The Repository Abyss Report: Into the Dark

Nobody Could Name the Golden Orb, So They Took It

August 26, 2026 · Runtime 15:51 · Watch on YouTube

In 2023 an ROV found a smooth gold dome stuck to a rock 3,250 metres down in the Gulf of Alaska. Nobody watching the live feed could say what it was - and the control room recording of them failing to name it is public domain, timestamped, and as far as I can tell has never been transcribed. It took 965 days and a genome to answer.

Chapters

Transcript

A robot finds something gold 0:00

Three thousand two hundred and fifty metres down, on a seamount in the Gulf of Alaska, a robot found something gold. It was stuck to a rock. It was about the size of a grapefruit. It had a torn hole in one side. And the people watching it happen, live, could not tell you what it was.

The room runs out of ideas 0:20

They were not amateurs. There were taxonomists in the room and more taxonomists in the chat, and between them they had the deep sea of four oceans in their heads. It took them about a minute to run out of ideas. This is the story of what that object turned out to be. It is also a story about a label, because when they wrote the thing down, they wrote it down wrong, and the wrong name is still sitting in the catalogue today.

EX2306, dive 07, Gulf of Alaska 0:47

The dive is E X two three zero six, dive seven, the thirtieth of August twenty twenty-three. The expedition was called Seascape Alaska five. The ship is N O A A Ship Okeanos Explorer, the vehicle is remotely operated vehicle Deep Discoverer, and the target was a seamount in the Gulf of Alaska, fifty-five degrees north, a hundred and forty-one degrees west, about five hundred kilometres from the nearest coast.

Two hours of nothing 1:16

Getting there takes about two hours, and almost nothing happens for any of them. The vehicle goes over the side, the last daylight is gone by two hundred metres, and after that the only light in the world is the one bolted to the front of the machine. One thousand metres. Two thousand. Three thousand. At three thousand two hundred and thirty-four metres it stops going down and starts going forward.

Basalt, glass sponges, black coral 1:43

What it finds is a slope of broken basalt with a thin dusting of pale sediment in the cracks. The white shapes standing on stalks are glass sponges, which build a skeleton out of silica and then stand on it like tulips. There are black corals, sea stars, urchins. For the archive this is a good dive but an ordinary one. The site has a name in the record. It is called Deep Discoverer Dome.

19:47 UTC: the first close look 2:10

Nineteen forty-seven Coordinated Universal Time. The camera is panning across the rock when someone asks for a closer look at a thing they had already been calling, casually, the yellow hat. Everything you are about to hear is the actual control room, from the dive recording. It is public domain, like everything else this expedition produced, and as far as I can tell nobody has ever transcribed it.

"Like it was a face" 2:43

Seventeen seconds later, the same voice tries again. And gets closer to the truth than anybody else will manage for the next two and a half years.

The object, at full resolution 3:00

Here is what they are all looking at, at full resolution, straight from the archive. A smooth dome, slightly over ten centimetres across, the colour of old brass under the vehicle lights. A tear near the base, big enough to see into. Inside, the same gold, but rougher. And it is attached to the rock, not lying on it.

Ten centimetres, measured by laser 3:23

Scale first, because scale is the one thing a screen destroys. The vehicle carries two lasers exactly ten centimetres apart, and that pair of red dots is the ruler in every frame it shoots. The object is a little wider than that gap. Ten centimetres is small. It is also enormous, for something that nobody in the room can assign to a phylum.

The expert was at lunch 3:47

There is a person the room would normally turn to for exactly this kind of object, and at that moment they had stepped out for lunch. Somebody says so, on the broadcast, with the timing you would expect. A few minutes later the expert comes back into the room and is asked, straight out, what everyone is looking at.

Sponge, coral, egg case 4:19

So the guessing starts, and it is worth listening to how it goes, because this is what expertise actually sounds like when it is working without a net. First a dead sponge attachment: the stump left where a sponge used to be. Then coral. Then an egg case. The science lead sums up the state of play.

The hole, and what it implies 4:52

The hole is doing most of the work in that argument. It is not a puncture and it is not symmetrical. It looks torn. And a torn hole in a smooth shell suggests a door, which suggests something that needed one.

"It's got us all stumped" 5:15

By now the vehicle has been parked in front of this thing for several minutes and the chat has stopped producing candidates. Somebody puts the position plainly, and then immediately does the sensible thing, which is to start thinking about the microscope back on the ship.

The rule that decided everything 5:39

And then comes the sentence that decides everything that follows. It is not a discovery and it is not a hypothesis. It is a rule of thumb, said out loud, on an ordinary afternoon, by somebody doing their job.

What a live broadcast cannot settle 6:10

Hold on to how unglamorous that is. Telepresence means a dive can be watched live by anybody, anywhere, and this one was: you can hear the host on the recording turning to the camera and explaining the suction sampler to the people tuned in. But nothing anybody saw on a screen that day was ever going to settle it. The only thing that could settle it was the physical object, in a jar, on a ship, going home.

The suction sampler 6:36

Which raises the practical problem. How do you pick something up when you do not know what it is made of? The manipulator arm can close on it, but if it is soft the arm will destroy it. They test that first, gently, and the answer comes back soft, and probably delicate. So they use the other tool. The suction sampler is a hose that pumps water through itself to create a draw, at a strength the pilot can dial up and down, and it empties into a rack of jars on the front of the vehicle.

Seven minutes, and not tidy 7:08

The collection takes seven minutes and it is not tidy. The thing does not want to come off. They raise the suction, they scrape, they shake the hose, and pieces break away and go up the tube one at a time. And the commentary keeps going through all of it, because on this ship the commentary is part of the job.

What do you write in the log? 7:36

Then the last piece goes in, and the science lead asks the question this entire episode is really about. Every sample gets an entry. Every entry needs a name. What do you write, when the honest answer is that nobody knows?

What the archive actually recorded 8:07

Biota, unidentified. That is the right answer. It is a good answer. And it is not what ended up in the catalogue. Pull the archive record for those same video segments today, and the keywords attached to them are these. E G G. Case, colon, egg. Gastropods. The guess that got made out loud, by people who said in the same breath that they did not know, is the guess that got filed.

Why a wrong keyword loses an object 8:36

That matters more than it sounds. This archive is thousands of hours long, and nobody watches it. People search it. What they search is the keyword field, which makes the keyword field the only thing standing between an object and permanent invisibility. A wrong label does not just sit there being wrong. It answers the question that would have found the object, and it answers it with somebody else.

The microscope on the ship 9:02

On the recording you can hear them plan the next step, and it is the obvious one. If this were a sponge, you would expect spicules: small structural elements of silica, which is what a glass sponge is built out of. Somebody suggests dissolving a fragment in household bleach, which strips soft tissue and leaves any skeleton behind. The examination that was eventually published does not report a sponge skeleton. It reports material with no recognisable animal anatomy at all.

Why DNA barcoding failed 9:34

So it went to the National Museum of Natural History in Washington, to the laboratory that N O A A Fisheries runs inside that building, and there they ran the standard test. D N A barcoding reads one short, agreed-upon stretch of genetic code and compares it against a reference library. It is fast, it is cheap, and it identifies almost anything. It failed. The likely reason is the most ordinary thing in the deep sea. The surface of the object was covered in microscopic life of its own, and the test kept reading the neighbours.

Sequence everything, sort it later 10:10

The escalation is expensive, and it is most of why this took years. Instead of one stretch of code, sequence everything in the sample, and then sort the animal out of the mixture computationally. That produced complete mitochondrial genomes, and a set of ultra conserved elements: stretches of code that barely change between species, and which are therefore very good at showing where something sits on the tree of life.

Spirocysts, and what they rule out 10:37

In parallel, somebody looked at the structure again, harder. The material was fibrous and layered, and packed into those layers were stinging capsules. That narrows it enormously on its own, because stinging capsules mean cnidarian: corals, jellyfish, anemones. A specialist at the systematics laboratory identified which type they were. They were spirocysts, and spirocysts occur in exactly one group, the hexacorals.

Relicanthus daphneae 11:07

Both lines of evidence land in the same place. The mitochondrial genome came back almost identical to a reference genome for a species called Relicanthus daphneae. It is a sea anemone, and calling it that undersells it. Tentacles over two metres long. A column a metre across. The longest spirocysts known in any hexacoral. It was described in two thousand and six, from the flanks of hydrothermal vents on the East Pacific Rise, and it has been taxonomically homeless for most of the time since.

It is the base, left behind 11:42

Now put those together, because this is the payoff. The golden orb is not an animal. It is not an egg, it is not a sponge, and it is not a young anything. It is the base. The part of the anemone that gripped the rock, left behind on the rock as a cuticle, with the animal no longer on top of it. N O A A is careful about the next part, and so is this. The anemone may have died. It may have moved. Nobody knows which, and nobody may ever know.

Three specimens, three oceans 12:15

One specimen is an anecdote, so the paper does not stop at one. A similar object had been collected in twenty twenty-one, from deep equatorial water, on an expedition of the research vessel Falkor. Same stinging capsules. Its mitochondrial genome landed in the same place. And they found the cuticle still attached, underneath a Relicanthus specimen from the Southern Ocean, which is as close as anyone has come to catching this happening.

It was already in the archive 12:45

Then there is the part that should be uncomfortable for anybody who works with archives, and I am including myself in that. In twenty sixteen, seven years before any of this, the same agency, with the same ship, filmed a living Relicanthus daphneae in the Mariana Islands region, with that golden base visible underneath the animal. N O A A published that frame in twenty twenty-six, next to the answer, as the illustration.

Eighteen segments in the whole catalogue 13:14

So I ran the obvious search. Ask the N O A A video catalogue for every segment tagged with the name of this animal, and you get eighteen. Nine moments, each filmed by two cameras, across four expeditions. The Musicians Seamounts, north of Hawaii. The Gulf of Mexico. Two more in the equatorial Pacific. The deepest of them is two thousand four hundred and forty-one metres, which is eight hundred metres shallower than the rock this thing was sitting on. None of them is in Alaska. And none of them is the Mariana footage.

USNM IZ 1699903 13:51

The object itself has an address now, which is the part of science that nobody makes films about. It is in the invertebrate zoology collection at the Smithsonian, catalogued as U S N M, I Z, one six nine nine nine zero three. That number is the whole difference between a story and a specimen. Anybody who thinks this identification is wrong can go and ask for the jar.

965 days 14:17

Here is the timeline in full. Collected on the thirtieth of August, twenty twenty-three. The preprint went up on the twenty-first of April, twenty twenty-six. Nine hundred and sixty-five days. And the director of the laboratory that did the work says, in the announcement, that he expected the routine process to clear it up, and that it did not. Two and a half years, for an object that is, in the end, a footprint.

Nobody bluffed 14:46

The internet wanted an egg. It got a base plate. But listen to what actually happened on that recording. Nobody bluffed. Nobody produced a confident identification they did not have. The room said it did not know, said so on a live broadcast, and then picked the thing up specifically because it did not know. That decision is the entire reason there is an answer to tell you. The only thing that went wrong is that the guess, and not the uncertainty, is what got written down.

Next: the squid with elbows 15:18

Next time, an animal that has been filmed fewer than thirty times in the history of the world. It has arms that bend at what look like elbows, and it hangs in open water four thousand seven hundred metres down, with filaments trailing metres below it. No intact adult has ever been collected: everything in a jar anywhere in the world is damaged, or a juvenile, or both. Which means almost everything anybody knows about it comes from footage, and that is a very different kind of evidence.

Description and sources

On 30 August 2023, dive 07 of NOAA's Seascape Alaska 5 expedition put remotely operated vehicle Deep Discoverer on a seamount in the Gulf of Alaska. At 3,250 metres it found a golden, dome-shaped object slightly over ten centimetres across, with a torn hole near its base, attached to basalt. The scientists on the ship and in the live chat worked through dead sponge attachment, coral, and egg case, and settled on none of them. They collected it with a suction sampler, and one of them said the sentence that decides the rest of the story: it is always worth a collection when our collective knowledge cannot identify it. In April 2026, after DNA barcoding failed and whole-genome sequencing did not, the answer arrived. The golden orb is not an organism. It is a remnant cuticle - the base that a giant deep-sea anemone, Relicanthus daphneae, leaves on the rock. This episode uses the dive's own audio for the moment nobody could name it, and it also follows what happened to the label: the guess that was made out loud is the guess that ended up in the archive's keyword field, which is how anybody ever finds anything in an archive nobody watches.

PRINT-READY, FROM THIS CHANNEL

Read This Dive, Vol. 01 — the method, and five dives read with it

https://therepository.gumroad.com/l/read-this-dive

PRIMARY SOURCES

Auscavitch, Reft, Collens, Mah, Best, Benedict, Rodriguez, Daly & Collins. The Curious Case of the Golden Orb - Relict of Relicanthus daphneae (Cnidaria, Anthozoa, Hexacorallia), a deep sea anemone. bioRxiv, posted 21 April 2026, CC0. https://doi.org/10.64898/2026.04.17.719276
NOAA. Scientists reveal identity of mysterious 'golden orb' collected during NOAA expedition, 22 April 2026. Specimen accessioned as USNM_IZ_1699903, Smithsonian National Museum of Natural History. https://www.noaa.gov/news/scientists-reveal-identity-of-mysterious-golden-orb-collected-during-noaa-expedition
NOAA Ocean Exploration. The 'golden orb', Seascape Alaska 5, 2023. https://oceanexplorer.noaa.gov/news/golden-orb/
Dive record: NCEI granule metadata for EX2306 DIVE07, 30 August 2023, 55.0150 N 140.8345 W, 3,234 to 3,249 m, site Deep Discoverer Dome. Video and frame grabs public domain, NOAA Ship Okeanos Explorer and ROV Deep Discoverer.
Relicanthus daphneae: described as Boloceroides daphneae by Daly, 2006, from the East Pacific Rise; tentacles over two metres, column a metre across, the longest spirocysts known in any hexacoral. See the mitogenomic treatment in Scientific Reports, 2019. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6890759/
Control-room audio transcribed for this episode from the published dive segments with whisper.cpp running locally. Quotes used on screen were checked by ear against the recording.

Abyss Report reads the deep ocean through the record that resolved it - instrument, date, position and the primary sources listed above. The evidence is the recording, not decoration.

About the material. Every moving shot in this episode is NOAA Ocean Exploration video from EX2306 dive 07, 30 August 2023, in the public domain, credited on screen with dive, date, depth and position read from the granule record. The nine passages of control-room audio are the dive's own soundtrack, played unedited, in the order they happened. Every diagram is drawn in code from the numbers in the sources above. No image in this video was generated.

#GoldenOrb #NOAA #DeepSea #OceanExploration #Alaska