Data, Driven: Spotify Under Surveillance

An academic team's look under the hood of the music-streaming giant arrives as worries over Silicon Valley snooping go mainstream. Angus Batey surveys the bookshop backlash against Big Tech

In engineering parlance, a "teardown" is an investigatory disassembly. The idea is that, by pulling a piece of technology apart, the curious outsider might be able to work out how it operates, and perhaps use the insights gleaned to develop or enhance a competing product. The five authors of Spotify Teardown: Inside the Black Box of Streaming Music do not mean for their investigation to enable rivals to the Swedish-based business, and there’s nothing in their dense yet relatively concise book that is likely to prove particularly useful to anyone contemplating competing with the system. Rather, their intent is to shed some light on Spotify’s inner workings so that customers, suppliers and anyone curious about the course of popular culture’s mediation by technology start-ups can better understand what Spotify is, what it does, and how it does it – and, perhaps, by laying some of this bare, to hint at what it all may mean for technology, music and culture.

The result is a work that is, by turns, both surprising and banal, provocative and benign, empowering and frustrating. Its greatest strength is its principal weakness: a rigid adherence to an academic methodology that keeps the authors at a respectful distance from their material, roots the discussions in broad and deep contexts, but means that pretty much every punch is pulled, every blow landed softened by caveats and circumlocutions. This is despite the fact that the team spent longer, worked harder and deployed more wit, imagination and diligence when picking at Spotify’s seams than looks to have been the case in the vast majority of the media coverage the company has garnered since it was founded in 2006. Yet the researchers seem determined to hide their lights under an overabundence of bushels. Too often in their determination to place each and every original observation into a rigorous investigatory framework, they make the reader put in far too much effort to extract the valuable material from their excavation’s tailings.

Over the course of four years, the research team (Maria Eriksson, Rasmus Fleischer, Anna Johansson, Pelle Snickars and Patrick Vonderau – all academics at Swedish universities) probed the Spotify "front end" and "back end" with ethnographic, cultural and technical tools, and staged a series of what they term "interventions", designed to tease out an understanding of how the streaming service works on a number of different levels, and from a number of different perspectives. Some of these interventions are obviously frivolous – such as their launch, at a tech-industry conference, of an app called Songblocker, which turns the volume down on the songs streamed by a user and plays only the ads, which can be read as an exercise in trolling the tech start-up world – while others, such as the establishment of an independent record company, so that the interactions between music-makers and the distribution ecosystem could be observed first-hand, display more obviously serious intent.

The results were uneven, and quite possibly unrepresentative: as the authors are careful to point out, the toe they were able to dip into Spotify’s raging torrent of digitally mediated audio can only give a very limited impression of the system’s inner workings. The failure of one intervention, an investigation into how Spotify users are profiled by the system, is laid bare in excruciating detail: imagine the report The Fast Show‘s airily unembarrassable Professor Denzil Dexter might have written on one of his doomed-to-fail experiments (with several pages of footnotes) and you’re most of the way there.

Which makes Spotify’s response to the book, and the research that lies behind it, all the more curious. Despite tentative to-ing and froing at an interpersonal level between individuals within the company and members of the research team from 2014, there was no decisive centralised replies to requests for engagement – that is, until a few days after one of the authors gave an interview to a Swedish business newspaper in 2017, during which they noted the widely known if infrequently mentioned fact that, when the company’s service was first launched, none of the music shared on it was licensed from the copyright-holders, and many of the actual files available in its initial Beta version had previously circulated on the Pirate Bay torrent network. Suddenly, Spotify took a close and detailed interest, which manifested as a cease-and-desist letter sent by the company’s lawyers to the researchers, and another one to a public fund that had contributed towards research costs, claiming that the academics’ methods violated Spotify’s terms and conditions, and suggesting that the grants they had made be revoked. The fund, to its considerable credit, told Spotify where to get off.

These are starting to become rather troublesome times for the 21st century’s tech titans. After a couple of decades of having things their own way, the likes of Google, Facebook, Uber and Apple find themselves in a number of regulatory cross-hairs. The EU has levied massive fines for breaches of competition rules; newly strengthened data-protection regulators are eyeing American data-surveillance businesses and beginning to flex their significantly enlarged muscles. Social media firms stand accused of facilitating not just hate speech and online bullying but being the media of choice for terrorists and killers to plan and organise atrocities; of providing the systems that allowed foreign actors to exacerbate intra-national tensions and possibly to swing an American election and a British referendum; and of not just baking these potentialities into their business plans, but doing nothing about it because despite the fine words in the mission statements (and the sorrowful ones when it comes time to publicly apologise), the only metrics that matter are growth and profit.

How did we get here? In Zucked: Waking Up to the Facebook Catastrophe, a book that is part memoir, part mea culpa and part manifesto, written by venture capitalist, Facebook investor and former Mark Zuckerberg mentor Roger McNamee, a few answers are offered. Try as hard as McNamee might – and he does – they’re not particularly encouraging. Western governments, lacking nuanced understanding of the digital realm and wholly unprepared to regulate it, allowed dizzying amounts of power to become concentrated in a tiny number of hands, most of them belonging to upper-middle-class white men with similar (and similarly narrow) life experiences, each of whom seemed to adopt a stance on public safety and accountability that boiled down to saying: "Trust us, we mean well." McNamee spends a little time wringing his hands over his failure to see how ridiculous this all was until it was too late, but the majority of his book relates the voyage of discovery he took following the 2016 US election and his analyses of where Facebook went wrong – and how the company and its far-too-powerful founder failed to adequately respond to each and every misstep.

McNamee and Siva Vaidhyanathan, the author of Anti-Social Media: How Facebook Disconnects Us and Undermines Democracy, differ on their prescriptions for reform. McNamee thinks it’s possible, but not without government regulation, and that in itself will require changes to the law (or at least its interpretation for the past half-century) because competition regulation – the toolset most likely to be able to make quick inroads into the big companies’ monopoly dominance – tends to work to protect the customer from price rises: this is not something that will happen when services are provided at no fiscal charge. Vaidhyanathan basically seems to think we’re stuck with Facebook as it is because it’s too convenient for people to turn off, so we need to buy (and read) more books and newspapers to compensate for its wall-to-wall streams of divisive fake news and its ongoing creation of atomising filter bubbles.

Both are in firm agreement that Zuckerberg’s techno-utopian response to every new calamity – to issue some warm and woolly apology, then go back to doing exactly what the company was doing before; suggesting that the answer to the problems caused by Facebook’s technology is to deploy more Facebook technology – has long since passed its sell-by date. Neither realistically believe that the company will voluntarily cease doing things that make it money, even after evidence continues to pile up about how destructive to nations, societies and individuals their profit-generating practices have become.

Happily for Zuckerberg, if rather less so for the rest of us, his company does not yet operate in an environment where the penalties for believing your own Kool-Aid-inspired daydreams can be measured in time spent behind bars. On the other hand, the tale of Elizabeth Holmes and her apparent Theranos fraud, told quite brilliantly by John Carreyrou, the Wall Street Journal reporter who did much to break the story, in the barnstorming Bad Blood: Secrets and Lies in a Silicon Valley Startup, suggests that there may yet be some consequences, even for the best-funded and highest-flying members of technocracy’s elite.

It is also perhaps instructive to note that Facebook has cancelled its unmanned aircraft project, Aquilla, after several years and millions of dollars failed to produce a vehicle that would carry out the program’s ambitious objectives. Aviation is a highly regulated, safety-first marketplace in which the disruptor mentality – summed up in Facebook’s one-time motto-come-mission statement, "Move fast and break things" – will not work.

Or, at any rate, it hasn’t worked yet: it is possible to see hints in Amazon’s very few public pronouncements about how it intends to open airspace in cities up for its proposed fleets of delivery drones that the traditional disruptor mentality – deploy the technology, then if it turns out the law doesn’t allow you to do so, spend heavily on lobbying for the law to be changed – may still be one that the tech giants will pursue, even in the massively safety-conscious aviation environment. Disappointingly, none of these books touch on what appears to be the single episode in Facebook’s history in which it has attempted to disrupt a regulated market and failed.

Spotify Teardown is a product of an earlier moment. The interventions were staged before the Cambridge Analytica scandal broke and "fake news" became an issue discussed in Parliament select committees; before social-media companies began to attract the kind of negative attention that sees McNamee’s and Vaidhyanathan’s books vying for shelf space with Jamie Bartlett’s fine, short, to-the-point The People Vs Tech; Shoshana Zuboff’s extensively reviewed breezeblock The Age of Surveillance Capitalism; P W Singer and Emerson T Brooking’s provocative and pointed LikeWar: The Weaponization of Social Media; and, the best of the lot, former FBI and US Army officer Clint Watts’ freewheeling, frightening, fascinating Messing with the Enemy: Surviving In a Social Media World of Hackers, Terrorists, Russians, and Fake News.

Spotify is not, in conventional terms, a social-media company, and despite Facebook’s evident keenness to acquire the streaming service some years ago, it remains largely ignored by these other authors. Zuboff namedrops the company once, in passing, in her 690-plus pages, and of the others with an index, none mention it at all. Yet Spotify Teardown nevertheless contains a significant amount of material that needs to be fed into the wider debates around internet-enabled technologies in which the user is not the customer but an intrinsic component of the end product.

Although it presumably makes more money from its paid-for service, the free version of Spotify has more users – and adheres to the same micro-targeted ad-centric business model pursued by Facebook, Twitter, Google et al. It requires its unwary users to make a similar series of compromises with their identity, their information and perhaps even with their mindstate. In one of Spotify Teardown‘s more emphatic and disturbing chapters, the authors point out how Spotify’s playlists (some human-curated, some automatically generated in response to user input) operate as (unintended?) mind-control experiments.

As they show, the system’s functionality is predicated on a series of assumptions based on the idea that achieving and maintaining a generalised and unspecific notion of happiness is the ultimate goal of every human being. And, in marketing its playlists, the picture of the audience Spotify presents, and seems to imagine as the one its users aspire to, represents a desperately thin slice of a young, western, middle-class demographic.

Just as troubling is the way Spotify playlists are eroding the notion of music as art with intrinsic value – not just commercial or monetary value, but creative and cultural value. To Spotify and its playlists, music not only becomes a data point to help sell targeted advertising, but a tool to assist the user in carrying out a task: "Travel", "Chill", "Workout". The idea that music may have a purpose or an effect beyond providing an audio backdrop to other things the listener might be primarily focused on is lost – if it were ever part of the plan to begin with.

A particularly valuable insight from the book, challenging otherwise perceptive points in McNamee’s and Vaidhyanathan’s work in particular, is the authors’ gentle but sustained unpicking of the notion of a "tech platform". This formulation, they suggest, is merely another means by which tech companies attempt to keep themselves more than a regulator’s arm length away from traditional media outlets. Describing themselves as "platforms" enables these businesses to insinuate that they should not be governed by the same laws that proscribe and regulate the activities of publishers. Like Facebook and Twitter, which use their "platform" status to avoid the responsibility for libel that a newspaper publisher has to accept, Spotify hides behind the designation when it wants to sidestep regulations that would apply to broadcasters – even though, by curating its playlists and presenting its service as an enabler to "music discovery", it has long ceased to be a neutral connector of listeners and music. With its ad-supported business model, its original industry designation as "retailer" rather than "broadcaster" seems increasingly inaccurate.

None of this, of course, should come as any great surprise. Spotify has risen to a centrally pivotal position in the global music industry without even trying particularly hard to hide its roots as something other than a music-focused business. The company was set up as a technical solution in search of a problem. The name is deliberately meaningless, so it could be applied to whatever market it ended up "disrupting". There is limited evidence to suggest either of the company’s founders were particularly passionate music fans – certainly far less than there is to confirm their track records in technology-centred enterprises. Other aspects of the face it presents the world are similarly conflicted. The company trades on and leverages its Swedish heritage, but, as the Spotify Teardown authors patiently piece together, after its numerous rounds of funding and its floatation on the US stock market, Spotify is to all intents and purposes an American firm today.

Despite the mileage it has gained by presenting its service as a means of "saving" the music industry, it is not only Taylor Swift and independent labels who have been critical of the pitiful royalties Spotify pays to those whose work flows through its digital pipelines. Earlier this month the company, alongside Google, Amazon and Pandora, launched a legal appeal against a decision by the US Copyright Royalty Board to increase the rates it pays to songwriters in the US – rates that Billboard says have reduced from 15 hundredths of a cent per stream in 2014 to 11 hundredths of a cent per stream in 2018. Not only that, but Spotify joined with Pandora and Google to publish a defence of this position that appeared to argue that paying higher royalties would cause harm to songwriters. The company is also suing Apple over the cut the hardware manufacturer takes on subscriptions bought via Spotify apps downloaded from Apple’s App Store, and is taking on Warner/Chappell after launching its service in India despite not securing a deal covering that publisher’s catalogue.

If, as McNamee fears and Vaidhyanathan believes, we are long past the point where sufficient users will voluntarily unplug themselves from the social-media matrix to give company boards pause for thought about what they’re doing, what happens next? The authors of Spotify Teardown pointedly avoid asking the question, though the book closes on a bizarre and sudden change of direction, with a letter sent in May 2017 to the Swedish data-protection authority suggesting that Spotify’s business practices may violate fundamental terms of the EU’s General Data Protection Regulation. The effect is as if the writers chose to end the song with a rising change of key that implies a further verse, yet to be written.

In February, Spotify announced it was profitable for the first time in its decade-plus history, but it expects to go back into the red in the next quarter and remain there overall for the financial year. That three-fronted legal fight needs to be seen in this context: as an aggressive move to increase the margins for a business that claims nearly 100 million subscribers and twice as many active users, yet is still unable to make money for its investors and shareholders. In this broader business context, Spotify Teardown will register as little more than a nuisance. But the picture the book paints of the company and its product, and the way Spotify has reacted to the research the book is based on, ought at very least give its users some pause for thought.

Spotify Teardown, by Maria Eriksson, Rasmus Fleischer, Anna Johansson, Pelle Snickars and Patrick Vonderau, is published by MIT Press

The Quietus Digest

Sign up for our free Friday email newsletter.

Support The Quietus

Our journalism is funded by our readers. Become a subscriber today to help champion our writing, plus enjoy bonus essays, podcasts, playlists and music downloads.

Support & Subscribe Today