There were no vending machines when I grew up as a ’90s kid in Guangzhou, only corner stores and hawkers. So my first encounter with a vending machine had to wait until fourth grade, after my family immigrated to Canada. I distinctly remember the feeling: delight. It was almost magical to see the machine whirl through its mechanical series of Rube-Goldbergian motions to produce a can of soft drink at my command, given through a firm press of the round, hard-plastic, glowing button. My fingers barely had to move an inch, and a machine large enough to crush me would quickly rumble to life in disciplined response. Since then, I remember always favoring vending machines over cafeterias for my recess soda needs.
Why was I drawn to that round, hard-plastic, glowing button? What pleasures lay in a button press at all? To peer behind the mundanity of this mystery, I present to you two fragments of technocultural development which, despite their occurring at distinct times and places in the recent past, share an intriguingly resonant imagery.
The first fragment echoes an event two decades ago, when IBM’s chess-playing supercomputer, Deep Blue, defeated the reigning human chess champion, Garry Kasparov, ushering in the age of machine dominance in competitive chess in 1997. In March 2016, Google staged a reenactment of this seminal moment in AI history — except this time instead of chess, they built an AI to play the board game Go, the oldest board game in human history, an intuitive strategy game deemed to be among the most difficult for computers to excel at. Google subsidiary DeepMind’s AI, AlphaGo, went on to beat 14-time world champion Lee Sedol easily, winning 4–1 to stun a Go world hitherto confident that Lee Sedol would easily trounce his AI opponent.
Controllerism acts as a release valve that reinscribes our mastery over “our” machines. “The future is still ours,” controllerism dramatically declares
The second fragment is less of an event and more of a scene: In dimly-lit but packed nightclubs across the globe and webcam-filmed bedrooms on YouTube, musicians since the mid 2000s have been playing electronic music from instruments attached to laptops that consist of nothing but a panel of white square buttons laid out on a grid. Called controllerism, their aesthetic style mixes the virtuosic performances of playing live instruments with the endlessly recombinant possibilities of music software, resulting in a promiscuous aggregation of disparate sounds and juxtaposed textures. Before Electronic Dance Music (EDM) culture became mainstream, people encountering the idea of DJs performing shows used to ask dismissively, “Aren’t DJ just pressing play? Are they actually doing anything?” As if in response to this challenge on their legitimacy, today’s DJs in the controllerism scene have made whole bodily choreographies out of pressing play, their fingers dancing nimbly across a field of buttons, sampling and remixing in real time, blurring the line between DJing and performing.
If you put a Go board and a grid-based music controller side-by-side, they look very much alike: both are large square grids divided into units of smaller squares, slabs erupting with perfectly identical ninety-degree angles. But what do an ancient board game and a contemporary musical subculture have to do with one another? The visual correspondence between Go and grid-based music controllers is no coincidence: they are both technologies used to produce and manipulate potential.
In times where we keep hearing canary calls of AI-driven automation, when machine learning and neural networks are getting billion of dollars in research funding so they can sooner replace us and our easily-exhausted meat brains at our jobs, we no longer seem to have the potential to become who we were supposed to be. While humanity continues to brew in our increasingly fascist stew of post-recession malaise, machines seem more lively than ever. If the triumph of AlphaGo over Lee Sedol signaled the emerging cognitive dominance of machines over humans — further stirring the pot in which our collective anxieties about the future threaten spillover — then controllerism acts as a release valve, letting steam out in a cathartic release that reinscribes our mastery over “our” machines. “The future is still ours,” controllerism dramatically declares, “but only if we open ourselves up to becoming-machine.” Weren’t machines supposed to take over the tedious work, so we could dedicate our time to play?
“With AlphaGo’s Victory, Are Robots Set to Take Our Jobs?” asked an article in the Wire not too long after AlphaGo defeated Lee Sedol in March 2016. Since then our anxiety of being replaced has not abated, but risen steadily: In the last year, searches on Google for “automation job loss” have quadrupled when compared to the average of the previous four years. Just last month in February, an article from Bloomberg titled “JPMorgan Software Does in Seconds What Took Lawyers 360,000 Hours” tells the story of Wall Street bank executives eager to replace the lawyers who used to interpret their loans contracts with faster and cheaper machine learning software. Who needs a team of expensive lawyers when a piece of software can do what they used to do many magnitudes quicker? It’s not just web journalists eager for page clicks stirring up needless panic. In 2013, a couple of Oxford researchers associated with the technocratic Martin School published a widely cited paper estimating that 47 percent of U.S. jobs are at risk of being automated. The conclusion is clear: Not only are jobs involving manual labour going to be replaced by machines, but traditionally prestigious white-collar jobs like lawyering are on the chopping block too.
When Deep Blue beat Garry Kasparov in 1997, people were genuinely impressed — conservative commentator Charles Krauthammer hailed the event as the harbinger of a “new and different form of being.” But they quickly went on to reassure themselves that, because the IBM supercomputer had relied on calculating all the possible moves during a turn and evaluating each outcome against one another, such a mechanistic and brute-force way of probabilistic computation was never going to be much of a threat to human creativity and ingenuity. Go is a different game: there are no queens, kings, rooks, or bishops; every piece is the same, and can theoretically be played on any of the 361 places on the 19-by-19 board. This means that there are more possible combinations of moves than there are atoms in the universe, a magnitude much too large for a computer to be able to exhaustively search and evaluate through. “It may be a hundred years before a computer beats humans at Go,” predicted astrophysicist Piet Hut in 1997. Go, in other words, was the ideal arena in which to prove that artificial intelligence technology can perform tasks requiring intuition and creativity better than humans can.
There are more possible combinations of moves in Go than there are atoms in the universe. Go, then, was an arena in which to prove that AI can perform tasks requiring intuition and creativity better than humans
Like IBM before them, Google’s main motivation behind developing an AI to play Go — and hosting a contest pitting this AI against the top human player — is marketing. There is no money to be made in Go, but there is lots of money to be made in healthcare. Google’s DeepMind Health division is now partnering with the UK’s National Health Service (NHS) to build a digital patient records system. This may never have happened without the publicity resulting from AlphaGo’s win — at the very least, the NHS would have asked: “What will you do with all this data?” Like IBM and their Watson publicity stunt on Jeopardy! (in which an AI trounced 74-time winning human Wikipedia Ken Jennings), Google had to show the world — or at least the healthcare industry — that its technology has matured to such an extent that it could replace humans in tasks hitherto not usually thought to be fully automatable, like finding, with the human eye, traces of cancerous tissue in images of body scans.
In Go, you want to play your pieces close to one another so they can support and defend each other, but to win you must spread out on the board and capture more area than your opponent, so balance is key. That any piece is functionally the same as another means that a piece’s value is not determined at all by any qualities intrinsic to it, but by its relationship to other pieces — by its place in the network of pieces on the board. This flexibility means that a piece in the right place at the right time can be extremely valuable, but also that the very next moment, it can become useless, its value ultimately indeterminate. The French philosophers Gilles Deleuze and Felix Guattari called Go a game of war without battles and without front lines — a contest in which the outcome is determined not as the result of some key epic engagement, but is being determined continuously at every point during every moment.
If chess is digital, then Go is analog: Instead of aiming to win set-piece battles that result in capture and removal, Go is more about manipulating a field of potentials around in order to gain an advantage that is everywhere and nowhere in particular. Ideally, you win without confronting your opponent at all, by simply surrounding them. In other words, Go simulates becoming part of and proliferating a network, that supposed key to success in the information age, where one node is nothing, but a bunch of connected nodes is everything.
In one of the most watched music videos on YouTube, a pair of hands — with the precise hovers of a classically-trained pianist, fingers strong and supple on the square buttons — are playing what looks a grid controller lit up in video-game hues of yellow and green. Like a disembodied Franz Liszt, these hands are both song and dance, flitting around the controller, and thus the screen, in an upbeat choreography of fingers. In less than three-and-a-half minutes, they run through an energizing routine of 39 pop songs, mixed by a then 17-year-old boy named Hugo Leclerc.
The video, now seen more than 38 million times, would catapult Leclerc — or Madeon, as he is known — to the top of the EDM world in less than a year. After debuting on BBC Radio One, Leclerc quickly received invitations to play at the largest commercial EDM festivals like Ultra, Coachella, and EDC. From there, his “complextro,” a sub-genre of texturally maximalist electro house made by quickly cutting back and forth between different instruments, found its way onto tracks for Lady Gaga and sports videogames like NHL.
What Madeon does in this video, if we look closely, is play clips of tracks from the music composition software Ableton Live with his physical Launchpad controller. To do so, he first had to sample sections of tracks that he wanted to play, and then “map” each clip or segment of sound to a button on his physical controller. Of course, Madeon could have put these 39 samples on a queue and have them all play one after another by pressing play once, or line them up and export them all together as a single track. But the whole point of the video is to show his virtuosic mastery of the grid by playing each clip after another in real time. That way, he can improvise by switching the order of the sampled clips, or the length of time they are played. Like Go, the grid controller lies as a glowing field of potentials, full of power standing ready, waiting to be unleashed.
“I have become a fetishist for button action,” Los Angeles beat-scene veteran and Flying Lotus homie Daedalus once said in an interview while talking about his instrument, the original grid controller Monome. If the number of buttons on a musician’s controller signals their masculinity (and yes, the vast majority of musicians knows for controllerism are men), then the 256-button Monome is the Chris Hemsworth to the Michael Cera of the 64-button Launchpad favoured by Madeon. On their website, the makers of the Monome declare that “by design the Monome grid does nothing on its own. You the user assign it purpose and meaning: instrument, experiment, tool, toy … choose your own adventure.” The user’s role is to “assign” purpose to the controller. By mapping each physical button to trigger something in their computer software, the user is to choose from and actualize an infinite set of potentials. Like dipping your head into digital wax, you the user are supposed to materialize your individuality in software form — or more precisely, it is through this act of mapping that one goes from “person” to “user.”
Compared to Madeon, Daedalus’s sound is even more chopped up. His samples are shorter and they are triggered more rapidly. Trained as a jazz bassist, Daedalus has developed his own virtuosic improvisational style that, with a non-stop stream of button presses launching sampled clips, erases the boundary between playing someone else’s music and playing your own.
Go simulates becoming part of a network, that supposed key to success in the information age, where one node is nothing, but a bunch of connected nodes is everything
Such a sample-heavy, postmodern style comes from older pioneers like the late DJ Rashad, a long-time veteran of the 1980s Chicago house scene as it morphed from booty house to juke before birthing the futuristic genre footwork, which was popularized in the earlier years of the 2010s by Rashad’s studio debut with experimental label Hyperdub. DJ Rashad took the technical possibilities afforded by this technique and made a career out of stringing together samples of tracks no more than a few seconds in length (with most being under a second), hammering syllables and words in repeated loops until their meaning melted away, leaving only sounds with echos of mood. Controllerism also traces its technological lineage back to an older generation of musical controllers like the Akai MPC (Music Production Centre). A favorite of DJ Rashad’s, it allowed musicians to cut up tracks into sampled segments and launch them from an array of 16 large square buttons at will. It was the MPC that first popularized this dual-movement of mapping sounds to individual buttons, and then triggering them to produce a track live, allowing electronic musicians to do something with their bodies in live performance instead of just pressing play and then standing around. No longer stuck with the linear certainty of a song playing itself out during live performances, electronic musicians could now remix tracks internally, swapping out one part of a song for another, ushering in a networked potential emanating from the MPC’s 16 square buttons.
By blurring the boundaries between sampling and composition into a show of virtuosic exuberance, controllerism performs an ideological function: In an age where machines are supposedly about to take over our jobs, the controllerist aesthetic declares, we can still stay on top and command the machines, if we just map the world onto an interface, and master the craft of manipulating it. Controllerism makes us desire interfaces.
All anxiety has to be dissipated somehow, or it will lead to dysfunction and paralyzation. It was Freud, after all, who said that our civilization (and especially the economy) is powered by the differential energy between desire and reality, between serenity and anxiety. If talk of AI replacing humans in our jobs is the source of our increasing anxiety over the technifying future, then controllerism provides a way for this anxiety to escape and be relieved. To retain our dominant position over AI, controllerism suggests, we just have to do what we were always better at: play.
Yet the flurry of fingers and dazzle of buttons on display in controllerist music belies the mundane nature of their performance. The virtuosic musician is doing something we do all the time: namely, interacting with interfaces. In the quotidian moments of our everyday existence, waiting for the bus or sitting in the office, are we not pressing buttons on our smartphones, clicking and launching algorithmic things on our systems, and navigating between various rectangular boxes on our screens?
Somehow, computing has become both the cause and the solution to the anxiety of labor automation. There is nothing capitalism loves more than to sell solutions to the very problems it produces. At the same time that artificial intelligence is overtaking humans in performing intuitive and creative tasks, we are told to master the interface — until it becomes intuition, second nature — as our salvation. Labor in post-industrial society consists of pressing the right buttons to do something to and with software, whether that is changing a value on a spreadsheet or copying a mask in Photoshop. And we’re supposed to have fun doing it too. Controllerism might seem to restore power to humans in their relationships to machines and to labor, by subjecting these functions to play: seemingly a uniquely human purpose, beyond the demands of labor, beyond the cold calculations of AI. But let us not forget: it was through the very moves of the world’s top human Go players, compiled and translated into machine-readable form, that AlphaGo was first taught how to play the game. We might see a boy-genius effortlessly make the interface do his every bidding, but behind the interfaces of our everyday lives are neural networks eager to absorb our inputs as learning, so that they might become more like us, and then better than us.
With the advent of the mass precarity under the gig economy, there is no longer any clear separation between work and play. For those still with full-time jobs, this means drinking the company kool-aid of frolicking in a kindergartenized workplace, complete with foosball tables and video games. For those juggling multiple gigs to make ends meet, this means seeing their downtime colonized by work time. For even when they are not actually working, they must be always ready for a new client to beckon from their phone. Click. Press. Drag. Drop. The immediate sensorial experience of precarity in the post-industrial economy is one in which the worker becomes a micro-investor, carefully investing bits of their human capital — i.e., attention and cognition — into this button or that button, this window or that one. The exuberant choreography of fingers in controllerist musical performances is, like the Balinese cockfight, a dramatized microcosm of labor in post-industrial society.
Controllerism as drama is heavily idealized. Elevating our interactions with interfaces from the drudgery of work to the virtuosity of art tells a tale that valorizes the human over the machine. In all these virtuosic performances, the human, through bodily mastery, has the ability not only to dominate and control the machine through their controller interface, but also can effectively maneuver in and through the complexities of the world by mapping it onto this interface. Like a mastermind, he sees the macro and can control it on a micro level, one button press at a time. But how many of us are really the ones mapping things onto interfaces, rather than the ones being mapped into them?
The author would like to thank Norah Lorway and Esteban Gonzalez for their comments on this essay.