Writing in 2004, essayist Geoffrey O'Brien called the personal mixtape "perhaps the most widely practiced American art form"
Discovering new music and studying the cultures they connect to has always been a very important part of my life. It's affected me on many levels; my career, the types of friends I have or the courses I took in University, where I travel, where I eat. There have been some really transformative moments where someone has shared a song or album with me - "hey man, you really need to check out this Ghanian afrobeat album, I think it's from the 1960's." These moments always seemed rare and required hanging out with people who had great taste in music and were constantly searching, like me.
Spotify's Discover Weekly has changed that for me. Discover Weekly is a playlist on Spotify created specifically for you using a set of A.I. based computational approaches. Spotify studies what you are listening to, the songs on adjacent playlists and other things like your taste profile.
And thanks to them because for the past year, it's like I've had the unique feeling of discovering new music I love, every week. ✌️
I understand how these recommendations are happening in a general sense, I have experienced the output of these techniques when using Amazon or Netflix. But when these techniques were applied to something I love as much as finding new music, the overall experience really caught me off guard. It inspired me.
I have been asking questions like:
- How else can technology augment the creation and consumption of art?
- What other kinds of computer recommendations could improve the quality of my human experience and my creative process? Would Final Cut Pro start making suggestions on how I could cut a film for a more optimal experience?
- How is curation going to evolve over time?
- What is the role of human curation vs. computer curation in our society's future? What is different between these forms of curation? How will they work together? Do we need DJs any more?
This is my good friend Seetoh and I at some weird bar in Toronto the night before he left us to live in Austria. I'm eating waffles and he's eating mac'n'cheese. The flash was pretty bright, but I pulled through:
At some point while in Austria, Seetoh came across one of my Spotify playlists called, 'airpods'. It is a 16 song playlist that I spent a few minutes creating for myself before stepping out of my apartment to meet a friend.
I described the playlist as, "stuff i listen to when walking around toronto thinking about the present and sometimes the future":
He then wrote me this email:
I saw your FB message awhile ago about having a human making you a playlist v.s. A.I., and it got me thinking.
So I made a playlist for you, in collaboration with Spotify's A.I.
Here's how I did it.
- I really like your airpods playlist, and have been listening to it a bunch. This gave me a human instinctual idea of what you like.
- I started a "Spotify Radio" playlist off airpods, which is a collection of songs Spotify thinks are similar to the songs in airpods.
- I then started creating my own playlist out of these songs called airpods - seetoh edit. Probably picked 20% of the songs from Spotify's first list.
- Once I had about 10 songs, I created another playlist and then used Spotify's recommendation engine to give me more new recommendations. I kept about 35% of the new set of recommendations.
- I repeated this process until I felt I had compiled 24 songs that I thought you would like. So in a lot of ways it's a collaboration between Spotify and I.
The rules I used were as follows:
- At least half are songs I have never heard before, as this gives Spotify a fair chance to "pick" the songs.
- Mostly artists that are not in your 'airpods' playlist
- Songs that are similar to the ones in your playlist, but not formulaically like the songs in 'airpods'
- music I would picture you listening to as per the description, "stuff i listen to when walking around toronto thinking about the present and somethimes the future"
And well, here is 'airpods - seetoh edit'. I really like it ❤️