Against Perfect Copies: Why Archivists Should Sometimes Prefer the Broken Link

In our field, the 404 error is the enemy. It’s the emblem of loss, the digital equivalent of a missing page torn from a book. The standard, almost sacred, advice is to capture everything: use the WARC file, the perfect snapshot, the immutable copy that stands forever against the tide of decay. We are taught to build fortresses of data, hermetically sealed and infinitely reproducible. But what if our obsession with the perfect copy is, in some crucial ways, blinding us? What if the broken link itself—the very thing we’re fighting—holds a vital piece of the record we’re trying to preserve?

The Unpreservable Context

Consider a government press release announcing a new public policy. We capture the page flawlessly—text, images, CSS. Ten years later, a researcher finds it in our archive. It looks pristine. But the original, at the moment it mattered, sat on a specific URL, nestled within a living website architecture. It was linked to from the homepage, commented on by blogs (whose links are now dead), and perhaps rebutted in a now-defunct forum. Our perfect copy tells the “what.” The 404s that now surround it, the dead pathways that led to it, told the “how” and “why.” They were its context. By preserving only the page, we’ve taxidermied a creature after removing it from its ecosystem.

This is the counterintuitive core: a perfectly preserved digital object is often a decontextualized one. It is removed from the flow of the web that gave it meaning. The broken links that proliferate around a topic aren’t just noise; they are a fossilized map of its influence, its connections, its contested place in the digital commons. A WARC file can freeze a page, but it cannot freeze the web’s reaction to it. That story is told in the pattern of absence.

This isn’t an argument for negligence. It’s an argument for a more nuanced documentation strategy. Perhaps, alongside our pristine captures, we should be deliberately archiving the experience of the rupture. Could an archival record include a screencast of a researcher navigating to a now-gone resource, encountering the 404, and following the breadcrumbs of the Internet Archive’s “No URL found” pages? Could we log the referrer headers from link aggregators at the moment of capture, not just the target? We need to preserve not just the nodes, but evidence of the edges, even—especially—when they fray.

The goal shifts, then, from creating an illusion of seamless permanence to documenting the true, fractured nature of digital existence. A link that works forever in an archive is a kind of lie. It suggests stability where there was only flux. The honest archive would show the click that leads nowhere, and would explain why that nowhere matters. It would acknowledge that the broken link isn’t merely a failure of preservation; it is often the most accurate record of what the web actually was: a temporary consensus of connections, constantly being unmade. In trying to save everything perfectly, we risk saving a beautiful, bloodless fiction. Sometimes, we need to save the scar.

Notes & further reading

A few pages I came back to while writing this: