Spotify as a Multi-Sided Market: An analysis of the evolution of Spotify’s platform

(1)

University of Amsterdam

Department of Media Studies

Master New Media and Digital Culture

Spotify as a Multi-‐Sided Market

An analysis of the evolution of Spotify’s platform

MA Thesis 2015-‐2016 New Media and Digital Culture

Date: June 24, 2016 Supervisor: dr. A. Helmond Second reader: dr. E. Weltevrede

(2)

1. Introduction

Being a music lover, I discovered different ways to collect and listen to music. In my early years, I used the vinyl records from my parents, which resulted in my love for music. I had bought cassette tapes before I found ways to record them myself, I had bought CD’s before I found ways to burn them myself, and eventually I found ways to download music, though illegally, which helped to develop my present broad taste of music. Being a computer literate, the fact that all the music was available for me for free allowed for a deep dive in music discovery. I used illegal file-‐sharing services in order to cope with my demands until 2009, when I came across Spotify’s music-‐streaming service. Despite many songs that were not available yet on the platform, it provided an easy way to listen to music. And there was basically no need anymore to download music illegally, while it did fill the (little) amount of storage capacity on my hard drives. Thus Spotify got me on board and it reduced (yet not ended) my illegal downloading activity and for the first time in years, I found a way to listen to music legally again. It made me wondering how Spotify had done this. And I surely was not the only one that “fell” for Spotify. Spotify’s platform completely changed the music industry in order to deal with the threats of music piracy (Cox) and in this thesis I will try to explain how Spotify made this possible and why it is still there.

While many paper have been written in the field of platform studies, these texts are rarely referenced in the emerging field of music streaming platforms. By using the different methodological approaches used in the analysis of platforms, I will give a basic framework for the description, analysis and positioning of this new sort of platform, the music-‐engineering platforms. I will base my analysis upon the business model of the multi-‐sided market that I will use as an analytical framework, and explain how the technical and expanding evolution of Spotify’s platform illustrates how this platform was able to rise and sustain in our contemporary society and to cope with the demands of their distinct groups of stakeholders. I will use the following research question, which will guide me through my research progress:

How does the evolution of Spotify’s platform over time relate to the way Spotify orchestrates its relationships with its most important stakeholders?

(4)

In the rest of this introduction I will reflect on the way digital music and cloud music services have affected the music industry. Chapter 2 will give a reflection on the way music-‐streaming services were likely to arise in our contemporary society, especially from a media-‐ecological perspective. In chapter 3 the different approaches within the field of platform studies will be described and discussed. Chapter 4 will indicate how these approaches can be combined and operationalized as basis for my research. Chapter 5 will give my analysis of Spotify’s platform development over time leading in to answering the research question put forward. Finally, a conclusion will be given resulting from my research on the rise of Spotify, and how this can be understood in the general context of platform studies and other approaches I discuss in this thesis.

1.1 Digital Music

The digitalisation of music has caused several problems for the music industry and it is struggling to stay in control on all fronts (Hesmondhalgh 57). When I speak of the music industry in this thesis my aim is to not only describe the recording industry which is many times mistaken to be the only group of companies within an homogeneous music industry, but also all organisations and individuals from other industries such as the live music and publishing industry (Williamson and Cloonan 314). The problem for the music industry was caused by the fact that digital music was easy to copy and circulate (Hesmondhalgh 59). Many platforms that supported unauthorized file sharing seized the opportunities this simple fact brought about. Hesmondhalgh describes four technological innovations, which were, according to him, responsible for the way digital music influenced our contemporary music industry. Firstly, the development of the MP3 format made it possible to compress audio files to make it suitable for computer storage. Secondly, the rise of high bandwidth connections, which made it possible for more and more people to download music online. The third innovation was the introduction of computers with increased storage capacity and possibilities to play music. And finally, the fourth innovation was the development of software that made it possible to convert audio files on CDs to the easily replicable MP3, and other formats

(5)

(Hesmondhalgh 59). All this created the fundamental base for illegal file sharing platforms to use peer-‐to-‐peer networks in order to make basically all music available for everyone for free. Obviously this was illegal and even to this day the music industry found itself in an on-‐going battle with unauthorised file sharing with a contemporary example such as Soulseek1_{.
The
digitalisation
of
music,}

although quite bad for the music industry, did create possibilities for music service companies to distribute music in a variety of ways. One early example is YouTube who created a streaming service to distribute music thereby bypassing legal concerns (Hesmondhalgh 63).

1.2 Cloud Music Services

Another significant development within the digital world, which was also important for the existence of a music-‐streaming platform such as Spotify, was the shift from software to services. In the past, people usually bought software that they could install on their personal computers. Nowadays, it is very common to subscribe to a platform and install a service. In this way, software has increasingly become a service that can be rented and does not need to be purchased (Kaldrack and Leeker 9). This development is what Kaldrack and Leeker call the Software as a Service (SaaS) business (10). The use of these services is no longer restricted to personal computers since the hardware has become more diverse with the introduction of laptops, smartphones, and tablets. And for these different forms of hardware, a variety of services have been developed and brought to the market. This development was enabled to rise due to the four technological innovations concerning digital music that I described in the previous paragraph and the SaaS business provided possibilities for the management of the increasing amount of traffic these technological innovations created (Kaldrack and Leeker 14).

The growth of the Software as a Service business goes hand in hand with a development coined as cloud computing. Cloud computing can be defined as “Internet-‐based computing, whereby shared resources, software, and information are provided to computers, and other devices on demand” (Kaur

(6)

373). Cloud services provide possibilities for large storage spaces which can be used for all types of data, for example music files, playlists, preferences, and other information which all can be accessed regardless the device of location (Morris n. pag.).

The introduction and development of cloud services and their supporting technologies provided opportunities for music streaming services that were not there before. The technical concept of SaaS and the growing capacities of cloud services enabled making music available to many more people on every location they want to and at their desire and at a cost level that made illegal downloading less attractive or desired. Hence it created a new form of music consumption, a new way of listening: ‘ubiquitous music listening’ (Morris n. pag.). A specific type of service that gained traction was the so-‐called music streaming service. In the next chapter I will describe how a media-‐ecological perspective helps to further explain the existence of the relatively new phenomenon of music streaming services.

(7)

2. Music as a Service

This thesis is organised around the way Spotify orchestrates its relationships with its most important stakeholders. Before I elaborate on the platform studies approaches and how these can reveal the specific way Spotify does this in the following chapters, it is necessary to explain how the development of the phenomenon of music streaming can be understood in the context of related aspects of our contemporary society. By investigating the way streaming-‐ services were able to enter and sustain in our society, it becomes possible to explain in more depth how Spotify became attractive for their distinct group of stakeholders in the first place because it is a music-‐streaming service. Therefore, I will use this chapter in order to explain how music-‐streaming services can be seen from a media-‐ecological perspective with a focus on the concept of new materialism. In paragraph 2.1, I will describe the relative media-‐ecological perspectives and how these can be operationalized in order to explain in how music streaming has become a common phenomenon in our society in paragraph 2.2.

2.1 Media Ecology and New Materialism

In the article “The intellectual roots of media ecology”, Casey Lum gives theoretical insights in what media ecology is. He describes media ecology initially as “the study of the interaction among various forms of media in the struggle for their own niche and survival in a complex ecology of social forces” (Lum 1). In the case of streaming services, the social forces of interest for my research in this chapter, consists of aspects that influence preferred modes of music listening in our society, such as the availability and accessibility of different music services in relation to the willingness to participate amongst music listeners and content providers on which I will elaborate later in this chapter. Many different scholars can be seen as influencing within the discipline of media ecology, such as Marshall McLuhan who has been known to study the relation between media and social change, and Neil Postman who is seen as the first that gave the media ecology paradigm a place within the field of media and communication studies (Lum 3). Also recently, various authors such as Michael Goddard provided a contribution on which I return in a while, what makes media

(8)

ecology a field of study with a theoretical perspective in development. According to Lum, the initial definition described above still gives a good perspective on what media ecology is, because it focuses on the most important aspect of media ecology, which is the complex relationship between media and the various forces in society (Lum 1). According to Postman, a medium is a technology, which affects the state of our contemporary culture, in a way that “it gives form to a culture’s politics, social organization, and habitual ways of thinking” (Postman 10). With this claim, Postman aims to say that media ecology is not only about an interest in the medium, but also about the ways in which interaction between a medium and people gives culture its character (Postman 11). In this way, Postman’s line of thought within the field of media ecology is connected with the concept of technological determinism. According to Dafoe (1049) technological determinism is about the assumption that technology drives societies social structure and cultural values (1049). But clearly technological determinism is more absolute in how technology influences how cultures and cultural values evolve over time than how Postman sees this. According to him there is an interaction between people and technology that leads to a kind of balance. It is less of a one-‐way street than the characteristics of technological determinism imply. Therefor, Goddard claims a critical reflection in this field was much needed. In his opinion this was achieved through the work of Matthew Fuller who claimed the study of media ecology lacked a focus on materialism, politics and complexity (Goddard 7). According to Fuller, we need to ask ourselves what the different material qualities of media are, and how their various rhythms, codes, politics, capacities, predispositions and drives can be interrelated to produce new patterns, potentials and dangers (Goddard 7). Goddard claims Fuller’s line of thinking is a crucial one, because its focus on the materiality, gives the academic field of media ecology a better connection with new media theory (Goddard 7). This is obviously also interesting for my research, since streaming services have a strong connection with the new material qualities of digital media and the possibilities for online music providers that were given by these developments. While Postman aims to emphasize the way how interaction between media and people gives culture its character, Fuller’s line of thinking is helpful to gain a more elaborated perspective within the study of the interaction

(9)

between media and people by focusing on the different material qualities of media.

Fuller’s focus on the material qualities of media connects to the concept of new materialism, which is described by Jussi Parikka in the article “New Materialism as Media Theory”. According to Parikka, the notion of materialism does not necessarily need to be connected to tangible objects. The complexity of computing and networks made place for a new conceptualisation of materialism, called new materialism. The concept on new materialism focuses on the way perception, action, politics, and meanings are grounded not only in human, animal, and tangible objects, but also in ‘things’ and even non-‐solid things (Parikka 96). This also includes non-‐touchable object such as “modulations of electrical, magnetic, and light energies, in which also power is nowadays embedded” (Parikka 96). In this way, Parikka claims we need to look at materialism in a more flexible way by proposing a multiplicity of materialisms, which makes it possible to analyse contemporary culture by focussing on a variety of specific forms of materiality. This makes it possible to include non-‐ solids and what he calls the ‘weird materiality’ of technical media, in order to understand how these new forms of materiality influence contemporary media culture (Parikka 99). I will use the concept of new materialism in order to answer on Fuller’s call who advocates media ecology needs to be more focused on material qualities of media. In this way, new materialism with its definition of extensions of types of materiality, including less tangible objects, enables me to analyse new media objects from a media ecological perspective what brings music streaming services into the picture and I will elaborate on that in the next paragraph.

2.2 Digital Music, New Materialism and Music Streaming

A significant moment in the age of digital music was the development of the MP3 format. Around 1995, the format came into use and it became a standard for music on the internet quickly from the moment the necessary encoding and decoding tools were easily available (Tilson, Sørensen and Lyytinen 4629). The digitalization of music through the use of the MP3 format enabled sharing music on a larger scale than ever before. The format was (and still is) easily replicable

(10)

and it removed the physically constraints tapes, CDs, and LPs had. Also, the MP3 format introduced a very high compression rate, which made it possible to store large selection of music files on computers. As mentioned earlier these possibilities that became available to many led to a high rise of illegal downloading and music sharing (Tilson, Sørensen and Lyytinen 4632). Most importantly, it created a transition from physical to less tangible music formats; a more manageable electronic file replaced CDs and vinyl records. This created a new form of materiality different from old music carrier formats. The intangibility of the MP3 format is difficult to define. Feathermann and Wells describe that the definition of intangibility has three dimensions, namely ‘physical intangibility’ which means the object has no access to human senses, ‘general intangibility’ which refers to the way that consumers can not define the object concretely in term of features and usage outcomes, and ‘mental intangibility’ which refers to the inability of consumers to gain a clear picture of the object in question (2). This definition shows it can be troubling to apply it to digital music since digital music has some access to human senses (the ability to hear sound), also it is possible to define it in terms of features and usage, and in a way people can also get a picture of digital music since the format is many times accompanied with pictures of album covers and other relevant pictures. From a media ecological perspective, it is interesting to take notice of the way these new material capacities of the digital music format created both new threats as well as opportunities for streaming services.

In the article “The Intangibility of Music in the Internet Age”, Styvén describes how providers of online music dealt with the intangible aspect of digital music and how these providers started to play a role in digital music consumption. Also, Styvén refers to the multiple levels of intangibility, which makes it hard to define digital music in terms of intangibility. According to Styvén, digital music as a market entity, can be partly tangible and partly intangible, but most importantly in the case of digital music is the fact that it changed a tangible aspect (physical CDs and LPs) into an intangible aspect, the MP3 format (55). The intangibility of online products became a central issue because it changed the way people listened to music. At the same time, people perceived digital music more as a service than a good, which resulted in a

(11)

difficulty of pricing (Styvén 54). The digitization of products led to an abstraction of the products, in the sense that people saw these products as less authentic and therefore less valuable than physical forms of the same products, such as CDs (Styvén 58). This was clearly a downside for music providers that embraced the opportunities the new format also had for them. The fact that people appreciated the digital format less than the physical forms was obviously an important factor for the battle between legal and illegal content providers, since most people perceived the intangible music format as an invaluable one as it was also available to download illegally for “free”. To this day, the new form of materiality of music carriers proved to be a challenge for online music providers selling musical content to consumers that would own and store it on their computers. The possibility to own and store music on computers seemed to be an important aspect of that challenge and music streaming services have found a way to adapt to the intangible nature of digital music and showed how a new form of materiality (the intangible format) could also create opportunities of how to deal with this differently.

What makes music streaming a phenomenon is the fact that music streaming can be seen as a service. This is in contrast with the Download-‐to-‐Own (DtO) business model, which is aimed at customers paying for songs to acquire property rights. With music streaming people use a service, which is characterised by yet another form of materiality, because people don’t acquire any property rights for the music they listen to (Doerr et al. 14). And in fact, the only possibly tangible aspect of the MP3 format, the electronic files on computers, have gone completely with music streaming services because one does not have these MP3 files on their computer anymore when music is listened through streams. Also the payment method for streaming services is different to the method used by DtO platforms. DtO platforms use a Pay-‐per-‐Download (PpD) method while some streaming services offered the possibility to stream music for free or to pay a monthly fee in order to listen to all available music from the platform in question (Doerr et al. 15). A well-‐known example of such a streaming service is YouTube. YouTube as a free service showed people could listen to the music they want for free. In order to cope with copyright infringements, YouTube was obliged to make deals with record companies such as Universal

(12)

Music, Sony BMG, EMI and the Warner Music Group in 2006, what decided that the record companies received a fee per-‐stream and a part of YouTube’s advertisement revenue (Kim 56). Recently, YouTube also provided a paid service called YouTube Red, what enabled paying users to make use of extra features such as the ability to play uninterrupted streams (without advertisements) and to use the service offline (“YouTube Red”). In contrast to DtO platforms, users have to be online (connected to the internet) to listen to free music streaming services. To resolve that limitation, streaming platforms started to make arrangements for listening to music offline. As the example of YouTube streaming service illustrates, many streaming services introduced the so-‐called ‘freemium’ model, which is a term coined by Fred Wilson. In a ‘freemium model’ consumers can use the platform for free and listen to the offerings of music while being online (‘free-‐tier’), and only if they pay a monthly fee they gain access to premium features of that particular platform, including the possibility to create offline playlists (‘premium-‐tier’) (Wilson). In addition, as subscriber to such a premium account, one can make use of community features and most importantly, one does not longer receive advertisements (Doerr et al. 16). The constant stream of ads (advertisements) was an important proposition in the business model for music providers to make the music itself available for “free”. Another important aspect for the existence of music streaming services is the availability of suitable infrastructures, which are fundamental for the distribution of music streaming services through all suitable types of hardware, including smartphones. Mobile devices are a significant new form of infrastructures that influence music listening practices. The development of this type of materiality, the one of the underlying infrastructure needed to listen to music, is also key to the success of music streaming services. As argued by Beer, the rise of mobile music technology has changed everyday practices (Beer 469). The infrastructures that support mobile music are part of the broader patterns of connectivity and informational extraction, which in turn is a part of the shift towards the harvesting of transactional information about our daily practices on which I will return in chapter 5 (Beer 479). The mobile devices with increasingly powerful microprocessors and storage capacity and their supporting infrastructures constitute a form of materiality, which is fundamental for the

(13)

existence of what I like to call ‘ubiquitous music listening’. This term is a derivative of what Mark Weiser coined as ‘ubiquitous computing’ what means that computing technology vanishes into the background, and is seamlessly integrated in our world, and therefore appears everywhere, connected with our daily practices (Weiser 94). For streaming services, the most important development in relation to ubiquitous music listening is the growth of wireless data services, what in turn has the effect that wireless carriers upgraded their mobile wireless networks to provide faster data rates for smartphones (Yang 344). The 3G cellular network exists for a while now, and more recently, wireless carriers started to provide the 4G cellular network, which supports higher data rates up to 42 Mbps (Yang 344). The increasing availability of the mobile wireless networks is an important factor for the growth of mobile cloud computing (Yang 353), which in turn is an important factor for the development of ubiquitous music listening within our everyday practices on which I will elaborate further in chapter 5.

From a media ecological perspective, the most important reason for the existence of streaming services is the sort of materiality digital music has evolved into, from vinyl records to CDs to MP3 files, to at this moment music as service with the intangible music format form as one can see it. Next to the fact, of course, the music providers were able to still make a viable business model out of it, while they enabled people to stream music for free. This came about for the consumers at the cost of having to bear with large amounts of advertisements of which, however, they could be protected from for a price for a paid subscription. The technical possibilities of music streaming with this new intangible digital music format proved to be a real threat for DtO platforms that use a PpD payment method, basically similar to the way people had always paid for physical music, but in this case for music one could not hold in your hands (like CDs) which made consumers less willing to pay. On DtO platforms, people pay for property rights to gain the possibility to own the music they buy and store it on their computers. Unauthorised file sharing also gave people the opportunity to store music on their computers for free, which became another, even bigger, threat for legal DtO platforms until this day. This also shows how the concept of materialism can be complex when applying it to digital music

(14)

since, in the case of DtO or illegal music downloading, people still have their music stored on their own computer, while the same format is being used as part of music streaming but in this model the consumer does not need to own it or have it on their computer or smart phone to listen to it. They can though, however, then they have to pay again for it. Music with property rights constitutes a specific form of materiality. Streaming services gave a new dimension to this material aspect. They decided to stop selling property rights and instead provided a service to listen to music instead of a tangible product that basically “carried” the music in it. In this way music streaming platforms showed how new material aspects of digital music created potentials and they started to rent out access to music.

(15)

3. Platform Studies

As I regard Spotify as a platform it is necessary to pay attention to the different approaches within the field of platform studies and the way platforms are conceptualised. In the last decade various contributions have been made, which can be coined as platform studies approaches. Since the word ‘platform’ can be used for many different objects such as hardware platforms, software platforms, internet platforms and in the case of Spotify music-‐engineering platforms, there is much difference between the approaches various authors have been advocating in order to analyse platforms. Not all approaches are suitable for my research what I will indicate, and therefore an important part of my research also consist of deciding what approaches do enable me to analyse Spotify’s platform. In this chapter I will describe the important theoretical and conceptual contributions in the field of platform studies that are relevant for the analysis of Spotify’s platform. Next to this I will describe how I think these approaches can be used in order to analyse Spotify’s platform and in particular to elaborate on the way Spotify’s stakeholder relations are related to the technical evolution of the platform. In paragraph 3.1, I will explain how platforms can be seen from a theoretical perspective and how studies of platforms were introduced. In paragraph 3.2 I will further reflect on a range of approaches in order to understand and explain how these approaches enable me to study Spotify’s platform.

3.1 Conceptualisation, Origins and Critique

As described above, the word ‘platform’ is connected to several connotations what makes it hard to conceptualize the ‘platform’ in singular way. Due to this fact, there are different approaches towards describing and explaining the concept of ‘platforms’. An important approach, which can be coined as a ‘platform politics’ perspective, comes from Gillespie who discusses the ‘platform’ as a theoretical concept in relation to internet platforms such a YouTube. According to Gillespie “‘platforms’ are ‘platforms’ not necessarily because they allow code to be written or run, but because they afford an opportunity to communicate, interact or sell” (351). Contemporary internet platforms face the same questions regarding their responsibilities as television networks did before

(16)

them, namely the questions of responsibilities towards their stakeholders (348). To address these responsibilities, platforms need to make use of different strategies, which aim to satisfy different stakeholders through different manners but with the same goal of attracting them and keeping them on board. To satisfy different stakeholders through different manners, the use of the term ‘platform’ seems to be an effective approach for content intermediaries such as YouTube. Gillespie claims, the use of this term becomes a “discursive resting point” (348) since there are benefits of making use of a term with different connotations. The broadly and discursive use of the term ‘platform’ reveals how many content intermediaries such as YouTube make it possible to present itself according to different principles to enact on their key constituencies (348). By using a successful example such as YouTube, Gillespie explains how they do that. Content intermediaries are often organised to present themselves strategically to three constituencies, namely users, advertisers, content providers and developers. This is what Gillespie means with the discursive use of the term, namely the fact that ‘being a platform’ enables these intermediaries to present themselves to those constituencies in different ways in order to become acceptable to each, and at the same time serve their own financial interests (353). Gillespie claims that the possible versatility of strategic approaches towards their key constituencies, establishes the satisfaction of the key criteria by which platforms are being judged (359). Since Spotify is also a content intermediary that has to enact on the demands of its distinct stakeholders, Gillespie’s focus on the way platforms communicate in various ways towards their stakeholders also becomes interesting while analysing a platform such as Spotify.

Platform studies were introduced in 2007 by Nick Montfort and Ian Bogost, who claimed that new media studies that are organised around the study of computers and culture, needed a deeper focus on the way platforms enable computational expression (Montfort and Bogost 2007, 1). They also published the book Racing the Beam (2009). In this book their aim was to “promote the investigation of underlying computing systems and how they enable, constrain, shape, and support the creative work that is done on them” (Montfort and Bogost, 2009, vii). According to Montfort and Bogost, every study of platforms

(17)

needs to be technically oriented, because if we want to research how the experience of developers and users who make use of the platform in question, relates to creativity, design, expression and culture, we need a serious investigation of computing systems (2009, 3). They distinguish five levels “that characterize how the analysis of digital media has been focused – each of which, by itself, connects to contexts of culture in important ways” (Montfort and Bogost, 2009, 145). These levels, which are layered upon each other, are (Montfort and Bogost 2009, 146-‐7):

-‐ Reception/operation, which can be analysed through studies of media effects.

-‐ Interface, which can be analysed through interface studies and other approaches such as visual studies.

-‐ Form/function, which can be analysed with ludology/narratology studies.

-‐ Code, which can be analysed with code and software studies.

-‐ Platform, which is the understudied layer from which the analyses needs to be fulfilled by platform studies.

Montfort and Bogost claim that a platform is the lowest level of the five, which is the most fundamental level and shapes the levels above (2009, 147). In this way they tried to explain the importance of the study of platforms, since the platform level influences all other levels. With their analysis they elaborated on how the material constraints of a particular hardware platform (in this case the VCS hardware of the Atari Video Computer) created possibilities instead of constraints for people that engaged with the platform (Montfort and Bogost 2009, 140). In this way they analysed a single platform “or a closely related family of platforms” (Montfort and Bogost 2009, vii-‐viii), in order to elaborate on it by investigating the material traits the platform embodies. While they base their analysis on particular sort of platform, which basically comes down to an analysis of hardware, they claim platforms can also be operating systems, programming languages or environments within operating systems (2009, 2). Dale Leorke claims that the Racing the Beam series produce a risk that might reduce platform studies in a standardised format that can be repeated on comparable platforms and, therefore, limits the approach rather than expands it

(18)

(258). According to Leorke, this is illustrated by the fact that the subsequent books in the series, which were released after Racing the Beam, apply the formula that was established by Montfort and Bogost by investigating the common material traits of a particular type of platform (260). While the platform studies series provided a framework for platform analyses, the specific method Montfort and Bogost were advocating to analyse platforms remained unclear. This is also described by Thomas Apperley and Jussi Parikka who claim that Montfort and Bogost did not contributed to a specific method for analysing platforms but rather just performed their analysis on the Atari Video Computer (2). Montfort and Bogost acknowledge this fact by calling out for contributions from all theoretical fields and research backgrounds to consider the platform topic more often and contribute to the understanding of platforms by doing studies which are centred around platform studies themselves (2009, 149-‐50). The approach that Montfort and Bogost are advocating is in my opinion too much based on the analysis of a particular type of hardware platforms, namely gaming consoles. Since ‘platforms’ nowadays embody a more varied range of objects such as Spotify’s platform, a platform that is difficult to compare with gaming consoles, I will focus on other approaches that I will describe in the next paragraph.

3.2 How Could Platforms Be Studied?

In this paragraph I will reflect on the approaches that can be useful to study my object of study. By reflecting on the various approaches I will explain how I follow certain authors in their approaches and how the combination of these approaches allow me to answer my research question. Firstly, I will describe how a historical engagement with platform can be achieved, secondly, I will describe the economic perspectives that allow me to study stakeholder relations, and thirdly, I will elaborate on software studies perspective that will help to elaborate on technical aspects of Spotify’s evolution, and how these are connected to the orchestration of stakeholder relations.

3.2.1 Historical Engagement With Platforms

Since my aim is to research how Spotify’s platform has evolved over time, is it necessary to develop a historical engagement with the platform. Therefore, I will

(19)

use this subparagraph in order to explain which approaches are taken into consideration for my historical analysis of Spotify’s platform. Various authors provided a contribution to the field of platform studies. One of them came from Anne Helmond. According to Helmond, the way to study platforms in a technically oriented manner (what Montfort and Bogost were advocating) can be achieved by analysing “a platform’s developer documentation, developer blog, company blog, privacy policy, terms of service or help documentation” (Helmond 2015a 18). By analysing these materials, Helmond was influenced by a software studies perspective from Kirschenbaum who advocates the method coined as ‘documentary analysis’. In the following quote, Kirschenbaum describes what his method consists of:

“Software is the product of white papers, engineering specs, marketing reports, conversations and collaborations, intuitive insights and professionalized expertise, venture capital (in other words, money), late nights (in other words, labor), Mountain Dew, and espresso. These are material circumstances that leave material traces -‐ in corporate archives, in email folders, on whiteboards and legal pads, in countless iterations of alpha versions and beta versions and patches and upgrades, in focus groups and user communities, in expense accounts, in licensing agreements, in stock options and IPOs, in carpal tunnel surgeries, and in the [former] Bay Area real estate market (to name just a few)” (Kirschenbaum 149).

Kirschenbaum claims the analysis of all of these documents and other traces is a fruitful method to recover digital histories (153). Kirchenbaum’s method of recovering digital histories is useful for my analysis since the analysis of these materials will allow me to engage with the evolution of Spotify’s platform from the perspective of the platform itself. In the chapter 4 I will further explain why Spotify’s own perspective is useful to reflect on.

Since the documentary analysis method is organised around the perspective of the platform owners themselves and therefore also restricted to their own perspective, it is necessary to reflect on other approaches. Another

(20)

approach to study platforms is advocated by Apperley and Parikka. They claim that the field of platform studies needs to be enriched by key themes within media archaeology (2). In their article, they focus on areas where the connection between media archaeology and platform studies could be interesting to bring new insights. The most important in relation to my research is the way in “which the archive can be used to reconstruct a platform“ (Apperley and Parikka 3). For Apperley and Parikka, the archives that can be used for platform research, can be called ‘paratexts’ (6). This term was coined by Mia Consalvo who described paratexts as the materials which “surround, shape, support, and provide context for texts” (Consalvo 182). They are actually texts about texts. Those paratexts consist of “delineated archives of developer interviews, end user responses, software and other material” such as journalism and marketing materials (Apperley and Parikka 6). In this way it is possible to reconstruct a platform and see how it evolved over time. To do this, researchers will need to make use of a large set of archives, which are available through publicly accessible paratexts (6). Since I need to make use of corpus materials that help me to explain how Spotify’s platform has evolved over time and there is not much written further about Spotify’s innovations, the paratexts which are available for me to conduct constitute an important part of my corpus materials on which I will base my analysis.

3.2.2 Platforms As Multi-‐Sided Markets

In order to elaborate on Spotify’s distinct stakeholder relations I will make use of economic perspectives that study platforms as ‘multi-‐sided markets’. Jean-‐ Charles Rochet and Jean Tirole approach platforms from such an economic perspective. They describe that most online markets such as platforms in software, portals, and media industries, bring together two or more parties (Rochet and Tirole 990). In this way they define multi-‐sided markets. Therefore, platform owners carefully need to employ strategies that successfully bring necessary parties on board (Rochet and Tirole 990). In many cases of multi-‐sided markets, platforms make profits out of some stakeholders, and have to take losses from other stakeholders (Rochet and Tirole 991). Another contribution

Spotify as a Multi-Sided Market: An analysis of the evolution of Spotify’s platform

Spotify as a Multi-­‐Sided Market

Table of Contents

1. Introduction

2. Music as a Service

3. Platform Studies

Spotify as a Multi-‐Sided Market