Le crate digging est-il tué par les algorithmes de streaming ?

In the beginning, humans discovered that we were guts and bones and cells that split into new cells over and over until death. This was the first of many blows to our egos, which constantly try to free us from the laws of nature to which we’re chained. The most recent acceptance to swallow is that humans are data: our lived experience is shaped by statistics like heartbeats per day, doubts per minute and median duration of an existential panic attack.

But such an acceptance allows us to rebel against the institutions that reap our data. You can falsify an online survey as a piss on corporate think tanks, or thumbs down a brilliant Youtube video just to see the rating count hit 69. Just as data is now embedded in human discourse, so too is a person’s need to defy all logic and expectation just because they can.

Which is why it’s so troubling to see the media praise computer-curated music playlists as a saving grace from the monotony of choice. Spotify’s Discover Weekly has been heralded by techies and music-heads alike as the first service to “crack the code” of human curation by producing playlists based on user meta-data.

"It’s good. It’s better than I thought it would be. They're as good as DJs — at scale." This blurb from tech pioneer Amil Dash inadvertently hits the controversy of Discover weekly- it implies that AI curators are not just bent on outclassing human DJ’s, but that their final goal is the suburbanization of human taste: a point at which an individual is a datafied node for a homogenized network “at scale.” So in an era where artistic taste is the next frontier for Silicon Valley’s regulation and datification habit, personal discovery and choice become necessary acts of rebellion.

Discover Weekly claims to be “human from the top-down,” which is partially true- their data stems from both an individual’s taste and the collective habits of their massive user base, which they synthesize into a playlist of songs oriented towards personal style and approved by thousands of fellow listeners. In this way, human data is employed to ease the discovery process and eliminate the labor of sifting through undesirables before coming across something special.

But, as history has shown time after time again, pre-approval is detrimental to cultural advancement. It’s the reason why punk began life slimy and sneering at its glamorized predecessors, and why many youth are skeptical of their parental unit’s taste in music. The success of Discover Weekly is rooted in the presumption that its listeners are content with their current musical palette, and that sounds challenging this viewpoint would cause unwanted distress. In this sense, algorithm-driven curators share the same ideology as your conservative grandparent: “Why change when everything’s so good now?”

Spotify does change, sometimes - the list of genres which it uses for categorizing music is ever-increasing, and currently holds over 1400, ranging from deep discofox to horrorcore. An algorithm analyzes the listening habits of those who are both listeners of the genre and Spotify users to infer how a genre might fit into the greater scheme of music history.

It’s an impressive system by Glenn McDonald, a champion of algorithmic discovery and the self-proclaimed “zookeeper” of Spotify’s playlisting functionality. He seems genuinely dedicated to offering a real discovery experience for Spotify’s user base. “If people are going to give up minutes of their finite lives to listen to something they would otherwise never have been burdened with, it better have the potential, however vague or elusive, to change their life,” reads an excerpt from a talk he gave at the EMP Pop Conference.

And while his goal is noble, simulating discovery has the potential to miss the mark like Columbus trying to reach the Indies and hitting America instead. Spotify’s algorithms also generate special playlists, titled “The Needle,” which seek out fringe-sounds (oftentimes local sounds from a perceived “exotic” location, such as Baile Funk from Sao Paulo) to place on listener’s doorsteps. “Art and joy always move faster than law. But eventually we always catch up. Everywhere can be a here now,” says McDonald on the system. So before artists in Sao Paulo can even begin to declare how they want their music interpreted by the world, algorithms have seized, categorized and relegated a sound to be at a specific place to fit the worldview of a specific listener.

McDonald is, understandably, upset that his system could be perceived as malicious. “Music recommendations are machines ‘versus’ humans in the same way that omelets are spatulas ‘versus’ eggs,” he states during his conference talk. But he’s playing with a dangerous comparison - anyone can ruin an over-easy egg with a mis-handled flip, or even shatter an egg before it’s ready to be cooked or sold, which perfectly embodies the world by algorithms: one where mathematics have constructed our future before we’re capable of doing so ourselves.

As described by post-contemporary thinkers like Armen Avanessian in Dis Magazine, “The present can no longer primarily be deduced from the past… it’s shaped by the future.” Predictive analysis on markets, war and social networks are leading to an existence predetermined by metadata and algorithmic thought. Those who rely on their Discover Weeklys will soon be fulfilling this outcome; by nesting in genre bubbles and mood clouds, they are forever drifting towards an identity defined by the quiet influence of machines and a multi million-strong neural network.

So it seems like the solution is to thrash, to dip and dive out of the eye of analysis and behave so erratically that our behavior transcends a computer’s ability to pinpoint our mental whereabouts.

YouTube videos are a great place to start- it’s stupidly easy to stumble into a community of drum’n’bass heads sharing vinyl rips from the halcyon days of raving, or a youtube channel dedicated to the best and worst of Italo-disco. At the moment, Youtube’s recommendation algorithm uses a mishmash of data relating to user interest, video quality and keywords to the point that videos recommended seem to “match” what is being watched using arbitrary measures. “Watching a radio mix from 1994? Here are 20 other radio mixes from 1994 (don’t worry about genre).” But these arbitrary selections yield so many interesting unrelated videos to the point that Youtube’s flaws are more cherished than its successes.

For a more human experience, internet radio has improved by leaps and bounds in the last few years and is overwhelmingly driven by human communities and talent. Apple music is leading the charge with its in-house platforms like Beats1 and OVO Sound. One listen to these stations and it becomes clear how much nicer it is to hear Kodak Black coming from Drake’s iPod than from track 23 of Discover weekly. On top of major institutional radio platforms, independent streaming services have become new hot spots for sharing sounds and ideas; stations like Radar Radio, NTS and The Lot still see a DJ’s as more than a skilled professionals. Their hosts have the power to send listeners on narrative journeys through lost cassette tapes and straight-from-the-forum music scenes worldwide.

Most archaic of all is the physical record store, also known as the place where the aforementioned radio DJ’s discover the bulk of their treasure. These institutions contain troves of music that may never end up on a streaming platform, begging to be discovered by someone with no intention of finding music relative to their own tastes. Discovering a Brazilian funk masterpiece in a sea of dollar-bins can yield a feeling of accomplishment and dumb luck that no algorithm would ever be shameless enough to allow itself.

There’s always the looming possibility that a truly-emotive artificial recommendation intelligence is on the horizon, and that even the most personal of human experience will never outmatch the intricacies of future deep-learning. Driving, manual labor and other caustic facets of human error are forever improved thanks to advancements in AI, and it’s not unruly to imagine that this same revolution could strike our emotions. Imperfection may soon be a relic of the past, but at least it got us this far- where it will take us lies in the hands of those who wield it with the same respect we show to our wrinkled flesh and withered bone. It’s not perfect, but it’s us.