National Security Intelligence

by Loch K. Johnson

  The Question of Dissent

  Especially tricky has been the question of how to represent dissenting views in an Estimate. The intelligence agencies sometimes have quite different perspectives on a world situation. Military intelligence organizations are notorious for their “worst-case” approach to estimating – a result, critics contend, of pressures on analysts applied by the Department of Defense and the military-industrial complex to justify larger military budgets by scaring the American people and their representatives in Congress with testimony about dire threats from abroad. Conversely, military intelligence officials often consider the CIA and INR as too “civilian” and unable to understand the true nature of foreign military threats.

  The clash of differing views among intelligence agencies can be healthy, if driven by an objective weighing of facts rather than policy bias. Debate among analysts can provide policymakers with a range of views, instead of just the lowest common denominator. Sometimes Estimates are diluted to a tapioca consistency that robs policy officials of the nuances they need to understand. A NIC vice chair remembers that NIEs were “all too likely to produce a hedged and weasel-worded result.”72 Further, agency dissents have been relegated to obscure footnotes on occasion, if included at all. The best NIC managers have been careful to ensure that dissents are stated at some length in the text of the NIE, not hidden in a footnote – if only to avoid the resentment of dissenting agencies that have had their findings and judgments shunted aside. Some dissenting agencies insist that their contrary opinions be highlighted boldly in the text, often in a boxed format obvious to every reader, and placed as well in the KJ section of the document – a useful practice that encourages debate.

  The Internal Liaison Challenge

  An additional responsibility of intelligence managers is to ensure that NIOs and other designated intelligence analysts maintain good liaison relationships with consumers. “The difficulty lies not only in predicting the future, in a world of many variables, incomplete data, and intentional deception,” writes an intelligence officer, “but in convincing policy makers that the prediction is valid.”73 Experience has shown that unless a policy-maker knows and feels personally comfortable with an NIO or other intelligence briefer, he or she is less likely to pay attention to an Estimate. Rapport between consumer and producer also provides analysts with a better understanding of the information needs of the policy departments, lowering the chances that intelligence reports will be irrelevant to current policy concerns and become merely “self-licking ice cream cones.”74 Too close a relationship, though, can undermine the objectivity of the intelligence process – the politicization danger.

  The Timing and Frequency of NIES

  An NIE can be written quickly, in two to four weeks (or less than a day in emergencies); in two to six months during normal times; or as long as three years on a slow track. Historically, Estimates have taken 215 days on average to produce: about seven months. Those studies readied on a fast-track basis during a crisis have their own name: a special NIE or SNIE (pronounced “snee”). During the Suez Canal crisis of 1956, the Intelligence Community produced a SNIE on Soviet intentions within a few hours. Analysts hope, though, to have at least three months to produce a thorough Estimate.

  From 1947 to 2005, the IC produced 1,307 NIEs, averaging twenty-three a year.75 The numbers have fluctuated over the years (see Figure 2.3), a reflection of the Intelligence Director's priorities and an administration's interest in receiving Estimates.76 Added into the mix are changing world circumstances that may or may not require the preparation of new NIEs. In times of war, for example, policymakers are likely to be focused on current intelligence that reports on immediate battlefield exigencies, with NIEs pushed to a back burner.

  Figure 2.3 Frequency of NIEs by year, 1950–2005

  Source: Central Intelligence Agency, 2006.

  NIE Hits and Misses

  The dean of CIA analysts, Sherman Kent, commented on the goal of in-depth analysis. “The guts of the matter,” he said, “is the synthesizing of the pieces and setting them forth in some meaningful pattern which everyone hopes is a close approximation of the truth.”77 Still, the end result remains something of a best guess, resulting from discussions among the top analysts in the Intelligence Community. As Kent once put it, “Estimating is what you do when you do not know.” One enters “into the world of speculating.”78 However shrewd the forecasts, they still remain hunches. It is better than blind luck, because the judgments are based on expert research and knowledge; but they are a far cry from certainty.

  At times, NIEs have been as accurate as an expensive Swiss watch; on other occasions they have been wide of the mark. Examples of successful predictions include: the likely conduct of the Soviet Union in world affairs (Moscow would try to expand, but would avoid the risk of general war);79 the launching of the Soviet Sputnik in 1957; the Sino-Soviet split of 1962; the Chinese A-bomb test in 1964; the development of new Soviet weapon systems throughout the Cold War;80 developments in the Vietnam War (1966–75); the Arab–Israeli War of 1967; the India–Pakistan War of 1971; the Turkish invasion of Cyprus in 1974; the Chinese invasion of Vietnam in 1978; the mass exodus from Cuba in 1978; the Soviet invasion of Afghanistan in 1979; the sharp deterioration of the Soviet economy just before the end of the Cold War (1984–89); the investment strategies of the Organization of Petroleum Exporting Countries (OPEC) consistently over the years; the rise and fall of political leaders around the world, including the breakup of Yugoslavia in the 1990s; the threat of “aerial terrorism” in 1995, presaging the 9/11 attacks (but, unfortunately, without details); and forecasting the difficulties of a post-invasion Iraqi society in 2002.

  Most of the Intelligence Community's major mistakes during the Cold War were about what the Kremlin intended, not what weapons systems the communist empire possessed. The ability to track the numbers and capabilities of Soviet weaponry was vital during the superpower confrontation, and remains so today with Russia. Arms negotiations with the Russians and others still depend on the ability of the intelligence agencies to detect any significant violation of arms accords – a process known as verification. “Trust but verify” was a famous arms control slogan of the Reagan Administration during the 1980s.

  Examples on the debit side of analysis include: the failure to predict the outbreak of the Korean War in 1950 or the placement of Soviet offensive missiles in Cuba in 1962; the reporting – especially by U.S. Air Force Intelligence – of a (non-existent) bomber and missile gap between the Soviet Union and the United States in the 1950s and early 1960s; underestimating during the Vietnam War the supplies coming to the Viet Cong through Cambodia; underestimating the pace of the Soviet strategic weapons program; faulty forecasts about the Soviet invasions of Hungary in 1956 and Czechoslovakia in 1968, the Arab–Israeli war in 1973, and the fall of the Shah of Iran in 1979; and a lack of precise predictions about the collapse of the Soviet empire in 1989–91 – although the CIA tracked its economic decline and rising political turmoil more closely than critics concede.81 More recently, mistakes regarding supposed WMD in Iraq in 2002 arose as a result of limited humint in the country, poor vetting of the few humint sources that were available (Curve Ball, for instance), and an overreaction to earlier underestimating of Iraq's weapons prowess in 1990.82 Attacks by “lone wolf” terrorists have also taken the United States by surprise from time to time, as in San Bernardino, California (2015) and Orlando, Florida (2016).

  In a nutshell, NIEs have been uneven in their capacity to provide officials with accurate forecasts about history's probable trajectory, especially when it comes to invaluable details. Long-range prognosticating is a skill that diminishes the farther one attempts to peer into the future. “The CIA Directorate of Science and Technology has not yet developed a crystal ball,” Senator Frank Church (D, Idaho) cautioned. “Though the CIA did give an exact warning of the date when Turkey would invade Cyprus [in 1974], such precision will be rare. Simply too many unpredictable factors
enter into most situations. The intrinsic element of caprice in the affairs of men and nations is the hair shirt of the intelligence estimator.”83 When it comes to predictions, Betts stresses as well that “some incidence of failure [is] inevitable.” He urges a higher “tolerance for disaster.”84 The bottom line: information is usually scarce or ambiguous, and the situation in question may be fluid and changing. Former intelligence officer Arthur S. Hulnick advises that “policy makers may have to accept the fact that all intelligence estimators can really hope to do is to give them guidelines or scenarios to support policy discussion, and not the predictions they so badly want and expect from intelligence.”85

  This realistic sense of limitations is unhappy news for presidents and cabinet secretaries who seek clear-cut answers, not hunches and hypotheses; but such is the reality of national security intelligence. It bears repeating, though, that having intelligence agencies closely examining world affairs is better than operating blindly. As a CIA analyst writes, “There is no substitute for the depth, imaginativeness, and ‘feel’ that experienced, first-rate analysts and estimators can bring to the often semi-unknowable questions handed them.”86

  Even if NIEs are less than perfect instruments for forecasting future events, they have the virtue of gathering together in one place a dependable set of facts about a situation abroad. This frees up decision-makers to focus attention on sorting out the disagreements they might have over which policy options to choose. Former NSA Director William Odom states this case: “The estimate process has the healthy effect of making analysts communicate and share evidence. If the NIEs performed no other service, they would still be entirely worth the effort.”87 Almost fifty years ago, Kent noted, too, that “the intelligence estimate will have made its contribution in the way it promoted a more thorough and enlightened debate.…”88

  The Iraqi NIE Controversy

  Caught up in the swiftly moving events that followed hard upon the 9/11 attacks in 2001, DCI Tenet never got around to ordering the preparation of an NIE on Al Qaeda or on suspected Iraqi WMD. Neither did the White House. Reportedly, advisers to President George W. Bush feared that a full-blown Estimate on the WMD question would reveal “disagreements over details in almost every aspect of the Administration's case against Iraq.”89

  The lack of an NIE on Iraqi WMD at the very time the United States was engaged in an important internal debate over whether to launch a war against Saddam Hussein's regime was unfortunate. Rumors about WMD in Iraq were rife and inflamed by references to “mushroom clouds” appearing on American soil, expressed by President Bush and National Security Adviser Condoleezza Rice.90 Tenet has admitted his error: “An NIE on Iraq should have been initiated earlier, but at the time I didn't think one was necessary. I was wrong.”91

  Senators Richard Durban (D, Illinois) and Carl Levin (D, Michigan), both members of the Senate Select Committee on Intelligence (SSCI), believed at the time that a NIE would be important in the debate over a possible war against Iraq. They insisted on a formal written assessment and persuaded SSCI Chairman Bob Graham (D, Florida) to send a letter on September 10, 2002, to Tenet requesting that an Estimate on Iraq be prepared as soon as possible.92

  Tenet replied that he would be unable to produce the kind of comprehensive NIE on Iraq that Graham sought, because of other pressing intelligence duties. Nevertheless, he promised to furnish, as soon as possible, an Estimate on the subject of WMD in Iraq.93 The DCI ordered a “crash project” to meet SSCI's request. The ninety-page Estimate went to the Senate about three weeks after the request – too hastily prepared, in the view of critics. One reporter called it “the worst body of work in [the CIA's] long history.”94 The document arrived at SSCI's quarters in the Hart Office Building in early October and Tenet came to brief the Committee's members on its main points. In retrospect, Senator Graham feels that the DCI seemed to skate over dissenting views that downplayed the Iraqi threat.95

  Senators Graham, Durbin, and Levin next sought to have the NIE declassified for public consumption, except for portions that might disclose sensitive sources and methods. They made the request on October 2, 2002, and two days later Tenet delivered an unclassified version of the longer document, this one twenty-five pages in length. The problem, from Graham's point of view, was that the new version “did not accurately represent the classified NIE we had received just days earlier.”96 Missing was the sense imparted in some passages of the longer, still-classified document that Saddam posed no immediate danger to the United States or his neighbors, if left alone. In Graham's opinion, Tenet had diluted the original document to keep in step with the opinion of the White House that Saddam was a great menace to the United States.97 Republican Senator Chuck Hagel (Nebraska) concluded that the condensed NIE was “doctored” to suit the political needs of the White House.98

  The “Key Judgments” section of the original NIE was not released until July 16, 2003 (the invasion of Iraq began four months earlier, on March 19, 2003). Only on June 1, 2004 – almost a year into the war in Iraq – did Tenet provide a more complete, but still redacted, version of the KJ section. In a report released in July 2004, SSCI concluded that the NIE's Key Judgments were, for the most part, “either overstated, or were not supported by the underlying intelligence reporting.”99 Only much later, in 2007, in a memoir published as war in Iraq lingered on, did Tenet acknowledge that “we should have said, in effect, that the intelligence was not sufficient to prove beyond a reasonable doubt that Saddam had WMD.” The DCI said that he now believed “more accurate and nuanced findings would have made for a more vigorous debate – and would have served the country better.”100

  Current versus Research Intelligence

  A vital question for any nation is how many resources should be devoted to the production of current intelligence, at the expense of preparing more deeply considered products of research intelligence like the NIE. In capitals around the world, most policy officials prefer to receive current rather than research intelligence. Indeed, policymakers in the United States recently rated NIEs eighth in value among the intelligence products they received.101 Former CIA senior analyst Mark M. Lowenthal writes that, recently, the Intelligence Community has “put its greatest emphasis on shorter, more current products,” a response to “a fairly consistent decline in policymaker interest in intelligence community products as they get longer and more removed from more current issues.”

  The upshot is that about 80–90 percent of the analytic resources of the U.S. Intelligence Community is now dedicated to clarifying for policymakers what happened today and yesterday, and what is likely to happen tomorrow – the essence of “current intelligence.”102 According to Lowenthal, the Intelligence Community has “gotten out of the knowledge-building business. Now it is: current, current, current.”103 A former CIA deputy director for intelligence points out, however, that “a bunch of research intelligence is done, not necessarily estimative – just everything we know about subject X. Then someone says, ‘It's about time we do a formal Estimate.’”104

  A Yardstick for Intelligence Reports

  Whatever the type (more current or more research oriented), all intelligence reports should attempt to honor the basic canons of professional analysis. Among the major hallmarks of an outstanding intelligence report, the first requirement is to get the facts right.

  Accuracy

  In 1999, the American comedian Jay Leno quipped that “CIA” must stand for “Can't Identify Anything,” after the Intelligence Community forwarded to a NATO bomber pilot coordinates for an arms depot in Serbia that turned out in fact to be the Chinese embassy. (The wrong map was actually sent to NATO by the National Geospatial-Intelligence Agency, not the CIA.) Several Chinese were killed in this accidental bombing and many Chinese continue to think the bombing was done on purpose. Obviously, the intelligence agencies must provide accurate information – within limits; no sensible policymaker expects IC clairvoyance about future world events.

  Timeliness

&nb
sp; Important, too, is the quality of timeliness. History runs on nimble feet, and if intelligence reports lag too far behind events, they are likely to be of little use to a decision-maker. The result may be the dreaded acronym “OBE” scrawled across an analytic report: “overtaken by events.”

  Relevance

  Policymakers have no interest in receiving an intelligence report on local elections in Greenland when their in-box is filled with decisions that have to be made about events on the Horn of Africa. Sometimes intelligence analysts wish to write about their own interests or speciality – perhaps derived from their PhD dissertation topic, say, “Rural Politics in Outer Mongolia.” If analysts are out of tune with the consumers they serve when it comes to topical focus, they may as well go on a fishing trip; consumers will have no time for their reports. The best intelligence reports are tailored to address the most pressing problems in a decision-maker's in-box. This reality underscores the value of close – yet non-political – ties between intelligence producers and consumers, brought about by forward liaison teams and periodic meetings between the two groups.