Deezer Flow

Deezer’s “Flow” is an intriguing idea. 

A unique, endless and personalised playlist for every user. The first suggestion in the main tab. It is a clear alternative to the endless lists of lists that other streaming services usually display.

Perhaps one bold suggestion is more valuable than a lot of scattershot alternatives and a solution to the “paradox of choice”.

I am curious how well it actually works.

Official description: “Flow is all your favorite music, mixed with fresh recommendations and songs you forgot you loved. This feature gets to know what you like and what you don’t and plays an infinite stream of music chosen just for you. All you have to do is press the button.”

Interesting features: the “angry face” icon gives some tools to shape the flow:


  • “Do not recommend this track”
  • “Do not recommend this artist”
  • Skip next track
  • Manage my exclusions
  • “Change mood”

The last is intriguing. I counted 47 moods: pop, hits, rock, electronic, happy hour, 2000, 80s, Party, Sertanejo (lol), French rap (hmm), Zen, Folk, Rainy Day, Disney, Road trip, Work out etc. Is it really a curious mix of genres, actual moods, contexts and time periods. 

They could use a better name than “Change mood” (Something like “Shape the Flow” is more enticing and correct) and the feature could have its own main icon, instead of being hidden in the “Angry face” menu.

One usability complain I have is that in flow mode there is no way to manage the queue: you cannot see upcoming songs, nor add a track to the queue.  It is a bold design choice, but it would be extremely convenient to add stuff to the queue and then automatically return to the flow. Most of the time I want to combine both things: add one or two particular songs, and then return to the flow. It is a shame to break the flow every time this happen.




The democracy of music as a streaming commodity

As someone who does not love podcasts, I feel slightly annoyed by Spotify directing so much energy towards them, but I can understand. Podcasts are “cheaper” and allow exclusivity. (Funny to think that music ends up being both a commodity and the more expensive product.)

Anyway, I am happy that music remains a commodity. Imagine how awful is music streaming resembled video streaming and its endless, relentless fragmentation of content libraries.

By the way, the commodity aspect of Spotify (and the others) never ceases to amaze me

To think that everyone in the world (with some regional exceptions) has access to a reasonable approximation of the entire universe of “commercial” recorded music for a small fee is still quite thrilling. 

Not a long time ago, accessing music was much harder. You had to live near a music store. Or a library. If you did not live in a major center, only when traveling you could visit a large store with wider selections. Every disc was a marginal purchase, and quite “risky” too, if you did not like the songs. There was a physical limit of how many discs to store in your house, or take with in the car or any kind commute. Discs or tapes would get lost or damaged. Or you would leave them at the summer place, and spend the rest of the year far from them. 

Anyway, it is fine that Spotify is pursuing this opportunity of “copyright arbitrage” and building of exclusivity walls,  but let’s not forget to observe this curious moment in history, where “everyone” has access to “everything” in music. How this democratic component affects the culture? Anything you enjoy you can share with everybody, while previously, music knowledge was much more stratified: you had to live near Tower Records, have a lot of empty bookshelves etc. etc. How does this affect the music we enjoy and identify with?

How does this affect the music we seek, enjoy and identify with? This seems an interesting topic to explore…

(Previously a twitter thread).

The Missing Songs

Have you ever tried to listen a favorite song in a streaming service only to find out it is no longer available? The first time I faced this “problem” was a few years ago, when I looked for songs from “Avonmore” – Bryan Ferry’s 2014 album, and did not found any. Since I recall listening to it on Spotify upon its release at the end of 2014, I was very surprised not finding it a couple of years later. Actually all I could find available was a related album: Avonmore – The Remixes, from 2016.

As far as I know, this disappearance was geographically limited: music streaming users in US for instance never saw the non-remixed Avonmore disappear. Fortunately the problem is no more, and I can stream “Loop de Li” and “Driving Me Wild” as many times as I wish. The most remarkable thing from the episode is the total “opacity” of the issue: there was no way to know why it happened, whether or when they would return, which “regions” were affected besides Brazil.

It just happened. And  a relevant album by a well know artist on a major label (“Avonmore peaked at number 19 on the UK Albums Chart and number 72 on the US Billboard 200” says Wikipedia) simply disappeared from streaming services. A related, less relevant remix album stayed there. And just as suddenly, the album reappeared years later. Is this normal?

I really appreciate that Spotify provides us the option to at least “see” unavailable songs, in a greyer text. (For what its worth, the Avonmore songs were completely absent, not present as grey ghosts).

This playlist by Shannon Liao is a clever idea: missing songs only! (For the record, 4 of the 42 songs appear to me as available. Progress!)

Thinking about missing songs always makes me remember this fantastic essay by Stephen Thomas Erlewine from 2016, “Why the Death of Greatest Hits Albums and Reissues Is Worth Mourning”:

Undoubtedly, modern listeners gain much from the history that’s just sitting on streaming services, waiting to be accessed. Amateur curators and music journalists are taking the time to trawl through the past to assemble shareable playlists that serve the same needs as old reissues. In some respects, this is an improvement over the old way of doing things—after all, it’s now possible to augment a collection of big ’70s hits with cuts by Zeppelin, Pink Floyd, or Paul McCartney and Wings, all artists who were unavailable for previous comps due to licensing restrictions—but the digital past is notoriously mutable.

Take “The Bomb in the Heart of the Century,” a terrific Spotify playlist of music from the year 1950 assembled by Michael Daddino a few years ago. Celebrated among music critics, this playlist is strikingly similar in breadth and ambition to a classic Rhino compilation. But within months of its debut, some of its featured songs were dropped from the service. Why bother with an archival project through a streaming service if it can all disappear without notice?

The myth of the Celestial Jukebox—that idyllic stereo in the sky that provides instant access to all recordings ever made—lies upon the assumption that because music is theoretically available for all to hear, the past is preserved and easy to access. Reality isn’t quite so simple. Since the dawn of recording, our musical history has always been inherently tied to the existence of physical, sellable product: The initial release created the history, and the reissue facilitated the writing of history, whether it was through carefully constructed archival projects or the existence of re-pressings of popular titles. Pink Floyd saw its catalog jump from EMI to Columbia back to Capitol/EMI, each getting somewhat ballyhooed and publicized reissues each time, the ad campaigns and endcap placements keeping the titles prominent; now they are just there, waiting to be called up by the user, if they care. When Van Morrison chooses to revive his catalog at this point, stories are shared on social media, then the music fades back into the vast digital chasm.

Finally, if I were to pick one unavailable song now, my choice would be Body Next to Body, the Falco/Brigitte Nielsen duet produced by Giorgio Moroder in 1987. It is nice to see it as song #3 of disc #2 of the FALCO 60 compilation and somewhere in the world someone might be streaming this towering achievement of 80s excess right now.


The music video is mesmerising (and streamable where I live):


How Much Music is Really Released per Year?

Every time I read a statistic about how much music is released for any given period I have more questions than I had before. 

For example, from a 2010 Billboard article: “In 2000, 35,516 albums were released. Seven years later, 79,695 albums (including 25,159 digital albums) were published”. I also have stats that the number of new releases was 98 thousand in 209 and 105 thousand in 2008.

80, 90, 100 thousand album released per year? Almost 2,000 albums released per week? Really? How much of this is rereleases, alternative versions, compilations etc. etc. How many actual new music is commercially released per year?

For a more recent statistic we were informed that “close to 40,000” tracks are now being uploaded to the Spotify platform daily. Of course the number is huge, as the barriers of entry are so low, but this also does not look like a meaningful statistic. 

It is more relevant to know that the universe of songs that have been streamed at least once during the year increased from 33.2 million in 2017 to 36.3 million in 2018. That’s a 3.1 million increase in songs per year. In the previous year the increase was of 4.9 million. These two numbers give an average of 4 million new songs per year: that’s still 11,000 songs per day, but much less than 40,000.

For a more granular look, let’s dig a bit into the Discogs database. I don’t know if it the most complete database, but it is quite comprehensive and easy to explore. Besides, being mentioned in Discogs serves a base level of relevance for any given record.

The total database has 12.7 million “releases” of all kinds, places of origin, formats and periods. A lot of releases are a repetition of the exactly same material. Searching for “Earthling David Bowie” yields 138 releases, some 70 of which are the same 1997 album we have in mind. The solution is look for what Discogs call “Masters”, where all this double counting is eliminated and there is one Earthling album to consider in our statistics. 

So, instead of the 12.7 million releases, let’s turn our focus toward the 1.8 million “Masters” available. Of these, 504 thousand were released in the US.

Let’s use the year 2000 for more focus. It is a good year because it is peak CD era and we can worry less about cassetes, LPs, downloads etc. 

In the 2000 Decade Discogs has 74,896 masters for the US. The year 2000 lists 8049 masters.. That’s 77% less than “35,516” albums as reported by Billboard. Narrowing down to releases labeled “CD” narrow it down to 4.785 releases. (The “Album” format list 3,207 releases.)

Other formats like singles, “Promo”, maxi-single, HDCD etc. are not that critical for this analysis.

So perhaps it is fairer to say that 5.000 relevant “records” were released in the year 2000. That still a 100 a week.

By the way, this is funny: there were some 10 MP3 releases in 2000!

A quick Twitter Thread: Pinned Tracks

A quick idea: Spotify doing something similar to the “Pinned Tweet”:

Everyday, every usar can “pin” one favorite song. 

The friends/followers can easily see other friends “pins”.

If the friend listen and like, they can “like” the pin. Those likes are counted. 

Every 24 hours, the user can pin a new song…

The list with the history of previous pins can be easily browsed.

A ranking of most liked pins for every day/country etc. can be easily browsed.

Of course, this a rough draft of the idea and it might not work, but maybe it does.

Maybe it would be a fun and focused way for people to recommend music.


T-shirt idea: “I would too like to hang out with Richard Page of Mr. Mister and the guy who used to drum for David Lee Roth”.

Ringo does deserve his place in the Rock & Roll Hall–his first five years as a solo artist produced as much good music as his mates; in fact, his best solo work ranks among the best solo Beatles–but the enduring charm of Ringo Starr is that he spends a good chunk of his time not hanging out with Grohl and Robertson but Richard Page of Mr. Mister and the guy who used to drum for David Lee Roth. He still sees himself as just another rock & roller and his never-ending All-Starr Band tours proves that it isn’t an act.

via Reasons To Be Cheerful, Pt 3

How Today’s Podcast Deals Look A Lot Like Yesterday’s Music Deals

Nice retrospective on Spotify efforts to bypass record labels. No wonder they have directed so much effort towards building their podcast business…

Over the past several years, Spotify has tried and failed to bypass record labels and build infrastructure for unsigned artists. They invested in DistroKid, an independent music distributor; launched a direct-upload tool for unsigned acts in the U.S., mirroring SoundCloud; and offered select independent artist managers six-figure advances to license their clients’ music directly to the platform without any third-party help. But Spotify learned the hard way that it’s difficult to gain influence in an environment like recorded music where incumbent corporations already control around 70% of the market, and has since scaled most of these initiatives back.

Monthly Listeners – 2016 vs 2020

Pasted Graphic

In September 2015, Spotify started to disclose the number of “monthly listeners” for every artist.  As David Turner writes: 

What are Monthly Listeners? It’s a collection of the unique number of listeners an artist got over a month. That means if you listened Travis Scott’s “Sicko Mode” once on a playlist, but never listened to him again that’s one listener, where myself, who is still playing Astroworld daily, is still only a single listener. The stat isn’t trying to capture total number of plays or potential money the artist could make, it’s simply a stat that show audience/reach.

Anyway, back in early 2016, I thought it would fun to compare the number of ML for a number of popular 80s bands. I checked the numbers for eight bands, the most popular was Police (2.29 million listeners), the least was Johnny Hates Jazz (83.9 thousand listeners).

What happened in four and half years? Listeners grew in average 433%, or 3.1% per month. Police grew the fastest (3.5%) and now has over 14 million listeners. In fact, the band’s Spotify About page highlights them as the 233rd most listened artist in the world. Duran Duran was in second place in 2016, but with the slowest growth in the group (2.5%) it now ranks third, below Tears For Fears. Pet Shop Boys remained four but the previous fifth, The Human League, was passed by The Outfield. Roxy Music (not really an 80s band, actually) and JHJ complete the list in both versions.

40 Years of Walkman

Matt Alt writes at the 40 year anniversary of the walkman:

The Walkman instantly entrenched itself in daily life as a convenient personal music-delivery device; within a few years of its global launch, it emerged as a status symbol and fashion statement in and of itself. “We just got back from Paris and everybody’s wearing them,” Andy Warhol enthused to the Post. Boutiques like Bloomingdale’s had months-long waiting lists of eager customers. Paul Simon ostentatiously wore his onstage at the 1981 Grammys; by Christmas, they were de-rigueur celebrity gifts, with leading lights like Donna Summer dispensing them by the dozens. There had been popular electronic gadgets before, such as the pocket-sized transistor radios of the fifties, sixties, and seventies. But the Walkman was in another league. Until this point, earphones had been associated with hearing impairment, geeky technicians manning sonar stations, or basement-dwelling hi-fi fanatics. Somehow, a Japanese company had made the high-tech headgear cool.

This bit is wonderful, the precursor to the current “ambient mode” of earphones:

Even stranger, by current portable-listening standards, were the Walkman’s headphone ports—plural—and a built-in microphone. The Walkman was initially designed to be used in tandem: a “hot line” button paused the music and activated the mic, letting two users chat even with headphones on. This specification had come at the insistence of Morita, who had irritated his wife by not being able to conduct a conversation while testing early prototypes at home.

The next part is a bit controversial… Could Akio Morita have launched a successful digital music player before Apple?

Jobs would get his wish with the début of the iPod, in 2001. It wasn’t the first digital-music player—a South Korean firm had introduced one back in 1998. (That Sony failed to exploit the niche, in spite of having created listening-on-the-go and even owning its own record label, was a testament to how Morita’s unexpected retirement after a stroke, in 1993, hobbled the corporation.)

“Why Sony Did Not Invent the iPod” would be a great essay topic.