No journalist has contributed more to our understanding of the streaming era than Liz Pelly.
A contributing editor at The Baffler and a lifelong DIY scene participant, she’s been investigating the inner workings of Spotify since 2016, writing a series of increasingly alarming stories that exposed the streaming giant’s black box of profit-seeking operations: mood-based playlists filled with mysterious fake artists, lean-back listening, algorithmic curation and ‘streambait pop’.
Her journalism has provided us with an arsenal of terms to better understand Daniel Ek’s dismal vision of context-free listening. And now she’s expanded her work into a book, Mood Machine: The Rise of Spotify and the Costs of the Perfect Playlist.
It’s essential reading for anyone who wants to understand how we came to live in a world where ‘coastal grandma’ is a genre and where ambient electronica playlists are filled with cheap stock music by unknown artists.
We asked Liz to explain the pivotal moments in this decade-long transition, whether Spotify has changed the sound of the underground too, and what Daniel Ek’s endgame might be. Get your hard hat on.
Elsewhere in this week’s episode, Chal runs down the best music she saw on a trip to Ljubljana’s MENT festival (3:49), we talk the films that could potentially tank a new relationship (9:47), Tom highlights a new release on CDR’s Pathways programme (13:00) and we even briefly discuss get-rich-quick schemes. Niche music recs and iffy financial advice? It could only be No Tags.
Some housekeeping: we’re going live again! We’ll be hosting an in-person No Tags as part of this year’s AVA London conference, in the British Library’s Eliot Room with special guest Paul Woolford. That takes place at 1PM on March 21st. Tickets for the conference are available now, and we’ve even got a 20% discount code for Taganistas to take advantage of.
Meanwhile, the second pressing of our book, No Tags: Conversations on underground music culture, has landed and should be arriving on your doormat any second now. If you missed out the first time, head to Shopify to place your order. The book can also be found in the following stores:
Bath: Magalleria
Bristol: Idle Hands, Rova
Manchester: Unitom
Leeds: Colours May Vary
Lisbon: Well Read
London: Artwords Broadway Market, Cafe Oto, Good News, Kindred, Magma, Photographer’s Gallery, Shreeji, Tenderbooks
Los Angeles: Ooga Booga
Stockport: Rare Mags
If you’re interested in stocking it, get in touch.
In the meantime, if you’re enjoying No Tags, please do rate, review and subscribe on your podcast app of choice. We’d also ask you to consider subscribing to our paid tier, which costs £5 a month and offers you nothing but the warm glow of philanthropy and a discount on our book.
CR: Liz Pelly, welcome to No Tags.
LP: Thank you so much for having me.
CR: Well, thank you for providing us, the listeners, with so many useful terms which have been really crucial to shaping a broader public understanding of what Spotify is and what Spotify wants.
TL: One thing I did want to ask straight away… as someone who’s run an independent record label for 15 years now, did you find writing the book as depressing as I did reading it?
LP: I feel like it came in waves. There were some moments where I was super in the weeds, going through the same material over and over to fact check, where I did have moments where I was like wow, I can’t possibly spend another moment being immersed in this way of thinking about music and culture. I just need to go do a DIY tour with one of my friends' bands, or book some shows in someone's basement… try to have a moment of palate cleansing. Luckily I did get to have a few moments like that while I was working on the book, but there were definitely some moments where it started to feel a bit grim.
CR: So over the years, you’ve examined some of the pivotal moments in Spotify's journey from its early form as this really handy app for free – well, pirated music, into a machine designed to keep people streaming, to keep people listening. Just one example in the book is this evolution of editorial playlisting into algorithmic listening – from the introduction of Discover Weekly to things like Spotify Wrapped, which is an entirely personalised read on what your music taste supposedly is.
How did that evolution change the jobs of people who were actually working at Spotify?
LP: I was trying to trace the periodisation of the past decade of Spotify, or trying to historicise it in some way, which is interesting because it is pretty recent history. And some of these changes or shifts were even happening during the course of my time writing about Spotify and the streaming economy. When I was writing the book in 2022, 2023, it had really only been about 10 years since the Spotify playlist as a cultural phenomenon emerged.
So I was trying to figure out what were the [key] points within this 10-year span where certain influential things happened. One of the things that came up a lot in the interviews that I did was the way in which the unexpected success of Discover Weekly changed the trajectory of the company. People I interviewed talked about how Discover Weekly's success created this new dynamic where the conversation was ‘OK, how do we make everything at Spotify more like Discover Weekly?’, it being their first big success with algorithmic recommendation.
There had been more rudimentary examples of algorithmic recommendation on the platform in the earlier years – artist radio, which was more like [early online-radio-cum-streaming-service] Pandora, or on an artist page you could press a button to get algorithmically recommended songs by similar artists. It was mostly being done using this pretty straightforward algorithmic process called collaborative filtering, which is finding people who like similar music to you and showing you stuff from their collections that maybe you haven't heard of before.
The Discover page was already on Spotify, but they found that a lot of people weren't using it and that people were spending more time on the Browse page looking at curated mood playlists and activity playlists. So they tried to figure out a way to bring the discovery aspect of the Discover page into the lean-back ease of the Browse page, and that first big success was Discover Weekly, this weekly playlist that you could pull up every Monday and would give you 30 tracks of music that you hadn't heard before. Which is so interesting now, because the way in which personalisation and algorithmic recommendation has changed is now so focused on your own listening history, it’s actually the total opposite to that original idea of the Discover page or Discover Weekly.
So the success of Discover Weekly was one thing that influenced the shape of the platform, and another would be the influence of TikTok later on in 2020, especially during the pandemic. I think there’s this general cultural shift that happened when platforms started realising that internet users and social media users were OK with this idea of a homepage or a feed where you’re not really putting in any input or making any decisions, and that people maybe even wanted this experience.
One of the former machine learning engineers from Spotify who I talked to referred to the TikTok feed as the ultimate distillation of lean-back listening, where you’re not really making any choices, and what you’re being recommended is just based on how much time you lingered on something or when you hit skip. Around 2018 and 2019, there was this really focused effort to rethink the Spotify homepage to try to shift people’s consumption over to this one interface where you would always get the perfect recommendation every time based on your listening history. A TikTok-like feed. I think that was another influential moment.
CR: You’ve mentioned lean-back listening there. That’s a term that’s coming from Spotify, right?
LP: It’s definitely a term that the music industry has started to use in the streaming era. I think the first time that I heard the phrase lean-back listening was in an interview with somebody who worked in digital marketing at a major label, for the first piece that I wrote about Spotify in 2016.
They explained to me that in the streaming era, the music business was starting to think more about this specific type of listener who was happy to put on a playlist and think more about mood or the activity and less about the artist. That’s certainly a focus of what a former Spotify employee described to me as the peak playlist era, circa 2016 to 2019.
This [peak playlist era] was a really helpful phrase for me in thinking about trying to historicise or periodise the past decade of streaming. There was this era where the curated editorial playlist on Spotify held a lot of cultural weight in the music industry – this would have been after 2015 but before Covid. In my book, I try to trace the rise of the playlist – the Spotify playlist in particular – as a phenomenon in the streaming era, and the rise of Spotify's playlist ecosystem and the different influences that pushed the company to embrace this more passive or lean-back listener as a new target user.
TL: In that period you're talking about, the peak Spotify playlist era, entire record campaigns were essentially built around playlisting. Everything seemed to centre around that first Friday your release was out and how many playlists spots you got, whether you were an artist signed to a major or an independent artist. It cannot be emphasised just how central playlisting was to release campaigns for a good two or three-year period. It can feel perverse looking back at it now, but it was a given then because those playlists were bringing so much revenue to labels and artists.
LP: Was that something that you were finding in your experience running a label?
TL: 100%. At that time, I was running my own label, but also managing campaigns for some much bigger artists, and across the board it felt like everything was driven towards playlisting.
And kind of perversely as well, it was a real source of pride. In some way it was the equivalent of post-gig Instagram DJ videos now, where artists would wait until Friday and then show off all the playlisting they got that morning. And to a lot of people that was how the success of a campaign, or certainly the success of the first week, was judged.
LP: What were the playlists that were considered target playlists for the type of music that you were releasing?
TL: So with my own label, Local Action, we release a lot of stuff but it's mostly underground dance music. So the highest targets we’d be pushing for… obviously there’s New Music Friday, but in terms of more genre-specific playlists it would be stuff like Metropolis and Pollen. Those were the real big ones to shoot for if you had a record that you felt had the legs to get to that tier, which was above smaller dance playlists like Altar. I’m aware of how silly all these words sound when you say them out loud! [Laughter]
CR: Tom, when we were talking about this before, you mentioned some of the tactics that Spotify employed then to make sure that playlists were the only focus of people's attention.
TL: Yeah, it feels like a significant thing that largely went unspoken in that period. There was a period in that peak playlist era where if a label or artist premiered music elsewhere before it was available on Spotify – whether it was traditional radio, or an NPR album premiere, or even SoundCloud – it was known that if Spotify spotted it they might pull playlisting from that track in response. Is that something that came up in your research?
LP: Something that came up was that artists were afraid to publicly speak negatively about streaming services for fear of retaliation or fear of being removed from playlists. And it sounds kind of conspiratorial now to talk about it, but actually, hearing stuff like this is a good reminder that these things were happening. There were consequences if you didn't play ball in the right way, which is super wild to think about.
CR: One thing that your writing has helped to bring to light over these years is the way that Spotify has actually affected the sound of pop music, particularly in the later 2010s. I’m referring to your concept of ‘streambait pop’, which is an adaptation of the idea of clickbait – content that’s data-driven or traffic-driven.
I'd be interested to know more about how you think this came about, but also, I find myself in some ways not really that bothered about this. The idea that pop music would evolve as a response to technical formats doesn't necessarily worry me. So I'm interested to know what you think the actual downsides are of pop music responding to format in this way?
LP: Well, I tried to be more intentional about the framing in the book. There's a chapter called ‘Streambait Pop’ that is an updated version of the essay I published in The Baffler in 2018, and I did try to be more direct in the book in explaining that this isn’t new and that the technologies and media formats through which music is created and released have always had an impact on the sound of music. Maybe this didn’t come through in the original essay, but I don’t think I ever necessarily was trying to suggest that this music is bad, or that music that sounds made for streaming is lesser.
But I think part of the goal is to show that this system, at the time, was always pitching itself as a neutral, data-driven meritocracy to independent musicians, but actually was a system that rewarded certain types of things. And there are certain types of music that are going to do well in this environment because they lend themselves more to the background, or to a mood playlist, or to a chill playlist, or to a sad vibes playlist.
Over the years, something I noticed, and sometimes felt regret over perhaps contributing to, was this dynamic of music journalists, when they didn't like something, calling it ‘streambait’ or something that sounds ‘made for Spotify’. Artists can have a whole catalogue of music, but often the thing that is rewarded by the systems of algorithmic recommendation or the playlist system is the most boring song in their catalogue, or the thing that sounds the most straightforward, the most palatable, the most background-music friendly. I did hear someone on a podcast describing music recently and they were like, ‘Yeah, it just sounds like filler music for your Daily Mix.’ And I think there's something true to that.
CR: Let us ask you about ‘perfect fit content’, because I think this section of the book is the most alarming. So around 2016, musicians and listeners started noticing tracks popping up on Spotify from labels called Firefly and Epidemic Sound. There are many more such labels, but these are two early examples. Their songs start popping up in these lean-back, mood-based playlists, and your journalism and digging over several years has exposed where these tracks have come from. It turns out they’re internally referred to as ‘perfect fit content’, or PFC, and are made specifically for Spotify’s playlist system and licensed to Spotify at a discount royalty rate.
It was a secret at Spotify for a long time. So tell us how you found out about PFC.
LP: Yeah, I feel like it was an open secret where people in the music industry knew there was something going on but didn't necessarily know the exact details. And now knowing how much of a revolving door exists between streaming services and major labels, it’s interesting to think about just how many people in the music industry have known about this over the past 10 years and just chosen not to talk about it, or maybe felt like they couldn't because of NDAs or whatever.
I first learned about this phenomenon in 2017 after I wrote my first article about Spotify. One of the first emails I got from someone was saying ‘Great article, but you should really research the fake artists on Spotify.’ I was like, ‘Oh, that's weird. That sounds kind of fake, but I'll check it out.’ But my friend David Turner was already doing some research on this and published a really helpful article right at the end of 2017 – so a really long time ago now. Through streaming data, he’d been able to piece together that there was this specific date when the Spotify playlist called Ambient Chill had most of its quote-unquote real artists removed from it, and the playlist was instead filled with tracks attributed to Epidemic Sound and Firefly Entertainment.
Journalists have been working on this story for a really long time. In 2022, there was a big investigation from the Swedish daily newspaper Dagens Nyheter, where they used copyright documents from Sweden to highlight that there was this really small group of songwriters that was actually responsible for hundreds of artist monikers and thousands of tracks all over the playlist ecosystem. But the question that still remained was: what is the deal that exists between these companies and Spotify? And how are these songs ending up on these playlists?
When I started working on my book, I knew something that I wanted to do was try to interview as many former Spotify employees as I could – and with this particular aspect of the story, I also really wanted to try to interview some of the musicians whose work was ending up on these playlists because they’re so mysterious. It was a real journalistic challenge, to try to get some of the fake artists to talk to me.
TL: Yeah, how do you try and find an artist that doesn’t exist?
LP: Well, really helpful to the process of reporting this part of the story was being able to review internal correspondence [at Spotify], Slack messages and internal documents. So the story started to crystallise through a combination of interviews and documentation of the practice internally at Spotify, and what I was able to learn from sources was that there is this small team of editors internally at Spotify responsible for looking after the lean-back mood music playlists. So, anything that would be a functional playlist for music for sleeping, meditating, focus music, studying, or any sort of ambient chill vibes playlist – essentially anything with chill instrumental music – is looked after by this specific team of editors who operate under this umbrella [which is] sometimes called 'perfect fit content’, and sometimes the team is called strategic programming or strategic content.
The way that they were describing PFC internally was ‘music commissioned to fit certain playlists and moods with improved margins.’ And talking to some of the musicians who did work for these specific licensors, they would explain to me an array of different practices, depending on which licensing company or which licensor they worked for. I heard things like, ‘I just make the tracks and submit them and get paid. I don't know what happens after that.’ Musicians would tell me about going into a session and cranking out 15 songs in an hour, just one take, one take, one take, trying to make as many as possible – with reference playlists of songs that had already done well within the playlist landscape.
Spotify has been responding to journalists saying that these playlists are specifically for lean-back purposes and account for a small percentage of total listening on the platform. And I think what’s important to keep in mind is that this way of thinking about value in the context of stream share or total streaming market share really warps our perception of what is a big deal and what’s not. Streaming services could say, ‘Well, it’s a small percentage of total stream share’, but even if it’s 1% or 2% of total stream share, Taylor Swift’s entire catalogue is less than 2%. In 2023, the entire genres of jazz and classical music accounted for less than 1% of total stream share on on-demand audio streaming services.
So 1% or 2% of total stream share might not be a lot to a streaming service, but it is a lot to musicians who have been trying to reach audiences through streaming services for years and finding it impossible to break through, or musicians who have found their work replaced by stock music on these influential playlists.
TL: And for a lot of independent artists making, let’s say, ambient music, these playlists were a significant percentage of their revenue from recorded income. And PFC takes that away, right?
LP: Since this story came out, I've heard from a lot of musicians who are grateful that this practice is having a light shone on it, because their streaming royalties were deeply affected by being removed from playlists, especially musicians in ambient music and the lo-fi hip-hop beats space.
TL: Has Spotify ever reached out directly to you about the book or your previous work?
LP: No.
CR: Ignore Liz Pelly and she might go away!
So we've got this music being made to order, and it's going onto these functional, lean-back and mood-based playlists, your ‘Beats to Study To’, sleep playlists and the like. I'm interested, as a fan of alternative, experimental, underground music – what is the result of this ‘perfect fit content’ music making its way onto playlists? What’s the bigger picture here in terms of an ambient playlist being replaced with PFC?
LP: I think there are a few things. I'm always thinking about the musicians in these musical traditions who are trying to figure out how to scrape together a living in the streaming era through different revenue sources. There’s a big impact on these musicians.
One of the many internal justifications for this practice that I heard from different sources was that internally, higher-ups would say things like, ‘We found a need for content. There’s not enough music that would do well on these playlists, so we had to commission this music because we didn't have enough for these playlists.’ But anyone who is connected to music culture at all will very quickly tell you that there’s no shortage of ambient, classical, jazz, and lo-fi music. There’s no shortage at all.
So the first place my mind goes is to the musicians operating within the worlds of classical, jazz, ambient and lo-fi hip-hop beats that haven't been able to meaningfully connect with audiences on streaming services. That’s a concern. But another concern is… well, I think I always think of it this way. I do a lot of lean-back listening in my own home. I have the radio on all day in my house. I listen to WKCR, which is the Columbia University radio station, they mostly play jazz and classical during the day.
CR: If that’s a university radio station and they’re playing classical and jazz in the day… is that also a ‘Beats to Study To’ channel?
LP: Well, in addition to playing this music, they also are really famous for their DJs who have a ton of jazz knowledge. There's one, he actually passed away a couple of years ago, Phil Schaap. He's a legend and he knows everything you could possibly know about jazz music, they are always playing archives of his programmes. So maybe this is more of a lean-back, passive listening moment, but you also have this DJ with incredible knowledge who is sharing every single detail about every recording session. There’s the opportunity for some musical connection to be made.
There’s no connection to be made when what you're listening to is just a bunch of stock music, other than a connection to the playlist itself. These are closed loops that point back to nothing other than the Deep Focus playlist on Spotify or the Study Jazz playlist on Spotify.
Creating this listening dynamic that points back nowhere other than to a button that you can press on a streaming service is a slippery slope. It really sets up listeners to not care at all about the artists they’re listening to and just be thinking about the vibe of the playlist. And so it’s not hard to imagine a scenario in which a listener who has been trained to think of their daily listening this way is totally fine with a playlist filled with generative AI slop that gets the job done.
TL: There’s a passage in the book where someone you anonymously interviewed says, ‘When you are a DSP and you have that much power and influence over people's education about music, it's a great responsibility. If I have a kid and I'm trying to teach them about the history of ambient music and go to Spotify, more often than not, what you'll find is PFC artists. I can’t teach my kid about jazz if they're listening to [Spotify playlist] Jazz In The Background.’ So it’s important to emphasise that we're talking about an erasure of culture here, an erasure of history and context.
CR: Do you think that Spotify has had an effect on the sound of underground or independent music more broadly?
LP: One of the interviews that really sticks out to me is one with someone who runs an independent record label. I think the quote was, ‘I don't know a single artist who listens to the Hanging Out and Relaxing playlist, but a lot of the artists that we work with are looking at their Spotify for Artists dashboard and seeing, ‘Well, this is the song that did well, so maybe I should lean more into the sound of this track.’ They had even noticed artists who aren't necessarily making pop music being influenced by the data they're being bombarded with daily through these apps.
Within independent music, maybe the influence has less to do with artists being nudged in certain aesthetic directions, but more to do with the fact that no matter what type of record you make, it's always the mellowest moment in the catalogue that seems to do well on streaming services.
Even when I was doing some of these interviews in 2022 and 2023 – because a lot of this reporting was happening in this post-peak playlist era, where you'd think the goalposts had changed – something that came up a lot with people from independent record labels was, ‘We've all given up on the whole playlist thing. We don't even think about it anymore because it’s just not part of our world.’
CR: Thinking about the future a little bit, Spotify is just one of many tech platforms that are steering us towards an increasingly automated internet experience, or life experience, where you've got bots talking to each other on X, you've got GPT scripts populating websites, stream farms and click farms… There’s this sense that human users are increasingly marginalised from this weird, simulated internet.
There were a couple of bits in your book that spoke to that. First, the co-president of Spotify describing his long-term view of Spotify as ‘self-driving music’, as if you could just turn it on and there’d be music for you – you wouldn't have to interact with it any more than that, it would just somehow know what you wanted. And you also describe going to an AI music tech fair, where people were showcasing their new ideas for AI-generated music tools. So what do you think the end game is for Spotify? What is it that they want to create in another five years?
LP: It seems to me that there are a lot of forces in the music business whose goal is to provide content as cheaply as possible. I hate the word ‘content’, but it’s what they use. They’ve been engaged in all these different processes of trying to reduce the cost of content, and it seems realistic to imagine that there will probably be other schemes that try in the coming years to reduce the amount of money they have to pay to rights holders.
If you’re a publicly traded corporation that has a responsibility to make returns to your shareholders, would you not pursue trying to make use of all the different tools of generative AI to provide content at a lower cost?
I’m not really in the business of future-casting. I don't know the future of music. All I know is that people who are trying to make money will figure out new ways to make money, and people who are trying to accumulate power will figure out new ways to accumulate power and influence.
We need to keep as big of a critical eye on streaming services as we do on the major labels. Right now, the amount of influence that Universal Music Group has over the entire recorded music business, through their deals with Spotify and other streaming services, is immense. Even just a couple of weeks ago there was a new contract signed between Universal and Spotify, and they really do seem to be positioning themselves as the ones who are able to rewrite the terms of how streaming services operate. It has an impact not just on Universal Music Group artists, but on anyone who relies on these tools in any way to piece together part of their livelihood. That is of equal concern looking forward.
TL: Piracy often feels like the elephant in the room in these discussions, and I was really glad that you address it early on in the book, because you can't have a conversation about the last 20 years of the music industry without accepting that recorded music largely lost its value around the turn of the millennium.
A lot of people, myself included, wrestle with the fact that they're philosophically pro-piracy, [as in] piracy as a democratising force, but it’s also file-sharing that created the landscape that Spotify and streaming services entered. To some degree, it’s the thing that got us in this mess. So I was interested in your wider worldview on this. Do you identify as pro-piracy?
LP: I think that I identify as not being against universal access to music, and I think there are better ways that we could imagine what that looks like.
A lot of people in the worlds of tech solutionism start rolling their eyes when I talk about the public library, but there’s a reason why there are 10 pages at the end of my book about public libraries. I know a ton of people, from when I started researching in 2021, at public libraries around the United States and Canada that have been working on local music streaming programmes, where they pay small flat fees to local artists to license their music for these small-scale, locally focused music streaming programmes.
When I started talking about this, tons of people I know started coming out of the woodwork, talking about how important the public library was to their own sense of musical education as a young person. And that was my experience too. I grew up in the era of Napster and music piracy mainstreaming, and I wouldn't be here talking to you all about music if it wasn’t for file-sharing opening up my sense of what music is. But similarly, going to the public library and borrowing a stack of CDs and listening to them all and then bringing them back was a really important experience.
So I don’t necessarily think that universal access to music is a problem. There are other ways that we could think about the library’s role, both creating these local streaming services, but also a lot of public libraries in the United States give people access to these services called Hoopla and Freegal, which are like library-funded music streaming services. I don’t know enough about how these services operate, but I do think the idea of your public library being what you turn to for universal music access seems so much more interesting than it being dictated by the major labels and streaming services, who are going to try to reduce the cost of content at every turn to shape what you hear.
To me, it’s not a problem of access to music. The problem is more like the way in which convenience culture has shaped a specific idea of what music is and what its purpose is.
CR: OK, last question as ever: please recommend us a film.
LP: Oh, that's interesting. My friends make fun of me because I see, like, two movies a year, I’m not really a big film person. But I saw two movies last year and they were really good. I saw Soundtrack to a Coup d’Etat, and I saw Union, the documentary about the Amazon Labor Union.
CR: That is so on brand.
LP: I thought they were both really interesting. I also saw a movie in 2023 that was really good about Robert Caro, called Turn Every Page. It was a big influence on me while I was writing my book. It’s about the relationship between Robert Caro, who wrote The Power Broker, and his editor, Robert Gottlieb – it’s about the author-editor relationship.
Sadly, Robert Gottlieb passed away last year, but it made a big impact on me because it chronicles Robert Caro's writing of The Power Broker and his books about Lyndon B. Johnson, for which he moved to Lyndon B. Johnson's hometown. Seeing that film inspired me to try to go to Daniel Ek's hometown. I was like, ‘Oh, I gotta go to Sweden.’
Share this post