In 1970, at the Carnegie Institution's Department of Terrestrial Magnetism in Washington, an astronomer named Vera Rubin sat down with a set of spectra from the Andromeda galaxy and found a problem she could not make go away.
Her colleague Kent Ford had built a spectrograph sensitive enough to measure the speed of stars in the galaxy's outer regions, far from the bright central bulge. The physics of what they should find was two centuries old. Stars orbit a galaxy's centre the way planets orbit the sun: the further out you go, the slower the orbit. Mercury circles the sun at forty-seven kilometres a second. Neptune manages five. The outer stars of Andromeda, sitting far from most of the visible mass, should have been crawling.
They were not. They were moving as fast as the stars near the centre. Plotted on a graph, the rotation curve should have fallen away like the tail of a comet. Instead it ran flat, all the way out to the edge of the light and beyond.
There were two possibilities. Either Newton was wrong, or Andromeda contained five or six times more matter than anyone could see.
THE MASS THAT DOES NOT SHINE
Rubin was not the first to notice. In 1933 a Swiss astronomer at Caltech named Fritz Zwicky had measured the speeds of galaxies in the Coma cluster and found them moving so fast that the cluster should have flown apart long ago. To hold together, it needed far more mass than its galaxies contained. Zwicky called the missing quantity dunkle Materie, dark matter, and published in a German-language journal that almost nobody read. He was famously abrasive. Colleagues found it easy to ignore him. The idea sat untouched for nearly forty years.
Rubin's curves were harder to ignore. Through the 1970s she and Ford measured galaxy after galaxy and found the same flat line in every one. By 1980 the evidence was overwhelming. Every spiral galaxy in the sky was embedded in a vast halo of something that had mass, that exerted gravity, and that gave off no light at all.
The current estimate is that roughly eighty-five percent of the matter in the universe is dark. Everything you can see through any telescope ever built, every star, every nebula, every planet, every galaxy, is the remaining fifteen. Astronomers have never detected a particle of dark matter directly. They know it exists because of what it does to the things they can see.
Rubin died in 2016. She never received the Nobel Prize, which is its own kind of invisible mass.
THE PRIVATE REPOSITORY
A GitHub contribution graph shows public work. By default, commits to private repositories do not appear on it at all. A developer can opt in to have them counted, in which case they show up as anonymised squares: a number with no repository name, no commit message, no detail of any kind. GitHub calls these private contributions. We have written before about what the graph counts and what it hides. Private work is the largest thing it hides.
Most professional software is private. The banking system, the logistics platform, the defence contract, the internal tooling at a company with twelve thousand engineers, the stealth startup that has not launched, the client work under NDA. A developer can spend the most intense five years of their career on code that will never appear on their public profile in any form. Their graph, viewed by a recruiter or a stranger, looks like a long sabbatical.
The public graph has a rotation curve problem. The visible stars do not account for the mass.
READING THE CURVE
Rubin's insight was that you do not need to see the dark matter to measure it. You look at the visible stars and ask what would have to be true for them to move the way they move.
The same reading works on a graph. Look at a developer's public contributions through a period when their private work got heavy. The side projects thin out first. Then the weekend commits stop. Then the open-source flares, the pull request to a library they use, the fix to someone else's documentation, go dark. The public sky dims in a particular way, and the way it dims tells you something. We have written about the dark spaces on a graph before. This one has a cause you can name.
That bending is the gravity of work you cannot see. A public graph that goes quiet after dense early years usually belongs to someone whose mass has moved somewhere the instrument cannot point.
There are other signatures. A developer whose public contributions are all reviews and no commits is often someone whose commits go elsewhere. A public graph that switches on every August and December, and only then, belongs to someone who codes in public only on holiday, because the rest of the year the code belongs to an employer. The private work never reaches the graph. The hole it leaves does.
WHAT THE ATLAS CAN DRAW
Codex Stellarum draws what the GitHub API returns. If private contributions are switched off, the atlas shows only the public light. If they are switched on, the anonymised counts join the sky as stars like any other, unnamed, which on a star chart is how most stars are anyway.
We could pretend otherwise. We could estimate, extrapolate, fill the halo with invented stars so the sky looks fuller. We do not, for the same reason the stats panel on every atlas shows fewer numbers than it could: a number we cannot defend does not belong on the page.
But a sky drawn from visible light alone is still a true instrument. Rubin's rotation curve was drawn from visible stars and it measured the invisible ones. An atlas of your public work does the same. Anyone who knows how to read it will see the dense early years, the point where the public light dimmed, the particular shape of the dimming, and will understand that the mass did not disappear. It went dark.
THE HALO
Every developer carries a halo of work that does not shine. The years at the company whose name is on the CV and whose code is in a vault. The product that reached a million people and was never open. The system that runs a hospital, an airline, a payment rail, and that no public profile will ever mention.
Zwicky's cluster held together because of mass nobody could see. So does a career.
Most of what you have built is invisible. The sky still bends around it.
See the visible part. Enter your GitHub username at codexstellarum.com and preview your atlas - free to try, prints from £25.