I went back to Spotify for this one killer feature

I've tried almost every streaming service out there. From the big players (Apple Music, Google Play Music, Spotify) to the lesser-knowns (Deezer, Tidal, SiriusXM) to the recently deceased (Songza, Rdio). I could write an encyclopedia about the pros and cons, the similarities and differences, the features and benefits of each service - and maybe one day I will - but right now, I'm only going to focus on one feature from one service: Discover Weekly from Spotify. That's because this single feature is what got me to give up all the other services and stick with Spotify, despite its (many) flaws.

Discover Weekly is a 30-song playlist generated by Spotify robots. Every Monday, you get a new 30-song playlist. Without getting into the nitty-gritty details of how Spotify chooses the songs (Adam Pasick at Quartz covers this in exhaustive detail, complete with flowcharts and infographics and everything), let me tell you that it feels like absolute magic. Best of all, it's available to everyone, even those who only sign up for Spotify's free account.

If you're new to Spotify, you won't have this feature for at least two weeks, which gives its algorithms a chance get to know you a little better. Now, I know you're probably thinking, "Wait a minute, I've seen the suggestions that 'algorithms' have surfaced for me before. Those suggestions are typically hot garbage." Well, you're not alone in thinking that. We've all been there. Watch one or two YouTube videos with Jimmy Fallon in them, and all of a sudden YouTube thinks you're a Jimmy Fallon superfan, surfacing everything he's ever been peripherally involved in next time you visit the home page. Buy a can opener from Amazon, and be prepared for the deluge of can opener ads on every site you visit for the next two weeks. Algorithms don't always get it right. But based on my experience, this one definitely does, and to prove it, I'm going to lay my personal music listening experience out there for you, guilty pleasures and all.

Finding Spotify's Discover Weekly playlist can be a multi-step process.

Finding Spotify's Discover Weekly playlist can be a multi-step process.

My first Discover Weekly playlist was based on weeks of varied listening habits that included everything from dance-punk (LCD Soundsystem) to Canadian alt-rock (Matthew Good) to indie pop (Vance Joy) and everything in between. I got a Discover Weekly playlist that included songs like these:

As you can see, it's a super-eclectic mix of familiar and new songs from a variety of genres. But what about my guilty pleasure songs? Wouldn't they ruin the algorithm somehow?

Let's use, as an example, 1996 smash hit and greatest song ever written, Ginuwine's "Pony." I listen to "Pony" an unhealthy amount. I play it in the car. I play it at work. I request it at weddings. I attend themed club nights where the DJs dress up like ponies and play Ginuwine's "Pony" every hour on the hour. Most algorithmic music services would see this and say, "Oh, this guy likes mid-90's R&B," and serve me up a ton of Jodeci or Brian McKnight or whatever. Spotify's algorithm knows better. In week two, nestled in my fresh batch of 30 songs, I got this gem:

Yes, that's right: a chill electro cover of Ginuwine's "Pony," something that maybe ten people on earth would actually want. It just so happens that I'm one of those ten people, and only Spotify's algorithm could have figured that out.

There's a deep philosophical divide between the music streaming services. On one side, you have Apple, putting most of its efforts behind human curation with their $3 billion acquisition of Beats Music and the subsequent creation of its Beats One radio station, as well as an army of curators developing hand-picked playlists. On the other, you have Spotify betting heavily on the algorithm, with its own high-profile purchase of 'Music DNA' company The Echo Nest and projects like Discover Weekly. In the middle, you have companies like Google who try to have it both ways, with a mix of algorithmic suggestions as well as human curation through their purchase of Songza, a playlist curation specialist.

You can think of the two options like this:

  • Apple Music is like that friend you have in college who works at the local radio station and always has his thumb on the pulse of what's cool and what's about to be. He burns you CDs with future Rolling Stone cover artists, bands that will win a Grammy two years from now, and songs that are about to get perfect 10s from Pitchfork. He doesn't care that you don't like Radiohead and will include their latest B-side no matter how many times you tell him not to. His CDs aren't about what you like, they are about what's objectively good.
  • Spotify is like that girl that has a serious crush on you in the 11th grade. She might not like all the same bands that you do, but she won't tell you that. She'll make you a mix tape that emphasizes your similarities, that makes you feel comfortable. She'll throw in a fresh band or two that she wants to expose you to, but in a way that still feels familiar. The more time you spend with her, the better she gets to know you, and the better her mix tapes get.

At the end of the day, that's the real argument in favour of the algorithm - curated services might have some playlists that are objectively better, but they require more work to find, and they don't reward you for spending more time with them the way algorithm-based services do. That's why I've chosen the service that gets better the more I use it.

You don't have to take my word for it - you can sign up for Spotify for free and try it for yourself. 

If you're planning on playing a bunch of tracks for an extended period of time that you really don't want affecting your recommendations, turn on "Private Listening."
I made the mistake of not doing this before using Spotify to DJ a party filled with college and university bros, and the following Monday, I was flooded with more EDM and trap anthems than I could handle. My Discover Weekly still hasn't completely recovered.