Abstract
A “knowledge project” is under way at the weapons labs. Its objective: to put everything on a quick-access intranet. Here, weapons designer Henry Street (with beard) is videotaped for posterity.
In Washington, the Cox committee alleges that the nation's nuclear weapons laboratories have been losing highly classified weapons data to the Chinese like water from a leaky bucket. (See “Call China's Hand,” page 21.)
“The People's Republic of China's penetration of our national weapons laboratories spans at least the past several decades, and almost certainly continues today,” the committee concludes.
Further, the committee judges that the Energy Department's counterintelligence effort is run so poorly that it cannot be “even minimally effective” until next year at the earliest.
Meanwhile, in California and New Mexico, the same laboratories are creating what may be the most attractive nuclear espionage target ever developed. All of the nation's most important nuclear weapons secrets are being collected, catalogued, indexed, cross-referenced, and put, as a neatly tied package, into a unified electronic database.
It's called the “Nuclear Weapons Information Base,” a point-and-click computer network of weapons knowledge so complete that its theft by foreign spies would constitute a loss of virtually every nuclear weapon design secret possessed by the United States.
The information base is a multimedia, how-to manual for every weapon in the U.S. nuclear arsenal, from blueprints and data from hydrogen bomb tests to video interviews with the physicists and engineers who designed them.
The system is intended to supply workers across the nuclear weapons complex–Livermore, Los Alamos, Sandia, Oak Ridge, Pantex, and others–with the accumulated knowledge of a half-century of weapons work on demand.
The database is part of the Stockpile Stewardship Program, which provides the labs with the world's fastest computers, most expensive lasers, and a host of other advanced and costly pieces of technology so that physicists may come to understand weapons physics from first principles, perhaps allowing them to design nuclear weapons without any fission-yield tests.
Hyper hyperlinks
Unlike the uncertainty surrounding the value of the weapons data that may have been lost to the Chinese, there would be no doubt about the impact if this database were compromised. It would explain everything; it represents the crown jewels of America's nuclear weapons program.
Is this all-encompassing database a good idea? Not every one thinks so, including some of those involved in its planning. The security issues have been hotly debated behind closed doors over the past five years.
According to Bill Bookless, a weapons physicist who is in charge of Lawrence Livermore National Laboratory's contribution to the network, some participants have argued that the security risk is simply too great and the project should be abandoned. But the majority, including Bookless, believed the security issues could be solved.
“We have reviewed this very carefully with internal and external review groups,” says Bookless, “and we have determined first of all that it is worth doing, the gain to our program is great–and we believe we can come up with solutions that will solve the key [security] issues.”
This description of the database, written by Bookless, was offered in the October 1998 edition of Stewardship News, an in-house newsletter for bomb designers at Livermore:
“Pulling together the archival data and files you need to perform your stockpile stewardship task has never been easier. You open the home page of the archive web server and click on one of the menu items, such as ‘NT (nuclear test)’ or ‘Ref (reference),’ you are presented with a list of topics specific to that area–perhaps some reference material and links to weapon systems, such as the W87.
“Another few clicks and you effortlessly locate the final report or raw test data from an event you are looking for. Next, to resolve those few questions raised by reading the report, you key a few words into the search frame. The search engine identifies links to the drawings, reports, photos, presentations, or even videos that match your criteria from resources across the complex.
“A few mouse clicks and the information you sought is up on your screen–all in a matter of minutes–and ready to save to your local disk for later use. This is the vision of the Nuclear Weapons Information Project (NWIP) and much of it is reality now.”
A basic version of the database is already in place. The format is familiar; it uses a standard Web browser (such as Netscape) that tens of millions of people use every day to cruise the internet. But in this case, the hyperlinks and the downloadable images are classified beyond Top Secret. The database includes:
▪ Reports, data, engineering drawings, documentary pictures, and details for many nuclear tests. This information is as classified as nuclear information gets. Data from old nuclear tests are used to calibrate and confirm new weapons simulation codes, particularly in difficult areas such as the rate of neutron production in an exploding primary. The test data are “the core gold mine for validating computer calculations,” according to one physicist.
▪ Full-text issues of classified periodicals, such as National Security Science and Technology Review and Defense Research Review.
▪ Full-text searches of several thousand online documents and reports.
▪ Searches through metadata, essentially bibliographies, for 500,000 documents. (One of the entries is “On Cell-Centered Lagrangian Hydrodynamics,” a report written by two researchers from Los Alamos, one of whom is Wen Ho Lee, the Los Alamos scientist suspected of providing weapons data to the Chinese government.)
▪ Several technical thesauri, much-needed items because different outposts in the nuclear weapons complex use different terms to describe the same thing.
▪ Tutorials, lessons, and conference and lecture transcripts.
▪ General information about how nuclear tests were conducted.
▪ Viewgraph collections–weapons scientists are said to be incapable of speaking without viewgraphs, and thus viewgraphs offer concise historical summaries of various projects.
Data … and memories
For most of their existence, the weapons labs frantically pushed forward with the next nuclear test, the latest weapon development. Keeping records was a low priority, especially in the 1950s and 1960s, when the arms race seemed especially urgent.
But an understanding of some of the fundamental principles in weapons physics was often inexact. To some extent, designers learned what worked through the trial-and-error of nuclear tests. As the weapons enterprise moved ahead, however, details of past tests faded. There was always a new weapon to invent.
But with the end of fission-yield tests in 1992, there came a realization that the past represented data to be mined and catalogued. Engineering blueprints–thousands and thousands of them–have been run through scanners, digitized, and stored on hard drives.
But the archiving, which began in 1993, has not been easy: Some Energy Department sites have as many as 300 different databases or catalogs of relevant data. And some data shelved in unmarked boxes have never been catalogued. Katie Walter of Livermore has written about the difficulties:
“Access to the full documentation today is sometimes difficult, in part because weapons-related data were often classified and/or compartmentalized to limit the risk of inadvertent disclosure or access.
“Also, older data are dependent on old computer codes, operating systems, or media that cannot be read, and old notes and memos are fading. But even more vulnerable is the critical knowledge still residing only in scientists' heads or stashed in individual repositories.”
An “individual repository” could be a filing cabinet or a safe in a scientist's office. Despite pleas from the labs' archives, these records sometimes go through the shredder when a scientist retires. One of the treasures discovered was a set of calibration data from nuclear tests that a scientist had stashed in his garage because he couldn't bring himself to destroy his life's work when he retired.
On the human side, teams are working hard to capture the knowledge recorded only in the brain cells of physicists and engineers. This downloading of human data is done on videotape, typically with retired researchers who have come back to the lab to be interviewed on camera.
Two or three of them get together and the dialogue takes off, says Keith Johnstone, who heads the database program at Sandia National Laboratory in Albuquerque. One engineer, he says, may tell another: “Jack, you designed this all wrong and this is what you should have done … and they go right into the design mode for three or four hours. That's amazingly valuable.”
Some individuals are walking storehouses of key knowledge, none more so than physicist Seymour Sack, a colorful perfectionist who played a role in the design of the egg-shaped primaries for the warheads of the MX, Poseidon, Minuteman, and Trident missiles.
“Seymour had a lot of tape time,” notes Kent Johnson, who heads up the history program at Livermore. Sack spent an entire day with a video crew inside a top-secret vault where exact replicas of every nuclear weapon are locked up.
“They taped him walking from one model to another,” Johnson said. “He explained why it was made this way instead of that way. It's one thing to document what you did but it's another to explain why you took this approach instead of some other approach.”
The videotapes are transcribed and indexed. A weapons researcher interested in the W87 (the MX warhead) might type in a search request for “Sack” and “W87.” In a matter of seconds, the video of Sack and the W87 model will appear on the screen, allowing the researcher to learn directly from the master himself.
Need to know
That this data, so conveniently sorted and assembled, is the perfect target for espionage has been a major concern since the project was first envisioned.
“A pyramidal need-to-know model is also being implemented,” says Katie Walter of Livermore. “Individuals authorized at the top of the pyramid may have access to nearly everything while those authorized at other levels have access only to information in a particular domain or perhaps about specific weapon systems. By enhancing its classified network infrastructure, Liver-more can balance the increased access to information against the increased threat of compromise.”
But, as lab officials acknowledge, defining a need-to-know mechanism in an online environment can be difficult. Each of the labs and production facilities maintains a database of its own information. But all of the local collections are to be networked “crosscom-plex”–but that linkage has been delayed while need-to-know procedures are worked out.
The difficulty, according to Livermore's Bookless, is determining who needs to know what. But Bookless says progress has been made, and that a pilot project, using non-classified information, is under way. In fact, he says, the need-to-know approach taken by the information project has become the model for the Energy Department's new “infosec” (information security system) in the wake of the espionage revelations.
On-site, the information database resides in closed networks that are isolated from the outside world by “air gaps,” meaning there are no connecting wires from the network to other computers.
A smile and a wave
As the current espionage scandal suggests, an air-gapped classified machine may sit in the same room as an unclassified computer. At least in the past, a person could transfer data from a classified computer to an unclassified computer by merely using disks.
Under Energy's new rules, classified and unclassified computers may still stand side by side. However, “they would be required to have separate and incompatible removable media,” says Livermore spokesman David Schwoegler. “A Jazz drive in one, Zip in the other, so that the classified medium cannot be inserted into the unclassified system.”
Another thorny problem: Sharing bomb blueprints across the complex. As Bookless wrote in last October's Stewardship News, “We have a vision for the future that any member of the DOE complex should be able to easily make a query from his desktop and immediately find online information anywhere in the complex.”
Sharing information means shipping it on private networks that are potentially vulnerable to interception. The labs already send classified data across the wires, from phone calls to large data files. All of the transmissions are encrypted using devices approved by the National Security Agency, which employ lengthy algorithms that change keys several times during each transmission.
The labs' security experts say it would take powerful computers a long time to decipher even a section of one intercepted message. But intelligence services routinely record and store intercepted messages, even those they cannot immediately decode. With computers gaining brute strength every year, what is unbreakable today may be decipherable a few years from now.
But, as the report of the Cox committee highlighted, human communications, not technical intercepts, may have provided China with U.S. nuclear weapons secrets. The weapons labs have from the beginning designed their security efforts to protect against theft from outsiders rather than from betrayal by insiders. “There was a lot of personal trust built into the culture,” as Livermore's Schwoegler puts it.
The barbed-wire fences and armed guards that surround the labs symbolize that philosophy. Guards carry automatic weapons to shoot intruders, such as terrorists. Meanwhile, thousands of employees walk through the gates daily with little more than a smile and a wave from those same ready-to-fire guards.
A computer disk in a shirt pocket would go unnoticed. In fact, they do go unnoticed because it is not against security rules to carry a disk around. At some entrances, there are no guards at all, but booths–electronic portals that allow employees to pass through if they know their personal identification number and if their weight approximately matches the figure stored in a computer.
The guards who roam the halls of Livermore's Building 111, where weapons physicists keep their offices, hand out security infractions to those who leave classified documents on their desks overnight. But the guards have no way of knowing if a physicist has taken a document home at night.
After much thought, the Energy Department is bringing together all it knows about designing nuclear weapons, every secret and every trick, and sending it out across the wires. The Energy Department is entrusting this information to a large number of people and hoping for the best.
Some have suggested that this knowledge, once collected, be destroyed. Livermore's Kent Johnson thinks that would be a mistake:
“I don't think the idea of destroying knowledge is a good thing for human beings to do,” he said. “You learn to deal with the knowledge.”