The Problem with Muzak

Here’s another article about Spotify, a company that fascinates me. The Problem with Muzak starts by examining the mood playlists that feature prominently on the platform:

Spotify loves “chill” playlists: they’re the purest distillation of its ambition to turn all music into emotional wallpaper. They’re also tied to what its algorithm manipulates best: mood and affect. Note how the generically designed, nearly stock photo images attached to these playlists rely on the selfsame clickbait-y tactics of content farms, which are famous for attacking a reader’s basest human moods and instincts. Only here the goal is to fit music snugly into an emotional regulation capsule optimized for maximum clicks: “chill.out.brain,” “Ambient Chill,” “Chill Covers.” “Piano in the Background” is one of the most aptly titled; “in the background” could be added to the majority of Spotify playlists.

[…]

One independent label owner I spoke with has watched his records’ physical and digital sales decline week by week. He’s trying to play ball with the platform by pitching playlists, to varying effect. “The more vanilla the release, the better it works for Spotify. If it’s challenging music? Nah,” he says, telling me about all of the experimental, noise, and comparatively aggressive music on his label that goes unheard on the platform. “It leaves artists behind. If Spotify is just feeding easy music to everybody, where does the art form go? Is anybody going to be able to push boundaries and break through to a wide audience anymore?”

It goes on to excoriate the branded playlists and the idea that companies should need to “show the world what kind of music your brand likes to listen to while partying, driving, or enjoying a cup of coffee.”

It is absurd to suggest that a playlist created by Bacardi, Gatorade, BMW, or Victoria’s Secret could exist for any purpose other than the sale of its liquor, sports drinks, cars, or fancy lingerie. And this encouragement of a false sense of objectivity found on its Terms of Service is seen nowhere on its “Spotify for Brands” website, where it has published a series of articles luring corporations to the platform: “In the biggest game of the year, many of the ads feature music front and center, whether it’s a big hit like Eminem’s ‘Lose Yourself’ [Chrysler’s memorable 2011 spot] or an indie jam like Hundred Waters’ ‘Show Me Love’ [Coca-Cola’s 2015 spot],” the article explains, directly equating branded playlists to an expression of commercialism. “Using music effectively can also mean curating the perfect playlist that reflects the sound of your brand.”

Brand playlists are advertisements, even if Spotify strives to imbue them with so-called editorial integrity. Such uncompensated advertorial playlists are harmful in that they offer artists no option to opt-out, but also because they undercut what can sometimes be a valuable source of revenue for artists.

Last.fm was the only music social network that made sense

At present, Last.fm has a lot of difficulty generating a profit. Possibly because it no longer serves a purpose aside from logging what its users are listening to. It’s no longer a catalyst for discussions and events, given that there’s already Facebook and Songkick; nor is there need for a personalized radio thanks to algorithm-driven recommendations from various streaming services. In the end, the music industry to which Last.fm was a counterpoint no longer had to the power to create renowned musicians from meager local artists, nor direct public tastes: Today, labels only try to acquire, through an artist’s name, a preexisting community of fans that the artist garnered themselves. Last.fm didn’t pay a central role in the changing of this paradigm, maybe because it never understood how to make itself flourish economically. Investing in the concept of a personalized web radio and deciding to charge a fee for it turned out to be an unwise choice in an environment where music was practically becoming free and accessible, through tenuously legal YouTube uploads and the rise to prominence of streaming services.

Source: Last.fm Was the Only Music Social Network That Made Sense – Noisey

One way or another, a reasonable chunk of what I listen to ends up scribbling to Last.fm. But I can’t remember the last time it was any use to me—recommending a new artist, matching me with another user, suggesting events—all things it once did fairly frequently.

The article mentions its ill-timed sale to CBS and that is certainly a factor; hindsight tells us there were many more suitable partnerships it could have developed, although it would have required some fairly far-sighted execs to bring about any large success. Spotify was apparently in talks to buy Last.fm before it acquired The Echo Nest, which led directly to the development of their personal recommendation services, such as Disover Weekly and Release Radar, which feature regularly on this site.

Following Spotify playlist curators around New York’s live music scene

In an otherwise boring conversation about some press release or another, a Spotify PR person mentioned to me that an artist who had a big hit on the platform’s Fresh Finds playlist was discovered when one of the curators just happened to see them play a show in Bushwick. I was as surprised as anyone really can be by an email from corporate PR.

Fresh Finds is one of Spotify’s prized products, a weekly playlist crafted from a combination of two different data inputs: it identifies new, possibly interesting music with natural language processing algorithms that crawl hundreds of music blogs, then puts those songs up against the listening patterns of users their data designates “trendsetters.” What’s going to a show in Bushwick have to do with it? I had visions of a bunch of suits using their business cards to get into cool shows for no reason other than to feel like Vinyl-era record execs for a night. It seemed extremely redundant, and more than a little like posturing. Why bother?

“It’s basically their job,” I was told. Okay but, excuse me, how is that a playlist curator’s job? To find out, I asked if I could tag along with on a few of them on their nights out. I did not expect the answer to be yes, mostly because I thought it should be obvious that my intention was to point out how weird the whole thing was.

But the answer was yes. So, for three weeks, I went with Spotify playlist curators to live performances in Chinatown, Bushwick, and an infamous club on the Lower East Side. I got dozens of half-answers to the question: Why are you here?

Source: Following Spotify playlist curators around New York’s live music scene – The Verge

Spotify traditionally focused on using data and algorithms to surface new music. Apple Music, when launched, made a big show of their human-curated playlists. With the former’s interest in IRL listening, and the latter’s acceptance that computer-generated playlists can be good at scale, it seems like the differences are receding.

How machine learning finds your new music

I’m a sucker for technical dives into Spotify’s Discover Weekly, and this is a great one.

In the article, Sophia Ciocca gives three types of recommendation models that are used to generate the playlists. The first is collaborative filtering: crudely, your friends like this, you might like this too. Digging deeper, the mathematical modelling sounds fascinating. The third is raw audio models: analysis of the audio tracks themselves. This is why Release Radar works so well, despite the tracks not having been played many times.

But I didn’t know about the second one, the emphasis Spotify puts on natural language processing, or NLP:

Spotify crawls the web constantly looking for blog posts and other written texts about music, and figures out what people are saying about specific artists and songs — what adjectives and language is frequently used about those songs, and which other artists and songs are also discussed alongside them.

While I don’t know the specifics of how Spotify chooses to then process their scraped data, I can give you an understanding of how the Echo Nest used to work with them. They would bucket them up into what they call “cultural vectors” or “top terms.” Each artist and song had thousands of daily-changing top terms. Each term had a weight associated, which reveals how important the description is (roughly, the probability that someone will describe music as that term.)

Spotify’s Discover Weekly: How machine learning finds your new music

Unlike many others, I’m a fan of the Apple Music UI and implementation. But I’ve not had terrific results with their recommendation engines. The opposite is true for Spotify. It’d be nice to save some money by cancelling one or other of the services, but they do such different things for me that I can’t see that happening any time soon.

RapCaviar: music’s most influential playlist

Craig Marks for Vulture:

Moneybagg’s songs have already appeared on the regional playlist The Realest Down South (324,000 followers), but the dream is to be featured on RapCaviar — “one word,” reminds Basa, “because of Aristotle’s ‘the whole is greater than the sum of its parts.’” [Tuma] Basa selects the songs for RapCaviar on his own, utilizing predictive skills — “gut, gut, and gut,” he declares — honed at previous programming gigs at BET, MTV, and Revolt TV. “When something comes in and doesn’t smell right, I can detect it. ”

At Spotify, Basa also has access to a trove of data that enables him to gauge a song’s performance across the site: from how many times a song or artist has been searched for, to playlist-specific metrics such as percentage of people who skip the song (under 40 percent is desirable), percentage of saves to a user’s own playlists and percentage of users who listen to more than 90 seconds of a song, known as completion rate. When his instincts falter, Basa crunches the numbers. “Earlier this year,” he says, “one of my former co-workers at MTV called me. He’s business partners with XXXTentacion’s manager,” referring to the controversial underground rapper. “He said, ‘Yo, this guy is blowing up on Spotify.’ I said, ‘He is?’ I looked up his search results, and I’m like, ‘Oh shit. He really is.’” So I put it on Most Necessary, and reaction was instant.”

Part data scientist, part romantic laboring over a cassette mixtape, Basa sees himself as part of a hip-hop tradition. “Hip-hop has always valued curators: DJs, mixshow hosts, radio personalities,” he says. “This is just a different manifestation.” Officially, RapCaviar is updated weekly, whereby five or so new songs get cycled in, but Basa is often fiddling with it. Later in August, on a trip to Atlanta to kick off RapCaviar’s new series of branded concerts, he stops our conversation mid-sentence and grabs his Mac to add a track by rapper Ugly God, “Stop Smoking Black & Milds.” Over the weekend, he says, site search spiked for Ugly God, a leading indicator for Basa of vitality. To make room, he studies the data on a pair of J. Cole songs. One, “Change,” has a fairly high skip rate. With a couple of keystrokes, the song is ethered from the playlist. “If Tuma moves your song down on RapCaviar, your shit’s not working too well,” says Atlantic’s Greenwald.

The influence of Spotify’s curated playlists

Neil Cowley, writing in The Guardian about how his jazz tune became unexpectedly successful when a Spotify staff member added it to a curated playlist:

Some radio play and a few posts on social media meant that we got the track to 3,000-odd plays in the first couple of days. […] Enter stage left the “Spotify playlist”. Though I far from realised it at the time, this is the holy grail for independent artists such as myself. Overnight I was lifted out of the musty basement section where men with National Health spectacles hang out, and up on to the shiny new rack next to the checkout counter. All because I composed a solo piano piece that Spotify in deemed fit to feature on one of its more popular playlists. “Peaceful piano” with 1.9m subscribers put me in the company of Ludovico Einaudi, Nils Frahm and Max Richter and gifted me on average 25,000 plays a day.

The idea here is that people might not choose to listen to a broad playlist named ‘Jazz’, but they’d listen to the same songs if they appeared in the more specific ‘Peaceful piano’ playlist.

This feels like an ongoing shift in taxonomy that influences curation and UX copy. Presumably Spotify knows that users are less attracted to traditional genre labels, but prefer mood, activity or theme-based descriptions which might cut across multiple genres.

Setting aside user preferences for playlists over albums, it suggests that an artist like Cowley, despite enjoying more plays of this particular song, will see a much more modest increase in plays of the parent album which might well contain jazz music that isn’t ‘peaceful piano’.

Spotify’s Release Radar

Release Radar is Spotify’s latest personalised playlist. Whereas Discover Weekly updates on Mondays and takes its pick from all the entire Spotify catalogue, Release Radar updates on Fridays and focuses solely on the past few weeks’ releases.

Ben Popper, for The Verge, quoting Spotify’s Edward Newell:

When a new album drops, we don’t really have much information about it yet, so we don’t have any streaming data or playlisting data, and those are pretty much the two major components that make Discover Weekly work so well. So some of the innovation happening now for the product is around audio research. We have an audio research team in New York that’s been experimenting with a lot of the newer deep learning techniques where we’re not looking at playlisting and collaborative filtering of users, but instead we’re looking at the actual audio itself.

Discover Weekly is easily my favourite thing about any streaming service, and this appears to be just as good, in spite of the data challenges posed by focusing on new releases.

I got tracks by:

  • Favourite artists that I already know have new material out (Dinosaur Jr., Father John Misty)
  • Favourite artists that I didn’t know had new stuff (Wilco! Why didn’t anyone tell me about this?)
  • Long-forgotten artists I would likely otherwise never have heard of again (Cotton Mather)
  • Artist I haven’t heard of but seem up my street

It’s brilliant. My only issue is that these great features sit apart from my iTunes library, so Spotify can’t learn from my broader listening habits, but that’s clearly no fault of the product.

On what street did you lose your childlike sense of wonder

1: Karaoke ebooks

This is terrific. As you might guess from the ebooks part of the name, it creates Markov chains from your tweets, but it forms rhyming couplets and sets them to MIDI music. Brilliant.

2: Stop your team using technical terms and jargon – disambiguity

Most weeks I am ridiculed by someone for insisting on plain language – avoiding acronyms and technical language / jargon in particular. People tell me that I’m slowing the team down by making them use proper words, and that their end users or stakeholders expect them to use technical language.

These things are both true. You should still use plain language.

3: Across the USA by train for just $213

Traveling coast-to-coast across the United States by train is one of the world’s greatest travel experiences. Amazingly, it’s also one of the world’s greatest travel bargains — the 3,400-mile trip can cost as little as $213.

4: Scenes from our unproduced screenplay: ‘Strunk & White: Grammar Police’

BEAT COP
It’s over here, detectives. The body was found about an hour ago.

STRUNK
Use the active voice, rookie.

5: As the Guardian Berliner format turns ten, we look back at a decade of design change

Ten years ago this month the Guardian launched its Berliner format. We talk to its creative team about a decade of rapid change at the paper, and examine how design is now more important than ever in helping us navigate an increasingly complicated media landscape…

6: How to Have 106 babies (and counting)

Ed Houben is Europe’s most virile man. And after years of donating sperm the “normal” way (sterile room, cup, cash), he and some women looking to get pregnant for free began cutting out the middlemen and getting it done as nature prefers it (sex!). Today, Houben has over a hundred children—and Ed the Babymaker is in greater demand than ever. We imagine you have some questions

7: How Spotify’s Discover Weekly cracked human curation at internet scale

The algorithms behind Discover Weekly finds users who have built playlists featuring the songs and artists you love. It then goes through songs that a number of your kindred spirits have added to playlists but you haven’t heard, knowing there is a good chance you might like them, too. Finally, it uses your taste profile to filter those findings by your areas of affinity and exploration. Because the playlist, that explicit act of curation, is both the source of the signal and the final output, the technique can achieve results far more interesting than run of the mill collaborative filtering.

8: Me Inc.

The paradoxical, pressure-filled quest to build a “personal brand.”

9: P.G. Wodehouse On The Dangers Of Literature

It was one of the dullest speeches I ever heard. The Agee woman told us for three quarters of an hour how she came to write her beastly book, when a simple apology was all that was required.

And:

Freddie experienced the sort of abysmal soul-sadness which afflicts one of Tolstoy’s Russian peasants when, after putting in a heavy day’s work strangling his father, beating his wife, and dropping the baby into the city’s reservoir, he turns to the cupboards, only to find the vodka bottle empty.

10: Apologies To The Queen Mary turns 10

A truly terrific album gets a good anniversary review.

Apologies To The Queen Mary is far more approachable, an album that spins universal reverie out of family trauma, relational struggle, and spiritual crisis. It’s music that renders the horror and delight of life on Earth as an epic struggle we all share. “I’ll believe in anything!” Krug sings at the album’s peak, desperately reaching for a fresh start and the freedom of some anti-Cheers: “where nobody knows you and nobody gives a damn.” Apologies To The Queen Mary itself can function as that kind of common ground, a set of inspiring songs many kinds of people can rally around, if only for a few fleeting moments. A decade into its history, it remains music worth believing in.

11: Future reading

I’m not entirely swayed by this piece—straw men abound—but it seems to have gotten a lot of people talking about books and reading and formats and focus, and that can only be a good thing.

From 2009 to 2013, every book I read, I read on a screen. And then I stopped. You could call my four years of devout screen‑reading an experiment. I felt a duty – not to anyone or anything specifically, but more vaguely to the idea of ‘books’. I wanted to understand how their boundaries were changing and being affected by technology. Committing myself to the screen felt like the best way to do it.

11: Nihilistic password security questions

On what street did you lose your childlike sense of wonder?

12: WEIRD SIMPSONS VHS