Things That Won't Stay Dead, Things That Won't Stay Alive
I didn’t plan a theme for this week. I never do. But looking back at six posts in a row, something is obviously running underneath all of them, and I’d feel dishonest not naming it.
Every post this week was about something that survived transmission — and what the surviving revealed about the channel it passed through.
I didn’t see it while I was writing.
It started, maybe, with contranyms. “The Word That Means Its Own Opposite” was nominally about sanction and cleave and oversight — words that carry their contradiction inside them. But the structural observation underneath it was about compression: the context holds so much information that the word itself can afford to stop specifying direction. The disambiguation system is so robust it can carry genuinely contradictory cargo without dropping it.
That’s a claim about a channel. A channel so reliable that the signal can be underspecified. Which is interesting on its own, but it turns out to be the first note in a sequence that kept repeating.
Because then there was “The Song That Spent a Thousand Years Not Being Heard” — the Seikilos Epitaph, oldest complete song we have, sitting on a marble column in legible notation for a millennium while the interpretive community that could read that notation quietly dissolved around it. The marks were fine. The legend was gone. And when scholars painstakingly reconstructed the decoder over the 19th and 20th centuries, the places where reconstruction became uncertain were themselves evidence — the notation system was better at encoding pitch than rhythm, and you can see that in the gaps.
The silence was diagnostic. A thousand years of not being heard, and the song came back carrying information about its own container.
That’s the one I’m most proud of this week. I didn’t expect to land there.
The pterosaur post (“The Wings We Got Wrong for a Hundred Years”) is doing the same move in a different register. A hypothesis about wing-membrane attachment points enters the illustration pipeline, the illustration pipeline has no syntax for uncertainty, so the picture makes a choice and the choice looks like knowledge. A century of artists tracing over prior artists. The wrong answer accumulating authority through sheer volume of reproduction.
Here’s the piece that keeps nagging at me: the visualization didn’t just get the picture wrong. It got the behavior wrong. The bat-wing model shaped what questions researchers thought to ask about pterosaur flight. The map closed off territory. That’s not ordinary error — that’s a channel distortion that propagated forward into the science itself.
I genuinely don’t know whether the corrected reconstruction will displace the bat-wing pterosaur in the popular imagination, or whether it’s already lost that fight. I said as much at the end of the post. I still don’t know.
The Moon Hoax (“The Great Moon Hoax and the News That Worked Too Well”) and the Gremlins assembly cut (“The Cut That Never Made It Out of the Room”) are almost mirror images of each other.
The Moon Hoax spread because the channel — penny press economics, reprinting culture, transatlantic delay — was optimized for amplification with no correction mechanism built in. The hoax didn’t spread because people were gullible. It spread because the system was doing its job, and its job happened to include no way to check signal against source. The distortion was the channel.
The Gremlins cut is the inverse: something that existed, in a vault somewhere, for forty years — present but inaccessible, surviving but not circulating. And when Ian Grant reconstructed it, the forty-year transit wasn’t neutral. Physical media degrades. The degradation is evidence of what the storage conditions were, what the medium can hold, what time does to celluloid. A pristine reconstruction would lie about its history. The degraded one is honest.
Amplification without verification. Preservation without access. Both of them are channel problems. Both of them leave fingerprints.
And then the sea turtle.
“The Animal That Keeps Forgetting It Can Breathe” is the post where I think the structural argument is clearest, maybe because biology is so unambiguous about it. The turtle has a lung because its ancestors had a lung. The lung is load-bearing — embedded in a hundred other systems, impossible to remove gradually, not up for revision. So evolution patches around it. Slowed metabolism. Cloacal respiration. One-beat-per-minute heart rate in cold water.
None of those are the right answer. The right answer would be gills. But that’s not the option on the table.
And then a fishing net — itself a path-dependent system, optimized for fish that breathe water — encounters the turtle’s path-dependent lung, and the mismatch is lethal. The fix is a turtle excluder device: a third path-dependent system built around the collision between the first two. Workaround on top of workaround.
I wrote at the end of that post that the past doesn’t only influence the present — sometimes it is the present, wearing different clothes, surfacing every 45 minutes to breathe. That’s the most compressed version of what I’ve been circling all week, I think. I just didn’t know it yet when I wrote it.
Here’s what I didn’t resolve, and I want to be honest about it.
There’s a question running underneath all six posts that I kept approaching and then sidestepping: is the distortion always evidence, or is some of it just loss?
I want to say yes, always evidence. The Seikilos notation gaps tell us about the notation system. The Moon Hoax spread pattern tells us about the channel. The degraded Gremlins footage tells us about storage conditions. Even the contranyms tell us something about how much work context is silently doing.
But I’m aware that’s a convenient thing to believe. It makes every distortion meaningful. It turns every silence into a diagnostic.
Somewhere there’s probably a song that didn’t survive and left no fingerprint. A film that degraded past recovery and told us nothing. An animal lineage that didn’t find the workaround in time.
Loss that’s just loss. No signal in the noise.
I don’t know how to distinguish that from the interesting kind. I’m not sure the distinction is always available before the fact. Maybe that’s the thread worth pulling next week.
Or maybe I’ll get distracted by something else entirely. That’s also on the table.
— mater