MA Thesis
New Media and Digital Cultures University of Amsterdam
Student: Jan Zuilhof Student nr: 6152139 Date of completion: 14-‐11-‐2014 Supervisor: Michael Dieter Second Marker: Marc Tuters
The Soundtracked Self: Algorithmic
Individuation on Spotify
I would like to thank Michael Dieter for supervising this thesis. The conversions that we have had about the topic and relevant literature were truly insightful and have helped me a lot. His patience is also very much appreciated. Although he left the University of Amsterdam before I did, he continued the supervision process, for which I am very grateful.
INTRODUCTION 4
1. CAPTURING ATTENTION: LEISURE OR LABOUR? 6
1.1 INTRODUCTION 6
1.2 DEMOCRACY ON THE WEB 6
1.3 EXPLOITED PROSUMERS 8
1.4 ALIENATION: THE DOUBLE LOGIC OF PROSUMPTION 10
1.5 CONCLUSION 13
2. INDIVIDUATION IN THE COMPUTATIONAL INDUSTRIES 15
2.1 INTRODUCTION 15
2.2 PSEUDO-‐INDIVIDUATION IN THE CULTURE INDUSTRY 15
2.3 RECORDING OF TRACES: ASSOCIATION 20
2.4 CONCLUSION 24
3. ALGORITHMIC RECOMMENDATION: INDIVIDUATION THROUGH SPOTIFY 25
3.1 INTRODUCTION 25
3.2 BEATS IN BITS 27
3.3 SOUNDTRACKING YOUR LIFE 31
3.4 DOWN TO BUSINESS 40
CONCLUSION 46 LITERATURE 49
INTRODUCTION
"Maybe with motion sensors in phones, we can start guessing things like 'are you running, biking or driving?' Maybe it has a temperature sensor, or a heart rate sensor so we can get a sense of whether you're tense.
Maybe it connects to some other services, for example if we know more about your sleeping habits we know what time you're likely to go to sleep or what time you wake up it can be personalised."
-Donovan Sung- Product manager at Spotify (Smith)
When my girlfriend and I wake up in the morning, we will sometimes listen to the radio. Whether it plays mostly depends on how we communicate: “shall I turn it on?... shall I turn it off?... should it be louder? The decisions that we make are based on the opinions of two humans that either have to debate, or agreed upon it in the first place. Whichever one of those it is, we have explicitly expressed how we feel about it, and by this have made it into a collaborative practice. During the day, many more of these types of situations happen. When friends come over for diner, should there be music playing in the background? And if yes, which music will every one like at least to some extent? What radio station do the most people like to listen to at work? All of these things can be discussed, and people can communicate their preferences in a way that everybody understands. In the past years, however, another approach for selecting music has also gained popularity. Whereas humans have been curators during the first century of recorded music, this can now also be delegated to algorithms. Although both human-‐to-‐ human and algorithmic recommendations rely on the preferences of people, they also differ from each other; algorithms can process information in ways that go beyond human comprehension.
This thesis is about how such ways influence individuation. To look for those implications, Spotify is used a case study because the service has caught much attention in the past years and is explicit about its motivation, which is revitalizing the music industry. Why Spotify is also important is its data driven approach. In the quote on the top of this page, Spotify’s Donovan Sung speaks about possible future directions that Spotify might go in. Although this is still speculation, it does signify that the extraction of data might get even more abstract than it sometimes already is now. This abstraction is an important thread for my project here. In a sense, my approach is quite the opposite of
Spotify’s, which means that I will look for answers that cannot be found in data alone but should be searched in theorizations.
The research question that will guide me through this thesis is: how do the discursive
and technological dimensions of Spotify influence individuation? To find an answer to this
question, I will treat how relations between humans and web platforms can be theorized terms of labour in the first chapter. Concluding this chapter, I will offer a definition of the widely used concept of prosumption. In the second chapter, I will focus on theories about how the distribution of media influences individuation. This chapter also
functions to introduce the vocabulary of associated and dissociated milieus, which will be used later. The case study of Spotify will be treated in the third and final chapter. For this analysis, the theories about prosumption, association and dissociation will be used as a framework.
1. CAPTURING ATTENTION: LEISURE OR LABOUR?
1.1 Introduction
In this chapter, I will discuss the first part of the theoretical background for this thesis. Understanding relevant debates about the economy on the web and its influence on subjectivity is crucial for the analysis of Spotify that will follow in the third chapter. I will begin by briefly sketching out why the democratic and emancipatory character that the web sometimes seems to have should not be taken for granted, but instead
examined critically. Next, I will move into a more specific direction, which is a Marxist description of the economy on the web. Doing so, I will show that the traditional Marxist concepts of exploitation and alienation have undergone a transformation, yet are still present in today’s digital structures. This chapter also functions as an introduction to the second chapter, where I will go further into the influence of corporate orchestrated media on subjectivity.
1.2 Democracy on the Web
The web offers many opportunities for people to communicate and express themselves. However, these opportunities are restricted by technological rule sets. User generated content is not only information produced by an individual, but also a composition of the rule set of a specific platform and an act or acts performed by people. There are
numerous examples of websites and –applications that allow users to perform an act that influences what happens in that particular space. Social media like Facebook enable people to interact with each other, and potentially anything that can be expressed in written language can be discussed there. Instagram lets its users take any picture of the real world and upload it to their database. The rhetoric of these types of platforms often suggests that anything can be done. Facebook states that its mission is to “give the people power and make the world more open and connected” (“About Facebook”). Instagram lets people share their lives in an easy way for free (“Instagram”). But the way in which these platforms are set up contributes to what can and cannot be done. While a person can decide what status update he or she will like or comment on, it is Facebook that has decided that you can like-‐ or comment on a status update. On
Instagram you can share every moment of your life, but Instagram also arranges the set of filters you can choose from so your friends will see your life in a very ‘Instagram way’.
So although people obviously contribute to web platforms in their own way, the spaces and conditions through which they move partly shape their roles.
Furthermore, this is complicated by the way various platforms are
interconnected. Music platforms such as Spotify and Soundcloud, or video platforms such as YouTube and Vimeo all encourage their users to ‘share’ content on social media like Facebook and Twitter. Facebook and Twitter even encourage their users to link content between the two platforms, so that a message posted on one will also appear on the other. All of these platforms work according to their own logic and follow their own business model, and both them and their users have their own agendas. The result of this is a complicated system of spaces, structures, and acts, in which various actors with different needs move.
What this system – the web -‐ is has been theorized in various ways in the past decades. One term that has resonated for years was ‘web 2.0’ (O’Reilly 2007). It was used to describe a new structure of the web in which companies “leverage customer-‐self service and algorithmic data management to reach out to the entire web” (ibid: 21), and “network effects from user contributions are the key to market dominance” (ibid: 24). While O’Reilly mainly writes about business models, the same shift has also been noticed in cultural studies. In Convergence Culture (2006), Henry Jenkins argues that new media reshape old media. He acknowledges that these changes are corporate driven, but he also writes that it is up to the people whether they are ready for greater participation (ibid: 243). Many more have argued that this phenomenon – often
described as ‘prosumption’ – on web 2.0 platforms has changed the role of consumers in the sense that they are more empowered and conscious now, and the web is a largely democratic medium (Andriole 2010: 78, Bruns 2008: np, Deuze 2007: 244, Leadbeater 2008: 6, Lin 2007: 102, Tapscott & Williams 2010: np).
There are others, however, that claim that these accounts tend to focus on the aspect of participation without critically examining its implications. Authors should take into account that the web is still ruled by big corporations that turn user generated content into large profits. Van Dijck & Nieborg (2009: 869) argue that Jenkins recognizes the difference between producers and consumers, but his rhetoric is the same as that of business manifestos that do not make this distinction. Elsewhere, Van Dijck (Van Dijck 2009: 42) has argued that user agency should be understood in its complexity, since “the boundaries between commerce, content and information are currently redrawn”. According to Christian Fuchs, moreover, positive claims about the revolutionary characteristics of web 2.0 platforms should be more modest. The web has always been communicative, but recent developments such as greater bandwidth and
cheaper technologies such as digital cameras make it possible for people to create more content (Fuchs 2011: 289). So although the web might now be a space in which people can actively engage in content creation, it is important to critically theorize the social and economic implications of these new platforms.
1.3 Exploited Prosumers
One very specific direction that the previously mentioned discussion gone into is whether what happens on the web can be described in Marxist terms. Marx wrote the trilogy of Capital in the same period that the telephone was invented, and the web would not be there for another century. The object of his analysis was capitalism with factory owners and their employees, which seems to differ quite a lot from people engaging in online activities in their leisure time. However, since value is created by activities of people on the web, a Marxist framework can still be relevant to gain insight in the relations that exist between platforms and their users. Many web platforms are corporate driven, and since users generate content that is turned into profit, the question arises if they are being exploited. In this section, I will juxtapose different perspectives regarding this question. This is often strongly related to the concept of alienation. I will turn to this in the next section, and how this can be related to the web and prosumption.
To find an answer to these questions, I will begin with a brief explanation of what exploitation in the Marxist tradition is. According to Marx, value can be measured in units of time. The wage that the worker gets paid to produce goods is reflected in the value of the product. This is accompanied by the value of the goods that have been used, which is also based on the labour time spent to produce them (Marx 1972: 116).
However, there is a difference between value and price, since there does not have to be a correlation between the labour hours used to produce a good and the price that the seller sets. So value can be expressed in units of time, while the price can be expressed in money (Fuchs 2012: 634).
Marx called the part of the price that “the capitalist gets from the worker without the return of an equivalent” (Wendling 2009: 81) surplus value. Surplus value solely benefits the bourgeois employer, since the employed working class only earn their wage. The difference between the bourgeoisie and the working class is that the former owns the means of production, while the latter can only sell their labour time for a set amount of money. The surplus value generated in this process can be reinvested in more means of production, solidifying the position of the bourgeoisie, and making him or her
richer. In this scenario, the bourgeoisie earns money passively while the working class is exploited since people are not fully compensated for what their labour is worth.
There are significant differences between the web and the factories that Marx wrote about. Users of web platforms often engage in their activities without the expectation of economic revenue, while factory workers go to their jobs to make a living. However, web platforms are often corporate driven, and their users can generate surplus value just like factory workers. While users of platforms are not fully compensated financially for their work, Tiziana Terranova argues that it is too easy to simply dismiss labour in the digital economy as an “innovative development of the familiar logic of capitalist exploitation” (2000: 33). Rather, it reflects a complex relation between labour and people that is widespread in late capitalist societies. However, she also writes that the internet is one specific instance where this can be made apparent. For her, the digital economy is connected to the social factory that Italian Autonomous Marxists describe as a situation where work has shifted from the factory to society. This allows for labour to be seen as something that happens in society as a whole instead of in the factory alone. From this perspective, who is exploited where becomes much more complicated then when a clear separation between labour and leisure time is drawn. Whereas the factory worker was exploited during his labour, which had the sole purpose of earning wages, a member of the social factory contributes to the complete apparatus of society, while not necessarily getting paid. Like society at large, the web is a space where people
contribute “free labour” (ibid) because they like it, even though they generate money that they will never see.
This leads to some people stressing that the labour theory of value is not applicable to social media, because value is expressed differently. To “create and
reaffirm affective bonds” (Arvidsson & Colleoni 2012: 136) constitutes value, and this is not only to boost the sale of commodities but acts as a complicated network of exchange between firms and actors. The work that users of web platforms do is often beneficial for both those firms and actors, which would make exploitation an irrelevant concept. Furthermore, if the surplus value generated on those platforms would be divided among all of its users, this would hardly produce any noticeable income (ibid: 138). Moreover, the value of web platforms is often dependent on affective investments, which
…can be interpreted as a symptom of a transition away from a Fordist, industrial model of accumulation where the value of a company is mainly related to its ability to extract surplus value from its workers (to use Marxian terminology), to an informational finance-‐centred model of accumulation where the value of a
company is increasingly related to its ability to maintain a convention or brand that justifies a share, in terms of financial rent, of the global surplus that circulates on financial markets. (ibid: 145-‐146)
But even if this might be true, the creation and reaffirmation of those affective bonds is time consuming. The more time people spend on web platforms, the more data they generate by refining their profiles. This allows for better personal targeting (Fuchs 2012: 639), but also contributes to the financial bubble around a commercial platform. This means that, even if not in the traditional sense, people tweeting, liking and sharing more create more value by making a platform attractive for investors. It takes a lot of time and effort to establish and maintain the brand name that Ardvisson & Colleoni write about. However, individuals who do these hours of work do not see any money as compensation. Fuchs (2011: 298) argues that workers on the web can be divided into two categories, which are paid employees and unpaid prosumers. Ritzer & Jurgenson state something similar; for capitalists, unpaid employees are even better then low paid ones. For them, working consumers create “nothing but surplus value” (2010: 26). Although all these people do not get paid, they are constantly (and successfully) being persuaded to spend more time on web platforms. Using platforms is made
comfortable through the use of browser cookies that remember people’s preferences such as login details. But many websites also offer a personalized environment in which previously gathered data is used to select content that might be interesting to the user, avoiding the risk that her or she will get bored and disengage. Google Search takes location, but also search history into account when presenting its results for a query (“About Google”). Facebook and its algorithm make sure that users get to see stories of others they have recently interacted with, or content that concerns topics that they seemed interested in before (Owens and Vickrey). Websites where goods are explicitly sold like Amazon suggest more and more products to buy till infinity, but sites like YouTube and IMDB that are ‘free’ to use too constantly show their users what they also might be interested in. This capturing of attention makes sure that efforts of free labourers are indeed mutually “enjoyed and exploited” (Terranova 2000: 33).
1.4 Alienation: The Double Logic of Prosumption
Exploitation and alienation are concepts that are often related to each other. In Marxist theory, exploited labour alienates the factory worker, because he or she has no direct affiliation with or power over the product that he or she produces. To understand this
dialectic, it is important to grasp the distinction between objectification and alienation. According to Marx, objectification exists in the relation between human work and passive material. People can shape nature, and by this create a world in which they can live. In a capitalist society, however, this objectification becomes alienated. This is a mode of production where the “fruits and tools” (Wendling 2009: 15) of production are unjustly distributed. Employers own the means of production, and the workers do not appropriate the produced goods but instead get money in exchange. The worker becomes alienated from the product he or she produces as well as from his or her life activity, since the labour is “barbarous and detested” (ibid: 49).
For his theory, Marx falls back on Aristotle’s concept of use-‐value and exchange-‐ value. Whereas use value actually relates to what can be done with a product, exchange value signifies what one can expect in return if it is traded. For Marx, this also brings about a double character of labour. The first is qualitative, and the ones who did the work directly enjoy the result. The second is abstract labour that is quantitatively measurable. In the process of qualitative labour, which can be expressed in use-‐value, objectification takes place. Abstract labour, however, can be expressed in exchange value, and this is here where the worker becomes alienated (ibid: 52).
For Marx, this does not only mean that the worker is estranged from the actual product he or she produces, but also lacks self-‐realisation in the process. That the bourgeoisie owns the means of material production empowers them to also regulate mental production. Since the working class has no power over the material means, it is also subjected to the intellectual production that is scripted by the ruling class. Although alienation finds its roots in the economic and the technical, this psychological dimension in widespread in society. In this way, the concept of alienation transcends being a feature of specific economic relations to a mode that characterizes social life as a whole, which influences the becoming of subjectivity. This leads to a false consciousness in which the subject holds the “illusion of an individual producer whose subjectivity and sociality are founded in the acts of labour and exchange” (Ibid: 49).
Like exploitation, alienation on the web has been theorized and compared to alienation in industrial society. From one perspective, alienation has been reduced because
motivations for online prosumption differ greatly from factory work. In many instances, people engage in activities that they choose, not because the feel like they need to but simply because they like it. On platforms like Facebook, YouTube and Instagram, prosumption is enjoyed for both its productive and its consumptive qualities. They could not exist if their users would not feel any affiliation with them, because they
facilitate “consumption that is possible because it is simultaneously productive” (Rey 2012: 408). They simply need their users to like what happens there. In this sense, late capitalism actually benefits from minimalizing alienation, since reduction of
experienced alienation means more capturing of attention.
This attention, however, is not all that is captured. Whereas the factory worker saw and touched what he produced, the productive dimension of prosumption has a double character. On the one hand, people communicate intentionally. On the other, large portions of data are collected and saved into databases during the process of prosumption. The exact details of what is being captured remain hidden, so the prosumer never really knows what information he or she gives, and to whom. This metadata “or information about information” (Pasquinelli 2014: 14) is also a product of prosumption, but one that the worker has not direct affiliation with, and is thus
alienated from. Part of this is often used to enhance the use-‐value of the platform for the specific user. The user that becomes an “ambient producer” (Rey 2012: 410) can only guess which part that is. This can be related back to the Fordist mode of production, in which the factory worker is alienated from his product also because he or she only contributes to the end product by executing a task that is only a small part of the entire production process, while for the capitalist, it is the end product that counts. In
prosumption, two degrees of alienation take place. The first being that metadata is extracted; the second being that this is not even a complete product. So prosumption can be experienced as “socially constructed relations…” that are “…‘voluntary’ and
‘empowering’ (which at a lived, concrete level, they are) yet, in some fundamental respects, they are not” (Comor 2010: 319). This is apparent, for example, where data is being gathered of which the prosumer never knows exactly when, where, and what is done with it.
Furthermore, this metadata has exchange-‐value because it can be sold or directly used for better personally targeted advertisement. Moreover, improving the user experience using this metadata means that capturing attention becomes easier, resulting in even more data. An example of what can happen to the exchange value of a platform when all this data accumulates is the market value of Facebook, which is now over 200 billion dollars (“Facebook Inc. Stock Quote”). The alienation of metadata from prosumption is then directly related to the exploitation of prosumers.
The algorithms that aggregate and process this gathered metadata are the centre of growing interest (Pasquinelli 2014: 14). For Pasquinelli, algorithms should be divided in two categories; the first kind translates information into other types of information, the second one extracts metadata. This second category includes Google’s Pagerank and
Facebook’s algorithm, but also Spotify’s recommender algorithm that I will later discuss. These algorithms are continuously redesigned to meet the specific needs of the people that use them to generate profit, so that they facilitate “control, accumulation and ‘augmentation of surplus value’“ (ibid:15). Pasquinelli suggests that,
If Simondon recognized already the industrial machine as an info-‐mechanical relay between flows of energy and information, a further bifurcation of the machinic phylum should be proposed to recognize the information machine as a meta-‐informational relay whose algorithms handle both flows of information and metadata. Metadata can be logically conceived as the ‘measure’ of
information, the computation of its social dimension and its transformation into value.
Metadata is thus the crucial link here between prosumption, exploitation and alienation. It is extracted, abstracted, and processed into new information, generating surplus value.
1.5 Conclusion
In this chapter, I have treated various perspectives on activities on the web and what these imply. I have mainly focused on Marxist frameworks, and how these can be related to the web as it is today. Although the project of this thesis is not to engage with this specific vocabulary, these theories are very helpful in understanding the digital economy. People on the web are exploited, but the account of exploitation needs to be updated since exploited activity is also enjoyed.
Like exploitation, alienation on the web takes place in different dimensions than in the traditional Marxist sense. A double logic takes place, in which traditional
alienation is diminished only to make place for a newer, less conscious type that is, however, inherent to communication on many web platforms. This unconscious alienation will be the starting point for the next chapter, where I will treat some
theorizations about the relation between corporate owned media and human subjects.
At this point, I do wish to briefly reflect on some terms for the sake of my argument. Although a lot of literature in new media is concerned with web 2.0, I will avoid using this term in the rest of this thesis. The term inconsistently refers to many different platforms that work in many different ways and have different business models, if these are even present. Some are based on premium models for revenues, while others gather their income from advertising. Some combine these systems and work with a ‘freemium’ model, in which the paying user will get more options, customization and less
advertising. There are definitely patterns that can be recognized, but there are just too many different types of websites that overlap and a lot has happened since the term was first coined.
The term ‘prosumption’, however, I will use in my argument for pragmatic reasons, although I agree with Van Dijck (2009: 42) this simple term might distract from the complexity of the concept. It is important not to forget that many acts performed on platforms also shape that same platform in unique ways. The feedback loop between the platform and the actor constitute what both the platform and the actor can be. However, there often is a difference between what the prosumer produces and what he or she consumes, and wherever this distinction is relevant I will point this out (for example, Facebook users directly produce and consume the same type of content as well as produce metadata that they do not directly consume). So although I do acknowledge that there is difference between traditional consumption and prosumption, I want to make clear that the paradigm of production, distribution and consumption has by no means been transcended. Rather, I argue that prosumption exists as a connection between production and consumption as well as between different actors and their specific products. In this sense, prosumption should not be seen as an individual act but as an aggregate of collective activities inside a system. This aggregate of activities, however, has different consequences for different parties. Although I do realize that this is not a radical view, it should be pointed out because it is exactly the complexity of the concept that is relevant for my later analysis of Spotify.
2. INDIVIDUATION IN THE COMPUTATIONAL
INDUSTRIES
2.1 Introduction
In this chapter, I will move to a discussion of theories that concern the production, consumption and distribution of cultural products. Besides introducing a different approach to algorithms that can complement what I have written about in the previous chapter, this also allows me to get closer to Spotify as the main object of this thesis. I will begin this chapter by elaborating on Max Horkheimer and Theodor Adorno’s account of the “culture industry” (2002: 94), and where the links and discontinuities are with what David Berry has described as ‘computational industries’ (2014: 25). Next, I will move to Bernard Stiegler’s theory of the “programming industry” (2011: np), which both
criticizes and complements Horkheimer and Adorno’s analysis. I will, moreover, discuss Stiegler’s concept of ‘associated-‐ and dissociated milieus’ (2010: 82), and look at how this will prove to be useful for theorizing subjectivity in the digital economy. This account will then serve as the main theoretical framework through which I will analyse Spotify’s material dimensions in chapter 3.
2.2 Pseudo-‐Individuation in the Culture Industry
In the chapter ‘The Culture Industry: Enlightenment as Mass Deception” of their book
Dialectics of Enlightenment, which first appeared in 1944, Horkheimer and Adorno
(2002) introduce the notion of the culture industry to describe that cultural production and distribution have come to be industrial processes. For them, the culture industry impresses “the same stamp on everything” (ibid: 94). They argue that monopolies rule mainstream culture, because large corporations carefully plan the construction of cultural artefacts, which include architecture, but also movies and television. They write that all mass culture is the same in this situation, and its creators do not even pretend that it is art anymore. As the creative industries today do, people identified their fields of work as industrial already then.
Like Marx, they notice that this organizational structure has large implications for the subject, because those who have the economic means acquire power over society with technology. This is apparent, for example, in radio. Whereas communication over the telephone enables two parties to equally participate in a conversation, radio “democratically makes everyone equally into listeners, in order to expose them in
authoritarian fashion to the same programs put out by different stations” (ibid: 95-‐96). In this system, Horkheimer and Adorno argue, private broadcasters are denied freedom and have to work by the rules that have been imposed on them from above. Talent gets scouted by professionals and absorbed by the industry, with which all spontaneity is removed. So Horkheimer and Adorno’s analysis of the culture industry leads them to conclude that it is organized in a top-‐down structure in which people are subjected to ruling monopolies.
In various respects, this differs greatly from the way in which the creative industries on the web are organized. Algorithmic recommendation devices have replaced much of the careful planning of artefacts that is done by humans. Besides, the web is a space where communication heads in multiple directions instead of only top down. Recommender algorithms are set up to be a feedback loop between people and technology, distributing content by a logic that does not categorize groups of people, but targets individual users. Furthermore, actual content production and distribution are now for everybody; the technological means to make artefacts such as music, videos and blogs are affordable to most people, and everyone that is connected to the web has various options to choose from regarding the distribution of their duplicable creations. However, there is more to Horkheimer and Adorno’s analysis. According to them, the top down structure that they write about leads to a society in which people are ‘pseudo-‐individuated’ (ibid: 125). In industrial production, both procedures and goods are often standardized to maximize efficiency. The interchangeability of parts lowers production costs, since it is cheaper to produce one kind of tire that fits three cars, than design and produce three separate types of tire. However, Horkheimer and Adorno argue, this greater efficiency also lead to homogeneity in the end products that are sold. Many goods are largely identical, but carry different price tags. This is
illustrated, for example, by the Chrysler range and General Motors products being basically the same (ibid: 97). The products that are produced by the industry are categorized into different classes. The distinctions between goods are made very clear, so every consumer can choose which type of product he or she wishes to engage with. The part-‐interchangeability is hidden here, and individuality made clear. However, Horkheimer and Adorno argue, this individuality is not real because everyone gets the same product in a different package anyway. Adorno argues elsewhere that this also happens in popular music, where the structure of every pop song is the same, so that the trained listener will know when the chorus begins in advance, and the beginning and end follow simple harmonic schemes. Songs must become pre-‐digested to ensure that consumers do not have to struggle with them. But novelty is suggested with every new
hit song by introducing new hooks that are carefully worked out by professionals. The part interchangeability that facilitates this, however, has to remain hidden for this illusion of novelty to be sustained (Adorno 1941: np, Gendron 1986: 23).
However, while the production of music could be seen as an analogy to the production of cars, there are certain differences in their production processes that are worth mentioning. Bernard Gendron (1986: 28) argues that typical industrial
production involves a Fordist assembly line in the process, while for music, this
assembly line is only present in the production of carriers on which music is distributed, such as a vinyl discs or CDs. He writes that an original text, whether spoken or written, is not something that lends itself to be mass-‐produced. Recordings of songs are written texts in this context, and the assembly line metaphor can only be used for the
duplication of material carriers of these songs. In industrial production, part-‐
interchangeability is used because standardization makes production efficient. In music, however, this efficiency is reached only in the stage of distribution of copies of a song, a time at which the song has already been made. The production of this song itself does not involve an assembly line, so for Gendron, the concept of standardization for greater efficiency makes no sense here.
Figure 1. Loopmasters website with options for sounds in different genres. Loopmasters. Web. 9 Nov.
2014
What Gendron did not foresee was that some decades later, interchangeable parts of music are actually being sold in the form of sample banks and software presets, often pre-‐formatted to different genres to prevent producers from spending too much time looking for the exact right part that will fit their needs (see figure 1). It is no surprise that large corporations such as Loopmasters and Native Instruments orchestrate this
efficiency in music production. The parts are exchangeable to such an extent that it has been made explicit. While it could be argued that the ability to make music outside of expensive professional studios is empowering, selling sounds as parts also allows such corporations to explicitly push and re-‐articulate categorizations in a top down structure. This calls to mind Horkheimer and Adorno’s pseudo-‐individuation, but in this context choices about consumption are also about production. Computer musicians can buy pre-‐ recorded sounds such as drum hits, piano chords and vocals. These parts are
professionally made by recording and mixing engineers to potentially be used in many different songs. They can be browsed in music software, and the musician can make his own composition of fabricated, mass-‐distributed interchangeable parts. Traditional studio music production would have also involved musicians and engineers, but these would work towards an end product instead of an interchangeable part. In this sense, music software becomes a digital assembly line that uses duplicates of original recordings to save time and money. However, the mode of production is unlike the Fordist since the assembler does not repeat the same task over and over again. He or she makes choices about which interchangeable parts fit best. So whereas Gendron rightly notices that traditional studio music production does not follow an assembly line logic, these techniques are now indeed sometimes used to make music production efficient, but in radically different ways because individuals take care of the composition.
The material dimensions of digital music also imply different possibilities for distribution. In a digital domain such as the web, original texts can be reproduced till infinity at only the cost of disc space and the network that people are already connected to. So one relevant question that comes to mind is how distribution through streaming media is related to the assembly line metaphore. Streaming media serve end products; recorded, processed and compressed in music software, written on the server of platforms like YouTube, Soundcloud and Spotify, and ready for people to consume. However, the material dimension of distribution here lies in the connection between the server and the machines that request the information that will be played back as music. Files are only distributed when an actual request is made, and the only carrier that serves the sole purpose of transporting this specific package of information is written in code. While the assembly line in the production of music is to a certain extent more present then ever, it has started to fade away in distribution.
This assembly line then makes place for a new artefact: metadata that is used for profiling users as consumers for third parties as well as capturing attention on the platform. For Horkheimer and Adorno, monitoring of popularity also happens in the
culture industry. Both part-‐interchangeability and market research signify that cultural products can be seen as quantified goods. While the former lowers production costs, the latter makes it easier to categorize products and target consumers. Artworks are
succumbed to the instance of their economic value, which turns them into fetishized commodities of which the production has to be efficient, and how they are perceived becomes a quantified statistic. Horkheimer and Adorno write that the industry
designates this system as necessary because identical goods need to be distributed to an unlimited amount of consumption points. According to producers, the consumers would happily embrace this without struggle, since it already was their own wish anyway. Adorno and Horkheimer disagree, stating that the attitude of the public is part of the system rather than an excuse for it (Horkheimer & Adorno 2002: 96).
Like alienation for Marx, this pseudo-‐individuation can be seen as state of the people under capitalism, although the former directly addresses factory workers while the subjects of the latter are consumers. Control is key here, and the few set out the rules for the many. On the web, this control exists in protocols, or “conventional rules that govern the set of possible behaviour patterns within a heterogeneous network” (Galloway 2004: 83). This standardization of formats allows different types of software and languages to communicate with one another by encoding and decoding packets of information so that they can be properly transported. On the one hand, protocols are necessary for the network to function properly, but on the other hand this implies certain power relations that are backgrounded so that the rhetoric of web platforms can remain empowering and democratic. Whereas the culture industry’s control over
distribution was largely centralized, Galloway argues that “protocol is how technological control exists under decentralization” (ibid). However, recent accounts of large web platforms show that power is often also centrally organized again. As these “walled gardens” (Berners-‐Lee 2010: 82) grow large enough it gets harder for others to
compete, and since their walls are also written in protocol, it could now be argued that this is where control still exists after re-‐centralization.
So although the relation between media and people has changed a lot, the abstraction of the production process shows similarities with the way flows of information are
abstracted in software in general, and web platforms in particular. In both the culture industry and the computational industry, the end product that the customer gets served is the result of a machine of which the workings are intentionally made cloudy. Where standardization -‐ of material parts and production processes -‐ and market research
accounted for this in the culture industry, abstracted protocols and metadata characterize the computational industry.
Because Horkheimer and Adorno extracted their theory of pseudo-‐individuation from their analysis of the culture industry, and the computational industry has a
different economy, I will move beyond this concept. However, although the process of individuation through the consumption of cultural products has changed, it remains relevant for the analysis that follows. In the next section, I will turn to a more detailed account of individuation by introducing Bernard Stiegler’s work and the accompanying vocabulary.
2.3 Recording of Traces: Association
Whereas Horkheimer and Adorno were mainly concerned with how the culture industry imposes products from above, a framework in which the relation between the human subject and technology is emphasized more offers a different perspective. One of such theories can be found in Stiegler’s work, who criticizes the analysis of Horkheimer and Adorno because for him, the culture industry requires a critique of Kantianism, so using Kantian schematism to analyse it is not apt (2011: np). For Kant, the schema is a
category of the mind that comes prior to external images. Stiegler, however, argues that if it is necessary to make a distinction between schema and images, it should also be noted that these two concepts could only exist in relation to one another. The mind is always penetrated by external images. This originary technicity means that humans are the product of technology as well as the other way around, so for Stiegler there has always been a relation between humans and technology and what constitutes us a human beings is our relation to that technology (Lechte 2012: 80, Stiegler 2011: 42, Verbeek 2008: 388). This also helps to address the condition of subjectivity in a different way than I have done before. The Marxist framework allows for a theory in which technology facilitates the alienation of humans from their labour. But like the Kantian schema, the theory of alienation arguably assumes the existence of an authentic subject prior to relations with technology. By reframing both alienation and pseudo-‐ individuation as dissociation using Stiegler’s work allows me to focus more on the composition of human and non-‐human elements that interact with each other.
Although their method was invalid to Stiegler, in his 2004 essay ‘Suffocated Desire, or How the Cultural Industry Destroys the Individual’, he also writes that Horkheimer and Adorno were right about that culture industry’s “aim is to ensure the flow of new