The Wizards of Langley

by Jeffrey T Richelson

  In September, as Hirsch was preparing to leave, he discussed the two panels’ findings with his successor—findings that probably reinforced some of her basic notions about what needed to be done.

  THE NEW NUMBER ONE

  In his search for a successor to Hirsch, Deutch looked in a number of places and asked the national labs, which included Sandia, Los Alamos, and Livermore, for nominations. Sandia’s nominee, and Deutch’s choice, was Ruth David, who at the time was director of Sandia’s Strategic Thrust in Advanced Information Technologies.5

  David was truly an outsider, although one with an impressive résumé. She held a doctorate from Stanford in electrical engineering and had taught at the University of New Mexico. Her academic credentials included a number of technical papers and coauthorship of two reference works on digital signal processing. At Sandia she had managed the development of data acquisition systems to monitor underground nuclear testing at the Energy Department’s Nevada test site, as well as the development of various engineering test facilities for Sandia programs. But she had never had any contact with the CIA, either as an employee or consultant.6

  David’s selection was announced on the last day of July; on September 15, ten days after Jim Hirsch’s last day as deputy director, she started work as his successor.7 Her deputy at the time was Gary W. Goodrich, who had become associate deputy director in October 1989. But he would not be around for long—December 31 would be his last day at the agency. On January 1, Pete Daniher, chief of the Office of Technical Collection, who had long coveted David’s job and whose chance to get it may have left the building with Jim Woolsey, became her deputy.8

  NEW OFFICES

  While still at Sandia, David had participated in the “Agile Enterprise Manufacturing Forum” at Lehigh University. The meeting focused on how corporations could deal with a rapidly changing business environment—forming teams quickly to solve problems and then moving on to the next problem.9 David brought the concept with her to her new job. When she arrived in Washington, she noted that the distinct organizations that made up the intelligence community faced rapidly changing priorities, tight budgets, and consumers who wanted intelligence tailored to their needs. Part of the answer, she believed, was to form new alliances within the community and with consumers as well as with individuals and groups from academia and think tanks. She was soon giving briefings on the notion of “Agile Intelligence.”10

  A key to being able to operate an agile intelligence community was to increase the directorate’s focus on information technology, as the blue-ribbon panel had suggested. She believed that it had not received sufficient attention but “touch[ed] every aspect of what we do.”11 Even as resources were becoming tighter, the volume of information was continuing to expand dramatically—due to the Internet, the loosening of political constraints in the former Soviet Union and elsewhere, and increased volumes of radio, television, and other telecommunications and of scientific and technical data. In addition, demand grew for the intelligence community to provide its product with greater speed. Further, there was the growing U.S. involvement in military and humanitarian missions in areas of the world not traditionally of great concern—which meant rapidly shifting priorities.12

  Traditionally, analysts received data only after the information had gone through one or more stages of human processing. Open source data, for example, were “once carefully selected, translated, edited, and organized by people who brought . . . a great deal of knowledge” to the task.13 David later noted in a 1998 speech that on one occasion the CIA received a collection of 100 diskettes containing information of potential interest. Examining their content required two to three dozen analysts and technical experts.14

  In today’s world, she observed in the same speech, an analyst might need to sort through a vast volume of data, possibly data whose content was completely unknown, looking for relevant information. Even if the general content was familiar to the analyst, the volume of data could make any search tedious and time-consuming. Or the analyst might need to identify a pattern in a huge data set. Such jobs might use so much of an analyst’s time that they simply could not be done. Automated processing, using keywords, was one solution.15

  Such concerns and her belief that improved information technology capabilities held the solution led David in 1996 to establish three new offices in the directorate—the Office of Advanced Analytical Tools (OAAT), the Office of Advanced Projects (OAP), and the Clandestine Information Technology Office (CITO).16

  The advanced analytical tools office was established as a joint effort with the Directorate of Intelligence—combining technology push with user pull. The CIA described the office as having been chartered “to investigate, develop, and deploy innovative information systems to enhance . . . capabilities to collect, process, and disseminate intelligence.” Solutions would “reduce information overload, increase analyst collaboration, improve the intelligence knowledge base, and automate foreign text translation.”17 Appointed to head the office was Susan Gordon, an intelligence analyst who had specialized in foreign weapons and space systems. By the time David departed the agency in fall 1998, the office had a staff of 100 and was focusing on five areas: information extraction, data mining, data visualization, machine translation, and security.18

  The office’s work on information extraction sought to give analysts a means of picking out data they would want to enter in a database from a collection of intelligence reports. As Gordon expressed the problem, “spotting trends in the data is an area where we really need extractors.” Data-mining efforts are geared to developing tools to employ keywords to build a database or bases from a number of large databases designed for uses different from those required by the analyst. Software that could prowl through a body of data and respond by isolating specific information for an analyst to review could ease the analyst’s task.19

  Developing machine translation capabilities had become more pressing, due to the multitude of languages that have become important in recent years. The CIA is short of human translators, particularly in languages such as Farsi. At the same time, there is no significant commercial market for such a translation capability in many of the required languages, which requires the analytical tools office to oversee development of such a capability.20

  The advanced projects office was established to overcome problems in inserting technology into the intelligence process, transferring technology from the research and development stage to operational use. It was to provide a bridge between development and use by “taking things out of R&D and deploy[ing them] quickly,” David recalled.21 Doing that job required the office to look ahead to what technologies would be needed for the collection and analysis of information, identify relevant commercially developed technologies, and seek to develop required technologies when none were commercially available.22

  The Clandestine Information Technology Office was established as a joint office with the Directorate of Operations. According to a CIA press release, the office would “address collection capabilities within emerging information technologies”—including fiber optics and the Internet.23

  ALIENATION

  Robert Phillips, who served in the directorate for over thirty years, “liked Ruth a lot,” but felt that as the deputy director for science and technology she was “way in over her head” and had “no feel for the politics of the agency.” There were “things going on in the CIA that none of her experiences ever prepared her for” and “things [that she did] that turned out awkward.”24

  One of those things was her creation of “new offices without [any] idea as to how to staff them” and pay for them. Phillips’s views were echoed by former ORD director Philip Eckman, who noted that David’s creation of the new offices was done “without any real understanding of or caring about the culture of the organization,” and without taking the necessary steps to ensure funding. Ultimately, the new deputy director found a directorate component that she felt could be cut back in order to help fund t
he new offices—the Foreign Broadcast Information Service. Her plans apparently called for a 20 percent cut in personnel and a 38 percent cut in the nonpersonnel budget. Approximately one-third of FBIS’s fourteen foreign bureaus would be closed as part of a plan to save $20 million.25

  FBIS was older than the agency itself—the “purveyors of fine open source intelligence since 1941”—according to an FBIS briefing slide. It had been nearly exempt from cuts, because its cost was minimal and it made an enormous contribution, according to Phillips and others. David believed the FBIS mission was important but not as important as the new offices she was establishing.26

  Naturally, many in FBIS were annoyed with their new directorate head. One of David’s early actions had been to establish a Lotus Notes system that enabled personnel to send her anonymous e-mail, which she promised she would read and answer. Her unpopular decision, at least with FBIS personnel, to slash funding for the unit, led to “some very cruel messages” calling her an amateur and an interloper.27

  David felt that if she could explain her views and how important the new offices were, the animosity could be neutralized. To get those views across, she began holding a series of town meetings. Fifty to sixty FBIS members were in attendance at one, and after her opening speech they launched a verbal assault, objecting that she had just arrived at the agency and didn’t know what she was doing. David, Phillips recalled, was “almost in tears.” In his view, the situation could have been alleviated if her deputy, Pete Daniher, who had been in the agency for over twenty years and had credibility with the dissidents, had spoken up in her defense. But Daniher, who was sitting with her, “never said a word.”28

  And it was not only agency insiders who objected to the planned cutbacks. Academics, who made great use of the unclassified digests of the foreign press, were not pleased.29 Congressional oversight committees were also supporters of FBIS, noting in 1997 that “comprehensive open source collection, translation, and analytic effort is crucial to the [intelligence community’s] ability to maintain global coverage,” and that “careful scrutiny of ‘closed society’ media . . . can also reveal valuable information on trends, new developments, and leadership plans.”30

  Earlier in the year, David backed off on plans to cut the broadcast service, and a CIA spokeswoman announced that FBIS would be spared from proposed funding cuts. She also noted that FBIS would continue to monitor, translate, and publish accounts from about 3,500 foreign broadcast and press outlets in fifty-five languages and newspapers—which represented “virtually 100 percent” of current coverage.31

  Congressional concern and interest were expressed in the House intelligence committee’s report on the 1998 Intelligence Authorization Act, published in summer 1997. The committee noted that FBIS’s “re-engineering strategy” called for “using more modern and commercially available technologies as FBIS’s operational linchpin and to transition from traditional large-scale, static collection and processing centers toward a more agile and less expensive architecture.” The committee applauded the effort to “adapt FBIS’s infrastructure and operating practices to incorporate new technologies and to meet intelligence requirements more efficiently.”32

  The committee did express concern about resource-allocation decisions that were being made “without fully taking into consideration ‘customer’ requirements.” In short, it was unclear to many FBIS customers what regions of the world would be “affected by significant decreases in collection, translation, and analytical activities.” It was necessary, the committee believed, that “open source collection . . . be driven by the direct input of major customers, particularly the all-source analysts who best understood where their information gaps lie.”33

  THE BEST SCIENTIST

  In a brief statement in April 1996, the CIA announced that ORD’s John Craven, then fifty-seven, had been named the agency’s best scientist. The statement said his “breakthroughs in areas of computer logic, digital signal processing and laser technology are truly remarkable.” The following year he was named one of the fifty CIA trailblazers.34

  That Craven was able to contribute even a fraction of what he has is astonishing. In 1968, he received a doctorate in solid-state physics from the University of Chicago; his dissertation topic was “The Fermi Surface and Band Structure of White Tin as Derived from de Haas–van Alphen Data.” After Chicago, he joined the CIA and was placed in a special career-development track for the agency’s top prospects. The six-month program took him from Cape Cod to a U-2 base to Strategic Air Command headquarters and onto a nuclear submarine. Then in 1971, a swimming accident left him paralyzed from the neck down.35

  Today, Craven lives in a modest apartment in a Maryland suburb of Washington, where he has an around-the-clock caregiver. Once or twice a week, he travels to the CIA for briefings. Mostly, he works at home, relying on a computer, a voice-activated phone, and a fax machine. He holds a pointer in his mouth to tap out letters, never more than fifteen a minute, on his computer keyboard. His phone has special encoding devices that allow him to conduct secure conversations with colleagues.36

  Those circumstances did not prevent Craven from doing award-winning work on three projects, one of which is classified. One project involved the use of microwave technology to create a 100-fold increase in the speed at which computers could operate. The speed of computers is limited by the power they consume and the heat they generate while operating. Microwave technology could operate at much lower power levels, conserving energy and reducing heat.37

  The focus of the second project was laser cross-links. One problem with using laser beams as communication devices over long distances was that they had to be perfectly aimed and were easy to interrupt, with even a slight misalignment resulting in the loss of the signal. Craven led a team of scientists who figured out a way to reduce those problems.38

  Craven told an interviewer, “Our charter is to push the state of the art,” whereas his “goal is not to nudge the state of the art but to try to make a quantum leap.”39

  NIMA

  On October 1, 1996, the National Imagery and Mapping Agency (NIMA) opened its doors with a staff of 9,000—more than any intelligence agency other than CIA and NSA. That day marked the conclusion of over five years of studies and debates over the organization of the U.S. imagery intelligence effort.

  In his April 1992 testimony before the House and Senate intelligence committees, then DCI Robert Gates noted that the Imagery Task Force he had established upon becoming DCI had recommended the creation of a National Imagery Agency (NIA), which would absorb the CIA’s National Photographic Interpretation Center as well as the Defense Mapping Agency (DMA).40

  The task force’s vision for an NIA was not as broad as what had been recommended by some people in congressional hearings and written into proposed legislation by both the House and Senate intelligence committees. The broader vision would have created an agency that would control the entire range of imagery intelligence—research and development of future collection systems, operation of current systems, tasking (the selection of targets), and analysis of the images collected.41

  During his testimony, Gates rejected the recommendations of both his task force and the congressional committees, noting he had no desire to establish a new, large agency. Gates was not alone in his reluctance to merge NPIC with DMA. Joint Chiefs chairman Colin Powell did not want to relinquish authority over DMA, which was vital to providing support to military operations.42

  Gates and Powell did agree to the creation of a small Central Imagery Office (CIO), which was established within the Department of Defense in early May. The new office was to address the problems perceived to exist within the imagery intelligence effort, particularly by many in Congress and the military. Those problems included a lack of coherent imagery management, imagery collection and dissemination difficulties, budgetary constraints, and changing requirements for the support of military operations.43

  The CIO was officially a joint CIA-DOD enterprise, c
hartered by both DCI and Defense Department directives.44 In contrast to the alternative national imagery agencies that had been proposed, the CIO was not designed to absorb existing agencies or take on their collection and analysis functions.

  Rather, the mission of the CIO included tasking of national imagery systems (assuming that role in place of the DCI Committee on Imagery Requirements and Exploitation, or COMIREX) to ensure more effective imagery support to the Department of Defense, combat commanders, the CIA, and other agencies; advising the Secretary of Defense and the DCI regarding future imagery requirements; and evaluating the performance of imagery organizations. The most important role assigned to the new office was ensuring that imagery dissemination systems were “interoper-able”—that an image transmitted on one system (an Army system) could be received on another system (a Navy system). During the Gulf War, there had been fourteen different imagery transmission systems in the Middle East theater, only a few of which were interoperable.45

  Creation of the CIO delayed, but did not prevent, creation of NIMA. In April 1995, then DCI-designate John Deutch told the Senate Select Committee on Intelligence that, if confirmed, he would “move immediately to consolidate the management of all imagery collection, analysis, and distribution.” He argued that “both effectiveness and economy can be improved by managing imagery in a manner similar to the National Security Agency’s organization for signals intelligence.”46