A few years ago, I did a lot of work on “nuclear knowledge management,” or NKM. There are a lot of definitions of what NKM includes, but my first brush with knowledge management came when, as a Department of Energy manager, I was visited by researchers from the national laboratories. They told horror stories of experimental facilities that had been shuttered and troves of valuable data that were haphazardly stored and that were deteriorating over time. The researchers had scattered and the information contained on antiquated magnetic tapes and other media were in danger of being lost. Because so many nuclear research facilities had closed, we were not even able to reproduce the work in the United States if we needed to.
While I always knew this problem was not unique to the nuclear industry, it seemed that the knowledge management problem for data from nuclear experiments was particularly acute. So many experiments had been abruptly cancelled, so many facilities had been closed, and so many technological approaches had been abandoned for one reason or another. As I spoke to people about NKM, I always acknowledged that the nuclear KM problem was part of a larger problem, but in truth, I came to think of the nuclear part of the problem as being unique.
It was therefore an eye-opener to read a recent report alleging that scientific data in general suffers substantial loss over just a couple of decades. I was further disturbed to see that the problem is a continuing one. Although I recognize that new electronic technologies can sometimes create as many problems as they can solve, I had hopes that data-sharing on websites, better communications, etc., might mean that today’s research stood a better chance of being preserved and usable over long periods of time. This is not so, according to the article. Much data is still stored on devices that become obsolete, and despite the wide reach of the Internet, researchers move, change e-mail addresses, and cannot always be found.
Certainly, not every bit of data retains its value over time. However, all knowledge is built on what came before, so the report in the article that 80% of scientific data is lost within 2 decades comes as a surprise and a disappointment. In my work on NKM, I tried to point out to others that the handling of data should be an integral part of research, and that organizations should be sure that old data is migrated to new storage media as necessary.
I realize that this is easier said than done, but too often today, the problem is simply not recognized until it is far too late. A good first step would be to establish a deliberate program within each research organization to ensure that researchers document and archive their data properly, and have a plan to handle the data over time.