Logo IMG


Bit Rot

Brian Hayes

Somewhere in a cobwebby corner of my computer's hard disk are a few manuscripts I wrote 15 years ago on my first PC. The word-processing software I used then was grandly named The Final Word. It was anything but. I've gone through a dozen word processors since then, and nearly as many computers. To keep older documents accessible, I've had to transfer and transform them repeatedly, from one disk to the next and from one file format to another. And still I have yet to find the Final Word. Sooner or later I'll be gathering up my digital belongings yet again and converting them to some new format. This time I'll have 12,000 files in tow. I can't wait.

But my personal data-migration problems are puny compared with those of corporations, universities, libraries and publishers. (Imagine the plight of the National Archives, the agency charged with preserving everything the U.S. government deems to be worth keeping.) And the material to be preserved is not just text. Obsolete storage media and file formats are just as vexing when the files hold other kinds of information, such as images, engineering drawings, the numerical results of scientific experiments, digitized audio and video, maps, tax returns, databases.

One cause for worry among archivists is the impermanence of digital storage media. In this respect civilization has been going downhill ever since Mesopotamia. Paper documents cannot match the longevity of the Sumerians' clay tablets, and magnetic media seem to be even more evanescent than paper. That's disturbing news, and yet I suspect that relatively few disks or tapes have yet died of old age. Long before the disk wears out or succumbs to bit rot, the machine that reads the disk has become a museum piece. So the immediate challenge is not preserving the information but preserving the means to get at it.

Occasionally, the rescue of some long-neglected digital resource calls for heroic measures, such as reconstructing an antique tape drive. But most file transfers and translations are routine; utility software handles the conversion, though often with a minor loss of information. Even when the process is easy and successful, however, file conversion is a nuisance. It's a lot like moving your household—more work than you expected, and a few dishes always get broken. As my stockpile of files for the digital U-Haul continues to grow, I dread the prospect more and more. I daydream of hiking into the woods as a cybersurvivalist, refusing ever again to upgrade my hardware and software. If I stock up on spare parts—80-megabyte disk drives, 30-pin SIMMs—I could live out my remaining years in a log cabin with a Macintosh SE/30.

Most likely I would not be alone in the woods, but before I begin hoarding the computer equivalent of canned goods it seems prudent to consider less-extreme alternatives. All I really want is some way of representing digital information that I can stick with for a while. I want a file format for the ages—a single format that will serve many purposes and continue to work with many combinations of hardware and software. Which format is that? I don't have a definitive answer—not even an answer that meets all my own immediate needs. But I think I know where to look for inspiration. Here's a hint: Among all the kinds of things stored in computers, the ones that are hardest to keep up-to-date and hardest to move from one platform to another are programs for the computer itself. Maybe programmers know something the rest of us ought to learn.

comments powered by Disqus


Subscribe to American Scientist