The Insecure Vault: Why We Must Archive the Deliberately Broken Link
For years, the cardinal rule of digital preservation has been clarity and authenticity. We seek to capture the webpage as it was, the dataset in its pristine, accessible state. Our holy grail is a perfect, functional replica. But what about the things designed to be incomplete? I want to make a case for a different kind of artifact: the deliberately broken link, the intentionally fragmented record. In our rush to preserve perfect ‘snapshots,’ we are systematically erasing a fundamental truth of digital existence—the truth of revocation, of retreat, of the digital door slammed shut.
Consider the standard web archive capture. It follows a link and tries to freeze everything it finds. But it rarely captures the experience of clicking a link that returns a 403 Forbidden error, a 410 Gone, or a custom landing page that says, ‘This content has been removed at the request of the author.’ That moment—the denial of access—is itself a crucial piece of public record. It is a data point about power, about changing norms, about regret or legal pressure. By only archiving the content that was *once* there (if we can get it), or by ignoring these paths altogether, we create a dishonest history where information only ever appears, never disappears. We smooth over the friction of the web.
This goes beyond takedowns. Think of a government dataset portal that publishes a ‘readme’ file explaining why certain fields have been redacted or entire tables withheld ‘for security.’ The standard archivist’s impulse might be to lament the missing data, to hunt for an unredacted copy. But the redacted document, in its official, barren state, is the *true* public record. It is a monument to a specific moment of bureaucratic calculus. Preserving *only* a hypothetical complete version would be an act of historical fabrication. The broken link, the withheld field, the deleted tweet—their absence *is* the signal.
Critics will say this archives ‘nothing,’ that it’s a waste of resources to catalog voids. But that’s the old mindset speaking. We’re not archiving nothing; we’re archiving the *performance* of removal. The specific language of a 410 error message, the exact timestamp a document vanished from an index, the legal citation used in a takedown notice—these are the forensic details of our digital culture’s metabolism. They show us not just what we knew, but what we were *allowed* to know, and when that permission was rescinded.
Our archives must become secure vaults for insecurity. They must hold, with equal reverence, the published report and the 404 page that later replaced it. To do otherwise is to pretend the digital world is a steadily accumulating library, when in reality, it is a dynamic, contested space where links rot, pages are memory-holed, and access is a fluid privilege. If we wish to preserve the real web, we must preserve its fractures. The broken link is not a failure of the archive; it is one of its most honest artifacts.
Notes & further reading
A few pages I came back to while writing this: