Cabinet of Small Machines

Every calendar you have ever used is a piece of quiet Newtonian propaganda. It carves the day into identical boxes — nine o'clock, ten o'clock, eleven — and asks you to believe that an hour in a dentist's waiting room and an hour falling in love occupy the same amount of time. The grid is administratively convenient and phenomenologically false.

Henri Bergson had a word for what the grid erases: durée, the lived, unequal thickness of experienced time, as opposed to the interchangeable units a clock hands out. A grid calendar is clock-time with a UI. It lets you schedule a life but not describe one.

This experiment replaces the boxes with a single vertical stream — one line the day moves down, events surfacing and receding as you scroll instead of sitting bolted into hour-wide cells. Nothing here solves scheduling. It asks a smaller, stranger question: if the interface didn't insist on uniform units, what would a day actually look like laid out honestly? Would deep work get to be as long, visually, as it feels? Would the gap after lunch — the underrated, undecorated openness of open time — get to look like relief instead of a blank cell waiting to be filled?

A stream can't be subdivided the way a grid can, and that's the point: it resists the grid's implicit argument that all hours are fungible and any of them can be moved or double-booked without loss. Durée doesn't optimize. It accumulates — and most people meet that argument daily, at nine a.m., without ever suspecting it's philosophy.

André Bazin called cinema “the myth of total cinema” — the dream of an image so complete it becomes indistinguishable from reality, watched together, in the dark, by strangers who agree for two hours to want the same thing. It's one of the more sincere claims art has made for itself.

None of that survives contact with a P&L statement, and this experiment doesn't pretend it does. It's a screen-allocation and profit model for a theater chain: sliders for ticket price, concession attach rate, staffing, screen count per title — the unglamorous arithmetic that decides whether Tuesday's 7:40 slot goes to the awards contender or the fourth showing of a marketable sequel. Move enough sliders and you feel a manager's actual week: the myth of total cinema, reallocated hourly, one auditorium at a time.

It would be easy to read this as cynicism — art demoted to inventory. The more interesting reading is that the ledger was always there, underneath the myth, and pretending otherwise doesn't make theaters more sacred, it just makes their operators less visible. A screening survives because someone did the allocation math correctly enough times in a row. The total-cinema experience an audience gets to feel unbothered by economics is a product of someone else being thoroughly bothered by economics on their behalf.

The model doesn't resolve the tension between the theater as ritual space and the theater as square footage generating yield per showtime. It just puts both in the same interface, so you can't hold one thought without the other sitting right next to it on the same screen.

Glass is a strange thing to render because glass, visually, has almost no properties of its own. You cannot describe what glass looks like without describing what's behind it, bent. Its whole identity is deference — refraction, dispersion, the way it takes an image of the world and quietly lies about where things are.

This experiment fakes that lie with SVG filter primitives — displacement maps, chromatic offsets, blur stacked on blur, no photons anywhere in the pipeline. It runs a reproduction of The School of Athens behind a pane of nothing and asks a feed of matrices to convince you there's a physical object there, distorting light the way physical objects do. It usually works, which should feel a little uncanny once you notice it: you're not looking at glass, or even a picture of glass. You're looking at an argument about glass, written entirely in mathematics that has never met a lens.

Gaston Bachelard wrote about the “material imagination” — the way certain substances carry a poetics prior to, and independent of, any actual sample of the substance in front of you. Glass has one too: clarity, fragility, the promise and the threat of seeing straight through something. What's interesting about faking it in code is that the fake only succeeds by honoring that poetics rather than the physics. Real refraction follows Snell's law precisely. This filter graph follows nothing but does it read as glass, tuned by eye, a displacement map dragged until it looks less like a bug and more like a window.

That's the quiet confession inside most convincing digital materials: they aren't simulations of physics, they're simulations of what physics feels like to an eye that already knows what glass is supposed to do. The realism is borrowed entirely from the viewer.