ALBUQUERQUE, N.M. — A melding of new data retrieval technology and the memories of veteran weapons project managers and designers is helping Sandia National Laboratories preserve 20th century technology to ensure responsible stewardship of the nation’s nuclear weapons stockpile into the 21st century.
The work is part of the Knowledge Preservation Project, an effort begun about two years ago at Sandia to capture the experience and ideas of retired or soon-to-be retired weapons engineers about arcane knowledge they shared with their colleagues but never wrote down.
“There were enormous lessons learned in jeopardy of being lost,” Keith Johnstone says about the approximately 50 weaponeers who have been interviewed on about 350 hours of videotape. Another 150 staff members or retirees have been identified for future interviews.
Johnstone and John Taylor, another Sandia weapons engineer, conceived the project as a way of assisting current and future weapons scientists who may need to maintain stockpiled weapons beyond their intended length of service without the benefit of having ever designed or tested a system. The information also can assist in recognizing somewhat old-fashioned attempts by non-nuclear nations to develop nuclear weapons capabilities.
Another benefit, says Johnstone, could be understanding technological approaches to problems that had to be discarded in an earlier age but one day may be applicable. “Science can lie dormant for years, and then something changes that allows you to implement it.”
“Within Sandia,” Johnstone says, “the stockpile is outliving the designers, production engineers, and technicians who assembled the components. In the past, all transfer of information took place through people. The human information servers, if you will, are leaving. That raises the questions, one, how do we capture and preserve this information, and two, more important, how do we provide access to it?
“We have lots of databases, designs, drawings, reports, and manuals. But how do you record experience, and most of all, how do you record ideas?”
Acting as moderator, Johnstone generally tapes sessions with a weapons specialist and one or two colleagues. “The interaction is very important,” he says. “You see the whole process of design and problem-solving come alive in front of you.”
Discussions in these sessions range from technology to policy. Sessions are scheduled in two-hour increments. Up to 20 hours total have been devoted to interviewing a single individual. Occasionally, panels of eight or nine speakers and an audience are assembled to discuss entire weapons systems or specific technologies.
“What we’re doing is trying to capture their ideas, but more than that, their psyches, to try to learn not just what they did, but why they did things the way they did; to find out what worked, what didn’t work, what might have worked had the supporting technology been more advanced, and so on,” says Johnstone.
Video was seen as ideally suited for recording the kinds of undocumented technical information the project was designed to preserve: the art as well as the science of weapons design, the seat-of-the-pants hunches, the intuitive leaps, the frustrations, the insights, and the synergies that have driven Sandia’s weapons work over the decades.
“But the trouble with video is that you often don’t know what you have on tape, and when you do know, you don’t really have a good way to access it quickly and easily,” explains Carmen Ward, who serves as project leader for the knowledge preservation team.
Johnstone says the team made a commitment to the participants that the information and insights they provided would be used, it would be accessible — at the desktop — to a new generation of weapons engineers.
To meet that commitment, project team member James Borders, who works in Sandia’s Engineering Information Management Department, came up with a software-based video retrieval system called RePAV for Relevant Point of Access Video. It is a system that makes it possible to search a video and access the relevant section based on keyword input.
RePAV works like this: The audio portion of a videotaped interview is transcribed and linked to a time code keyed to the video. Meanwhile, the video is converted to a digitized format that enables almost instant access to any frame. When you type in a keyword, the software finds the word in the transcript, matches it with the time-coded information, then accesses the video segment where the word is spoken.
“The system is pretty good now, but it’s only going to get better,” Johnstone says. “Data compression, storage capacity, processor speed, resolution, networking, digital imaging — we’re seeing breakthroughs in all of these areas on a regular basis.”
While RePAV was designed to give researchers access to weapons design data, the technology has potential applications far beyond Sandia. Several national organizations that store vast amounts of videotape have expressed interest in the RePAV system, Ward says, and research and archival facilities the world over could put the system to immediate use.
“More and more information is captured only on video. The RePAV technology and the need have come together. I think this will change the way people gather information,” Johnstone says.