Periodically worries surface about the length of time a data recording medium can be in use for. Hard drives fail. CDs fail over time and so forth. This last year commentators have said that its not the working life of the media that's the issue; it's whether devices will still be working in, say, thirty years time that can actually mount and read the media. Anybody with reels of old paper tape now has a lost archive of data stored on it because, for all practical purposes, paper tape readers are no longer being made. Ditto punched cards.

In general though, I don't think this is a problem for business. How many businesses actually want to keep data for longer than fifteen or so years? There might be sentimental value in fifteen year-old company data but there is unlikely to be any business value whatsoever. Any data that might need to be kept that long can be stored in an off-site vault by some records retention company, like Iron Mountain. Its job includes having a sufficient stock of reading devices such that it fifteen years time it still has usable ones.

Its job is also, I'd hazard a guess, to foresee the looming end of a particular medium/device combination's usefulness and migrate the stored data to a new recording medium.

The data retention requirements of governments, of museums, universities and other institutions can be considerably longer. They may see themselves as having to hold on to data forever. When the data is held on paper, engraved in stone, written on parchment, pictured on canvas and so forth then the retention period is as long as they can keep the medium usable. When the medium starts failing then the data has to be captured and stored anew. Inevitably that means electronic and digital records.

Hence the worry for them that, in terms of their retention timescales, media like CDs have an ephemeral life.

But this relatively ephemeral is not a tragedy. In fact it is a triumph. Because once data is captured and held electronically then its potential storage life is infinite precisely because it can be migrated to new media as required. THe digitalisation abstracts the information and removes it from reliance on any one recording medium. Because it is electronic it can be transferred. Bits are mobile. Marks engraved in stone, ink lines on parchment, brush strokes on canvas; these are all static. immobile, forever at the mercy of the longevity of the medium on which they reside.

Once data is engraved on stone and the stone fractures then the data is lost. It's literally irretrievable. Once data is stored electronically then it can be recreated. Thus, conceivably, somewhere in the world's computer infrastructure, there is a piece of data held on a Sun/StorageTek tape that started out its life as a punched card. It was then migrated to paper tape, moved on to hard drives, became archived to an early tape format and then progressed to a later one.

This movement from medium to medium is the essence of longevity. Once a piece of data is committed to a medium from which it can't be readily retrieved then its archival storage strategy is flawed. There always has to be the potential for movement to a new medium for a truly long retention period.

For thousands of years wandering Arab tribespeople crossed middle eastern deserts from well to well with no charts, no written down instructions. All navigational data was passed on by word of mouth from one generation to another. It was a successful information strategy. It worked. So too, in our electronic universe, data longevity is best ensured by being able to pass it on, from one generation of recording media to another. Media and device life is not the issue in data longevity; it's being able to migrate data from old and fading medium/device combinations to new ones and doing so when needed. That's the duty of care owed to archive data by custodians of electronic data.

Recording media does not endure. Bits can.