Spotify Discover Weekly, Distilled
the machines that are making our mixtapes
Estimated reading time: 3 minutes, 53 seconds
What we’re pouring (poring) over: Discover Weekly
Discovering music is not supposed to be a job for algorithms
Why is this a cool topic? The Spotify Discover Weekly playlist is impressive technology: you get a new playlist each week with songs you’ll probably like. What’s not to like? Literally! They make sure there’s nothing not to like!
But the algorithm driving the playlist can actually make your music less interesting in the long run. We’ve outsourced music discovery to a machine, taking all the fun out of it.
What to know about Spotify Discover Weekly: Like most things we use these days (Netflix, Amazon, Zillow, Tinder, you name it), Spotify employs a recommendation engine, which is basically an algorithm that tries to predict your preferences based on what you’ve liked before.
Spotify learns what you’re likely to enjoy based on your listening activity and song attributes (like BPM), and comes up with a weekly playlist of songs that are just outside your bubble but still similar enough that you feel like Spotify knows your taste.
It’s a dilemma for the engineers at Spotify fiddling with the algorithms: How do you balance similarity with novelty?
Over time, similarity wins out. The algorithm does too good of a job, curating your playlists so well that you never actually hear anything that challenges your taste. The playlist stays safely and firmly in the realm of what you’re already familiar with.
The problem: Recommendation engines aren’t inherently bad. Spotify Discover Weekly is just trying to help you enjoy music! Amazon has a cookbook you might like based on your recent purchase of a spatula! LinkedIn wants to show you some job openings you’re qualified for! Netflix urges you to watch Formula 1: Drive to Survive (I urge you to watch it too)!
But these systems make it too easy to get stuck in a rut if we let them. Recommendation engines are not built for the random circumstances that set you up to stumble on something different, hard as they might try to recreate the chaotic listening experience of shopping at a Forever 21 in the early 2000’s. Where were you when you heard your favorite songs for the first time? Probably not on Spotify.
So these days, it takes some extra effort on our part to branch out. We have to try harder to burst our bubbles. And it’s not easy. (Raise your hand if you’ve looked at a Lollapalooza lineup and felt alarm at the number of artists you fail to recognize ✋)
But if we rely too much on the machines like Spotify Discover Weekly, we rob ourselves of the purest joy in life: hearing a lot of crap on the radio, and then hearing an awesome new song that we didn’t expect to like. Then listening to it on repeat for the next week.
What to do now with this information: A few ideas:
Take a more active approach to Spotify: Don’t get too passive by letting the machine have all the fun! Spotify is a great way to manually search for new artists and releases the same way you would have on YouTube or iTunes back in the day.
Shake it up with Pandora: I recently re-adopted Pandora Radio, which I hadn’t used since high school. Pandora created “stations” before the days of Spotify, and it’s much more randomized and radio-like.
Use the Soundhound app: It’s a lifesaver for when you’re out and about (like at a store or a bar) and you hear a good song you don’t know.
Talk to your friends: I don’t know the last time I simply asked someone what they’ve been listening to, but it’s the best way to find my next obsession.
And algorithms make all aspects of our lives predictable, not just the music we listen to. So make it chaotic, unpredictable and impersonal.
Pick up a physical newspaper or magazine that hasn’t adjusted a news feed based on your personal views. Ignore the Netflix recommendations for Tiger King 2 that appear at the top of the TV screen and try a weird documentary instead. (But don’t ignore the recommendation if it’s for Drive to Survive.) Go to the bookstore and see what the staff has been reading.
Sometimes, the best things in life are the things that we (or an algorithm) never could have predicted.
Go to the source:
How Spotify’s Discover Weekly cracked human curation at internet scale (The Verge)
Spotify’s Discover Weekly explained — Breaking from your music bubble or, maybe not? (Medium, The Sound of AI)
We Need to Make Our Own Playlists Again. Our Lives Might Depend On It. (Esquire)
This week’s beverage pairing
Bohemia Pilsner, Cervecería Cuauhtémoc-Moctezuma
I was lucky enough to spend the holidays with my family in Cancún, where Chris and I accomplished our mission to find Bohemia beer somewhere. Even though you can find it in the US, there is something about drinking a good cerveza in Mexico that feels different. Bohemia is light and drinkable like other popular Mexican beers (Modelo, Tecate) but flavorful and malty since it’s a German-style pilsner.
We found this one at a Señor Frog's, where according to them “the only rule to follow is to follow no rules!”
Other random things I have on my reading list this week
A Day in the Life of a Reporter Covering the Elizabeth Holmes Trial (The New York Times): I continue to be obsessed with this trial as we wait for a verdict. Also, one time at a networking event this same reporter (Erin Griffith) tapped me on the shoulder and informed me that my name tag was stuck in my hair.
Kim Kardashian West Is the Real Spider-Man Villain (The Cut): Kim accidentally shared a spoiler from the new Spider-Man movie on her IG story. Proceed with caution.
The year of garbage internet trends (Vox): “Most viral fads are meaningless, as explained by sea shanties.”
As we head into 2022 (just now I tried to type 2020, then 2021, then I realized it’s 2022, which is insane) I’m grateful for each of you for continuing to read this incredibly sporadic newsletter.
I have no idea how often I’ll keep writing it or what it will look like in the future, but it’s fun, so I keep doing it. Hit me up with any ideas on what you want to see more of!
On tap for next week: The tongue map. Remember that from grade school? More in my next Distilled!