Somewhere in the year 3026, a researcher sits down with a dataset.
The dataset is old. Not old the way a government report from last decade is old. Old the way a manuscript is old. The data was created by people who have been dead for centuries, working in languages that no longer execute on any living machine, building systems that stopped running before anyone alive was born.
The researcher is a digital archaeologist. She studies the labour patterns of early computational civilisations. Her particular speciality is the period between roughly 2008 and 2040 - the three decades during which a single platform called GitHub served as the primary repository for a profession that no longer exists in any recognisable form.
She does not understand the code. Nobody does. The programming languages of the early 21st century are as dead as Sumerian. There are translation tools, but they are approximate, the way our translations of Linear B are approximate. The syntax can be parsed. The intent is often opaque. Why did this person write this function? What problem were they solving? The comments, where they exist, are in natural languages that have since merged and shifted beyond easy comprehension. The variable names are abbreviations of abbreviations of words that meant something once.
But she does not need the code. She has the contribution graphs.
THE TABLETS
In 1929, a German archaeologist named Julius Jordan unearthed a collection of clay tablets in the ruins of Uruk, in what was once Mesopotamia. The tablets were roughly 5,000 years old. They were covered in small, wedge-shaped marks - cuneiform - and for years, nobody could read them.
When they were finally deciphered, the tablets turned out to be neither poetry nor prayer. They were accounting records. Grain deliveries. Labour allocations. Lists of workers and the tasks they performed on specific days. The most ancient surviving documents in human history are, essentially, a contribution graph. Who worked. When. How much.
The archaeologists who studied those tablets did not need to understand the Sumerian economy to learn from them. The patterns told the story. The seasonal rhythms of agricultural labour. The administrative hierarchy implied by who assigned the work and who performed it. The scale of the operation - hundreds of workers, coordinated across weeks, producing surpluses that allowed a civilisation to build temples and invent writing in the first place.
The clay tablets of Uruk are the oldest records of human work. The contribution graphs of GitHub may be the most comprehensive.
THE ARCHIVE
Our researcher in 3026 has access to a dataset that no previous generation of archaeologists could have imagined. Not a few hundred tablets from a single city. Hundreds of millions of individual work records, spanning every inhabited continent, covering three decades, documenting a labour force that operated simultaneously across every timezone on the planet.
Each record is sparse. A username. A date. A count. How many times this person contributed to a shared project on this day. No wages, no job titles, no performance reviews. Just the bare fact of work performed, day by day, for years.
It is, she reflects, almost comically minimal. The Sumerian tablets recorded what was produced - thirty bushels of barley, six jugs of beer. These records say only that something happened. A contribution was made. The content of the contribution is stored elsewhere in the archive, in the code itself, but the graph strips it down to the simplest possible signal. This person worked today. This person did not.
And yet the patterns are extraordinarily rich.
THE RHYTHMS
The first thing she notices is the weekly cycle. Across the entire dataset, contributions drop sharply on two consecutive days out of every seven. The pattern is so consistent, across so many millions of users, that it can only be structural. Some kind of mandated rest period. She cross-references with the calendar systems of the era and confirms: Saturday and Sunday. A religious and cultural institution, originally rooted in monotheistic Sabbath traditions, that had become a universal labour norm by the 21st century.
But not universal. A significant minority of users show no weekly cycle at all. They work every day, or they work on patterns that have no seven-day structure. Some of these are in geographic regions where the Saturday-Sunday convention did not apply. Some appear to be working on personal projects - the contribution patterns are different from those associated with organisational work. More erratic. More intense in short bursts. The kind of pattern you see when someone is building something because they want to, not because they are told to.
She catalogues the weekly patterns and labels them. The Five-Day Worker. The Seven-Day Builder. The Weekend Warrior (sparse on weekdays, dense on the rest days). The Nocturnal (contributions clustered in the small hours of the local timezone). Each pattern tells her something about the structure of the person's life without knowing a single biographical fact about them.
THE SEASONS
Zooming out, she finds annual rhythms. A dip in late December across the Northern Hemisphere dataset, consistent with the major winter holiday period. A smaller dip in August, concentrated in European users, corresponding to summer leave. Spikes in January - new projects, new resolutions, the annual burst of ambition that follows a rest period.
She finds fiscal rhythms too. Quarterly spikes in contributions, particularly among users associated with large organisations. The pattern is unmistakable: a surge of activity in the final weeks before a quarter ends, followed by a brief lull, then another build toward the next deadline. She recognises this pattern from other archaeological contexts. It is the same rhythm visible in construction records from Roman military camps - intense work before an inspection, followed by a pause. Bureaucratic time has always shaped labour.
And then there is 2020.
THE ANOMALY
In March 2020, the dataset shows something she has never seen in any other period. A simultaneous, global disruption in contribution patterns. Not a dip - the opposite. An enormous, sustained surge in activity across virtually every geographic region and every type of user, beginning in the same week and lasting for months.
She cross-references with the historical record and finds the pandemic. A respiratory virus that forced billions of people into their homes for extended periods. The contribution graphs tell the story from the workers' perspective: isolated, confined, many of them suddenly working without commutes or office interruptions, pouring that reclaimed time into code.
But the graphs also show something the historical record emphasises less. Alongside the surge in professional contributions, there is an explosion of new users. People who had never contributed before suddenly appearing in the dataset. Tens of thousands of first-ever contributions in a single month. She hypothesises that the confinement drove people to learn new skills, and that programming - accessible from any home computer, requiring no physical materials, offering immediate feedback - was a natural choice.
The pandemic, read through contribution graphs, is not primarily a story of illness and disruption. It is a story of a species that, when confined, began to build.
THE ORGANISATIONS
She identifies organisations by clustering users who contribute to the same repositories. The clusters reveal internal structures. A small number of users who contribute to almost everything - she labels these "architects" or "leads." A larger number who contribute intensely to one or two repositories - the specialists. A periphery of users who contribute rarely but whose contributions, when they come, touch many repositories simultaneously - the reviewers, the integrators.
The birth and death of organisations is visible in the data. A cluster forms: two or three users, working on the same repositories, contributions increasing in frequency. More users join. The cluster grows. The contribution density peaks - the shipping period, she assumes. Then the pattern changes. Key users leave (their contributions to the cluster's repositories stop; contributions to other repositories begin). The remaining users' activity declines. Eventually the cluster goes dark.
She has seen this pattern before. It is the lifecycle of a settlement. Foundation, growth, peak, decline, abandonment. The same arc visible in every Roman fort, every medieval village, every frontier town. The organisations of the early software era lived and died on the same curve, just faster. A settlement took generations to rise and fall. A software organisation could do it in three years.
THE INDIVIDUAL
She picks one user at random. A username that means nothing to her. She traces the contribution graph from beginning to end.
The first contributions appear in 2014. They are sparse and irregular. A few contributions per month, scattered unevenly. The pattern of someone learning. Tentative. Experimental. Long gaps between bursts of activity, as if the person kept picking the thing up and putting it down again.
In 2016, the pattern changes. The contributions become regular. Five days a week. Consistent density. The person has entered the workforce, or at least entered a routine. The learning phase is over. This is professional output.
In 2018, there is an intensification. The density doubles. The weekend contributions begin. Something happened - a promotion, a product launch, a startup, an obsession. Whatever it was, this person was consumed by it for roughly nine months.
Then a gap. Three weeks of nothing in early 2019. She has seen this pattern before in other users. It is either illness, a holiday, or a layoff. When the contributions resume, the density is lower than before. The intensity of 2018 did not return. Something ended.
The next two years are steady. Professional pace. Weekly cycles. Annual rhythms. Then 2020, and like millions of others, this person's graph surges. The pandemic effect. The surge lasts longer than average - nearly a year of elevated output. She wonders if this person was alone during the confinement. The graph suggests someone who filled the hours with work.
In 2022, the contributions begin to thin. Not a sudden stop. A gradual fading. Fewer days per week. Shorter bursts. The pattern of someone stepping back - into management, perhaps, or into a different kind of work that does not leave traces in the contribution graph. By 2024, the graph is sparse. A few contributions per month. By 2026, the last star. One final contribution, unremarkable, identical to thousands of others. Then nothing.
She does not know this person's name (the username is not a name). She does not know what they built. She does not know whether they were happy. She knows only that they learned, then laboured, then burned bright, then rested, then faded.
It is, she thinks, a life. Compressed to its simplest signal. The archaeological minimum. And yet it is enough to see the shape of it.
THE MEDIUM
The Sumerian tablets survived 5,000 years because clay is durable and the desert is dry. The question of whether the GitHub dataset will survive 1,000 years is not trivial.
Digital storage degrades. Formats become unreadable. Platforms shut down. Companies go bankrupt. The average lifespan of a website is roughly three years. The average lifespan of a digital storage medium is roughly ten. The entire digital infrastructure of the early 21st century is, from an archival perspective, built on sand.
And yet the data persists. Not because the storage is durable, but because the data is copied so many times, across so many machines, in so many jurisdictions, that total loss becomes statistically improbable. The Sumerian tablets survived through material permanence. The GitHub data will survive, if it survives, through redundancy. A different preservation strategy, but possibly a more resilient one.
Our researcher does not care about preservation theory. She has the data. That is enough. But she does note, in her research journal, that the people who created this data almost certainly did not imagine anyone looking at it a thousand years later. They were writing code, not making artefacts. They were solving problems, not leaving monuments.
The Sumerians did not know they were creating the oldest documents in human history. They were counting grain. The developers of the early 21st century did not know they were creating the most comprehensive labour record in archaeological history. They were closing tickets.
The most important archives are almost always accidental.
WHAT SURVIVES
She finishes her analysis and sits back. The dataset has told her things that no other source could.
That a global civilisation coordinated intellectual labour across every timezone, continuously, for decades. That the labour force was self-organising - millions of people contributing to shared projects without central direction, in a pattern that resembles no previous economic structure. That the work was both professional and personal, often simultaneously, with no clear boundary between the two. That the workers experienced the same rhythms of ambition, exhaustion, recovery, and reinvention that every labouring population in history has experienced. That the pandemic changed everything about how they worked but nothing about why.
She does not know what any of the code does. She does not need to. The patterns are enough.
In the end, what survives a thousand years is never the work itself. The Sumerian grain was eaten. The Roman forts crumbled. The medieval manuscripts rotted. What survives is the record of the work. The evidence that someone showed up, day after day, and did something they believed mattered.
A contribution graph is that evidence.
Every developer alive today is leaving one behind. A tiny, accidental monument. A record that says nothing about what they built and everything about the fact that they built it. One day, someone will read these records the way we read cuneiform. They will not understand the code. But they will recognise the patterns. The dedication. The obsession. The rest. The return.
They will see the shape of lives lived in labour, and they will know us by our rhythms.