Gates (who deals with this episode at length in his forthcoming book), recalls discussing the PDB report with President Bush at Kennebunkport on August 17. According to Gates, the president asked whether the report should be taken seriously, and he responded that, once the date was set for the Union Treaty to take effect, the conservatives would face their last chance to take action. Since the conditions that the intelligence community had set as critical warning markers had been met, he added, the warning should indeed be taken seriously.

In any case, part of the reason why the coup was not detected with certainty was precisely because it was poorly planned and thus lacked many of the tangible signs of a major military operation. In January 1991, for example, the intelligence community had provided good warning of the Soviet crackdown in Vilnius, the prediction having been based in part on intelligence observation of the preparations being carried out by the Soviet military, such as the movement of troops, establishment of communications links, and other traditional warning signs.

In contrast, many military and KGB units were not even told of their mission in the August coup. The plotters also failed to take effective steps to capture Yeltsin (indeed, Yeltsin reported to the Kremlin on the morning of the coup and was turned away) and little or no effort was made to take potentially hostile radio and television stations off the air. Much of the Soviet military leadership itself did not know of the coup until after it had begun.

Once the coup was underway, the intelligence community quickly determined that, precisely because the signs of adequate preparation were missing, the plotters had little chance of success. This analysis enabled the United States to adjust its position quickly as the crisis evolved. In his first response to the coup, President Bush had referred to the plotters' action as "extraconstitutional"; by the end of the day, the president denounced the coup as illegitimate and endorsed Yeltsin's demand that Gorbachev be restored to power.

Lessons To Be Learned

There are a number of lessons to be learned from the myths concerning the intelligence community's performance during the final years of the Soviet Union. They pertain to observers of the intelligence community in and out of the government, the intelligence community itself, and policymakers.

1. The Danger of the Conventional Wisdom. The notion that the CIA seriously misread the situation in the Soviet Union quickly became accepted as fact -- although there does not appear to have been any attempt by those making such a pronouncement to investigate the matter seriously. Even after CIA officials disputed the claims of Senator Moynihan, no major newspaper or other organization appears to have bothered to launch an in-depth investigation of his serious charges. Others have been quick to accept the essence of Moynihan's claim without demanding evidence or analysis to support it.

Aside from the unfairness of stigmatizing those who provided intelligence analysis, such unwarranted claims create a presumption that drastic changes are needed. But "fixing" things that are not broken can make things worse than they are.

2. The Limitations of Intelligence in the Policy Process. One reason why critics claim U.S. intelligence failed is that the United States did not achieve all of its objectives in responding to the collapse of the Soviet Union -- specifically, preserving the integrity of the USSR and the Gorbachev regime. But the decision to follow this policy had little to do with an intelligence failure, and the intelligence provided to U.S. officials should have made them aware of the risks of this policy.

Undesirable outcomes are not necessarily the result of faulty intelligence. Intelligence analysts do not make policy; policymakers make policy. Most supporters of democracy and responsible government prefer it this way.

3. Intelligence Analysts Are Not Psychics. Policymakers sometimes complain that intelligence analysis is not sufficiently explicit to get their attention. Just as intelligence analysts may miss warning indicators in the noise of day-to-day events, policymakers may not appreciate the significance of an intelligence product that is presented to them in the context of their routine business.

At times, the inability of intelligence to make the necessary impression on policymakers may reflect the fuzziness resulting from the process of coordinating national estimates, with compromise language replacing sharper language. But the complaint may also reflect an unrealistic expectation by some policymakers that specific future events can be precisely predicted well in advance.

Precise turning points in tidal changes such as the collapse of the Soviet Union depend on the confluence of many individual events that can not be predicted. As a result, it often is not possible to write the kind of specific, unconditional intelligence forecast that would cut through the noise (e.g., "hardliners will attempt a coup on the morning of August 19, and this will ultimately result in the collapse of the Soviet Union") -- especially if one demands such a definitive forecast weeks, months, or years in advance.

George Kolt's "The Soviet Cauldron" is an example of brilliant intelligence not because it made a correct exact estimate of when the Soviet Union would collapse or a coup would be undertaken, but because it laid out the basic factors that were driving events in Soviet politics in mid-1991, and the possible logical consequences, in succinct, blunt terms. The memorandum presented evidence of several necessary conditions that existed for a coup, and then left it to the policymaker to either refute this evidence, alter the conditions, or consider the risk the United States was willing to tolerate in maintaining its existing policy of supporting Gorbachev. Contrast this analysis with a case in which an intelligence failure does appear to have occurred, such as the fall of the Shah, when the intelligence community reported that Iran did not appear to be in "a revolutionary or even pre-revolutionary" condition.

Clinton administration officials today must make a similar evaluation of any evidence they receive suggesting that democracy and reform in Russia may soon collapse. They have to determine whether they are willing to accept whatever case the intelligence community makes, based on their own assessment of the potential risks and available options. But, as in the case of the fall of Gorbachev, the ultimate evaluation will be made by policymakers, not intelligence analysts.

4. The Intelligence-policymaker Relationship Needs To Be Improved. It is clear from our research that one problem limiting the usefulness of intelligence was that senior policymakers generally had to deal with intelligence analysts at arms length, through constrained channels of communication, and intermittently. As a result, policymakers were limited in their ability to appreciate why a particular conclusion had been reached, the assumptions or conditions on which it was based, and why there might be differing views within the intelligence community. The intelligence community can at least begin to address this problem by rethinking the intelligence process from its end.

Since World War II that process has been roughly as follows: Policymakers present the intelligence community with a problem; the intelligence commodity develops a consensus through a process of evaluating evidence and debating hypotheses; and policymakers are then presented with one set of conclusions that reflects the larger wisdom of the entire intelligence community (although a member of the community can note a dissent if it disagrees strongly with the consensus). The significance of this approach is reflected in the fact that community-wide products are generally accorded higher status than, say, products produced by a single agency (e.g., NIEs have higher standing than Intelligence Assessments produced by the CIA's Directorate of Intelligence).

This concept, which perhaps reflects the way people used information and analysis in the mid-twentieth century, may no longer be valid. Indeed, it runs counter to the way in which most consumers of information behave today. Because people have greater access to information and are generally less willing to trust authorities, they demand the opportunity to question the analyses with which they are provided, and to shop around for sources of expertise and analysis. They make their final decision based on a number of factors -- some objective, such as the track record and credentials of a source, and some subjective, such as prejudice or self-indulgence.

Indeed, as noted above, policymakers in the real world make decisions by considering a multitude of views. A more realistic approach to providing intelligence support to top policymakers on broad policy issues such as the future of Gorbachev -- or, today, the future of Yeltsin -- would allow such officials greater interaction with the analysts. The intelligence community could facilitate such direct interaction through the use of modern secure communications systems and data retrieval systems -- technologies that were notably absent when the classical concept of intelligence estimation was developed in the late 1940s.

Rather than attempting to generate consensus, the intelligence community should facilitate a process in which the intelligence consumer is fully aware of the different views held in the intelligence community and the reasons for these differences. Policymakers, who will decide which truth they will adopt anyway, should be the ultimate coordinators of intelligence, knowing that they will be held responsible for their decision.

Essay Types: Essay