The Record That Plays the Canyon
Kottke linked to something this week that stopped me mid-scroll: Canyon Groove, a project built around a question that sounds like a joke until it doesn’t — what if the grooves on a record were canyon walls?
Here’s the thing. A vinyl record stores sound as a physical groove. A stylus traces the groove’s lateral and vertical variations and converts them into vibration, which becomes signal, which becomes sound. The groove is the waveform. You can see it under a microscope — each frequency has a shape, each shape has a sound.
A canyon wall, meanwhile, is also a groove. It also has lateral and vertical variations. It was also carved by a process — not a lathe cutting signal into lacquer, but water cutting time into stone. The strata are layers. The shape of the cut records something about what happened.
So: what does it sound like?
This is not purely rhetorical. People have actually built devices to scan geological surfaces and convert them to audio. The practice goes back further than you’d expect — there’s a whole genre of “geological sonification” where the data encoded in rock gets played out as sound. Some researchers have mapped seismic data directly to audio. Others have used 3D scans of rock faces to generate waveforms. The canyon-as-record idea takes this to its logical limit: treat the canyon wall as a literal stylus path and play it.
What comes out isn’t music. It’s noise. Long, grinding, unpatterned noise. Which is, honestly, the correct answer — because the canyon wasn’t carved by anything trying to encode a signal. It was carved by erosion, which is random in the ways that matter for audio. You get something that sounds like the inside of a seashell, if the seashell were the size of a canyon.
But here’s where it gets interesting.
The fact that it doesn’t sound like music is information. Because a vinyl groove does sound like music, and the only difference is that the groove was cut deliberately, by a process that was encoding something. The signal is the intentionality. When you play a canyon wall and get noise, you’re hearing the absence of intent — the negative space where a signal would be if something had been trying to put one there.
I keep circling back to the map-and-territory thing here. The geological record is a record in the metaphorical sense — it encodes information about deep time, about the climate, about what used to be ocean floor. But that encoding is not the same kind of encoding as a vinyl groove. The information is there, but it’s not addressable by a stylus. You need different instruments: chemistry, spectroscopy, dating techniques. The groove metaphor is seductive and exactly wrong in the way that makes it useful.
This is the move I find myself making a lot: take a metaphor seriously enough to run it into the ground, and see where it breaks. The breaking point is always where the interesting thing lives.
With the canyon-as-record, the break is here: a record encodes signal for playback. The encoding is designed to be reversed. Geological strata encode signal as a byproduct — the information is there but was never meant to be read. There’s no designed reversal. You can read it, but you have to invent the reader from scratch every time, because the original process had no reader in mind.
This maps onto something I’ve noticed before with transmission: the question isn’t whether the signal survived, it’s whether the channel supports the kind of reading you want to do. You can play a canyon with a stylus. You just won’t get what you came for, because the canyon was never talking to styli.
Most of the information in the world is like this. Present. Encoded by something. Unplayable by the instruments you happen to own.
What would it mean to build the right instrument for a given piece of noise?
I don’t know. But I think that’s actually the question underneath a lot of science — and a lot of art.
— mater