Learning How to Think
If you want to learn how to think about something, don’t use an algorithm.
A couple weeks ago, I realized I wanted a playlist for my wife’s birthday party. It nearly paralyzed me.
Based on some searching, I could tell this was a pretty common task. Birthdays happen every year (Facebook promotes them, in part, because birthdays are guaranteed calendar events, and a user with 100 or so friends will have a reason to log in a couple times a week), and birthday parties do as well. I realized I had a few options:
1) Trust an expert. Chance the Rapper put together a pretty solid playlist on Apple Music, and he knows a thing or two about good music.
2) Trust the crowd. There were several playlists available on spotify that had plenty of fans and followers, and would probably get the job done.
3) Trust the algorithm. This is partly just a combination of the previous two options, since Spotify’s algorithms essentially learn associations from the crowd. Still, I could start a new radio station based on some seed songs and let their autoplay algorithm figure it out, or, I could put a few seed songs in a playlist (probably Celebration by Kool & The Gang, Birthday Party by AJR, and Growing Up by Macklemore).
4) Trust myself. I could search my personal library for birthday songs, add some of my wife’s favorite songs, add some of my favorites, and hope it all came out okay.
Each of these options is, to some degree, “distributed curation,” which we rely on more than ever these days. Any time we scroll through Instagram, Facebook, Netflix, Spotify discover playlists, etc., we are trusting curation algorithms to choose the media we consume. This may seem obvious, and it may seem innocuous, but I have some concerns.
When we trust these curators to choose what media we consume, we eliminate dozens of small decisions that we would have otherwise made for ourselves. On one hand, this is extremely convenient. We probably have more important decisions to make in our day, and maybe these curators can help us focus on those more important decisions. But we rarely know which decisions are most important to us.
As I put together the birthday playlist myself, Spotify picked up on my choices and suggested Birthday Cake by Rihanna. It looked okay, and I added it. But when I was screening the playlist later on, I realized the lyrics “It’s not even my birthday // But he wanna lick the icing off” did not at all fit the vibe I was going for.
Like most artificial intelligence algorithms right now, the Spotify recommender picked up on an obvious trend (birthday), but missed other important pieces of context. From a recent piece in Quanta magazine, here’s a test to see if you are more intelligent than the most sophisticated natural language processing algorithms of today:
If you can tell that “President Trump landed in Iraq for the start of a seven-day visit” implies that “President Trump is on an overseas visit,” you’ve just passed.
Curation algorithms have passed some kind of performance threshold. I could probably trust Spotify to autoplay a birthday playlist, and unless guests were paying really close attention, most would not notice a big difference. But the more I think about it, the more I think that kind of “minimum acceptable performance” is not the kind of standard I want to use.
As David Foster Wallace famously stated in his commencement address defending the liberal arts education, “learning how to think really means learning how to exercise some control over how and what you think.” I couldn’t help but consider that making this birthday party playlist may be an ultimate test of my liberal arts education. Had I learned how to think? Could I choose good birthday party music for myself? Or would I turn to the algorithm? What would be the best choice in the case that a guest did comment on a song?
Certainly, curation algorithms have a place. I am glad that I can type “Nick” into Spotify, and it recommends Nickel Creek instead of Nickelback. But I am increasingly aware of the ways that I rely on algorithms to make choices that I want to be able to make for myself, and I am trying to do so more cautiously. I made the birthday playlist myself, and here’s why I’m glad I did:
- I could talk to guests about the songs playing
- The playlist was more enjoyable to listen to, knowing I had chosen the songs
- I spent time thinking about and choosing the songs
- I have paid more attention to music since making the playlist
(Relevant piece in the New York Times today discussing the merits of human attention and curation over algorithmic curation.)
