The Myth of the Seamless Migration: On the Violence of Data Conversion

In the world of digital preservation, there is a recurring fantasy, a comforting bedtime story we tell ourselves when faced with obsolete formats and dying systems. It’s the story of the seamless migration. The notion that with the right tools and enough effort, we can gently lift data from its aging digital vessel and place it, pristine and fully formed, into a shiny new one. The bits travel in a state of grace, untouched by the chaos of change. It’s a clean, bloodless transfer. I believe this to be one of the most seductive and dangerous illusions we entertain.

Data conversion is not a gentle migration; it is an act of translation, and like all translations, it is inherently violent. Every time we convert a document from a proprietary word processor format to a plain text file, or rasterize a complex vector drawing into a PNG, we are making a series of brutal choices. We are deciding what is essential and what is expendable. The meticulously tracked changes in that document, the hidden layers of that drawing, the interactive elements of a Flash animation—these are often the first casualties. They are deemed non-essential to the "core" content, the semantic husk we seek to preserve. But are they? The context, the process, the very feel of the original artifact is stripped away, leaving behind a ghost.

The Loss of the Ineffable

This process becomes particularly fraught when we consider digital art and early web culture. Preserving a GeoCities page as a static WARC file captures the text and the images, but it murders the blaring MIDI music that auto-played, the scrolling marquee text, the garish, table-based layout that shuddered as it loaded over a 56k modem. These weren't bugs; they were features. They were the essence of the experience. By converting the dynamic, messy, and often infuriating original into a clean, accessible, but static snapshot, we haven't preserved the thing itself. We have preserved a museum diorama of it, a specimen pinned under glass. The life is gone.

We console ourselves with the concept of 'significant properties'—the core attributes of a digital object that must be maintained for it to remain authentic. But who defines what is significant? The archivist? The original creator? The potential researcher fifty years from now? This is not a technical question; it is a deeply philosophical and curatorial one. Every conversion is a bet on the future, a wager that our present-day judgment about what matters will align with the needs of an unknown future. It is an act of profound interpretation, not a simple mechanical task.

This isn't an argument against migration. It is a necessity; digital entropy demands it. But we must approach it with the solemnity and humility of a restorer working on a fresco, not the brute efficiency of a file-copy operation. We must document our conversions meticulously, acknowledging what was lost, what was altered, and why. We must save not just the converted file, but, whenever possible, the original in its native format—the digital equivalent of keeping the papyrus scroll alongside the printed transcription. The myth of the seamless migration leads to complacency. Acknowledging the violence of the process makes us better, more honest stewards of a fragile and translated past.

Notes & further reading

A few pages I came back to while writing this: