Intelligence failure is inevitable

Predicting the future is one of life’s mysteries, which is still unreachable and, from that point of view, it is impossible to expect intelligence agencies to be infallible in assessing adversaries and their future activities. Since the 9/11 terrorist attacks in the United States (US), the interest of politicians, scholars and media for intelligence failures has dramatically increased as they would like to understand the reasons for past failures. Moreover, the attack was a milestone, not only for the US intelligence, but also for some other foreign intelligence agencies. Respective governments refocused intelligence activities on asymmetric threats, mainly terrorism. Such reform of the intelligence system was intended to optimize the system and prevent future intelligence failures. However, while such changes are likely to improve the ratio of success to failure, it is unfeasible to expect that any system can exclude its failures completely. Alternatively, as Russ Travers stated, ‘despite our best intentions, the system is sufficiently dysfunctional that intelligence failure is guaranteed.’[1] Therefore, this essay will argue that the occasional intelligence failure is enviable and natural.[2] On the one hand, it is unreasonable to expect intelligence to predict or prevent all surprises and on the other hand, it is also untrue that intelligence is wrong all of the time. If the latter was the case, then the sole reason for the existence of the system would be brought into question and further discussion would become irrelevant. In addition, as far as the former part of the statement is concerned, it is actually a challenge to find a scholar who would oppose it. Even Stephane Lefebvre and Loch Johnson, who argued that intelligence failures are actually policy or bureaucratic failures, did not assert that failure is avoidable.[3] Sources and reasons for failure, which assure that intelligence failures will definitely occur in the future, will present the core of this discussion. As policymakers can only be considered responsible for policy failures, they and their failures will be discussed in the essay only in the sense of having any impact on the intelligence system and its elements. Therefore, this essay will focus on reasons for intelligence failures and their inevitability.

To open, it is important to define intelligence failure, its limitations and when something is either an intelligence failure or just another event. A certain unexpected event becomes intelligence failure when an intelligence agency is not capable of providing an accurate, relevant and timely intelligence warning to policymakers, and it results in an intelligence surprise, which has a direct or indirect negative impact on the agency’s nation and/or its national interests abroad. With this definition, all other unexpected events that are not part of an intelligence agency’s area of responsibility and interest will be regarded as events and nothing more. There is also an important distinction between intelligence and policy failure; the latter only happens when policymakers disregard accurate intelligence warnings, while it is only considered an intelligence failure when policymakers receive an incorrect warning and act upon it. The latter also applies to a vague intelligence product disregarded by decision makers, as they were not persuaded to use it.[4] Both intelligence and policy failure can lead to an intelligence surprise and, in the worst-case scenario, to a surprise attack.

History is full of intelligence failures and the details surrounding them, but how to completely prevent them has yet to be discovered and cannot be expected in the near future. By examining some strategic intelligence failures, such as the German attack on the Soviet Union (1941), Pearl Harbour (1941), Yom Kippur War (1973), fall of the Shah of Iran (1979), failure to foresee the Soviet collapse (1989), failure to foresee the 9/11 terrorist attacks on the US (2001), wrong evaluation of the presence of WMD in Iraq (2003), and failure to foresee Georgian-Russian conflict in South Ossetia (2008), it is evident that there is a serious problem for intelligence communities to detect, identify, assess and timely warn policymakers on forthcoming threats.

Whenever an intelligence failure occurs, theorists, analysts and commissions struggle to find who is responsible and in most cases they determine the failure to be on the side of the intelligence community. In contrast, Johnson strongly defends the intelligence community as he believes that ‘often ‘intelligence failures’ are really policy failures’.[5] Johnson’s theory generally blames policymakers for failures and he somehow disregards the complex intelligence process, known also the intelligence cycle, where errors can arise at any stage. In addition, organizational or system abnormalities and an adversary’s denial and deception, which can have an influence on an analyst’s perception, amplifies the probability that the failure will actually take place inside an intelligence organization. Most intelligence failures are likely to occur within the intelligence process, as according to Michael Herman in intelligence analysis, he considers this to be ‘the weak link of the process’.[6]

At a time when the interrelationship among political, economic, military, social, and cultural facto had become increasingly complex, no agency was postured to conduct truly integrated analysis. From the vantage point of 2001, intelligence failure is inevitable.[7]

Intelligence failures mainly take place because of analysts’ cognitive limitations, permanent anomalies of an intelligence system and policymakers’ direct and indirect involvement in the intelligence process. Therefore, to understand an intelligence failure’s inevitability, it is crucial to understand both the analyst’s and intelligence system’s behaviour. Intelligence failure can also occur at other stages of the intelligence process, but not as often. Moreover, external factors, such as an adversary’s deception and denial should not be underestimated as they can also have a significant impact on the perceptions of analysts, agencies and policymakers. Politicization also has an effect on the intelligence system and this issue will be discussed in more detail within the context of institutional failures.

Lefebvre stated that ‘the kernel of the problem [intelligence failure] with respect to intelligence analysis, [...], seems to reside in an analyst’s mind—in his or her thought processes—and with the work environment’.[8] Discussions about improving intelligence analysis almost never refers to the improvements of analysts’ thinking, only to changing the products’ format and improving the analytical process and communication with consumers. Richards Heuer was one of the first scholars who became aware of analysts’ pitfalls. In his research, he discovered that intelligence analysts have cognitive limitations, which derive from a human mental process or cognitive thinking, starting with certain assumptions developed through education, training and experience. Analysts perceive what they expect to perceive and form their opinions quickly; they are resistant to change. Therefore, new data is usually modified to fit into the existing ‘puzzle’. Even when better information becomes available, an initial impression, based on incomplete or ambiguous data, is likely to persist.[9] As a result, the analyst is unlikely to recognize a strategic shift in somebody’s behaviour as it does not match his or her perceptions. For instance, before the armed conflict in Georgia in August 2008, foreign intelligence agencies failed to anticipate a strategic shift in the country’s behaviour. Prior to the conflict, it is likely that no foreign intelligence agency predicted any possibility that Georgia would attack the separatist region of South Ossetia, which was strongly backed by Russia. Reasons for this can be found not only in their cognitive thinking but also in other analyst’s failure, for example in consensus intelligence, conventional wisdom or even mirror-imaging as ‘we would not do such stupid thing, so they would not do it either’.[10] The probable ambiguity and ambivalence of information also had a certain role to play in the aforementioned failure.

Ambiguity of evidence presents an important obstacle to accuracy and analysis, and is another reason that often leads to analytical error. Ambiguity can be caused by either a lack or excess of data. As Johnson noted, ambiguous and conflicting indications lead to ambivalence, which presents another challenge for analysts. They try to solve the problem with cautious estimates, which are not useful for decision makers.[11] Analysts can express a higher degree of certainty in their products, but this leads to intelligence failures becoming even more evident, especially in short-term assessments. Not only are long-term strategic assessments less accurate and highly speculative, they are initially less important and receive little exposure to policymakers. Analysts in these assessments cannot usually predict strategic shifts in an adversary’s behaviour, or any other significant changes, as the assessments are based mainly on previous trends. Ambiguity and ambivalence is more prevalent in long-term assessments due to the lack of information, which results in few indications to alert analysts. When data that is more accurate is collected, they have time to modify their assessments without policymakers noticing. In contrast, short-term assessments have to be more accurate in the first place, since their failure can have immediate and devastating effects on a nation. As analysts are aware of the latter, they write assessments with less specific certainty in order to avoid accusations that their assessments were wrong. As already mentioned, these products are usually useless for decision makers.

It is expected that once an analyst’s failures are known, they are solvable. Unfortunately, this is not the case. Although agencies can be aware of analysts’ weaknesses, which are probably still less common, their preparedness to resolve it is still questionable at this moment. The increased awareness of analysts’ cognitive thinking and biases can extensively facilitate the reduction of analytical failures, and with additional training and work experience will likely limit them, but not eliminate them as they are a part of the person’s mental process and his personality. This awareness can also have a negative effect since some analysts would become too cautious in their writing. Some would too frequently warn policymakers without any reason and cause the so-called ‘cry wolf’ effect, while others would wait for perfect information before releasing their intelligence product, by which time it would be untimely and irrelevant.[12] Even if we presuppose that educating and training analysts minimize the probability of intelligence failures on an analytic level, there are still other factors that make intelligence failure inevitable, the primary being bureaucracy.

In general, bureaucracy presents irreparable anomalies and is one of the main reasons for intelligence failures within an intelligence system or organization. Known also as organizational or institutional failures, it takes different forms, including the lack of collaboration and sharing of information, forming new levels of supervisors and coordinators, decentralized and incompatible computer systems and their failures, and the complexity and enormity of the intelligence system.[13]

Complex, fragmented and enormous intelligence communities often lead to a lack of control, coordination and oversight and this is likely to increase the probability that failure will actually take place. Lack of cooperation and sharing of information, also known as compartmentalization, is a result of competitiveness among national intelligence agencies and was probably a crucial reason for the surprise attack in the US in September 2001. Basis for the problem can be found in the state budget and amount of money that each agency would have received. Since the criteria for obtaining more money from the budget is often also an agency’s effectiveness and relevancy, they have to present themselves in a favourable light. That naturally results in protecting its information and only showing it to policymakers who will praise the agency when its intelligence is accurate and relevant. To illustrate this point, within the US intelligence community, the protection of sources and methods has been a hindrance to information sharing, especially between its two main ‘players’ the Central Intelligence Agency (CIA) and the Federal Bureau of Investigation (FBI) and they use the ‘need-to-know’ rule to avoid sharing information with other national agencies.[14] This became officially evident after the National Commission inquiry reported on the terrorist attacks upon the United States (in further text referred as the Commission). Because of poor collaboration there was no link between domestic and foreign intelligence, in other words between the CIA and the FBI, and the 9/11 planning fell into this void. In summary, the FBI looked for sleepers’ cells and the CIA were focused overseas for possible threats to the US. Nobody focused on the foreign threat to domestic targets and therefore ignored foreigners who had infiltrated and then attacked the US.[15] As a response, David M. Walker recommended stronger information sharing and to adopt the ‘need-to-share’ rather than the above-mentioned ‘need-to-know’ principle.[16]

Dissemination is one of the conditions for a successful implementation of the ‘need-to-share’ principle, but it still presents quite a significant challenge to intelligence agencies. It can cause intelligence failure if accurate and relevant intelligence warnings are not delivered to policymakers in a timely manner. The most common error at this stage is communication failure. For example, in the Pearl Harbour case it was discovered that although an intelligence warning was written about the Japanese attack, there was a communications failure as the telegram arrived too late.[17] Today finished intelligence is normally not sent by telegrams anymore, but via secret computer network, which presents the intelligence community’s nervous system. Unfortunately, having a fast and secret network does not guarantee that data will be actually delivered to policymakers in a timely manner. National secret networks can be down because of normal system and power failures or cyber attacks, which are often part of an adversary’s Computer Network Operations (CNO). They have a significant impact on national security, especially during crisis or war, as they thwart the transfer of information to policymakers or military commanders. Such situations are likely to lead to an intelligence failure, especially if the intelligence community did not plan an alternative means of dissemination. The problems discussed in the previous section also provide another reason for computer system related errors. Melvin Goodman noted that before the 9/11 terrorist attacks, the lack of collaboration and information sharing meant there were no computer links between agencies. The State Department’s computers were not linked to the CIA, the FBI or the Immigration and Naturalization Service and as a result, a joint network was nonexistent.[18] By the end of 2008, there was still no progress on that issue and this will be discussed later in the text.

A failure to identify all new threats is unavoidable for intelligence agencies. In general, intelligence agencies have always had a problem identifying new threats and therefore are unable to implement preventive measures. Responding mainly to intelligence surprises, they are consequently always behind their adversaries. Reasons for this originates from the lack of collaborative strategic analysis on potential future threats, poor collection planning, limited financial and personnel resources, over-tasked analysts with current intelligence, complex and enormous intelligence community, and lack of policymakers’ interest. For example, before September 2001, terrorism had not been the highest priority of the US intelligence community, despite intelligence warnings. It is not known how much of this was due to an intelligence or policy failure, yet when the terrorist attacks actually happened, both were surprised and reacted immediately.

Intelligence failures are more likely to happen in the intelligence community, which has a broader area of responsibilities. Bigger countries, such as the US and United Kingdom, have national interests all over the world. Although their intelligence agencies have to be focused on foreign threats, they cannot neglect domestic threats. In addition to all of the above-mentioned analytical and institutional problems, an intelligence agencies’ wider area of responsibility is often due to policymakers’ demanding to be informed about everything and everyone. This aggravates agencies’ efforts to be focused on more probable threats and constantly monitor them. Because of policy, it is very likely that relevant information about imminent threats will be overlooked and intelligence failures will occur, especially when successful adversary’s denial and deception are implemented.

Denial and deception have an adverse impact not only on agencies, but also on its elements, such as intelligence collectors, analysts, managers, and elements outside the agencies such as the public, media, and policymakers.[19] In general, democracies employ denial and deception mostly during wartime, authoritarian regimes use it as a regular instrument of governance, and for non-state actors, such as criminals, rebels, and terrorists, denial becomes a part their modus operandi.[20] The problem for analysts is that they are never sure when an adversary has actually used it or not, and therefore ambiguity is even higher. The denial and deception threat cannot be eliminated, but it can be improved by increased awareness.

After the 9/11 terrorist attacks in the US, politicization of intelligence or subordination of intelligence to policy, has emerged as an area of central concern as it presents a significant and often unavoidable reason for intelligence and policy failures.[21] It is well documented that policymakers’ influence intelligence, but is very difficult to prove it, especially when a Committee or Commission is not interested in exploring the question of executive responsibility.[22] A reason for this lack of interest is likely due to the executive often appointing members of the investigative body, who report to the executive, and hold membership at the pleasure of the executive. At least this was the case in UK, where the Intelligence and Security Committee had no interest to explore the politicization of intelligence in the case of pre-war UK intelligence on Iraq’s Weapons of Mass Destruction (WMDs).[23]

Can reforms prevent future intelligence failures and surprises? The Orthodox School is pessimistic and supports an opinion that intelligence agencies are unlikely to detect surprise attacks due to inherent limits in the intelligence process and that there is no real ‘fix’ of intelligence. It is worth noting that Orthodox scholars believe any procedural or organizational adjustment will not eliminate errors associated with international uncertainty, normal human psychology and organizational life. Conversely, the Revisionist School is more optimistic and believe the most important factors, which lead to intelligence failures, can be addressed by reform, as intelligence failures occur due to the broken intelligence system.[24] In agreement with the latter, Stephen Marrin noted, ‘even if intelligence failure and corresponding policymaker surprise are inevitable, the process of intelligence collection and analysis can and should be reformed and refined to improve its product on the margins’.[25] Nevertheless, the below mentioned ascertainment about a ‘success’ of the US intelligence community reforms actually speaks more in favour of the Orthodox School. In November 2008 Edward Maguire, Inspector General in the Office of the Director of National Intelligence (ODNI) issued a report which revealed that the US intelligence community still has the same problems which had probably resulted in the biggest surprise attack on US soil.[26] Despite an increase in the US intelligence budget by fifty percent, which is now $45 billion, and a signed legislation in 2005, which should have ended ‘turf’ battles, there were no significant improvements in the US intelligence community.[27] It is evident from the report that the culture of protecting information remains a problem. The result is almost no sharing of information or collaboration among agencies. Inter-agency rivalry and suspicion are still present and consequently agencies continue to control and limit access to information and products essential for analysis. In addition, the US intelligence community has 100.000 personnel and the information technology systems that should link those personnel are largely disconnected, incompatible and still vulnerable.[28] When discussing bureaucracy and evidence, the report confirms it is impossible to fight it. According to the report, the ODNI is just an additional layer of bureaucracy, on top of the earlier system, with no contribution to its efficiency, as the Director of National Intelligence spends too much time briefing the President rather than managing the intelligence community. To summarize, during the last seven years, reforms within the intelligence community have been unable to solve the many problems it had prior to the 9/11 attack in the US. As Goodman noted, ‘the September 11 attacks exposed the inability of analysts and agents to perform strategic analysis, challenge flawed assumptions and share sensitive secrets’ and it looks like everything is still the same.[29]

In conclusion, occasional intelligence failures are inevitable and natural. They mainly occur because of analysts’ cognitive limitations, permanent anomalies of an intelligence system and policymakers’ direct and indirect involvement in the intelligence process. Analytical and institutional failures provide the reasons why intelligence failures will definitely occur in the future. Intelligence failures can occur at each stage of the intelligence process, known also as intelligence cycle, but most likely during analysis. An analyst failure is due to their cognitive thinking, ambiguity and ambivalence of collected data, personal interests in writing reports subjectively and trying to influence or please policymakers. An intelligence organization or system is responsible for failures due to a lack of collaboration and sharing of information, forming new levels of supervisors and coordinators, decentralized and incompatible computer system and its failures, and the complexity and enormity of intelligence systems. Intelligence failures are likely to occur more often within the intelligence community, where the area of responsibilities is broader. Dissemination of intelligence can cause intelligence failure if accurate and relevant intelligence warnings are not delivered to policymakers in a timely manner. Failure to identify new threats is unavoidable for intelligence agencies, due mainly to a lack of collaborative strategic analysis on potential future threats, poor collection planning, limited financial and personnel resources, over-tasked analysts with current intelligence, complex and enormous intelligence community, and a lack of policymakers’ interest. Deception and denial have an impact on intelligence collectors, analysts, managers, and elements outside the agencies, such as the public, media, and policymakers, as they are unaware when they have been used to influence them. Intelligence professionals or policymakers due to their personal interests and benefits cause politicization of intelligence and this makes final intelligence products biased or simply inaccurate.

Only radical reforms jointly supported by policymakers and the intelligence community have the possibility of improving the ratio of success to failure, yet they will not exclude failures completely. The authority would have to encourage professionalism among policymakers, intelligence managers and analysts, and try to eliminate or at least reduce competitiveness among agencies, which is the main obstacle for better cooperation. Moreover, an establishment of joint and compatible computer system networks could improve dissemination and information sharing, which should be based on the need-to-share principle. Dissemination should have alternative ways for the distribution of agency’s intelligence products and a system, which confirms that consumers have received the product. Reforms can be successful, first, when the intelligence community is simple and effective, and second, when control, coordination and oversight can be easily established. As improving intelligence requires an objective evaluation of intelligence and at the same time readiness of intelligence officers and policymakers to accept critics and reforms, it represents a Sisyphean challenge for reformers. Therefore, the probability that intelligence reforms would succeed is very low, as there is usually not enough resolve among policymakers and intelligence professionalism because of their personal, bureaucratic and political interests. Consequently, the latter ensures that intelligence failures will occur in the future.

Bibliography

Books

Gill Peter, Phythian Mark . Intelligence in an Insecure World. Cambridge: Polity Press, 2008.

Herman, Michael. Intelligence Power in Peace and War. Cambridge: Cambridge University Press, 2007.

Heuer, Richard J. . Psychology of Intelligence Analysis. Washington, DC: Center for the Study of Intelligence, 1999.

Lowenthal, Mark M. Intelligence from Secret to Policy. Washington D.C.: CQ Press, 2009.

Shulsky, Abram N. , Schmitt, Gary J. . Silent Warfare - Understanding the World of Intelligence. Washington D.C.: Potomac Books Inc., 2002.

Journals

Betts, Richard K. ‘Analysis, War, and Decision: Why Intelligence Failures Are inevitable.’ World Politics, 1978: 61-89.

Godson Roy and Writz James J. . ‘Strategic Denial and Deception.’ International Journal of Intelligence and CounterIntelligence,2000: 424-437.

Goodman, Melvin A. ‘9/11: The Failure of Strategic Intelligence.’ Intelligence and National Security, 2003: 59-71.

Heuer, Richard J. ‘Limits of Intelligence Analysis.’ Orbis - A Journal of World Affairs, 2005: 75-94.

Honig, Or. ‘Surprise Attacks—Are They Inevitable? Moving Beyond the Orthodox-Revisionist Dichotomy.’ Security Studies, 2008: 72-106.

Johnson, Loch K. ‘Analysis for a new age.’ Intelligence and National Security, 1996: 657-671.

Lefebvre, Stéphane. ‘A Look at Intelligence Analysis.’ International Journal of Intelligence and CounterIntelligence, 2004: 231-264.

Marrin, Stephen. ‘Preventing Intelligence Failures by Learning from the Past.’ International Journal of Intelligence and CounterIntelligence, 2004: 655-672.

Internet sources

Buncombe, Andrew. ‘At 8.46am on 11 September, the US became a nation transformed.’ Website: The Independent World. July 23, 2004, http://www.independent.co.uk/news/world/americas/at-846am-on-11-september-the-us-became-a-nation-transformed-756027.html, [accessed April 19, 2009].

Maguire, Edward. ‘Critical Intelligence Community Management Challenges.’ Website: Office of Director of National Intelligence (ODNI), November 12, 2008. http://www.fas.org/irp/news/2009/04/odni-ig-1108.pdf, [accessed April 9, 2009].

Parry, Robert. ‘Another Bush Intelligence Failure.’ Website: consortiumnews.com. Apr 6, 2009. http://www.consortiumnews.com/2009/040609.html, [accessed April 9, 2009].

Pence, Mike. ‘Text From the Congressional Record.’ Website: C-Span Congressional Chronicle November 3, 2004. http://www.c-spanarchives.org/congress/?q=node/77531&id=6609442, [accessed April 9, 2009]

‘The 9/11 Comission Report.’ Website: The National Commission on Terrorist Attacks Upon the United States. 2004. www.9-11commission.gov/report/911Report.pdf, [accessed April 11, 2009].

Travers, Russ. ‘The Coming Intelligence Failure: A Blueprint For Survival.’ Website: Central Intelligence Agency (CIA). 2001. https://www.cia.gov/library/center-for-the-study-of-intelligence/csi-publications/csi-studies/studies/97unclass/failure.html, [accessed March 15, 2009].

Walker, David M. ‘Reorganization, Transformation, and Information Sharing.’ Website: The U.S. Government Accountability Office (GAO). 2004. http://www.gao.gov/new.items/d041033t.pdf [accessed April 11, 2009].

---

[1] Travers, R. (2001), p. 35

[2] Heuer (2005), p. 76, Betts (1978), p. 88

[3] Lefebvre (2004), p. 233, and Johnson (1996), p. 25

[4] Gill and Phythian (2008), p. 113

[5] Johnson (1996), p. 25

[6] Herman (2007), p. 227

[7] Travers (2001), p.35

[8] Lefebvre (2004), p.235

[9] Heuer (1999), p.1, Heuer (2005), p.76-86

[10] Lefebvre (2004), p.233-234, Shulsky and Schmitt (2002), p.65-69

[11] Betts (1978), p.69-71

[12] Herman (2007), p.233

[13] Buncombe (2004), Pence (2004)

[14] Goodman (2003), p. 64

[15] 9/11 Commission Report (2004), p. 277; Gill and Phythian (2008), p. 119-120

[16] Walker (2004), p. 5. United Stated Government Accountability Office and David M. Walker, Comptroller General of the United States, were tasked to prepare recommendations for reorganization, transformation and information sharing based on the Commission Report.

[17] Herman (2007), p. 225

[18] Goodman (2003), p. 68

[19] Denial and deception is a combination of information operations, where an adversary with denial blocks correct information that can be used by ‘us’, and with deception causes that ‘we’ believe something that is not true. Denial and deception is successful when the deceiver has succeed that the target believes a ‘cover story’ rather than the truth. It depends on an adversary how often it will use denial and deception. Source: Godson and Writz (2000), p.427-437

[20] Godson and Writz (2000), p.427-437

[21] Politicization of intelligence or subordination of intelligence to policy can emerge in both, overt and subtle ways by policymakers or intelligence professionals, ‘when [the latter] allow bureaucratic or personal incentives to influence their estimates and reports.’ In some cases, policymakers exert so much pressure on intelligence analysts, usually indirectly with questioning the report so many times, that the analysts finally write it in the way that will please policymakers. Analysts can also initiate distortions in order to curry favour with policymakers or by their attempts to influence policy. Source: Johnson (2007), p. 144 and Turner (2005), p. 112

[22] Lowenthal (2009), p. 140-143

[23] Gill and Phythian (2008), p. 92-96

[24] Honig (2008), p. 76-99

[25] Marrin (2004), p. 658

[26] Maguire (2008); The report has a title ‘Critical Intelligence Community Management Challenges’

[27] Parry (2009)

[28] Maguire (2008)

[29] Goodman (2003), p.62