Thinking about playlistsSeptember 5, 2010
I love playlists. I live and die by them, and make new ones almost daily. My car doesn’t have an MP3 input and I have a daily commute, so a good chunk of my music listening is in the form of burned CDs—de facto sub-75-min playlists. And I realize it’s antediluvian, but I still trade mix CDs with many of my friends (via snail mail, no less; I think we all love the charm of the hand-made packages in the post), and those CDs are one of my favourite modes of music discovery.
Almost all the playlists I make are custom, largely by necessity: the songs are usually hand-selected, and they are frequently also hand-ordered. In time for this weekend’s London Music Hack Day, The Echo Nest debuted a powerful and flexible set of tools to algorithmically generate playlists, and I did a little gedanken experiment to compare the playlists you can currently generate with these tools with the kinds of playlists that I make.
Here are some examples of playlists I’ve made or updated recently, ordered roughly from least to most amenable to automating:
New music: I have a playlist called ‘Current’ where I throw recently downloaded music for further listening.
Albums: If I download an entire album, I’ll keep it together, at least for the first few listens (and then I decide that I really only like “Sprawl II” off the new Arcade Fire album).
Artists: Today I will listen to every Elliott Smith song I own.
Playlists by geography: I have a playlist called ‘CanCon‘ that I made for a friend of mine who just moved to Canada. Amazingly, this looks reasonably easy to do with Echo Nest’s new APIs, although it might require a bit of careful tweaking to include my hometown of Toronto, since it’s well south of the 49th parallel.
Workout playlists: Recently, I’ve been doing musical sprint intervals: moderately-high tempo songs intermixed with short, loud, fast songs by punk bands like the Ramones or Pansy Division.
Playlists of bands with upcoming shows: Boston-based concert tracking service, Tourfilter, has a monthly residency at a local bar, at which I DJ’ed a few months ago. All of the songs are by artists that are playing in the Boston-area in the next month or so.
Songs I can play on bass: Sadly, a very short and slowly-growing list right now (“Green Onions,” “Seven-Nation Army,” and a handful more).
Playlists for other people: Playlists or mix CDs made I’ve made for friends of music that I think they’ll like, based on what I know of their tastes.
Playlists by mood: Usually not just ‘happy’ or ‘sad,’ though. I have a recent playlist I made as a soundtrack when I was feeling melancholy and restless (lots of Waterboys, Sea Wolf, Frightened Rabbit).
Playlists by theme: As an example, I made a playlist of ‘embarrassing’ music for a friend of mine, which was mostly songs at the intersection of nerdy, funny and bawdy (think The Bloodhound Gang’s “The Bad Touch”).
‘Best of’ lists: Like most music geeks, I like making lists of the stuff that I like best (although I guess if I was a real music geek, I’d describe it as ‘the best music’)
What do I feel like?: Quasi-random concatenations of whatever I feel like listening to on a given day.
The first half are pretty straightforward. The second half get a little tougher—some of them are nearly algorithmic, but only if you happen to be me. The thought processes behind the last two are opaque even if you are me. Coming up with those last few seems very close to a musical Turing Test; not that I’d put that beyond the ability of people like the Echo Nesters, although there might be a few existential crises along the way.