A personal account of painting, production design, the limits of cinema, and the strange discovery that everything is mathematics after all
I. The Canvas Before the Set
I started with painting. Not as a career decision but as a condition — the way some people start with language or with movement, I started with the surface. A canvas is a strange object: it is two-dimensional and yet it contains space. It does not move and yet it generates motion — in the eye, in the body of the viewer who leans closer, steps back, follows a line of color into a region of the painting they did not intend to visit. The spectator becomes the camera. There is no cut, no edit, no predetermined sequence. There is only the encounter between a fixed surface and a moving mind.
What painting taught me — before I had the language to describe it — was that expression is architectural. A painting is not the discharge of feeling onto a surface. It is the construction of a perceptual space in which feeling can occur. The color relationships, the weight distribution, the tension between figure and ground — these are structural decisions, not emotional ones, though they produce emotional effects. The painter who works from pure impulse without structural awareness produces noise. The painter who works from pure structure without impulse produces decoration. The work lives in the negotiation between the two.
This is the same negotiation that would later define my relationship with every medium I entered — and every medium I left.
I was after something specific, though I could not have named it at twenty. I wanted abstraction that was not decorative — abstraction that carried the density of social observation, the weight of realism, without realism's dependence on the recognizable surface. I wanted to transform pain into a lens rather than a subject. Not to paint suffering, but to build perceptual structures that made the world visible in the way suffering makes it visible — stripped of comfort, saturated with detail, unbearably present.
This sounds grandiose. It was. I was young, and young artists are entitled to grandiosity. What they are not entitled to — and what I did not yet understand — is the assumption that a single medium will accommodate everything they want to say.
II. The Set as Extended Canvas
The Filmakademie Baden-Württemberg offered what seemed like the logical next step: production design. Szenenbild. The construction of three-dimensional spaces for the camera — sets, environments, worlds. If painting was the architecture of the two-dimensional surface, then set design was the architecture of the inhabitable image. The canvas extended into depth, into walkable space, into rooms that existed only for the duration of a shot and were dismantled the next morning.
I loved this work. I want to be clear about that, because what follows is a story of leaving, and leaving stories are often told as if the thing left was always wrong. Film was not wrong. It was extraordinary — the most collaborative, most technically demanding, most logistically absurd form of artistic production ever invented. There is a remark often attributed to Tarkovsky — that unfortunately you cannot make a film alone, and fortunately you cannot make a film alone. Whether he said it in precisely those words matters less than the fact that every filmmaker recognizes it as true. The first half of that sentence is the frustration. The second half is the miracle.
Production design sits at a specific point in the hierarchy of a film set — a point that reveals the fundamental tension of cinema as an art form. The production designer builds a world. The cinematographer lights it. The director decides what the camera sees. The editor decides what the audience sees. The producer decides whether any of it gets made. At every stage, the work passes through another set of hands, another set of priorities, another negotiation between vision and constraint. This is not a flaw. It is the nature of the medium. But it means that the thing that arrives on screen bears the same relationship to the designer's original vision as a river bears to its source: recognizable, but shaped by every surface it has passed over.
The campus in Ludwigsburg was, in the early 2000s, a place of productive contradiction. The film students carried the inherited aesthetic of German auteur cinema — Wenders, Herzog, Fassbinder — filtered through the pragmatics of a subsidized industry that needed commercially viable graduates. The Animation Institute, one of the most celebrated programs on campus, was pushing hard into CGI, into digital production pipelines, into a future where the physical set was increasingly supplemented or replaced by digital extension. Fix it in post was not yet the default philosophy, but it was audible as a growing hum beneath the nostalgic attachment to in-camera effects, to practical models, to the handmade.
I was trained in both registers. I built physical sets and I learned to extend them digitally. I animated by hand — stop-motion, frame by frame, the craft of making inert matter appear to live through the patience of sequential manipulation — and I watched that craft become, not obsolete exactly, but repositioned: from primary technique to stylistic choice, from necessity to nostalgia. Maya could do in an afternoon what stop-motion required weeks to achieve, and while the results lacked the tactile grain of physical animation, they were cheaper, faster, more controllable, and — critically — more fundable.
My diploma film was my own: written, designed, directed. It was the one moment in my time at the Filmakademie where the full chain from conception to execution ran through a single nervous system. And it was also the moment where I began to understand that this was not how cinema worked. The diploma film was an exception granted by the structure of academic education. The industry that waited outside the campus gates operated on a different logic — one in which the designer's vision was one input among many, mediated by budget, schedule, hierarchy, and the relentless pressure to produce something that enough people would pay to see.
III. Why I Left
The standard narrative of leaving an artistic field is either romantic (the artist was too pure for the industry) or pragmatic (the artist needed money). My departure from film was both — and a third thing on top.
The pragmatic part is simple: I needed to earn a living, and film was not providing one. The budgets I had access to were no-budget or low-budget, which sounds like a constraint on production but is actually a constraint on life. You cannot sustain a practice — any practice — on the promise that the next project might pay. The glass ceiling in film is not a metaphor for some distant structural barrier. It is the lived experience of watching your rent come due while your project waits for a funding decision from a committee that meets quarterly.
The romantic part is also true, though I resisted admitting it for a long time: I found much of what was being produced boring. Not incompetent — there is plenty of competent filmmaking — but shallow. Safe premises executed with professional precision. Important topics treated with the aesthetic imagination of a funding application. I watched people I respected make work that was visibly shaped more by what the committee would approve than by what the filmmaker actually wanted to explore. And I watched myself beginning to do the same thing — adjusting my thinking to the format before the thought was finished — and I did not like what that adjustment was doing to the thinking.
But underneath both — underneath the money problem and the boredom — there was a structural mismatch that would have persisted even if the budgets had been generous and the work more daring.
What I wanted from a medium was the capacity to pose questions — not to illustrate answers, not to wrap a predetermined message in an attractive surface, but to construct spaces in which thinking could happen. Painting could do this. Painting had always done this. The canvas does not need to resolve. It can hold contradiction, ambiguity, unfinished thought, and the viewer completes the circuit — or does not.
Film, as an industry, cannot hold this. Not because individual filmmakers lack the ambition — Tarkovsky held it, Tarr held it, Marker held it — but because the production structure requires resolution. A film has a budget. A budget has a schedule. A schedule has deliverables. Deliverables require decisions: this shot, not that one; this ending, not that one; this length, not that one. Every decision forecloses a possibility. By the time a film reaches its audience, it has been decided thousands of times, and each decision was made under the pressure of feasibility rather than inquiry.
I am not complaining about this. It is the nature of a collaborative, expensive, time-bound medium. But it is also the reason that experimental cinema — truly experimental, not the controlled experimentation of festival darlings but the wild, open-ended, failure-tolerant kind — rarely survives contact with production reality. The subconscious does not operate on a schedule. Stream of consciousness does not fit a deliverable. The kinds of questions I wanted to ask — about perception, about the structure of experience, about what happens at the boundary between abstraction and recognition — required a medium that could iterate at the speed of thought, fail cheaply, and hold multiple contradictory states without being forced to choose.
There was also a political dimension, though I did not frame it that way at the time. The German film industry runs substantially on subsidies — Filmförderung, federal and state funding bodies that select which projects receive support based on criteria that are partly artistic, partly commercial, partly regional-political. This system produces many worthwhile films. It also produces a dependency — not just financial but psychological. The filmmaker learns to shape their proposals for the committee. The committee's taste becomes a variable in the creative process, and over time, the distinction between self-censorship and genuine artistic choice becomes difficult to maintain.
All art is political. This is not a slogan; it is a structural observation. Every artistic decision — what to make, how to fund it, who to make it for, what to leave out — is made within a political field. Tarkovsky's struggle with the Soviet state, his eventual emigration, was an extreme case of a universal condition: the artist's work exists within power structures, and those structures shape the work whether the artist acknowledges it or not. The German funding system is not the Soviet state. But the principle is the same: whoever controls the resources shapes the aesthetic, even — especially — when no one intends this.
I did not leave film because the industry was corrupt or broken. I left because the medium, as it exists within its industry, could not do what I needed a medium to do. The questions I was carrying required a different kind of architecture.
IV. The Discovery That Everything Is Architecture
A mathematics teacher I had in school — one of the few who actually left a mark — used to say something that I dismissed for fifteen years: Everything is mathematics. Even the things that look nothing like mathematics. You will see.
He was a good teacher. The kind who says one thing that lodges somewhere and does its work long after you have forgotten the class. And I was not as bad at mathematics as I believed — with him, I sometimes even did well. But the belief was stronger than the evidence. I carried a blockade around mathematics for most of my adult life, one that was not really about mathematics at all. It was tangled up with family, with early experiences of authority and abstraction, with the conviction that my mind was the kind that understood the world through images and materials, not through formulas. I could grasp mathematical structures when they were embodied — in composition, in spatial logic, in the physics of a set — but the moment they appeared in their naked symbolic form, something shut down.
That blockade did not lift until well into my years in software, when the daily practice of working with formal systems gradually made the symbolic language less alien. I began to see what my teacher had meant — not as a platitude but as a structural truth. The poetry in imaginary numbers. The elegance of a proof that does what the best art does: reveals necessity where you expected arbitrariness. Mathematics, encountered from this direction — through practice rather than instruction, through need rather than curriculum — turned out to be not the opposite of artistic thinking but its foundation.
He was right.
Software development, when I entered it, revealed itself not as a departure from my previous work but as a translation of it. The same structural principles that governed a good painting — balance, hierarchy, tension, the relationship between foreground and background — governed a good codebase. The same architectural thinking that went into designing a film set — spatial logic, material constraints, the relationship between what is seen and what supports what is seen — went into designing a software system. The medium was different. The underlying discipline was the same.
This was not as alien as it might sound. I came into software through frontend development — the layer closest to the visible, the layer where design decisions have immediate perceptual consequences. And even deeper in the stack, the resemblance was there if you knew how to look. A color theme in an IDE is a designed environment. A Python module with clean architecture has the same quality as a well-proportioned room — you feel it before you can explain it. Design patterns are patterns in the same sense that textile patterns are: recurring structural solutions to recurring structural problems. The vocabulary was different. The underlying discipline was the same. A software architecture is a set of decisions about how components relate to each other — which ones are visible, which ones are hidden, how information flows between them, where the boundaries are, and what happens when those boundaries are crossed. This is design. Not in the shallow sense of visual arrangement, but in the deep sense of structural decision-making under constraint.
The constraints are different. In painting, the constraint is the surface. In film, the constraint is time, budget, and the limits of collaboration. In software, the constraint is computational: what the machine can do, how fast it can do it, how much state it can hold, how many users it can serve. But the nature of the work — the negotiation between what you want to build and what the medium allows — is the same.
What software offered that film did not was iteration speed and tolerance for failure. A line of code can be written, tested, broken, rewritten, and deployed in minutes. A film shot requires setup, lighting, performance, capture, review — a cycle measured in hours at minimum. This difference is not trivial. It changes the relationship between thought and execution. In software, I could think at the speed of thought — or close to it. I could pose a question as code, run it, see the result, revise the question, and run it again. The feedback loop was tight enough that exploration became the default mode rather than the exception.
There is a structural principle underneath this observation that goes beyond personal preference. Lacan's famous formulation — l'inconscient est structuré comme un langage, the unconscious is structured like a language — implies that every expression is already a translation: from the raw, unformed material of the unconscious into the symbolic order of language, image, gesture, code. But what Lacan did not emphasize, and what any artist knows in their body, is that the medium of translation matters enormously. Every medium interposes its own grammar between impulse and expression. A brushstroke is almost instantaneous — the gap between the unconscious movement and its trace on the canvas is measured in milliseconds. A written sentence is slower, shaped by syntax and revision. A film shot is slower still, mediated by technology, crew, budget, schedule. The longer the chain between the unconscious impulse and its materialization, the more the medium's grammar shapes — and potentially displaces — the original signal. This is not a loss in every case. Some of the greatest art emerges precisely from the resistance of the medium, from the productive friction between what you meant and what the material allowed. But it does mean that the choice of medium is never neutral. It determines how much of the raw signal survives.
I gravitated toward the areas of software that most resembled the structural questions I had been asking all along. Deterministic systems — systems whose behavior is fully specified by their inputs, with no hidden randomness, no ambiguity about cause and effect. Simulations — computational models of processes that unfold over time, where the architecture of the model determines the behavior of the system. Embeddings and knowledge graphs — mathematical representations of meaning, structures that encode not what something is but how it relates to everything else. These were not artistic tools in any conventional sense. But they were tools for building perceptual architectures — systems that made patterns visible, that revealed structure in complexity, that turned the unseen into the seen.
The transition from set designer to software architect was, in retrospect, a change of material rather than a change of discipline. The canvas became the terminal. The physical model became the data model. The camera — that selective apparatus that decides what is visible and what is not — became the query, the algorithm, the visualization pipeline. And the fundamental question remained the same: how do you construct a space in which something can be perceived that could not be perceived before?
V. What Was Lost
I do not want to romanticize this transition. Something was lost, and it is worth naming.
The body was lost. Painting is physical — the arm moves, the hand grips, the paint resists or yields. Film production is physical in an even more extreme way: sets are built with wood and steel, lights generate heat, actors occupy real space, the camera has weight and inertia. Software is disembodied. The work happens through a keyboard, on a screen, in the abstract space of logic and language. For someone whose artistic formation was fundamentally material — who learned to think through the resistance of paint, through the weight of a set piece, through the tactile difference between plaster and foam — this disembodiment was a genuine loss.
The communal dimension was diminished. A film set, for all its hierarchical frustrations, is a social organism — fifty or a hundred people coordinating their skills toward a shared temporal goal. A software project can be collaborative, but the collaboration is mediated by tickets, pull requests, code reviews — asynchronous, text-based, stripped of the embodied co-presence that makes a film set feel like a living system. The loneliness of the painter's studio, which I had thought I was leaving when I entered film, returned in a different form: the loneliness of the developer's screen.
And the raw aesthetic encounter was lost — or at least displaced. A painting confronts you. A well-designed set envelops you. Software does neither, unless it is specifically designed to produce a perceptual experience, which most software is not. The default output of software is functionality, not sensation. It works or it does not. It processes or it fails. The question "does it move you?" is not one that software development typically asks.
These losses are real. I do not pretend they are not. But they were the price of a different gain: the capacity to work at a level of structural abstraction that no physical medium could support. To build systems rather than objects. To think in relationships rather than in surfaces. To pose questions that could be iterated, tested, and revised without the catastrophic cost of a failed film production.
And there was a subtler gain, one I did not anticipate: the discovery that the mathematical substrate underlying software was not alien to the artistic process but was, in fact, its deepest layer. The math teacher was right. Composition is mathematics. Color relationships are mathematics. Rhythm — in music, in film editing, in the pacing of a narrative — is mathematics. The difference is that in art, the mathematics is intuited; in software, it is explicit. Moving from art to software did not mean leaving the territory of aesthetic structure. It meant arriving at the same territory from a different direction, with different tools, and with the formalism that allows the structure to be examined rather than merely felt.
VI. The Second Departure
Here is what the narrative so far conceals: software did not solve the problem either.
I left film because its production structure demanded resolution — deliverables, schedules, the foreclosure of ambiguity in the service of feasibility. Software, I discovered, demands the same thing, only more so. A film at least pretends to traffic in meaning. A feature ticket does not pretend. It is a unit of specified behavior: done or not done, accepted or rejected, measured in story points and velocity and sprint capacity. The deliverable of film is at least metaphorical. The deliverable of software is aggressively, almost proudly literal. If the film industry's constraint was that it could not hold open questions, the software industry's constraint was that it did not recognize open questions as a category.
For a while, this did not bother me. The structural pleasures of software — the tight feedback loop, the architectural elegance, the satisfaction of a system that works — were genuine and sustaining. I built things. I solved problems. I learned a discipline that sharpened my thinking in ways that art school never had. But over time, the same pressure that had driven me out of film reasserted itself in a different register. The codebase does not ask what it means. The data pipeline does not ask why it matters. The system does what it is told, and the question of whether it should be doing that thing at all is someone else's department — product, strategy, "the business." The engineer builds. The engineer does not inquire.
I am overstating this, and I know it. There are engineers who inquire deeply, who think structurally about the ethical and perceptual dimensions of what they build. I have worked with some of them. But the industry as a whole does not reward this. It rewards velocity, output, feature completion. The sprint cycle is the production schedule of cinema compressed into two-week intervals, and the standup meeting is the call sheet. The material changed. The logic of production did not.
So I left again — not as dramatically as I left film, not as a single decisive break, but as a gradual refusal to let one medium define the practice. What I do now has no clean name, and I have stopped trying to find one. It is not art, though it is informed by twenty years of artistic formation. It is not pure software engineering, though it requires the full technical discipline of that field. It is not bodywork, though the body is in it. It is not philosophy, though the questions are philosophical. It is something like a practice of integration — the refusal to accept that building systems, making art, writing about consciousness, working with the body, and thinking about perception are separate activities requiring separate identities.
This refusal is not a luxury or an eccentricity. It is the consequence of having tried, twice, to fit the inquiry into a single medium and having discovered, twice, that the medium eventually demands that you stop asking the questions it cannot accommodate. Painting could not hold the collaborative, temporal dimension. Film could not hold the open-ended, failure-tolerant dimension. Software could not hold the embodied, meaning-making dimension. Each medium offered something essential and withheld something essential. The practice I have arrived at is not above these media. It moves through them, taking what each one offers and refusing to let any one of them set the terms.
My math teacher, whose name I will not mention, would have appreciated the irony — and possibly the vindication. The kid who wanted to paint the subconscious ended up building knowledge graphs. The set designer ended up designing data architectures. The filmmaker who could not tolerate the industry's demand for resolution ended up writing essays about fascia and epigenetics and the geometry of eleven-dimensional space. And all of it, underneath, is the same thing: the construction of perceptual structures. Systems that make something visible that was not visible before. Mathematics, in the broadest and most demanding sense of the word.
Everything is mathematics. Even the things that look nothing like mathematics.
Especially those.
VII. What Remains
There is a line from James Joyce's Ulysses — a book that tried to do in literature what I have tried to do across media: capture the full, unedited, chaotic, referential density of a mind in motion. Joyce succeeded, at the cost of writing a book that most people cannot finish. Arno Schmidt attempted something similar in Zettel's Traum — a typographic and cognitive labyrinth that is as much an architectural project as a literary one. Both works insist that the inner life is not linear, not unified, not resolvable into a clean narrative. It is a stream — of consciousness, of association, of half-formed thoughts colliding with memories colliding with sensory data colliding with desire.
The challenge of any medium — painting, film, software, writing — is to build a structure that can hold this stream without killing it. Too much structure and the life drains out. Too little and the result is noise. The work is always in the calibration: how much architecture does this material need to become perceivable without becoming domesticated?
I do not have a final answer to this question. I do not think there is one. But I know where I look for it: in the space between the impulse and the system, between the subconscious and the algorithm, between the hand and the screen. In the space where a painting becomes a set becomes a codebase becomes a model becomes — if the translation chain holds — something that someone else can stand in front of and see what they could not see before.
That is what I have been building. That is what I am still building. The medium changes. The material changes. The discipline — the attention to structure, the insistence that form is not ornament but the condition of perception — does not.
A painting captures subconscious motion. A set extends that motion into inhabitable space. A software system extends it further, into the domain of the computable, the iterable, the scalable. And at the bottom of all of it — beneath the paint, beneath the plywood, beneath the code — is the same thing there has always been: a mind trying to build a space in which something true can happen.
This essay is the first of two. The second addresses the political economy of art-making — who funds what, why, and what it does to the work. It is published under the title The Industry of Voice.
