants	Participants who did not talk enough about listening patterns to be coded into any of the three categories.

Most participants value the opportunity to explore and discover music, yet they also remain attached to the music they are already familiar with. Table 1 hence hints at the importance for participants to have personal music collections, and hence archivable as an affordance. In the below analysis, I first discuss practices of exploring music, pointing towards explorable as an affordance, and next discuss practices towards playing the same old songs, pointing towards archivable as an affordance.

Concluding discussion

Therefore, what does the Spotify-machine ‘want’ and what do humans want when it comes to exploring music and archiving music? Figure 2 depicts explorable and archivable as affordances relative to the materiality and machine agency features of Spotify and human practice. Whereas participants value the opportunities to explore and discover music, their practices are equally directed towards archiving music.

OPEN IN VIEWER

Figure 2.

Explorable and archivable as affordances of Spotify, analysed as imbrications of materiality, machine agency and human practice.

The bidirectional arrow between explorable and archivable in Figure 2 refers to how the two are not trade-offs: what users archive relates to what they have explored, and what they explore relates to what they archive. The latter is even more the case after the introduction of personalized recommendations, where Spotify’s recommendations are based on individual taste profiles: your playlists and the music you play serve as input to the recommendations you receive.

The central claim of this article nevertheless remains that Spotify appears skewed towards prompting users to explore rather than archive music. This claim can be countered as Spotify grants people opportunities to save music and create playlists. Yet, it remains a legitimate claim, particularly if considering how the participants talk about music becoming ephemeral or transient. Add to this the seemingly stronger emphasis on machine agency for exploring. Whereas saving music and managing playlists depend on human action, Spotify acts as an autonomous agent with regard to tailoring and prompting recommendations towards users.

An important theoretical contribution of this article is delineating how objects with machine agency are different from material objects in terms of affordances. The action potentials of material objects can be considered as symmetrically constituted by what the objects provide relative to an active being. The action potentials of objects with machine agency interfere with this symmetry: the machine is designed and programmed to act on behalf of the human being, making certain affordances more perceivable than others. An affordance cannot be reduced to an object’s features or reduced to how people choose to use an object. Still, when objects are programmed to act autonomously, we may raise the question of whether or not this comes with the cost of reduced agency to humans.

Much as affordances cannot be reduced to what people choose to do with objects, the personal meaning of music cannot be reduced to what people choose do with music (cf. Hesmondhalgh, 2008). Both types of enquiries need to address how technologies, and cultural services and products, are embedded in economic and political contexts. This has implications for the analysis of musical consumption and for the analysis of the business logic of music streaming services.

First, with regard to musical consumption, access to vast libraries of music per se is no social equalizer. The accounts of the participants suggest that they manage and curate music and hence make their own archives to cope with the volume and speed of cultural products (cf. Featherstone, 2000: 162–163). Yet, the participants’ ways of managing music reflect how they consider music to be important in their everyday lives. Hence, they likely express a stronger will to archive compared with people for whom music plays a less significant role. This will is coupled with an interest and ability to invest the required time and effort to curate and archive. This study only hints at how the will and ability to archive (and in a sense to resist the push to explore) varies. Studies that build on empirical data from more heterogenous participants are needed to address questions of the limits of people’s ‘will to archive’.

Second, with Spotify as a case, it becomes central to discuss how the business logic of the service is reflected in the design of the service. If Spotify pushes certain behavioural patterns, can these patterns be explained partly as consequences of the interests of the different stakeholders involved in delivering the service? This question means moving from how to why Spotify is skewed towards exploring. Spotify argues that it is a strength that people increasingly use Spotify to discover music and that users listen to a greater variety of artists, and moreover points to their programmed features (e.g. Discover Weekly and Daily Mix) as ‘driving this growth in artist diversity’ (Erlandsson and Perez, 2017: unpaginated). Yet, the why-question calls for investigation: for whom are such user-patterns more desirable? Which stakeholders benefit more when people explore more?

With regard to information providers and search engines, Gillespie (2017) argues that ‘those interested in having their information selected as relevant will tend to orient themselves toward these algorithmic systems, to make themselves algorithmically recognizable’ (p. 64). This article asserts that Spotify appears to encourage (or push) users to explore, and does not delve into a discussion of the power to define what users are encouraged to explore, or if and how big and small music providers adjust to be picked up as relevant. Future research needs to examine the logic of streaming, and how different players respond to this logic. Such efforts may benefit from steering clear of the black box metaphor and instead follow Bucher’s (2018) lead and ask ‘how and under what circumstances different aspects of algorithms and the algorithmic are made available – or unavailable – to specific actors in particular settings’ (p. 55). Reverse engineering, experimenting with data inputs and data output, can give insight into how systems work (Bucher, 2018; Kitchin, 2017). In addition, studies of how music industry actors (such as composers, musicians, labels, producers, publishers) relate to and adapt to music streaming services are needed to provide insight into the economics and politics of the production and distribution logics of streaming.

Networked systems are designed by humans situated in an economic, cultural and political context. My contention is that we can only properly examine the human-machine networks of which we are now an entangled part by granting interest in both the human and the machine part of these equations. This is even more important as the technologies we use are increasingly embedded with autonomous components, designed to act on their own, and in the case of Spotify appear to direct towards one behaviour over another; explore over archive. The notion of affordance is pertinent for such analyses, yet requires investigation of what happens in the space between materiality and machine agency on one hand and human practice on the other hand.