The fire at a state data center in South Korea in September 2025 proves human fragility in the face of digital storage. Between risky sovereignty and secure dependency, what price are we willing to accept for our digital heritage?
On September 26, 2025, 750,000 South Korean civil servants lost all their work files following a fire in a government data center. Nearly 900 terabytes of state data vanished, with no possibility of recovery. This catastrophe resonates with the personal experience we have all lived through: the loss of a Word document we worked on at length, a thesis, sometimes an entire dissertation. Digital technology makes us experience this strange duality, which resonates with these words by Paul Valéry: “We civilizations now know that we are mortal”, and our data is equally so.
Digital data possesses this paradoxical characteristic of being infinitely reproducible while remaining infinitely fragile. A perfect copy can be created in a fraction of a second, identical to the original down to the last bit. Yet, without preventive duplication, its disappearance is absolutely irremediable. This fragility gradually fades thanks to cloud services (Google Docs or Microsoft 365 for example) which orchestrate automatic backup for us across multiple, geographically distributed servers. So, we lose fewer documents than before. But the risks are no less great, they are of a different nature. Let’s take stock.
I observe that we have collectively accepted a transfer of responsibility. Our most personal documents no longer reside in our carefully organized folders on our bedroom shelves, nor in the physical archives of our institutions. Now, they inhabit an abstract space, managed by private companies to which we pay a subscription, the freemium model offering us just enough free space to get us used to it before making us pay.
This massive migration to the cloud, initiated about fifteen years ago, marks an anthropological rupture that Sherry Turkle analyzes in Reclaiming Conversation: The Power of Talk in a Digital Age (2015): we become “connected but alone”, dependent on infrastructures we no longer control. We no longer manage our data ourselves, we have abdicated this fundamental responsibility. Physical media, hard drives, USB keys, still represent a form of persistent mass storage (even if obsolete over time), but how many of us still really use them?
This progressive abandonment of our digital sovereignty produces perverse effects. To save on subscriptions, many regularly delete their old emails for example, sacrificing their digital heritage on the altar of immediate economy. This practice appears to me as a form of voluntary amnesia, a renunciation of our own history. As Kenneth Goldsmith writes in Uncreative Writing (2006), our contemporary writing now encompasses data, texts, sounds, images and videos; it is our entire heritage itself that is thus weakened, because it is now almost entirely digital.
And furthermore, questions of longevity remain unanswered. The only sure solution is to manually copy one’s media in duplicate every 5 years onto new media, which represents a new organization. For will Google, Microsoft, Amazon still exist in fifty years? Let’s take an example, Megaupload, brutally shut down by the FBI in 2012, clearly reminds us of the precariousness of these services: millions of legitimate users, businesses as well as individuals, lost their data overnight, without any recourse. The fire at an OVH building in Strasbourg in 2021, which instantly and irremediably took offline more than 400,000 websites (I detail this in the article IT Responsibility and Backup) is also chilling. The only ones who could recover their websites were those who had regularly backed them up themselves.
The geographical location of our data also raises major legal issues. State documents stored under foreign jurisdiction potentially violate national sovereignty laws and principles. And if an authoritarian drift occurred in the country hosting our data, blackmail would become formidably effective, immediately and totally.
I advocate for taking back control of our digital sovereignty, being aware of the responsibilities it implies. This reconquest requires the emergence of new professions, new skills, new vigilance. Regular data replication, daily, weekly, monthly or even annual depending on their criticality, in sites at least 900 kilometers apart seems essential to me. This distance is not arbitrary: it corresponds to the maximum radius of major natural or industrial disasters, notably nuclear.
The required investment represents the price of our freedom, our independence and the future existence of our civilization. Jacques Derrida warned in Archive Fever (1995): “There is no political power without control of the archive, if not of memory.” Control of our digital archives conditions our ability to write our history, to transmit it, to inscribe our very existence in time. The West built its domination on its written culture, its meticulous archives, its ability to document and preserve. Faced with extremely rich but vulnerable African oral cultures, colonization substituted a dominant written narrative, erasing with unspeakable violence centuries of traditions. It was like the OVH fire, which erased the very trace of immense civilizations.
The Korean case unfortunately illustrates the failure of poorly assumed sovereignty. The virtuous will to preserve national data collided with disastrous cost-cutting: deemed too voluminous (860 terabytes), the data was not duplicated. A simple fire was enough to lastingly destabilize the administration. This negligence reveals a fundamental misunderstanding: sovereignty without taking responsibility becomes a greater vulnerability than dependency.
Our relationship to time and transmission is at stake in these apparently innocuous technical choices. Without structured and sovereign archives, we tip toward what Walter Ong calls a “secondary orality,” an immediate, ephemeral communication without concern for the trace. Exchanges proliferate but disappear, creating a society of perpetual instant, amnesiac by construction.
This question runs through all levels of our social organization. Already at the family level, transmission is deteriorating. Shoe boxes filled with yellowed photographs in attics have given way to cloud accounts likely to disappear with their owners. Our grandchildren will probably find no trace of our digital lives, erased by service changes, increasing subscription costs, or simply the planned obsolescence of formats.
Civilizations, organizations and individuals who know how to preserve their digital traces will tomorrow impose their narrative on the world. Those who have neglected this responsibility will find themselves, like African oral cultures facing colonial writing, dispossessed of their own history. The battle for digital memory determines the power relations in our future and future anthropology.
The choice between sovereignty and security is the sign of an intrinsic tension in our digital age. AWS, Azure, Google Cloud offer remarkable technical security, multi-site replication, high availability, specialized expertise. But this security comes with a structural dependency whose long-term implications we poorly measure. In twenty, fifty, a hundred years, who will control our archives?
The alternative exists: build robust sovereign infrastructures, invest in training and skills, accept the costs of redundancy. The Korean disaster teaches us that the half-measure, proclaimed sovereignty without adequate means, constitutes the worst of options. We must choose: either delegate consciously with all the risks that entails, or fully assume our heritage responsibility.
I plead for the second path, despite its constraints. Because preserving our digital autonomy means preserving our ability to tell our own story, to transmit it to future generations, to exist as subjects and not as objects. The apparently laborious side of backup management is actually an existential issue: remaining masters of our personal collective destiny, in our new universe, in which data and power inevitably merge.
In the 21st century, the vast majority of human production is made with digital tools and circulates in digital form: writing, photography, sound, video, multimedia. Our memories, our works, our social and institutional archives have become data, almost always stored somewhere other than with us, by companies to which we have entrusted their keeping.
I use the word “heritage” for lack of a better one, aware of what it carries of patrimony and transmission through the fathers. What is at stake here is something broader and more alive: access to human production, past and present — cultural, artistic, social, intimate. Heritage has a cultural, political, economic and historical value; without it, societies have no history.
The digital world arrived very fast, and we have not had the time to take its measure. We believe it to be immaterial, when in fact it rests on physical supports — fragile, located somewhere, under a given jurisdiction. Preserving our data is not merely a technical matter: it is a new responsibility, one we are only beginning to grasp. And it touches our very identity, for what we entrust to infrastructures we do not understand — subject to foreign laws, or exposed to failure and political pressure — is the memory of who we are.
This is where heritage and sovereignty meet: it is by taking care of our heritage that we become sovereign, and we can be sovereign only on the condition that we take care of our heritage. One does not go without the other. This plays out at every level—from the national level and the government’s infrastructure decisions, to organizations and local communities, all the way down to the more everyday level of our own actions.
How, then, are we to identify, preserve and transmit this digital heritage without giving up our sovereignty? This section offers food for thought, along with methodological, technical and strategic markers for taking hold of these questions concretely.