Inheritance: The Arctic Code Vault
On an island in the Arctic Ocean, there is a mountain with a door in it. The island is called Spitzbergen. It belongs to Norway, although it's closer to the North Pole than to Oslo. For part of the year, the sun never rises, and for another part it never sets. There are glaciers, reindeer, and polar bears. Beyond the little town of Longyearbyen, people are advised not to travel without protection against them. For most of the 20th century, people came here to dig for coal. They tunneled into the mountains, following black seams through the rock. But in mine number three, the seams were so thin that modern machines couldn't fit. So the miners kept working with equipment that was 50 years old. Then in 1996, mine number three closed. The machinery stopped and the miners left. Years later, though, people began bringing things back into the mountain. Some of them brought film, reels of it, sealed in boxes and carried deep underground. Some of it holds images of paintings. Some of it is the founding documents of nations. And in July of 2020, a shipment arrived containing something rather more difficult to explain. Computer code. Millions of software projects written by people all over the world had been transferred onto 186 reels of film. Each reel was about a kilometer long. Unrolled end to end, they would stretch more than 100 miles. The people who brought the code to the mountain hoped that it would be protected for the next 1,000 years. They left it there for someone. I'm Daina Bouquin and this is Found in the Machine. Most of the software you use every day was not written by one person. It was written by many people at once, often strangers, often in different countries, adding to the same work and fixing each other's mistakes. A lot of that work happens out in the open where anyone can read it, use it, or make it better. It's called open source, and it runs underneath almost everything. Your phone, websites you visit, the machines behind nearly everything you do online. Much of it lives on a website called GitHub. GitHub is a kind of shared workshop for code. People keep their projects there along with a record of all of the changes made to them. A programmer on one side of the world can propose a fix to something written on the other side, and the people who run the project can look it over and fold it in. When GitHub started planning its vault, tens of millions of people were using it this way. So in November 2019, GitHub announced that it would preserve millions of public software projects inside a mountain in the Arctic. It gathered advisors, historians, archivists, linguists, archaeologists, people who think for a living about what lasts. And then they had to decide what to keep. The obvious answer is to keep the important things, the famous projects, the software that everything else depends on, and that's usually how saving things work. The things we remember an era by are mostly the things people decided were worth remembering. One of the advisors, though, a historian and author named Ada Palmer, pointed out a problem with that. She said, we have plenty of letters from wealthy Renaissance aristocrats. What historians would love to have are ordinary people's shopping lists, and almost none of those survive. Shopping lists are made to be thrown away. So GitHub decided to try to keep the shopping lists on purpose. The ordinary things. On February 2nd, 2020, it took a snapshot. Every public project on the site that met a few simple rules was copied at once, frozen in that moment. The rules said nothing about quality. If someone had worked on a project in the past 80 days, it went. If at least one person had marked it as worth keeping and it had been touched in the past year, it went. If a project was popular enough it went no matter how long it had been. The famous software is in there. But so are class assignments. Things that were started and abandoned. Weird ideas that might someday turn into something useful. There was code written by students building robots trying to pass information from a camera to their machines. Not everything made it. Private projects weren't included, and plenty of forgotten projects didn't meet the rules. But millions did. The code captured on that day in February went to Drammen, a city outside Oslo, to a company called Piql, and into a machine. The machine is called a piql writer. Inside it, there is a green light and a chip covered in millions of tiny mirrors, each one much smaller than a grain of sand. The mirrors work something like the slats of a window blind. Each one can tilt. Tilted one way, it sends a speck of green light down onto the film. Tilted the other way, it sends the light off into the dark. Arrangement after arrangement builds up a frame, a pattern of light and dark. Then the mirrors move and the film moves 20 frames a second. At that pace, one reel takes just under an hour. All 186 reels make up about a week of light. The film is old technology, just a strip of polyester, 35 millimeters wide, the same size as movie film. It's coated with gelatin holding tiny crystals of silver salt. Where the green light lands, the crystals change, but it's nothing you could see yet. The film has to go through chemicals the way a photograph does. The touched crystals then can become grains of metallic silver and the rest wash away. That's what the code becomes. Grains of silver sitting on plastic. Hold a strip up to a window and it looks gray. Under a magnifier, the gray comes apart. Each frame has millions of tiny black and white squares packed edge to edge. Someone with a camera and the right instructions can convert those squares back into files. And once the writing is done, the machine can be switched off. The film doesn't need power to remember. It can sit in a box while the computers that made it go obsolete. And the computers that replaced those go obsolete too. But someone would have to know what they were holding. So the makers gave every reel the same beginning. First, a length of empty film. Then a single frame with nothing on it but a solid black rectangle in one corner. A mark to measure from. After that comes pages a person can read, actual letters, photographed onto the same film as the code. This is a guide, instructions that can travel with the thing that they explain. The man in charge of the program, Jon Evans, was a software engineer and he was also a novelist. Now he had to help write for a reader who might have never seen a computer. So the guide starts small. All you need, it says, is a source of light and some way to magnify what's in front of you. The guide tries to imagine who is reading. Maybe they have computers far beyond ours, but maybe they have none at all. To explain what a file is, the guide asks the reader to think of data as sand and a file as a bag that holds it. By the time anyone reads that, they will already be holding the sand, the grains of silver on a strip of plastic in their hands. At one point, the guide stops and acknowledges how hard all of this might be. It says, quote, if this complexity is difficult and costly for you, we apologize. The guide went on to the film in five languages: English, Arabic, Chinese, Hindi, and Spanish. And the first file on every reel holds the Universal Declaration of Human Rights in more than 500 languages. If a reader knows one of those languages, it might help them find their way to the others. No one knows which language, if any, the reader will understand. On July 8th, 2020, the film reels were taken to the old coal mine, to a chamber deep in the frozen mountain. Outside, it was the season when the sun doesn't set. And later that month, people began noticing something new on their GitHub profiles. A small badge that said their work had gone to the Arctic. It was a thank you and a way for GitHub to draw attention to the project. On a forum for student robotics teams, people were comparing notes about what had made it into the archive. One of them posted under the name bobby sq. They had written a little program that passed what a camera saw along to a robot, and it had made it into the vault. And it didn't have a README. That little document that usually explains what a piece of software is and how to use it. The student addressed whoever might find it, saying, quote, I would also like to apologize in advance to the aliens, post-society humans, or evolved dolphins that find my code. Sorry, there's no readme.md yet. Maybe I'll head up there and write it down for you someday. In 2022, GitHub went back. They brought a reel of books and documents a person could read by eye about computers and how they're made. They brought a full-text snapshot of Wikipedia in all five of the archives languages. They brought a complete stack overflow data dump. And they brought a steel vault for the film. It was about 3,000 pounds and etched with art. The earth on the front and the Milky Way on the back. Alexander Rose of the Long Now Foundation had told them, quote, if you don't make it beautiful, it's for sure doomed. For decades, mine number three gave up what was inside of it. People came in on their shifts and carried the coal out. Now it holds things. Boxes to leave alone and a door to close behind you. No one has set a date for the reader to arrive. A thousand years is how far the makers hoped the film could reach, but someone could need it sooner. Someone could open a box just to see and put it back. And if they do, they will have to find the beginning. A length of empty film. A black rectangle in one corner. They will need some light and a way to magnify what comes next. "You may be reading this one year from now or one thousand. But either way, we hope its contents and perhaps the very concept of open source are useful to you." I'm Daina Bouquin and this is Found in the Machine. If you enjoy my stories, please share them or leave a rating or a review wherever you listen. They really do help people find the show and they mean a lot to me. And if you'd like more content from me, you can join the show's Patreon. Patreon members get access to Found in the Margins, my mini bonus episode series that alternates with the main feed. The most recent Margins episode was called Rope Mother. You can find the Patreon and more ways to support the show by going to found in the machine . com / support. Thanks for listening, guys.
{show notes}
Deep inside an abandoned coal mine in the Arctic, there are 186 reels of photographic film holding millions of software projects. Some of that code runs the internet. Some of it is unfinished homework.
In 2020, GitHub put it all there, hoping it would survive for a thousand years. But how do you preserve software for someone who may never have seen a computer? And why go to such extraordinary lengths to save it in the first place?
In this episode
- The Arctic Code Vault: How an abandoned coal mine on Spitsbergen became a home for millions of software projects
- Piql's photographic film: How microscopic mirrors, green light, and grains of silver can preserve digital information for ages
- A guide for an unknown reader: The instructions, translations, and apologies left for whoever might eventually find the archive
- An unfinished project: A student's unexpected contribution to the vault, and a message for the evolved dolphins who might someday read it
Episode Music:
- Jay Opie / Nihilore, CC BY 4.0
Additional Reading
bobbysq. (2020, July 19). [Reply to the discussion “Robotics repositories in the Arctic Code Vault”]. Chief Delphi. https://www.chiefdelphi.com/t/robotics-repositories-in-the-arctic-code-vault/386554
Evans, J. (2022, September 20). “If you don’t make it beautiful, it’s for sure doomed”: Putting the Vault in GitHub’s Arctic Code Vault. The GitHub Blog. https://github.blog/open-source/maintainers/if-you-dont-make-it-beautiful-its-for-sure-doomed-putting-the-vault-in-githubs-arctic-code-vault/
GitHub. (2020, November 5). A guide to the GitHub Code Vault. https://github.com/github/archive-program/blob/master/GUIDE.md
Metcalf, J. (2020, February 3). The Arctic Code Vault starts production and your open source projects are being archived. The GitHub Blog. https://github.blog/open-source/maintainers/the-arctic-code-vault-starts-production-and-your-open-source-projects-are-being-archived/
Metcalf, J. (2020, July 16). GitHub Archive Program: The journey of the world’s open source code to the Arctic. The GitHub Blog. https://github.blog/news-insights/company-news/github-archive-program-the-journey-of-the-worlds-open-source-code-to-the-arctic/
Piql. (2026, October 19). piqlFilm. https://piql.com/technology/film/
Piql. (2026, October 19). piqlWriter. https://piql.com/technology/writer/
Sabliński, J., & Trujillo, A. (2021). Piql. Long-term preservation technology study. Archeion, 122, 13–32. https://doi.org/10.4467/26581264ARC.21.011.14491
💚
This show is independently produced. You can support the show by joining the show's Patreon. Patreon members get access to "Found in the Margins" bonus episodes and help make the show possible. You can also donate via Buy Me a Coffee.
If you enjoyed this episode, please consider leaving a rating and review on Apple Podcasts or Spotify. And don't forget to sign up to receive Notes from the Machine with each episode.