FAS Homepage | Government Secrecy | S&G Bulletin ||| Index | Search | Join FAS


Issue Number 43
January 1995

Experimenting with Intelligence Estimates

In an uncommon innovation, the intelligence community has asked several outside contractors to produce intelligence estimates in parallel with estimates that are being prepared on the same topics inside the intelligence community.

Intelligence officials hope this experiment will teach them about alternative methodologies in preparing an estimate and presenting its findings, said Gregory F. Treverton, Vice Chairman of the National Intelligence Council, an official advisory group to the Director of Central Intelligence based at CIA headquarters.

Intelligence estimates, which are among the most important products of the intelligence community, are intended to inform policymakers about the future of some policy-relevant topic or area.

The experiment in parallel estimates should illuminate the limits of classified intelligence and the comparative value of non- traditional sources. "The idea that classified intelligence sources can give you purchase on lots of important topics is a fiction," Treverton told S&GB on December 28. The long-term political economy of Russia, for example, is an important topic that is essentially opaque to clandestine intelligence collection.

The parallel estimates project may also have an impact on the deliberations of the new Commission on intelligence roles and capabilities, since the estimates will provide an empirical indication of how well certain intelligence functions can be performed outside of the existing bureaucracy.

"The next step will be to involve outsiders directly in the writing of estimates, not in parallel projects," according to Treverton. "It seems obvious that we need to interact with outsiders in a whole series of ways."

Two parallel estimate projects are currently underway. The Hudson Institute has a contract to produce an estimate on the future of the European economy over the next 5 to 7 years and its political implications. This project is nearing completion. Science Applications International Corporation is preparing an estimate on the future of the Russian economy, which will probably be finished in spring 1995 (the estimate, not the economy). Both contracts were awarded following a competitive bidding process.

In a article entitled "Estimating Beyond the Cold War" (Defense Intelligence Journal, Fall 1994), Dr. Treverton discussed the motivation for soliciting estimates from outside the intelligence community. He wrote that "experts are hard-pressed to conceive of discontinuities; they tend to imagine the future as a bounded version of the present, about which they know so much. The current [estimative] process is very much driven by experts and, thus, it will be intriguing to see if outsiders do any better at assembling teams to think creatively about discontinuities."

Security Policy Framework Reorganized

A new interagency group known as the Security Policy Board, established by Presidential Decision Directive 29, is supposed to develop consistent and cost effective government security policies across the whole range of security disciplines, including classification, personnel security, physical, technical, and information systems security. It held its second meeting in December.

The new Board, supported by a full-time staff of 20 civilian and military personnel, plans to dissolve or consolidate a variety of moribund security organizations throughout the government. A new Security Policy Forum, supported by policy development committees, will become the locus for formulating new policies under the Board's overall guidance.

The Board has already moved to take control of security policy development government-wide. "The Board Staff will be a mandatory coordination step for all agencies, departments, components and programs that prepare implementing or supplemental directives, regulations or publications" on security, according to one Board document.

It is too early to tell whether the new Board will actually do anything about the defects and anachronisms of current security procedures, or whether it will simply be another way station on the endless detour that is security policy today.

The Board's staff director, Peter D. Saderholm, did not respond to requests for an interview. But another official agreed to provide copies of Board documents on a confidential basis. A rather dull 42 page Board report entitled "Creating a New Order in U.S. Security Policy," 11/21/94, is available from S&GB for $2 to cover postage.

Automating Declassification

Recent drafts of the pending executive order on classification call for new technological approaches to declassification, as well as the establishment of a government-wide database of information that has been declassified.

Acting in the belief that the executive order may actually be signed some day, the National Archives and the Information Security Oversight Office asked the MITRE Corporation to investigate the options for automating the declassification process, and for establishing an electronic database of declassified records.

The result is an October 1994 report entitled "Interagency Declassification Support System: Estimated Costs and Implementation Considerations," by Howard E. Clark, et al, of MITRE. The executive summary of the 150 page document, which will soon become available on the Internet, is available from S&GB.

The exhaustive study concludes generally that there is a need for greater automation to overcome enormous backlogs in declassification activities. Depending on the desired level of support, establishing an automated declassification capability would cost each agency between $846,000 and $3.8 million over five years (Federal Computer Week, 12/5/94, p.8).

"Automated declassification" here does not imply any kind of artificial intelligence approach to declassification. Nor does it necessarily entail changes in declassification standards or practices such as bulk declassification. Rather, it means computerized tracking, processing (redaction), and interagency coordination of declassification actions. It is believed that this would increase staff efficiency while laying the foundation for an electronic database of declassified documents.

Making declassified documents available to other agencies on- line, it is also argued, would minimize the number of cases where agencies are inconsistent in declassifying the same material.

But advocates of greater openness say government inconsistency here often works to the citizen's benefit. The National Security Archive made this point in a report it provided to MITRE identifying some 600 documents that had been differently declassified by one or more agencies, with the net result that more information was released to the public. Likewise, writer Richard Reeves in his 1993 bestseller on President Kennedy explicitly recommended multiple declassification requests for the same document to take advantage of inconsistent declassification policies.

Other noteworthy observations in the MITRE report include the following:

~ "Some agencies indicated that giving the public ready access to declassified materials in electronic form creates a possibility that the materials could be examined [by computer] and analyzed for patterns that would lead to deducing classified information.... This is a potential risk. Agencies will have to weigh it against the government's stated desire to get more of its information out to the people of the United States, who are its rightful owners.... The `mosaic' problem may be one that the government must indeed live with if it intends to make electronic declassified information widely available." (p. 11-16)

~ Referral of documents from one agency to another "seems to be a major cause of the apparent `footdragging' for which the government is sometimes criticized... Giving agencies the authority to review and declassify information for which other agencies have equities would help to eliminate referral delays. Agencies would be given declassification guides from other agencies for reference purposes.... MITRE asked some of the agencies for their opinion of this concept; none of the agencies asked thought it was a good idea." (p. A-32)

~ "Most agencies stated that their [declassification] backlog is due to inadequate resources, particularly staff. They also stated that it is unlikely they will be given more staff." (p. A-33)

~ "At one agency... it was estimated that a [declassification] reviewer spends up to two hours per day cutting, marking, stamping, photocopying, and filing paper to support the redaction process." (p. A-32)

~ "We estimate that the federal government employs between 600 and 800 full-time equivalent declassification staff." (p. A-36)

~ "Some agencies use black-out marking to indicate redacted material. Others believe that this creates a negative reaction in the requester's mind, and use white-out marking as an alternative." (p. 3-12)


Opennet, the bibliographic database of declassified, publicly releasable Energy Department documents, is now available for public access through the Internet, Secretary Hazel O'Leary announced on December 7.

The initial database contains basic bibliographic data on more than 265,000 documents, including document location and accession numbers, enabling users to search and request previously declassified documents directly. Additional information on all documents that are declassified and released after October 1, 1994 will be routinely added to Opennet.

Opennet does not offer the full-text access to declassified documents that is envisioned by the MITRE report noted above. On the other hand, Opennet has the great advantage that it actually exists. Moreover, "in the future," according to DOE, "the Secretary's vision is to further expand Opennet" to include full- text access. However, "we are still a long way from that point."

Currently, Opennet has "a lot of growth potential," i.e. it leaves a lot to be desired. An Opennet document search using the term "space nuclear power" identified 542 relevant documents. However, the current Opennet software will only display 40 of them. The remainder must be separately requested from the system operator. The display capability will soon be upgraded to 200 titles, DOE indicated, and eventually more.

Opennet can be accessed through the DOE "home page" on the world wide web at http://www.doe.gov, or by telnet to cupid.osti.gov, or by direct dial to (615)241-3906. The DOE press release on Opennet with related instructional materials is available from S&GB.

Disinforming Congress

Defenders of the status quo in intelligence often invoke the hoary cliche that while intelligence failures are publicly trumpeted by the media, the public never finds out about all of the intelligence community's successes. "You're never going to see a list in the newspaper of all the good things they have done," said incoming House Intelligence Committee chairman Rep. Larry Combest lately. (Los Angeles Times, 12/29/94). There is no doubt some truth in this assertion, even though it is usually used perniciously to delegitimize public discussion of intelligence issues.

But the opposite phenomenon is more serious: The intelligence community is in the habit of proudly declaring its successes to Congress, while suppressing knowledge of its failures, according to the November 30 House Intelligence Committee "Report of Investigation: The Aldrich Ames Espionage Case" (available from the Goverment Printing Office).

"While there is rarely a reluctance to provide information [to Congress] when there is an intelligence success, the Committee's experience when there is bad news has been uneven at best," the Report stated (p.40).

Despite an exhaustive series of Congressional hearings and investigations into counterintelligence matters in the late 1980s, "no information was provided about the loss of important assets in the Soviet Union," the Committee reported, citing "a pattern of a lack of candor by senior CIA officials in answering questions of Committee members."

Congressional oversight was effectively derailed at a time when U.S. intelligence capabilities in the Soviet Union were being crippled. Congress remained in the dark even though the Soviet purge of U.S. assets was so drastic and unsubtle, as Aldrich Ames stated in a revealing interview with Congressional investigators, that by 1986 "The KGB might as well have taken out an ad in the New York Times saying, `we got a source'." (Report, App. 3, p. 31)

The failure to notify Congress happens to be a violation of the law. "The Committee finds that the CIA failed to fulfill its statutory obligation to keep Congress informed of the lost Soviet assets and subsequent investigation. While candid and complete information should not be dependent on asking `the right question,' in this case the right questions were asked," the Committee report said. (p. 71)

But legal requirements aside, the failure of the intelligence community to properly brief Congressional officials-- even on a classified basis-- is potentially as disturbing as the espionage scandal itself, since it means that basic oversight mechanisms are not functioning reliably.

Secrecy and Government Bulletin is written by Steven Aftergood and published by the Federation of American Scientists, 307 Massachusetts Avenue N.E., Washington, DC 20002. For more information call (202)675-1012 or send email to saftergood@igc.apc.org

FAS Homepage | Government Secrecy | S&G Bulletin ||| Index | Search | Join FAS