The Unbearable Lightness of Being Saved: On the Tyranny of Perfect Metadata

We are taught, from our first foray into digital preservation, that metadata is king. The mantra is repeated in conference halls and best-practice guides: without rich, structured, and meticulous metadata, an archived object is a ghost, a thing without context, a digital brick without the mortar of meaning. We are told to describe, to tag, to categorize, to relate. We are told that perfection in this descriptive layer is the highest calling, the only way to ensure future usability. What if this relentless pursuit is not just a Sisyphean task, but a fundamentally flawed one that actively hinders the very accessibility we seek?

The common advice creates a tyranny of the perfect. It sets an impossibly high bar for entry into the preserved record. Consider a small historical society, a lone researcher, or an overwhelmed city clerk. Faced with the daunting prospect of creating Dublin Core-compliant records for thousands of images or documents, they often do nothing. The psychological weight of doing it "right" paralyzes action. The result isn't imperfect metadata; it's no preservation effort at all. The perfect has become the enemy of the good, the enemy of the saved.

This obsession also risks creating archives that are only intelligible to other archivists. We layer on technical jargon, controlled vocabularies, and complex schemas that we assure ourselves are for the user's benefit. But are they? Or are we building a labyrinth only we know how to navigate? We preserve the object, but we bury it under a mountain of our own professional scaffolding. The future researcher we claim to serve may not care about the precise EXIF data of a image; they may simply need to see the faces in the crowd. Our perfect metadata can act as a gatekeeper, obscuring the raw, human content with a veil of technical correctness.

The Case for the 'Dumb' Save

Perhaps we need to champion the 'dumb' save—the act of preservation stripped of its descriptive burden. A JPEG thrown onto a hard drive with a filename like '1985_street_fair_001.jpg' has more future potential than the same photo that was never saved because the archivist didn't have time to ascertain the photographer's full name and a precise geotag. The former has a fighting chance. It has content. It exists.

This is not an argument for chaos, but for a radical re-prioritization of effort. The primary, most heroic act is the act of saving the bits themselves—preventing the irreversible slide into digital oblivion. Descriptive metadata is a secondary, important but not primordial, layer. It can be added by crowds, by AI, by future scholars. It can be imperfect and iterative. But the object itself must first exist.

By fetishizing metadata, we have inadvertently built a system that values description over substance, process over content, and perfection over existence. It is a counterintuitive truth: to make our digital heritage more accessible, we must first learn to save it imperfectly, to embrace the 'good enough' save, and trust that the meaning we so desperately want to pin down today might be better found by the curious minds of tomorrow.

Notes & further reading

A few pages I came back to while writing this: