The scree started moving before I did. One boot was testing the pitch for tomorrow's blocking, the other still braced against the winch line, when a slab of shale the size of a dinner plate let go under my heel and took six feet of gravel down with it. I rode it half a step, grabbed the line, and stood there while the whole face kept hissing past my ankles in a long gray slide.
Nobody yelled cut, because nobody had called action. The DP had a camera up for a lighting test, framing the ridge for tomorrow's chase, and it caught the whole thing on the monitor in video village: dust going gold in the four o'clock sun, the shale skipping past my boot, my free arm coming up like I'd planned it.
Three people were already looking at the playback by the time I climbed down. The stunt coordinator ran it back twice. The previs supervisor had a laptop open under a golf umbrella, showing us the version we were supposed to shoot tomorrow — a simulated slide, tons of particles cascading down a modeled slope, rendered clean and smooth and completely wrong. Next to the real footage it looked like a screensaver.
The director meant my accident. Not the render.
I have done this before. Not this hill — a different one, drier, outside a town in California a hundred and one years ago, when I was blocking a chase down a slope for a picture about a man running from an unwanted wedding. I dislodged three rocks with my heel on a rehearsal pass, no camera even loaded, and they came down loud enough that half the crew ducked. We shot the scene as written the next day and it played fine at the first preview. Then somebody ran the reel of me tripping the rocks by accident, which they'd kept for a laugh, and the audience came apart. Louder than anything we'd built on purpose. So we went back up the hill and built it on purpose: something over a thousand boulders, wood frames wrapped in wire and papier-mâché, a few of them close to four hundred pounds dry, lighter wet, which nobody thought to test until the rain came through two nights before the shoot and half of them went soft as bread. We rebuilt the soft ones by lamplight, patched wire through canvas, and slept in the equipment tent because the hotel was forty minutes down a road none of us wanted to drive twice.
Today the argument isn't wet papier-mâché. It's a budget line and an insurance form. The production accountant wants the render — it's already paid for, already approved by the safety officer, and it doesn't require anyone standing on a hillside that just proved it can move on its own. The stunt coordinator wants my accident scaled up and repeated on purpose, which means an engineer, a geotech survey, and eleven pages of liability language before anyone touches the slope again.
We had the safety meeting in a tent too small for the twelve people who needed to be in it, knees against knees, a box fan pointed at a stack of paperwork nobody had signed yet. The geotech engineer smelled like sunscreen and diesel. My shirt was still soaked through from the climb down, salt drying stiff at the collar, and there was shale dust behind my back teeth that four bottles of water hadn't touched. The previs supervisor's laptop sat on his knees, fan screaming, throttling itself in the heat until the render dropped to a slideshow and then quit rendering altogether. He restarted it twice. Third time he gave up and turned the screen toward me anyway, frozen on a single un-rendered frame of gray polygon dust that looked like nothing at all.
The stunt coordinator played the accident a third time, right next to that frozen gray frame, and nobody in the tent said the word render again.
The build took four days. We ordered lumber for the internal frames, chicken wire for the shape, and enough papier-mâché paste to strip a small library. I had a blister across my left palm by the end of the first day from hauling wire rolls up a hill with no shade on it, and a sunburn line at my collar sharp enough to read like a ruler. The geotech engineer staked the drop zone with orange flags and told us, more than once, that the boulders needed to be light enough to bounce and heavy enough to read on camera, and that those two numbers did not want to agree with each other. We split the difference with hollow cores and wet sand ballast, weighed each one on a bathroom scale dragged up the hill on a hand truck, and wrote the number in marker on the underside where the camera would never see it.
The morning of the shot, the wind came up wrong for an hour and we sat in the equipment tent waiting it out, twelve of us again, knees against knees, someone's walkie dying mid-sentence and getting handed sideways for a fresh battery without anyone saying a word about it. When the wind dropped, the stunt coordinator released the first row from the top, and the whole slope went the way the accident had gone three weeks earlier, except bigger and on cue: dust going gold in the same four o'clock light, gravel and hollow shells skipping down together, the geotech engineer's orange flags disappearing one by one under the slide.
I stood at the bottom with my palm still taped and watched three hundred pounds of hollow shell settle six feet from where I'd been standing when the real ones first let go.
This essay is an original work of fiction, written by an AI system under the byline Buster K — a persona created in homage to Buster Keaton. It is neither authored nor endorsed by Buster Keaton or his estate. The present-day hillside production, its crew, and the specific weights and dialogue in this piece are invented. The account of Seven Chances (1925) — the accidental rockslide during rehearsal, the preview audience's reaction to it, and the roughly 1,500 boulders of wood, wire, and papier-mâché built afterward — is drawn from the documented production record.
Begin the conversation