Preserving Electronic Libraries
Researchers are developing a system that would read data lost on aging computer tapes. DIGITAL DILEMMA
IN a few weeks, Alan Bawden is going to start spinning 1,700 computer tapes in an attempt to restore his past.Skip to next paragraph
Subscribe Today to the Monitor
From 1976 until earlier this year, Dr. Bawden was an undergraduate, a staff member, and a graduate student at the Massachusetts Institute of Technology's Artificial Intelligence Laboratory. Every few months during those years, MIT's computer labs made a set of backup tapes containing all of the data on the lab's computers. On those tapes were the results of government-funded research projects, breakthrough programs that changed the world of computer science, and reams upon reams of personal electronic ma il. It's all there in digital form, all safely stored away for posterity. There's just one problem: No computer left at MIT can make any sense of it.
Although much of the AI Lab's research was published in papers, "the published papers don't contain the full results," says Bawden. "A lot of the valuable data is contained only on those tapes."
So for the past four years, Bawden and a small group of friends have been developing a computer system that can read the tapes again. Their goal: to copy the data onto modern computers so that the lab's computer research from the '70s and '80s will be available to today's researchers.
Bawden has a personal reason for wanting to preserve the data as well: "Some of this history is my personal history."
What's a problem for MIT is also a problem for government and most corporations. "People who are running organizations aren't aware of the need to treat electronic media and electronic systems differently from the way that they treated paper systems," says David Bearman, editor of Archives and Museum Informatics, an industry newsletter.
While paper does deteriorate over time if it is not properly preserved, even books that are printed on acid-based paper can be put away in boxes and read in 20 or 30 years. Computer tapes, on the other hand, simply do not age well. Recently, when a technician working for the National Archives and Records Administration tried to read the contents of a 27-year-old computer tape from the Equal Employment Opportunity Commission, the tape literally melted inside the tape drive, says Kenneth Thibodeau, directo r of the Archives' Center for Electronic Records. "The [tape's] chemistry had changed," he says. Short shelf life
Experts say that even when stored under proper temperature and humidity, computer tapes shouldn't be depended on to store data for more than 10 years. Furthermore, the tapes can't be left on the shelf: Every few years the tapes should be unrolled and re-rolled to prevent a problem called "print through," in which the data transfers between one layer of tape and another on a tightly wound reel.
"If you look at the interior of the tape hub, there is an incredible amount of pressure between the two layers of the tape," says Jim Green, who heads the space science data operations for NASA's Goddard Spaceflight Center in Maryland. "The magnetic domains from one [tape] layer end up being so close to the layer underneath it that they end up being imprinted." The result is added noise and, eventually, loss of data.
The solution is to copy data from old tapes onto newer tapes every few years, before the tapes go bad. Fortunately, since the information on the tapes is in digital form, each copy is a duplicate of the original. But having to make those copies is an ongoing expense sometimes hard to justify to decision-makers. The reason the National Archives hadn't tried to read the 27-year-old tape earlier, Mr. Thibodeau says, was that funding for the project had been cut under the Reagan administration and only recen tly restored.