Early one morning in February 1966, Cleve Backster was alone in his office at the Backster School of Lie Detection, drinking coffee and thinking about setting his secretary’s Dracaena cane plant on fire. During normal business hours at the school, Backster taught NYC police and FBI agents how to administer polygraph tests – knowledge gleaned from his work with the Counter-Intelligence Corps during WWII, and later as an interrogation specialist for the CIA, where he introduced the use of lie-detection machinery. But after pulling an all-nighter in his office, Backster found himself wondering: “Would this plant respond like a human does if I hooked it up to a polygraph machine? What would happen if I set one of its leaves on fire?”
This is the story that came to mind during my first listen to Emptyset’s Blossoms – the duo’s second LP on Thrill Jockey, and latest in a series of delightfully raw and rackety releases produced over the course of a decade. James Ginzburg and Paul Purgas have worked on the bleeding-edge of electronic music genres, crafting an evolving catalogue notable in part for its razor-sharp precision, both conceptual and compositional. Most recently, 2017’s Skin featured an exploration of microtonal vocal and acoustic stringed elements, while the metallic drones of Borders were generated from tactile and gestural source points. Unlocking the hidden sonic potential of architectural spaces, distorting custom analogue instruments using other analogue devices – these are the types of systems that have remained fairly constant in the group’s work to date.
A few days ago, I listened to Blossoms over headphones as I worked in my yard, ripping out a summer’s worth of overgrown deadly nightshade and smartweed. As I pared back tendrils of lemon balm choking out a patch of seven-foot-high cosmos, the need to temper my unruly yard in suburban Detroit was mixed up with thoughts of violent action taken against potentially intelligent species. It’s easy to imagine these compositions as having been cultivated from plant-derived interactions, easy to imagine that the album and track titles were grown from the metaphorical mulch of Emptyset’s decade-long harvest.
Blossoms was not created using the sonic impulses of plant life, however. Machine learning and a custom software model were used to synthesise (or "seed") new work from their prior output, along with new improvised sounds. Over two years, Ginzburg and Purgas studied the growing technology and worked with programmers at the forefront of neural network artificial intelligence systems. In March of this year, a breakthrough provided the type of range and complexity the group required.
The structural unfolding of Blossoms seems to mimic the processes of an intelligence gaining traction as the machine learning system finds patterns. Like any good neural system, it uses a logic that humans otherwise might not. We hear a neural network “becoming”, catching up with itself as a structure locks into place. There is something not-quite-but-almost voice-like in the openers ‘Petal’ and ‘Blossom’ as we listen in on exchanged phrases working in a logic loop that begins to mimic itself in an increasingly frenetic conversation, returning to the rhythm and fact-checking with confidence: “Am I a thing? Yes, I am a thing.”
‘Pollen’ disperses the grains through a saw-like resonance into something more traditionally, if experimentally, song-like. ‘Blade’ slides particles into a reverberant field of pretty-but-sinister ambience, like a territorial entity tentatively accepting its surroundings.
Blossoms moves into a strong back half with ‘Axil’. The record’s world expands, as knocking bass thrums get pushed further into a distant somewhere, and the highs pierce a little closer. Then it all locks into a bit of a chugging banger before the clatter and blast of ‘Filament’, and then the muffled sounds of ‘Bulb’ – which seems to mimic something struggling to get to the surface. ‘Stem’ breaks itself apart, then returns to a version of the contagious chugging pattern of ‘Axil’. Is it pareidolia (the act of hearing things in static) or is that a voice chanting “gamma ray” in the mix? Purposeful or not, it would be totally appropriate given this morsel relayed to me by Purgas and Ginzburg: “In order to create a final sound, we placed the pieces within reverb that we generated from an impulse taken in the Trawsfynydd Nuclear power station in Snowdonia. […] It seemed appropriate to use this reverberation signal as a container to work across the record.”
But getting back to Cleve Backster: after clipping the polygraph leads to the leaves of the office Dracaena (sometimes known as “happy plant”), he tuned his thoughts to acquiring a match, to see if the presence of fire caused any kind of electrical reaction. According to Backster, the very thought of fire-plus-plant sent the equipment meters spiking. His conclusion was that his secretary’s plant must be anxious – and obviously, telepathic. The theory of “primary perception” (the claim that plant life has psychic abilities) was born. But when other scientists attempted to replicate Backster’s results, they came up short. Was signal bleed from faulty equipment to blame? Or, as some people wondered, were Backster’s own psychokinetic signals influencing the galvanometer and the results? Was the plant ever really attempting to communicate at all?
We’ve come a long way since the speculative psychic experiments of Backster, but I can’t help but wonder what he’d think of the idea of seeding a machine to grow art. Call me old-fashioned (and I am by no means a Luddite), but I’m usually a little leery of AI-derived compositions. For whatever reason, that’s not the case here. It might be that I’ve learned to trust Emptyset’s methodologies or possess a simpatico appreciation for the conceptual leanings behind their productions. Regardless, however you or I might feel about almost-literal computer music is beside the point. The strong sense of perpetual emergence – of listening in on an intelligent system gaining confidence – makes Blossoms an especially remarkable listen.
