Machinations
by Jacob Niedzwiecki
I'm very grateful to dance+words for not just convening some fabulous and smart individuals for illuminating roundtables, but also making space for a wider group to listen and respond. As a (former) professional ballet dancer, I came of age creatively in the genteelly shabby manse of a graying imperialist artform, with senior instructors who longed for the glory days when ballet was — like space exploration, American literature, or chess — heavily funded by government as part of the civilian conflict space of the Cold War. As a choreographer, I leaned away from that institutional context and towards more street-level settings, inspired by authors like William Gibson1, who once said with regards to technology that "the street finds its own uses for things."
I’ve lived through the complete upending of two creative fields, when the street found its own uses for things, software-wise. The first was large-scale amateur music piracy with apps like Napster and BitTorrent, and the second is large-scale amateur image generation using generative AI models like Dall-E or Stable Diffusion. (We could also call this the enclosure and exploitation of the entirety of the visual output of human culture to date — your mileage may vary.)
Unlike writing, music, or visual art, dance remains largely untouched by computational exploitation, though Fortnite and TikTok (heralded by Second Life and Vine) have nibbled around the edges. Fortnite is the most popular multiplayer video game in the world, and player avatars can bust out basic dance moves or pay for fancier ones. The game's developers, taking advantage of the lack of copyright for choreography in American law, have ripped off and sold popular dance moves without crediting or compensating their originators, who can actually be traced, because these moves spread via TikTok. There are also participatory dance video games like Dance Dance Revolution and Beat Saber, which license and pay for the music they use. Clever users have created open-source clones and mods (modifications) for these games so they can dance to music they like and share choreographic scores for those bootleg songs, but the motivation is to avoid the music licensing regime, rather than anything to do with the choreography.
To some degree, the lack of computational exploitation of dance is because of a lack of demand. Even a fabulous prize like the New York Public Library’s comprehensive video-based Broadway Archive, if stolen, laundered, and fed into training a generative AI, would let a nefarious online brigade… flood the world with knockoff musical numbers? Step three, profit? Bonne chance, là.
And yet. Every time one has that impulse regarding the future — that things will, probably, be alright — it’s worth checking what shape of curve we’re on. For example, climate change is on an exponential curve, which is a mathematical way of saying things will get much worse, much faster, than primates can really imagine. On any exponential curve, we're Wile E. Coyote, still hovering in the air out of disbelief at how quickly the road eroded from beneath us. And so the development of machine learning ("AI") in late capitalism seems to demand some aggressive engagement with its possible futures, which are in fact already present.
The technical challenges of archiving the digital were well-understood by the professionals I listened to at the Toronto and Montreal sessions. I want to pose a few more speculative questions for roundtable participants. These could serve as prompts, not for "future-proofing" an archive, but for designing and negotiating an archive that can continue to function meaningfully deep into the twenty-first century.
How might you distinguish between human and automated users of your archived material? Will you?
How will you discriminate between onsite and remote users of your archived material?
What will you do when non-profit or academic users launder your archived materials into datasets for training commercial AI?
When labour becomes expensive, employees turn to automation. What conditions might lead corps de ballet dancers to be more expensive than robots? (Disney already uses robotic performers in stunt work.)
If you become trusted by vulnerable artistic communities (i.e. drag queens experiencing stochastic terrorism in the US, indigenous dancers maintaining private cultural traditions), how will you maintain that trust under pressure? Will you maintain warrant canaries or fight subpoenas?
Will you distinguish between dance created entirely by humans, and dance created using generative AI?
Will you archive dance forms or trends that exist only online on video, like TikTok challenges or Fortnite avatar moves?
How will you navigate mandates like the EU's right to be forgotten?
Regardless of who, if anyone, monetizes it, and who remembers it, dancing is a primeval impulse, and dancers will keep moving. Whether we'll invite machines into the cypher is yet to be seen.
1 Famous for inventing the term 'cyberpunk'. Aging as a dancer is very cyberpunk: like Gibson's character Johnny Mnemonic, there is information trapped in my nervous system that I cannot expunge. The exhortations of my pas de deux teacher — "If you drop the woman you're lifting, you had better hit the ground underneath her to break her fall" — still occasionally force me to suppress that impulse when escorting my mother along an icy sidewalk.
Jacob Niedzwiecki (@jakemoves) is a queer artist who works in code, media, and movement. He frequently collaborates with dance and theatre artists as a sort of technological dramaturg, and has created web- and app-based experiences with TIFF, Outside the March, adelheid, and others. His custom app for the immersive show MeetMe won a Montreal English Theatre Award in 2023. He lives between Toronto and Montreal.