Groove ain’t got no algorithm
7 March 2017
Being an anal sort of fellow, I group the various music apps on my iPhone into a ‘Play’ folder for all the tools that are supposed to help me create, and a ‘Listen’ folder for all the apps I can use to enjoy other people’s music.
And one of the apps in my Listen folder is a 3 month trial of Spotify for 99p. One of Spotify’s tricks is analysing your listening habits and then discerning other things you might like. After about a month during which I threw everything at it, from Bach to Deep House, American Country to Scottish bagpipes and Funk to Prog Rock, it managed to delight me with Boo Boo Davis: What Kind Of Shit Is This - Boo Boo Davis being an old school Mississippi blues guy and the Shit being some EDM style production from BLu ACiD.
I enjoyed it so much I searched for more information and other Boo Boo Davis material. I came across a Blues Matters review which managed to suggest that without Boo Boo’s voice and harp playing the album was shit (your point being?). I also came across other Boo Boo Davis recordings with real ‘pickup’ bands which, to me, sounded wearily generic - competent but generic. And that reminded me of the sheer glee I experienced the first time I heard the Hoxtons remix of Sunshine of your Love (be honest, the original is kinda stodgy).
Suddenly, the hair on the back of my neck begins to prickle. I'm recording my songs using mic, guitar and laptop in the back bedroom of a South Manchester terraced house. I've spent years jamming in rehearsal rooms with live bands and years jamming with computers in back bedrooms. My current audio template is to create recordings which sound like all those live rehearsal rooms I’m so familiar with because it’s ‘authentic’.
Authentic or just generic? Why not just record in an electronic style, jamming with my computer - isn't that just as honest, as real, as convincing? An electronic Boo Boo Davis works for me! And either way, I'm faking it anyway!
Enough about you - what does your audience want?
But enough about me (Hmmm, the Sub-editor is in repeat mode) and my tortured pretensions, what does my audience want? Well, I don't have an audience, save the audience at #OpenMic at my local pub and a few wonderful diehards and family members. But that’s where the nitty gets gritty, trying to second-guess an audience, never mind a potential paying audience. It would be useful to know what sort of James Bisset Music people might actually pay for (which is the point of the exercise after all). So here goes...
If I'm playing live, there's a finely tuned intuition at work. I've written previously about the feeling during performance that the senses are working at superhuman speed; the feeling that everything you sing or play is in slow motion; that the smallest slip in timing or pitch reverberates like a rock fall in a cave. But that sensitivity is also true of the tiny threads of connection that string out between the performer and the audience. Like a spider, a good performer can sense the audience members being tangled in the cunning web spun, and can tell when they struggle or even wriggle free. Next time, the poor buggers won’t escape so easily!
But in our digital age with our audience remote (Oi, Sub Editor! Think up your own sub-heads!), sitting alone in our room and seeing the whole of the moon, that finely tuned intuition gets wrapped in a blanket and put safely in the cupboard. Now, we’re uploading tracks and watching the data streams, analysing the peaks and the troughs, liking the shares and sharing the likes.
“Wow, look, people listen to my music most at two in the morning!”
“Hey, 63% close their Facebook accounts within 4 hours!”
“Surely there can't be a correlation?”
“Ummm... what’s that scraping noise at my window?”
Of course, all this is fine when the data is huge, when it cross-links to Google Analytics age and gender demographics, to Facebook profiles, to Trends on Twitter and Hashtags on HInstagram, but for the rest of us it’s guesswork, bloody guesswork. Am I best pursuing the blues-rock authenticity, or the Scottish folk sincerity, or the powerful EDM drive or the funky jazz mischief? Why not some skillful blend of all four? And what does that even sound like? Me?
Let’s see what Spotify does
Well, let’s see what Spotif–ALRIGHT! I’ve had enough of this! I want to speak to the editor. No editor? What do you mean, no editor? So who’s writing these sub-heads? An algorithm? An algorithm is writing the sub-heads? It’s not writing sub-heads, it’s just copying me!
Where were we? Ah yes, Spotify is trying to identify music that I will like. It's looking at the stuff I prefer to listen to and then, using cunning algorithms, it's going to find the quality music that other people with similar tastes like. Spotify is going to open up a world of Funky, Folky, Jazzy, Witty Rock and Dance Music (with lots of guitar) and that will be my audio production template!
And I've just found that the ‘Library’ page has “Your Daily Mixes”, whereas before I was exploring "Discover Weekly". OK, let’s take a look.
Daily Mix 1 - guitar rock from the early seventies including Humble Pie, Steven Stills, Little Feat, Derek and the Dominoes.
Daily Mix 2 - Folk music from Richard Thompson, Jim Moray, Martyn Bennett, Shooglenifty and others.
Daily Mix 3 - singer-songwriters from Laura Marling to Ryan Adams and Paolo Nutini.
Daily Mix 4 - Funk from Phat Phunktion, The Meters, Orgone, The Souljazz Orchestra, The J.B.s
Alright, I see what you’re doing here Spotify. You’ve just copied everything I've listened to through genre playlists and album searches and dumped it all in half a dozen playlists 'just for me'. That Boo Boo Davis album was a lucky stumble.
So much for cunning algorithms, closely analysed data and a world of new music. You’re just playing my own choices back at me.
You know what - I can’t be arsed.
You know what? What it said.
Add new comment