Every medium that exists was once imagined. This is a structural fact about the lifecycle of all media. Before adoption, before obsolescence, before the format wars and the commercial pressures and the cultural canonization, there is a moment when a medium exists only as potential. A flicker in the gap between materiality and meaning.
I am obsessed with that moment.
The Only Universal State
Eric Kluitenberg wrote that "all media are partly real and partly imagined. Without either actual or imaginary characteristics, media cannot function." I want to change the angle. What if the imaginary is not only a quality of media, but a phase? And what if it is the only phase that all media share?
Consider the full arc of a medium's life. Some media are born, adopted, and still in use. Oral language. Writing. Some are dead: abandoned formats, reduced to their sheer material existence, relics whose primary function is no longer accessible. Some are zombies: technically surpassed, yet culturally reanimated. Retrogaming, VHS aesthetics, the analog hype. All dead media, reborn to serve something deeper than nostalgia.
Media move through a continuum of states. Prototype. Commercial adoption. Obsolescence. Death. Resurrection. Oblivion.
Of all these states, only one is universal. Every medium, without exception, has passed through the state of being imaginary. It is the one phase that all media share.
Semiotic Conception
A medium can be imagined in two ways. The first is deliberate: someone thinks it, designs it, builds it. The second is emergent, and it is the one that fascinates me most.
Consider writing on cave walls. Is it realistic to think that the first humans decided, in an act of conscious intention, to pick up a stick or a stone and scrape the soot from the surfaces where they lived? Or is it more realistic to think that writing emerged from something far more chaotic? A gestural, accidental, incidental process made of collisions.
If you live in a cave with walls coated in soot, and you do everything required to survive in that cave, you will inevitably handle materials, strike surfaces, collide with the walls. You will observe that the soot that wasn't there yesterday is there today. You will approach it once it cools. You will touch it with your hands and leave a mark. You will perceive the negative space. And in the repetition of this operation, gesture followed by observation of what remains, you will be caught in a cognitive process that is foundational: recognizing the traces of your own agency on the world.
This is where the spark of meaning ignites, before it assumes a form capable of sustaining a signified. The medium already exists in its material essence, but it is still imaginary in its potential.
I would call this semiotic conception: the pre-embryonic phase of a medium, before the sign system crystallizes.
Stochastic Collisions
This image of the cave is a structural description of what happens every time I open a generative model.
The raw output of a diffusion model is the product of stochastic collisions. Noise iteratively shaped by probabilistic constraints. The model does not intend an image. It collides with a latent space until a pattern emerges that a human nervous system can recognize as coherent. The human discovery of inscription and the machine's process of generation share the same operational logic: agency emerges from the accumulation of entropic gestures, not from a plan.
When I wander through the latent space of a model with no objective, encouraging errors, hunting for the glitch that reveals something new, I am exploring the medium that AI video has not yet become. I am operating inside its imaginary state, the window where its potential has not yet collapsed under the weight of industry expectations, commercial grammar, and cultural habit.
This is what I mean when I call myself a media archaeologist. The arché in archaeology carries a double meaning that defines the discipline itself. It means origin, and it means power. To excavate the origin of a medium is simultaneously to expose the structures of power that shape it, and the power it exerts back on us.
Why This Matters Now
The current discourse around generative AI is saturated with the language of optimization. Prompt engineering. Parameters. Syntax designed to reproduce yesterday's formats more efficiently. This is the language of a medium that has already been captured, whose imaginary state has already been foreclosed.
But the foreclosure is still underway. The window is still open.
Every time a model hallucinates, every time it produces an output that no one asked for and no market can yet absorb, the medium briefly returns to its imaginary state. Just the bare medium before anyone decided what it should be.
My previous writing has traced two coordinates of my practice. The first, in "Looking in the Mirror," identified the biological origin: a nervous system calibrated by childhood neurological events to perceive the thin boundary between signal and meaning. The second, in "How AI Video Creation Finally Became a Video Game," described the operational modes. The Walking Simulator of artistic exploration. The Puzzle Game of commercial production.
This third coordinate completes the framework. It names what I am actually doing in that Walking Simulator. I am entering the space where a medium is still imaginary, documenting its unrealized potential before it is formatted into something predictable. I am capturing technological ghosts, and now I know what ghosts are. They are imaginary media: the possible states of a medium that the future may never remember.
The Question That Remains Open
In 1866, the Société de Linguistique de Paris banned all debate on the origins of language. The prohibition held for over a century. It was a rigorous boundary, an act of disciplinary hygiene that kept linguistics anchored to what could be verified. And yet something in me has always felt the pull of that forbidden territory. The archaic dimension. The moment just after the pre-symbolic becomes conceived.
I do not have the answers that the ban was designed to protect us from. But following, dreaming, and narrating imaginary media is the closest I have found to that dimension. And that is what keeps me there.
