For Peer Review
Total Page：16
File Type：pdf, Size：1020Kb
View metadata, citation and similar papers at core.ac.uk brought to you by CORE New Media and Society provided by Repository of the Academy's Library “Splendid Isolation”: The Reproduction of Geographic Inequality in Spotify’s Recommendation System Journal: New Media and Society Manuscript ID NMS-20-1205 Manuscript Type: Original Manuscript extreme metal, Hungary, inequality, network analysis, popular music, Keywords: Forrecommendation Peer systems, Review Spotify, algorithms In this paper, we intend to tackle the issue of spatial inequality of music scenes on streaming platforms, through a network analysis case study of Hungarian extreme metal bands’ connections on Spotify, and by giving an overview of the role of algorithmic agents that shape and guide consumers’ access to musical content, and discovery of new music. We argue that the primary determinant of a given band’s international Abstract: connectedness in the ecosystem of Spotify is their international label connections. Those bands who are on international labels, have more reciprocal international connections, and are more likely to be recommended based on actual genre similarity. But bands who are signed with local labels or are self-published, tend to have rather domestic connections, and to be paired with other artists by Spotify’s algorithm according to their country of origin. https://mc.manuscriptcentral.com/nms Page 1 of 20 New Media and Society 1 2 3 4 Introduction: Inequality in the digital music industry 5 As in various aspects of everyday life, in the operation of cultural industries, including the music industry, 6 7 the reproduction of social inequalities is a determinative factor. The unequal distribution of resources, 8 income and potentials can be discovered in the dimensions of ethnicity, nationality, geographical location 9 and gender, among others (Berkers and Schaap 2018). Inequalities created and reproduced in this field 10 affect all elements of the value chain of the music industry: they appear either in the spaces of production, 11 in the careers of musicians, creators or in the distribution networks and on the side of the consumers. 12 With the digitalization of the cultural industries, time to time resurfaced the idea that with the 13 ubiquitous presence of the internet, the music industry would become more level, equal and democratic, 14 so all those inequalities would slowly disappear (Hesmondhalgh 2019). As the advent of the internet was 15 accompanied by the expectation that the network will moderate offline inequalities, so were the first 16 17 music-centered social platforms generating similar hopes. The MySpace-boom was fueled partly by the 18 faith that the platform will open up unprecedented opportunities for everyone, regardless of location, 19 ethnicity or nationality. Similarly,For streaming Peer platforms Review have taken up the role of the revolutionary social 20 equalizer later. However, reviewing the data and analyses and meta-analyses regarding the current state 21 of the cultural industries, this desired condition has not been fulfilled. Instead of democratization and 22 equalization, rather the restructuring of power structures and hierarchies take place. In virtually all 23 segments of the digitalizing society (Eubanks, 2018), so too in the digital music industry, reproduction of 24 already existing social inequalities is still prevalent. Among other factors, female creators are still 25 26 underrepresented (Wang and Horvát, 2019), and artists operating from a central geographical location 27 have still way more opportunities to distribute and communicate their work (Verboord and Noord, 2016). 28 One of the most important fields of the reproduction of structural inequalities in the realm of 29 digital music is the selection and curation of the available musical catalog. The selection of musical content 30 by various methods and practices is prevalent since the advent of the music industry. Gatekeeper 31 decisions have determined in the pre-digital commerce the catalogues of brick and mortar stores 32 distributing items of recorded music (vinyl, cassettes), the representation of creators the broadcasting of 33 their works in media, including the live shows and the DJ selections, and those fundamental mechanisms 34 remained unchanged in the digital music industries. In the currently dominant streaming ecosystem still 35 36 gatekeeper decisions determine the selection and recommendation of musical content, only the nature 37 and agency of those gatekeepers have changed radically (Barzilai-Nahon, 2008; Wallace, 2018). The role 38 and place of the previously reigning big publishing companies have been taken by digital platforms 39 (Hesmondhalgh, 2019), that became new gatekeepers in all branches of the cultural industries by 40 aggregating content from various sources, and building on the wealth of data collected from their users’ 41 interactions and behavior (Vonderau, 2019). 42 In this paper, we wish to analyze how geographic inequalities are reproduced on the leading 43 digital music streaming platform by algorithmic agents that shape and guide consumers’ access to musical 44 45 content, and discovery of new music. We intend to demonstrate the workings of this process through a 46 network analysis case study. Our main question is: how spatial isolation of Central and Eastern European 47 (particularly Hungarian) extreme metal scenes is represented and reproduced by the “related artists” 48 feature of Spotify’s recommendation system? Throughout the case study and analysis we emphasize the 49 importance of reproduction, by arguing that patterns of unequal distribution of music consumption and 50 discovery are not novelties of the digital space, but rather have their roots in the already existing 51 inequalities of the music industry. 52 53 54 55 56 57 58 59 60 https://mc.manuscriptcentral.com/nms New Media and Society Page 2 of 20 1 2 3 Personalization and algorithmic recommendation systems in digital music 4 Communication of a musical catalogue, or a particular music towards the consumers in the physical space 5 6 of brick and mortar stores have been realized typically by guiding the focus of the customer gaze. 7 Recommendations, shelf barkers, sale signs tried to grab the attention of the consumer and to achieve 8 that they choose one particular item from the available hundreds or thousands. 9 In the musical ecosystem available on the digital visual interface – displays of different forms and 10 size – nevertheless, it is not possible anymore to present the catalogue and direct the attention of the 11 user in such ways. The range of visible options is fundamentally limited: even if it is possible to arrange 12 twenty to forty items on a larger screen, the same cannot be done on a small smartphone display, only a 13 couple of albums or songs can be shown, which is only a fraction of the options to be seen in a physical 14 store. Under such circumstances it is of paramount importance, what is included and what is not in this 15 16 limited selection, and if the user interacts with the interface (clicks, chooses, skips an item, makes a 17 purchase), what further options, paths are offered to them by the system. A particular paradox is created 18 in this context. We access the endless catalogue of the “infinite shelf”, as Anderson (2006) called it, though 19 an interface which offers usFor a way smallerPeer initial poolReview of choices than a brick and mortar store with 20 definitely finite shelf space. By the convergence of streaming services (Amazon Music, Apple Music, 21 Google Play Music, Spotify) platforms less and less aspire at selling individual music tracks to costumers, 22 but offer music as a service, as a „utility” (Goldschmitt and Seaver, 2019), and by doing this, they connect 23 the previously separated practices of music purchasing and listening. On a convergent platform, a user 24 25 can be a customer and a listener at the same time. 26 In navigating the vast digital catalogues, curatorial assistance is essential for the users (Jansson 27 and Hracs, 2018), and the same applies to the provider side. Although streaming platforms often employ 28 human and robotized curators as well, Spotify for instance in compiling playlists, and Pandora in analyzing 29 and coding musical, aural traits (Bonini and Gandini, 2019; Fry, 2019; Morris, 2012) managing the choices 30 and deemed preferences of millions of users with human operators exclusively is practically impossible, 31 so this task is outsourced to the non-human algorithmic agents (Ricci et al., 2015). This, naturally, does 32 not mean that algorithmic recommendation systems are purely robotic agents, as they are designed, 33 coded, run and maintained by human beings. Thus, the difference between human and non-human agents 34 35 should not be radicalized, as (Goldschmitt and Seaver, 2019): 65) put it “...these music discovery tools 36 should not be understood in terms of the popular opposition between people and algorithms, but rather 37 as sociotechnical systems that rely on and reinforce particular ideas about human and machine capacities 38 in relation to music.” 39 This process in the digital realm, in which human-designed, robotized algorithmic filtering and 40 recommendation systems play a central role, is called personalization altogether (Prey, 2018). At the core 41 of the conception of personalization is the idea that users, if possible, should encounter only contents 42 that correspond with their presumed taste and expectation the best. However, the name 43 44 “personalization” itself is rather misleading, as the goal of the platform strategy is not to serve all the 45 possible individual demands but, based on the available personal data – browsing history, previous user 46 behavior, social media connections and interactions and other contextual factors, such as check-ins – to 47 achieve a flawless, “frictionless” stream of content consumption (Prey, 2018; Rónai, 2020) which doesn’t 48 necessarily is the most diverse content possible (Snickars, 2017). In executing this strategy, algorithmic 49 recommendation systems get the main role.