The (Human) Music Genome Project

high_fidelityBack in the day, I worked in a music store.  And, though it was camouflaged as a homogenous chain store, a cadre of employees actually made it a pretty awesome place to randomly find yourself shopping for music (trust me, the only way you’d find yourself there would be randomly).  The group of us working there really knew our music, and the way that we liked to strut our stuff was to provide personalized music recommendations to customers.  More specifically, we’d ask for a customer to name a few songs or albums that they have liked, and we would in turn provide a few recommendations of things they had never heard but would probably like.  We were a collection of khaki-panted mini-pandoras.

Except any of you who have worked in retail know that’s not the entire story.  Because saying that we provided recommendations of music we thought they would like is not exactly the entire truth.  Full disclosure, we provided recommendations that were at the intersection of what we thought they would like and what we thought they should like.  We were mini-pandoras with not-so-mini agendas– and some (well, many) bands just didn’t make the list of said agenda.

But I think that our agenda-led recommendations created more serendipity and true discoveries than some “Pandora purist” if-y0u-liked-this-you’ll-like-this recommendations ever could.

Picture 2While at dinner with friends last night, someone found out that I like music a lot and started to ask me what music I might recommend for her (yes, I know, I’m no longer in High School but still act like I’m working at the record store).  I asked her, as I would have back in the day, to give me the names of a few bands or albums that she has liked a lot.  She gave a really broad brief– Coldplay and U2, basically.

Now, if you take Coldplay and plug it into Pandora, you get suggestions such as those in the screen grab above.  Fine songs all, but the algorithm leads you to bands that are neighbors in the same musical sub-division.  There’s nothing in Pandora, that I’ve seen, that will tell you to jump in the car because it’s time to take a musical trip clear across town.  There might be some surefire songs to add to your iTunes shopping cart, but there’s little likelihood of a truly serendipitous departure (amidst the digital deluge).

So, for this post, I thought I would test my agenda-tinged recommendation skills by seeing if I could offer a Coldplay/U2 fan songs that she would like… in the hip-hop world (gasp).

As any comedian/DJ/record-store employee knows, it’s always good to start off when a safe bet.  Even when your safe bet artist is in the midst of a (possibly staged) microphone-grabbing nervous breakdown, when there’s a rap song that features the lead singer of one of the bands you were given as input, it’s a pretty easy bridge (heck, Pandora could probably come up with it).

  Homecoming, Kanye West (featuring Chris Martin)

Okay, no more cheating.  In the vein of arena-filling, anthemic songs Lupe Fiasco’s Superstar slots in nicely.  To add some additional music cred that is almost certainly unwanted by non-record store employees, I’ll pick the remix that features Young Jeezy and T.I.

 Superstar, Lupe Fiasco (remix featuring Young Jeezy and T.I.)

Lastly, to wrap it up with the music world’s favorite former BCG consultant (yes, really), a song by John Legend that includes the kind of sweetly-biting line that makes a list of great lines from break-up songs: “I wish you the best… I guess.”

 Everybody Knows, John Legend (RAC Mix)

Three songs that provide a bit of a leap out of the obvious.  Are they sure-thing recommendations?  Not at all.  But I think they’re the kind of connections that are a lot less common– not just because of the demise of record stores that I used to work at and frequent, but because of the digital connections that make it all to easy to hang out comfortably in our subdivision of known preferences.  In the coming years, where are we going to find our agenda-led-but-kinda-surprising discoveries?

2 responses to “The (Human) Music Genome Project

  1. Interesting issues raised here, Johnny. And they don’t seem confined to just the world of music. How can online retailers replicate the customer service that is (or used to be) provided in bricks and mortar stores? Is this even possible? Online retailers certainly seem to be trying–check out the Netflix Prize that was awarded recently, if you haven’t already read about it. If Deep Blue can defeat Kasparov, perhaps computing power is capable of providing “better” product recommendations than quirky human beings. But is your point that quirky customer service has a value somehow? That “better” product recommendations might not in fact be better–exactly because they are soulless, and un-quirky? Could quirkiness be modeled mathematically somehow, or would that be like trying to create a bachelor who is not an unmarried male?

  2. Pingback: Day 27: A Song You Wish You Could Play « Then, Now, Soon…

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s