top of page

Discover Weakly: The Irony of Algorithmic Music Selection

Perhaps the only thing more anxiously awaited by Gen Z’s around the world than presents wrapped up underneath a domesticated evergreen is the summary of our music listening habits wrapped up on our smartphone music apps.

I woke up one morning, early this December, to a roommate of mine guffawing at the fact that Spotify had crowned Drake her ‘most-listened to artist of the year’.

“Drake? There’s actually no way. Who decided this?” she exclaimed.

I didn’t really know what to tell her, except that it wasn’t so much a question of who but of what and how.

Music listening has become a practice much more to do with quantification and calculation than I’m sure our grandparents, or even our parents, ever imagined it could be.

Technology has obviously changed a lot about the way we perform our mundane, everyday practices, but who woulda thunk I’d be sitting around waiting for Monday, for my Discover Weekly to be updated, so I can listen to some new music already!

This blog is influenced by a discussion with UBC Music professor, Dr. Nathan Hesselink, over what has changed, and what we have lost, in our unyielding pursuit for optimally efficient music listening practices.

As someone who doesn’t even own a cell phone, let alone engage with music apps, his insights largely represent those of a pre-Web 2.0 era. Much of our discussion was prefaced and fuelled by my attempted explanations of what algorithmic music selection is — what it does, how it works, why it matters — and, as such, I figure this is where we ought to begin as well.

Algorithmic music selection is, in the simplest of terms, a computer "telling" you what music it “thinks” you should like. Now fancied up a bit — the curation of suggested music by means of an algorithmic process.

So how does it work?

I'm not too sure myself, so it's probably best to turn to the experts.

Spotify explained to The Verge in 2015 that it “has built a taste profile for each user based on what they listen to. It assigns an affinity score to artists, which is the algorithm’s best guest of how central they are to your taste.”

Specifically for their Discover Weekly algorithm, this “taste profile” is used to identify playlists and users akin to your music listening habits, which is then used to curate a weekly playlist of songs, pulled from said playlists, which you haven’t yet listened to.

Spotify boasts that, at least due in part to this, “ten billion times a month, listeners across both Spotify and Spotify premium stream a new artist they had never heard before.”

This is obviously incredible. Dare I say, revolutionary. Ten billion times per month is the kind of statistic that warrants a spit-take of some sort.

But I’m going to ask you to wipe your chin off and look beyond the numbers for a moment. Beyond the cost-benefit-type calculations which will always conclude, not only the niftiness of algorithms within the practice of music listening, but its absolute necessity.

After all, the sheer volume of music content available to us today is nothing short of daunting. The ever-increasing variety, alongside methods of distribution and access provided by the Internet, is multiplying faster than drama amongst Kardashians, and we can all barely keep up with them.

So obviously, the desire to have things narrowed for us is not without reason. However, I ask you to put all of this aside and think about what is really happening when your Discover Weekly is curated and listened to, and what it truly means to you — at least for the duration of your reading this (thanks for that by the way).

I urge you to begin by asking yourself what it even means to "discover."

Personally, when I ponder this word, an element of surprise is suggested to me; a kind of wondrous shock — not necessarily, but perhaps even, ‘without expectation’.

I think of discovering a new book to read by going to the bookstore and browsing spines, reading inside covers, back covers, and flipping to random pages — knowing it is a book I wish to find, but without expectation of which one I’m truly looking for — which one that is actually going to grab me, and invite me in for more.

To discover does not imply aimlessness of a search, but it is also not looking for something you already know of.

Herein lies what about algorithmic music selection rubs me the wrong way: the excitement of “discovery,” lost in the weekly silver-platter-presentation of worth-sampling music, cooked up by a series of computer calculations.

I expect to hear songs I will like on my Discover Weekly, and because of this, my enjoyment of them is exponentially dulled.

In fact, the music tends to blur-together and become background-noise as a result of my relinquishing the need to actively and truly listen to it. After all, I already “know” I’m going to like it.

Think of hearing a song you love come on the car radio, if you can remember such a time, and how much more thrilling it is than plugging in your phone and putting it on yourself.

Expectations fulfilled are adventures denied. You can quote me on that.

So how did we find new music before we had algorithms to do it for us?

Dr. Hesselink gushed about perusing record stores much the same way I do bookstores, and about discussions of music with students and friends, which often lead to mentions of artists he’d not yet heard of or listened to.

We shared anecdotes of being in the right bar at the right time when a band was in their early years, and of falling in love with the opening-bands at concerts.

Of course, the variety of and access to music was much more limited when he was my age — having only a handful of radio stations even available, let alone that played music he fancied, and so discovering new music was a much less daunting task. Barely could it be considered a task at all.

However, there is something to be said about the honesty of these methods of discovery, and the impact they have on your life.

I’m sorry to be the one to say it, but the feeling of having a song come on at a party that lights your fire, sparks an uncontrollable dance around the ping-pong table, and has you rushing to the DJ to ask what song it is, or succumbing to the secret Shazam, is simply not a feeling that can be replicated by listening to an algorithmically-curated playlist.

Or maybe it can for you, and I’m being completely self-indulgent and cynical.

To be fair to myself, I am in no way asking anyone to boycott Discover Weekly, or any other form of algorithmic music selection. In this day-and-age, these types of technology are the answers to many music-lover’s prayers.

But with these answers comes follow-up questions, and I urge you to ask them.

When you’re next listening to music suggested to you by an algorithm, if this is something you do, ask yourself, “Do I like this song? Why do I like this song? What about this song is ‘good’?”

You don’t need to be able to identify instances of rhythmic play, or ingenious compositional form. Just try to be conscious of the feelings the songs evoke, what places or times they remind you of, and whether or not you’d actually ever want to listen to them again, or show them to someone else.

And, if you’re really gung-ho, take a trip to the CD or record store once and awhile and award yourself the time to simply browse.


bottom of page