The politics of intelligence failures: power, rationality, and the intelligence process

ABSTRACT

This article develops a new approach to analysing intelligence failures. Rather than looking for their causes, intelligence failures are here analysed as part of a politics seeking to reify the value of rationality and the taming of power. To analyse this politics, the article draws on Bent Flyvbjerg’s notion of an asymmetrical relation of power/rationality, according to which power has a productive role that is inseparable from claims to rationality. The asymmetrical relation of power/rationality is used in order to challenge the instrumentalist language that pervades much of the literature on intelligence failures and what can be learned from them.

Introduction

The subject of intelligence failures lies at the heart of intelligence studies, providing a crucial link between academic research, policy, and practice.¹ By studying the causes of such failures, researchers are said to be able to contribute to the development of intelligence work and give advice on how to avoid future failures.²

While notoriously complex and highly dependent on specific contexts, in a more general sense, intelligence failures are commonly linked to the intrusion of power into the intelligence process. Intelligence, in this sense, rarely if ever works as a purely rational process in the way it ideally should. This process – including the key stages of collection, analysis, and dissemination – risks getting distorted as interests other than purely instrumental ones interfere. Power is, in this sense, almost exclusively seen as a problem in intelligence studies. As such it is commonly opposed to rationality, which provides the basis for various solutions. On this view, while a greater degree of rationality can help improve the various stages of the intelligence process, as well as the connections between them, power corrupts, distorts, and contaminates.³

The view of power as problem and rationality as solution permeates much of the intelligence studies literature, not only that which directly concerns intelligence failures. As such it also offers a useful starting point for interrogating the subject of intelligence more broadly, by questioning a narrow yet dominant instrumentalist take on what intelligence is all about. In particular, it can be used as a springboard for problematizing attempts at reifying ideas about the very possibility of a fully rational intelligence process. As this article will go on to demonstrate, explanations of intelligence failures subscribing to the need of improving this process through increased rationality and the lessening of power play a key role in such reifications. Moreover, these explanations tend to reaffirm the researcher’s task as being one primarily of contributing to the general effectiveness and fine-tuning of the intelligence process.⁴ This is problematic, at least if the aim is to pursue a more radical line of questioning that does not simply rehearse old wisdoms but opens up new space for thinking about intelligence. Hence, rather than taking the instrumentalist means-end logic of intelligence as a given, this article seeks to make that logic into an object of study and critique. It does so by analysing intelligence failures as part of a politics.

My analysis of the politics of intelligence failures draws on Bent Flyvbjerg’s notion of an asymmetrical relation between power and rationality.⁵ In brief, according to Flyvbjerg, this asymmetrical relation means that power has a constitutive function that shapes the meaning of rationality. Moreover, it implies that rationality has no value on its own, independent of the contexts in which rationality, or the lack thereof, is ascribed to decisions and events. To appreciate the significance of these contexts for how something comes to be seen as rational, Flyvbjerg argues, one has to understand the productive role of power. Rationality is thus seen as context-dependent rather than context-independent, and the primary context of rationality is power. Crucially, this asymmetrical relation differs from a more conventional understanding of rationality and power as symmetrically interlinked. The symmetrical relation implies that power and rationality can be negotiated as two equal sides, for example by drawing on rational thinking and methods to advance more efficient ways of using power to achieve certain end goals. It is precisely this instrumentalist take on power and rationality, and how it pertains to the intelligence process that this article calls into question by drawing on Flyvbjerg’s understanding of power/rationality asymmetry.

To explore how power and rationality can be seen as asymmetrically related, the article turns to the example of the US 2002 National Intelligence Estimate (NIE) on Iraq’s WMD programme leading up to the second Gulf War. While much has been written on this case, it remains useful because of the diverse accounts that exist of it, not least regarding the role of power in the intelligence process. For the purposes of this article, these diverging accounts of power become interesting to look closer at because they show, not only how power plays a constitutive role in shaping the meaning of rational behaviour as the process unfolds, but also how attempts are made to in hindsight rationalize away power’s intrusion in the intelligence process as mere aberration. Such attempts, it is noted, even if trying to divorce power from rationality can also be seen as an effect of power. Moreover, these attempts play a key role in reifying a very particular image of the intelligence process as predicated on the notion of a fully rational process that, at least potentially, is freed from power. Relying on such a narrow understanding of power and how it relates to rationality, it is argued, should not be seen as an objective representation of intelligence, but rather as an integral part of a politics that aims to reinsert belief in a particular way of thinking about what the intelligence process is and should be.

By adding the asymmetrical reading of power/rationality to the symmetrical reading, this article’s main contribution consists of providing an alternative take on both intelligence failures and the intelligence process. Doing so is important, considering how often power is blamed for intelligence failures, yet without elaborating upon the meaning and role of power. It is also important in order to challenge the widespread assumption in intelligence studies that power can simply be ‘used’ in more or less rational ways, which suggests that the solutions to the problems that emerge in light of intelligence failures are more straightforward than they probably are. Hence, this article is guided by the suspicion that the problem of power runs deeper and is much harder to grapple with than ideas about improving the effectiveness of the intelligence process by increasing the level of rationality suggest. A first step in fully appreciating this more fundamental problem of power is to advance a more sophisticated understanding of the productive role power has in the intelligence process. In this article I do this, then, partly by drawing on Flyvbjerg’s analysis of power/rationality asymmetry, and partly by turning to the case of the US intelligence failure on Iraq’s WMD programmes. In relation to the latter, crucially, my ambition is not to add a new account or a new explanation. Rather, I seek to draw on already existing accounts in order to demonstrate how there are, indeed, different ways of thinking about what constitutes an intelligence failure, and, most importantly, what they reveal about the power/rationality interplay.

The article begins by giving an account of prevalent understandings of intelligence failures and the instrumentalist language these understandings often rest on. The second section looks at the relationship between power and rationality, contrasting symmetrical and asymmetrical readings. The article then explores how the asymmetry between power and rationality plays out, first in relation to accounts of the intelligence process as it unfolds, in section three, and, in section four, by considering attempts to rationalize away power’s intrusion and thereby reify the image of an intelligence process supposedly free from power. The conclusion discusses some of the main implications of my analysis of the politics of intelligence failures, including the question of what it means for the idea of ‘learning’ from such failures.

The instrumentalist language of intelligence

Why do intelligence failures occupy such a central role in intelligence studies? One obvious reason would be that they reveal the high stakes of intelligence work, and the potentially severe consequences of getting it wrong, as illustrated by some of the most commonly used examples: Pearl Harbor, 9/11, and intelligence on Iraq’s WMD programmes prior to the 2003 US-led invasion. Another possible reason relates to the limits of intelligence as a form of knowledge, as it focuses on estimates rather than facts. The commonly used quote by General Michael Hayden offers the most succinct definition of intelligence in this respect: ‘If it were a fact, it wouldn’t be intelligence’.⁶ Given that intelligence is not about facts, and more about uncertainties and ambiguities, which require assessments and estimates, making sense of failures is difficult. It also points to the difficulty of preventing such failures from happening in the future, simply because intelligence is not a science making definite predictions.⁷

If the future is always uncertain, and if intelligence work deals with probabilities rather than certainties, speaking of ‘failures’ might not be very helpful in the first place. And yet the topic of intelligence failures has been central to the advancement of intelligence studies, and continues to be a subject that much of the field revolves around. In a very general sense, it does so by relying on what can be described as an instrumentalist language of intelligence. This language goes back to Sherman Kent’s seminal works, in particular his Strategic Intelligence for American World Policy (1949). For Kent, ‘intelligence is not knowledge for knowledge’s sake alone … but for the practical matter of taking action’.⁸ Knowledge, in this respect, must be guided by a ‘closer approximation to truth’.⁹ Such approximation is attainable through the use of reason and the scientific method: ‘the two instruments by which western man has, since Aristotle, steadily enlarged his horizon of knowledge’.¹⁰ Intelligence scholars are sometimes reluctant to use the word ‘truth’, preferring instead the seemingly less absolute term ‘proximate reality’.¹¹ Nevertheless, it is the aspiration to truth, or reality, that provides a key tenet of the instrumentalist understanding of intelligence.

The instrumentalist means-end logic of Kent’s early work, including his views of the value of social science methods for improving intelligence, has had an enormous impact on intelligence studies. This is evident, for example, in Robert Jervis’ work on intelligence failures, and his search for potential remedies and improved rational assessments.¹² For Jervis, in order to get ‘closer to the truth’,¹³ it is, for example, important to look not just for positive but also negative evidence that might disprove a certain hypothesis. Furthermore, one has to become aware of the psychology behind the inferences made, in order to determine whether there were factors at play other than purely rational ones, which may have contributed to the making of a particular assessment.¹⁴

The idea of improving the intelligence process, and making it more rational, is a recurrent theme in the literature on intelligence failures. Even if the ‘failure’ is linked to a specific end product, which is decoupled from the process of producing it, insights about the failure can still feed back into the continuous development of the process. Whether linked to a simplified image of a ‘cycle’ that ties together the ways in which ‘information is acquired, converted into intelligence, and made available to policymakers’,¹⁵ or to something more complex,¹⁶ references to a process make it easier to think about how intelligence ideally should be organized, but also to detect possible errors by looking for deviations from the process.

The point of referring to a process, then, is to have some sort of model or idealized image to fall back on. There is a process in which trust can be placed, and the main goal for everyone involved is to make it work as smoothly as possible. Doing the latter means, also, that the participants are able to play their part. As Loch Johnson puts it, what intelligence ‘largely boils down to’, in this way, ‘is the quality and matching of people charged with making the machinery work’.¹⁷ By contributing to the machinery and making it work, the people, or the subjects of intelligence, help serving the bigger purpose of national security. Deviating from the process means, not only that intelligence might ‘fail’, but also that national security is jeopardized. The importance of highlighting flaws in the process and making it work more efficiently thus becomes a matter of national security, which the intelligence process is ultimately supposed to serve.¹⁸ While acting in accordance with the prescribed process increases the likelihood of success, deviating from it increases the risk of failures. As Mark Jensen puts it: ‘Naturally when the IC executes its prescribed processes as flawlessly as possible, it makes a major effort towards reducing the number of potential failures’.¹⁹ Hence, learning from failures becomes an important task for intelligence agencies as well for national security more broadly. As Stephen Marrin observes:

The identification of causes of past failure leads to kernels of wisdom in the form of process modifications that could make the intelligence product more useful. A more effective, more accurate intelligence capability may still be vulnerable to the cognitive and institutional pathologies that cause failure, but a self-conscious and rigorous program based on the lessons derived from the existing literature would strengthen the intelligence product. This might lead to greater policymaker ability to respond to challenges, and thereby contribute to the national security of the United States.²⁰

Given that intelligence is not about facts, but more about uncertainties and ambiguities, which require assessments and estimates, making sense of failures can, however, be tricky. The discrepancy between facts and intelligence means that there is no obvious procedure for determining what has gone wrong.²¹ Intelligence failures are not like plane crashes, with a black box revealing the causes behind the accident. Moreover, having an estimate that turns out to be inaccurate, does not necessarily mean that the process of producing it was flawed. Often, however, it is the suspicion of wrongdoing in the intelligence process itself that prompts various investigations in the aftermath of a failure.

Explaining intelligence failures may take a variety of factors into consideration. Different explanations will highlight different causes as the most important ones. Yet a recurring theme in the literature on failures is the negative impact of power on the intelligence process. This is clearly illustrated in the many accounts given of the production and consumption of the 2002 US National Intelligence Estimate (NIE) on ‘Iraq’s Continuing Programs for Weapons of Mass Destruction’. Arguably the most common understanding of the failure linked to this NIE relates to the intrusion of power in the form of the Bush administration looking for a particular desired outcome, i.e., one that clearly pointed to Iraq’s development of biological and chemical weapons, to justify the subsequent US-led invasion in 2003.²²

As in the case of the Iraq intelligence failure, power is commonly seen as a problem since it interferes with the intelligence process by contaminating and distorting it. At the same time, as is often pointed out in the intelligence studies literature, the distorting effects of power can never be entirely eradicated. This is a point made, for example, by Richard Betts in his discussion of the perennial ‘enemies of intelligence’. In Betts’ view, the intelligence community (IC) must simply accept the continuous intrusion of power and, in response, find an appropriate balance between politicization and depoliticization.²³ While power contaminates the intelligence process, the remedy is to make the process less politicized, which for Joshua Rovner means, using ‘information objectively in order to calibrate means and ends’.²⁴

Understanding power in negative terms suggests that there is a potential even if not realistic chance of developing a fully rational intelligence process. Along these lines, Peter Gill also notes how ‘intelligence provides the basis for policy or decisions; if people, organizations and states are to act “rationally”, they will do so after canvassing fully the costs and benefits of the alternative courses of action open to them’.²⁵ The potential to act rationally on the basis of intelligence is difficult to realize, however, when those in possession of power use intelligence to prevent, compel, or persuade others to act in a certain way that serves their specific interests.²⁶ In other words, the intrusion of power into intelligence provides a key hurdle that needs to be overcome in order to develop an intelligence process that provides a solid basis for rational decisions.

The problem of power is in this way closely linked to the harmful effects of politicization, the term most commonly used to describe how intelligence as something objective and neutral gets distorted. This may happen through either direct or indirect manipulation by policymakers, or, reversely, through the subversion of policymaking by intelligence professionals. Jensen refers to politicization as ‘the practice of intelligence professionals bending intelligence to meet decision-maker needs or decision-makers focusing on selected intelligence products or passages thereof for their own political purposes …’²⁷

Even if politicization is commonly seen as a harmful manifestation of power interfering in the intelligence process, not everyone agrees it should be seen as unequivocally ‘bad’. Former Director of Central Intelligence, Robert Gates is often referred to as someone who advocates a closer link between the IC and policymakers.²⁸ And Nathan Woodard suggests that there is a form of ‘positive politicization’, which relates to how intelligence practitioners, rather than always striving to remain ‘neutral’, should also be encouraged to take a stand when presenting their findings and making policy recommendations.²⁹

Irrespective of whether politicization is seen as ‘good’ or ‘bad’, the different views of its role in the intelligence process tend to confirm the instrumentalist language. Insofar as politicization is seen as an expression of the problem of power’s intrusion into the intelligence process, the remedy is either to temper it through depoliticization or to politicize more effectively. The ‘problem’ of power is in this way not linked to power per se but rather to the ways in which it is used. Power, it thus seems, can be used in more or less ‘rational’ ways. To be used rationally, power has to be mobilized in order to generate a successful outcome of the intelligence process, while the irrational use of power risks leading to failure.

Symmetrical and asymmetrical understandings of power/rationality

The idea of finding more efficient ways of using power, or a more effective balance between politicization and depoliticization, rests on the assumption of a symmetrical relationship between rationality/power. According to this assumption, power and rationality can be said to represent two different sides that are negotiated in relation to one another. While power distorts and contaminates, rational thinking and methods can be drawn on to temper power’s harmful effects. The intellectual origins of this symmetrical view can be traced back to the Enlightenment tradition, and specifically the Enlightenment notion of rationality as something ‘context-independent’.³⁰ On this view, the meaning of rationality does not change depending on the context in which it is used; its meaning essentially remains the same regardless of context. As Flyvbjerg puts it, there is in this way a common assumption underpinning many references to rationality, namely that: ‘We know what rationality is, and rationality is supposed to be constant over time and place’.³¹ Most often, rationality is linked in this sense to the use of a priori concepts in the production of knowledge. There is, thus, a supposedly rational way of producing new knowledge that is independent of the context in which the knowledge production takes place. As such, rationality can also be seen as a remedy to power, by applying rational thinking and methods on the use of power, thereby taming the latter from a position that is supposedly free from power.

Against the symmetrical understanding of power/rationality, Flyvbjerg proposes an asymmetrical reading. By this he means that rationality should be understood as context-dependent rather than context-independent, and, crucially, that the primary context on which the meaning of rationality depends is power. Based on his case study of city planning in the Danish town of Aalborg, and his reading of key thinkers such as Michel Foucault and Friedrich Nietzsche, Flyvbjerg shows how the workings of power trump what could be considered more rational solutions and decision-making. Using Aalborg as a ‘laboratory’ for investigating how democracy works in practice, he demonstrates how rationality often comes up short against the workings of power as a number of key decisions are made.³² Rationality per se constantly proves insufficient when trying to give an account of how decisions were made as part of the planning process. Accordingly, power and rationality do not seem to constitute two equal and comparable sides that can be negotiated in relation to one another. Moreover, increased knowledge does not seem to generate more power, or more rational ways of using power.³³ Rather, power produces what is meant by rationality, and determines the value of knowledge.

To understand Flyvbjerg’s notion of an asymmetrical relation between power and rationality, and appreciate its relevance for thinking about intelligence, two assumptions in particular are important. The first relates to how ideas about what is rational and irrational, in any given context, have to be produced rather used instrumentally. Rationality is therefore nothing per se, whether linked to a priori concepts or a specific method of thinking along, for example, Cartesian lines. Rationality should rather be seen as an integral part and/or as an outcome of the process of rationalization, which ascribes reason, or the lack thereof, to human action and decision-making. The second assumption relates to the actors who make these decisions and participate in the intelligence process. Instead of looking at how these actors can draw on and implement the objective use of reason in order to arrive at what can be regarded as the most rational decisions, emphasis shifts to how subjectivity is produced in relation to their prescribed roles in the process. Moreover, it is important to look for potential conflicts between the roles they are supposed to fill on one hand, and on the other hand their concrete experiences of being entangled in relations of power. Along Foucauldian lines, these relations are seen as inseparable from the formation of subjects. Hence, rather than being something a subject possesses, power can be said to generate subjects and bring them into being in specific contexts.³⁴ Those contexts will determine the conditions under which ‘rational’ decisions can be taken and, indeed, what is meant by ‘rationality’ in the first place.

Discourse and context

On the basis of the asymmetrical understanding of power/rationality, rationality is thus seen as context-dependent rather than context-independent. This means that power cannot be ‘used’ in more or less rational ways, simply because there is no way of applying rational thinking and methods beyond or independently of power. The asymmetrical reading thus goes against the instrumentalist take on both intelligence failures and intelligence studies more broadly. It can even be said to run contrary to some of the assumptions underpinning ‘critical’ approaches to intelligence studies that stress the importance of historical and cultural contexts, as well as the role of discourse in the production of knowledge.³⁵ For example, according to Hamilton Bean, discourse can be seen as a bridge between knowledge and power, as objects and practices are ascribed meaning through discourse in ways that reflect power by privileging some meanings over others.³⁶

Turning to discourse, and looking at how discourses of intelligence stem from specific cultural contexts, may offer a way of coming to terms with some of the negative effects of power. Doing so includes, for example, identifying ‘dysfunctional, disempowering, and inane organizational practices, as well as the antidemocratic influence of intelligence secrecy, centralization, repression, and distortion’.³⁷ Studying intelligence along these lines can thus be said to enable a greater degree of sensitivity to the role of context, as well as a more ‘reflexive’ approach to studying intelligence. According to Peter de Werd: ‘Reflexivism aims to expose the institutionalized practices and socialized assumptions such as worldviews, thought styles or paradigms that shape the understanding of what to look for and hence what data and information is found’.³⁸

While a reflexivist approach encourages increased awareness of the contexts in which new knowledge is produced, it also stays rather close to the instrumentalist language of intelligence. This is because it tends to argue in favour of drawing on discourse and reflexivity as tools to be used, in an instrumental way. For example, as pointed out by de Werd: ‘Actors need to be reflexively aware of the dynamics between various knowledge regimes that constitute the broader social system or security ecosystem (for example, of a nation state)’.³⁹ Or, according to Bean, ‘critical/cultural perspectives are uniquely suited to the study of intelligence’, since they ‘explain how groups can express shared knowledge in ways that do not presume access to empirical “truth” in the face of ambiguity and uncertainty. Instead, these perspectives encourage adequate deliberation among stakeholders to ensure findings and guide action’.⁴⁰ Increasing awareness of how knowledge is produced with respect to cultural and historical contexts, or in relation to different ‘regimes of knowledge’, thus seems to hold the promise of providing better guidance for actors when navigating the challenges of intelligence work.

The key difference, perhaps, between these reflexive, and critical/cultural approaches, and the asymmetrical understanding of power/rationality drawn on in this article concerns the status of the actor, or, as I prefer to put it, the subject. In this respect, the former approaches move ambiguously between, on the one hand, the assumption of an already present actor who can use his or her reflexive skills in more or less effective ways, and, on the other hand, the subject as constituted through discourse. Of these two assumptions, the notion of an asymmetrical relation between power and rationality rests on the latter. Hence, the subject’s ‘presence’ is seen as possible, only in relation to a particular context, which is inseparable from power. For the purpose of this article – rethinking intelligence failures as a politics that challenges the traditional instrumentalist language of intelligence – the difference between these two assumptions is important. This is because in order to go beyond the instrumentalist language, the subjects of intelligence have to be seen as entangled in power relations, not as operating in a space that is seen, even if somewhat ambiguously, as prior or external to power. Hence it also requires taking seriously Foucault’s understanding of the ‘productive aspect of power’.⁴¹ Power, on this view, has to be analysed as primary and constitutive in relation to both knowledge and subjectivity. As such it permeates society and is ‘rooted deep in the social nexus’.⁴² Power is thus understood in relational terms, as intertwined with the meaning of objects and becoming of subjects, who undergo continuous change and transformation as the relations of power play out.⁴³

In accordance with Foucault’s understanding of the productive aspect of power, the importance of discourse and context cannot be explained by how the actors ‘use’ them as tools, in more or less instrumental ways, for example by becoming more aware of the cultural and historical context of a specific case. Rationality loses its instrumental role in this way, and is better grasped as part of a politics that produces subjects and reifies certain ideals linked to the production of knowledge. As will be demonstrated later in this article, the politics of reifying such ideals becomes especially evident in attempts to rationalize away the intrusion of power and thereby find ways of explaining why power and politicization should not be seen as a serious concern for intelligence work. Before looking at such attempts, however, it is useful to consider some examples of power’s intrusion into the intelligence process.

The intrusion of power

As mentioned in the introduction, while plenty has been written on the Iraq intelligence failure, it remains useful to consider precisely because of this voluminous literature. It means that the same ‘case’ can be linked to different accounts of what has happened, which in turn highlight different aspects of the productive role of power. Some of the most palpable examples of power’s intrusion in this specific case are found in personal accounts related to the subjects involved in the intelligence process. It is especially interesting to note here the clear sense of frustration expressed at being involved in a process that seems anything but rational. This is the case, for example, in the memoirs of Tyler Drumheller, Chief of the CIA European division at the time of the Iraq NIE. Drumheller writes:

I watched my staff being shot down in flames as they tried to put forward their view that Saddam Hussein had no weapons of mass destruction. I watched as politicians, using intelligence that was at best questionable, spoke passionately about the impending threat of nuclear attack by Iraq before we sent our men and women out to die in the Middle East. I railed inwardly at the stubbornness of this country’s leadership as it ignored voice after voice that warned of the perils of an ill-prepared war in the Arab world. Eventually I had to accept that nothing we said or did was going to change the administration’s collective mind.⁴⁴

In Drumheller’s account, the intelligence process is permeated by power in a way that makes it out of reach for rational thinking and methods to be invoked as the basis of solutions. When being prevented from developing and implementing such solutions, there is a clear risk that the subjects start losing faith in the process. With Drumheller, this is precisely what seems to happen when his colleague and friend ‘Bill’, who had heard from his European contacts that they had a senior source inside Saddam’s government, tries to bring this new information to his superiors. According to Drumheller:

As intelligence officers, he [Bill] and I wanted to know what the Iraqi knew. He was the closest thing anyone had to a solid source in Baghdad, and our instincts told us we should use him to the maximum to find out what was going on inside Saddam’s regime. Instead, we were forced to wait while officials at the CIA, and by extension the administration, debated whether it was worth pursuing him, even though his information was of immeasurable consequence to the Iraqis, our allies, and us.⁴⁵

If simply being ignored is one effect of power, being disciplined when the ‘right’ intelligence is not produced and delivered is another. The disciplining effect of power is well illustrated in Bob Drogin’s account of Curveball, the Iraqi defector who supplied the German BND with fabricated intelligence on Iraq’s chemical and biological weapons programme. The flawed intelligence collected from interviews with Curveball was shared with the US IC and played a key role in Colin Powell’s infamous UN Security Council speech prior to the invasion in 2003.

The significance of Curveball for the actual decision of going to war is contested. Irrespective of its impact, however, there are interesting elements of this story that point to the workings of power. Crucially, these elements are not simply about the information provided by Curveball, and how it diverged from what was later proved to be the case. They relate, also, to how different kinds of behaviour were rewarded or punished. For example, attempts to cast doubt on the information provided by Curveball, not least after the invasion and in the hunt for actual proof were easily quashed. Consider the fate of ‘Jerry’, a midlevel CIA analyst, and initially a true believer in Curveball and the author of intelligence reports arguing for the existence of bio-trucks as mobile labs. Increasingly, however, Jerry becomes sceptical of Curveball’s credibility. Drogin writes about Jerry: ‘He had only seen what he wanted to see. He had lied to himself. His expectations had blinded him, twisted his logic, distorted the evidence in front of him’.⁴⁶ In what seems an honest attempt to tell ‘the truth’ to his own organization, namely that the problem with the intelligence process was located not in Baghdad but at Langley, Jerry faced a difficult time back at the CIA:

He found someone else sitting at his desk when he came to work, his belongings stuffed in a box. He was reassigned to an office cubicle reserved for visitors, an isolated booth without a classified computer or secure telephone. He was given no work. People barely acknowledged his presence.⁴⁷

Representations of failure concern in this way not just the making of bad decisions on the basis of poor intelligence. They also relate to the creation of spaces and ‘spatial relations’ in which the subjects of the intelligence process are (un)able to interact.⁴⁸ Drogin gives another vivid account of these spaces and how they can work in order to discipline the subjects involved. When the CIA-appointed chief of the Iraq Survey Group, David Kay returns to Langley, his main objective is to fulfil the promise he had given himself when commencing his work: to ‘speak the truth’.⁴⁹ Having previously received whatever resources he deemed necessary to carry out his tasks, back at the CIA headquarters to report on the outcome, his treatment is altogether different:

On previous visits … [Kay] was assigned a parking place in the VIP lot in front of the old headquarters building, or a prime spot in the executive garage. Now the guard directed him to the general CIA parking lot, a brutal hike away from the far side of nowhere on a frigid winter morning. (…) America’s top weapons hunter no longer merited a seventh-floor office down the hall from [Director of Central Intelligence] Tenet and his covey of close aides. Kay now was consigned two floors down to a tiny, windowless office without a secure phone or classified computer. It lay at the end of a distant, deserted corridor undergoing construction. All the nearby offices stood empty, the halls dusty and silent.⁵⁰

Drogin’s account of Kay’s fate points to the importance of space and spatial relations in the intelligence world. These spaces play a crucial role, for example, in determining access to various levels of secrecy, and how subjects may interact. As such they also illustrate clearly the role of power in shaping the intelligence process as it plays out, and how the production of knowledge hinges on a wide range of elements not normally taken into consideration when focusing narrowly on the use of rational thinking and methods.

Power and rationalization

In contrast to the clear and obvious intrusion of power into the intelligence process, accounts provided by Jervis, as well as of George Tenet, Director of Central Intelligence during the production of the NIE, point in a very different direction. According to both Jervis and Tenet, the NIE was not necessarily irrational, even if inaccurate. For example, in his memoirs, Tenet claims that ‘the intelligence process was not disingenuous nor was it influenced by politics. Intelligence professionals did not try to tell policy makers what they wanted to hear, nor did the policy makers lean on us to influence outcomes’.⁵¹ Crucially, taking into account how Iraq had been pursuing WMD programmes in the past it was in fact fully rational to assume they were doing the same again. ‘In retrospect’, as Tenet puts it, ‘we got it wrong partly because the truth was so implausible’.⁵² And as explained by Jervis: ‘It is particularly difficult for analysts to get it right when the truth is implausible’.⁵³ The rational thing for Saddam Hussein to do, one thus could assume, was to develop Iraq’s stockpiles of chemical and biological weapons. Arguably it was also rational, at least from the US perspective, for Saddam to deny the existence of these stockpiles, as part of a deliberate deception strategy. Yet neither of these supposedly ‘rational’ actions was pursued; there were no stockpiles and there was no deception.

What was missing from the intelligence process leading up to the NIE were trustworthy sources and hard evidence. While HUMINT sources played a major role in producing the flawed NIE, much of the intelligence provided by these sources was fabricated. This includes the intelligence coming from Curveball, whose credibility was not properly scrutinized by the US IC. They also failed to consider the different logics and rationalities at play in Iraqi politics and society at large, which could have explained the seemingly ‘irrational’ behaviour of Saddam, and the untrustworthiness of Iraqi defectors such as Curveball. According to Tenet:

We divorced technical analysis from our understanding of Iraqi culture … and this hurt us in central ways. We failed, for example, to factor in how the regime’s harsh treatment of its citizens would make truthful reporting to superiors on the status of weapons programs less likely. We did not fully consider the impact of nearly a decade of international sanctions, UNSCOM inspections, continuous overflights, and US military actions.⁵⁴

By paying closer attention to the role of context, including that of culture and contemporary Iraqi society, a more accurate assessment would have been possible. The problem underlying the Iraq intelligence failure could, in this sense, also have been rather easily fixed. This point is shared by Tenet and Jervis, namely that the reliance on better methods could have produced a different outcome, but also that there was no severe form of politicization affecting the intelligence process. According to Tenet, ‘the intelligence process was not disingenuous nor was it influenced by politics. Intelligence professionals did not try to tell policy makers what they wanted to hear, nor did the policy makers lean on us to influence outcomes’.⁵⁵ While largely agreeing with Tenet’s assessment, Jervis also suggests that there was a general ‘atmosphere that was not conducive to critical analysis, encouraged excessive certainty, and eroded subtleties and nuances’.⁵⁶

Even if the atmosphere of politicization may have coloured the intelligence process, a more ‘rational’ way of handling this process was possible, for example by questioning pre-given assumptions, and turning these assumptions into testable hypotheses. Doing the latter would also have required looking for evidence that could be used in order to disprove a certain hypothesis. The intelligence process could then, potentially at least, have been less susceptible to power. A higher degree of rationality might have had the effect of minimizing the distorting effects of power. The risk of failure would then have decreased by allowing the process to play out in a more rational way, free from the harmful effects of power and politicization.

Determining what in hindsight can be seen as rational and irrational along the lines of Tenet and Jervis presents an opportunity to reinsert belief in the smooth functioning of the intelligence process. The ‘failure’ can then be seen as important for precisely this reason; it acquires value and meaning not primarily by determining the causes behind the failure, but rather by pointing to the necessary adjustments that can work as remedy. In doing the latter, making sense of intelligence failures becomes an important tool for reifying the image of the intelligence process as potentially fully rational and freed from power.

Reifying this image of the intelligence process also entails maintaining a strict separation between policy and analysis, so that ideally there should be policy without analysis, and analysis without policy. Depending on which of these two ‘sides’ that interferes with the other, the failure is not necessarily one of intelligence but can also be seen as one of policy. Yet only rarely is the term intelligence failure replaced by ‘policy failure’. For example, Drumheller, whose memoirs was referred to in the previous section, explains in an interview how: ‘It just sticks in my craw every time I hear them say it’s an intelligence failure … This was a policy failure … I think, over time, people will look back on this and see this is going to be one of the great, I think, policy mistakes of all time’.⁵⁷ The point of stressing the failure of intelligence, or as Jervis does in his book, addressing the question why intelligence fails reproduces the idea that there is intelligence on one hand, and policy making on the other. The ‘failure’ can then be explained by pointing to how one side interferes in the affairs of the other. Even if focus shifts from one side to the other, for example by emphasising policy failure rather than intelligence failure, the separation between them remains.

While the separation of policy and analysis seems rather straightforward when following Tenet’s and Jervis’ representations of a weak form of politicization, it becomes far more difficult to maintain in representations of a stronger form of politicization. In the case of the Iraq NIE there are numerous examples of accounts of excessive interference from policy makers in the production of intelligence. One such example relates to the creation of the Office of Special Plans (OSP), a separate intelligence unit set up within the Pentagon, providing interpretations and analyses that gave support to the assumption of Iraq acquiring WMD:s. Scott Lucas notes in relation to this and other examples how there was an important bureaucratic dimension of politicization that emerged in this context:

This bureaucratic reconfiguration, which amounted to creation of the ‘right’ evidence or interpretation if that was not forthcoming from the Agency and ignoring evidence and interpretation that did not support a desired policy, was rationalized across the Executive Branch.⁵⁸

While the Bush administration’s politicization of intelligence on one hand can be seen as ‘exceptional’, on the other hand ‘the possibility remains that, if a future administration decides to repeat the attempt, it may once again effect the necessary bureaucratic shift in power’.⁵⁹ The possibility of such a bureaucratic shift in power highlights the ways in which politicization can be seen as part of the ‘normal’ functioning of intelligence. Thus, even if the aim of politicizing intelligence for the purposes of regime change is exceptional, the method of fulfilling it can be seen as more mundane with respect to the intelligence-policy interplay. The potential for actualizing as well as rationalizing a power shift along these lines is very much an integral part of that interplay.

When analysing the intrusion of power as something that follows the mundane procedures of the intelligence-policy interplay, it becomes much harder to separate power from rationality. It also becomes more difficult to ‘use’ rationality in order to tame power, or, indeed, to rationalize away the workings of power as a mere aberration. Rationality should indeed be seen as important, but precisely insofar as it is made meaningful in the context of power. Rationality is, in this sense, not harmonized with but subordinated to power. ‘More’ rationality is therefore not going to eradicate the problem of power. Expressing faith in rationality rather perpetuates the illusion that power is a problem that can be resolved at all, and, moreover, that there is such a thing as a fully rational intelligence process that is free from power to begin with.

Whether power is either rationalized away when assessing the outcome of the intelligence process (Tenet and Jervis), or depicted as an integral part of how this process plays out (Drumheller and Drogin), depends on the accounts given and the positions from which those accounts are offered. Hence, these representations can also be seen as expressions of power, more precisely of the interplay of forces that act and react on one another. Drumheller explains in the introduction of his book, how he ‘felt driven to set the record straight’, and thus countering other accounts of events.⁶⁰ He writes: ‘I think it is important that the public knows the games that are being played with the agency at a time when this country needs a highly functioning intelligence community more than ever’.⁶¹ In Drumheller’s view, those accounts did not recognize the serious efforts that were made by him and others who were out in the field trying to provide useful intelligence. Tenet, in his turn, expresses similar motives for writing his autobiography, to counter a dominant narrative that centres on his alleged and infamous remark to President Bush that the intelligence available on Iraq’s WMD was a ‘slam dunk’. Both Tenet’s and Drumheller’s accounts can in this way be seen as symptoms of the asymmetry between power and rationality, and insofar as they contain any expressed beliefs in the possibility of harmony of power/rationality serving a reinvigorated intelligence process, such harmony has to be grasped as conditioned by a prior asymmetry. In this way, these accounts become valuable not by incorporating them within an instrumentalist language of intelligence, in order to objectively determine the actual causes behind the ‘failure’ and then trying to fix the process on the basis of what is found. They are valuable, rather, as illustrations of the asymmetrical relation between power and rationality, hence also of the politics of intelligence failures.

Conclusion

Shifting focus from the symmetrical to the asymmetrical reading of power and rationality means that all parts of the intelligence process, including its subjects, the spaces in which they interact, and the language used to represent the meaning of failures, are permeated by power. On the basis of this asymmetry, the idea of drawing a clear line between ‘intelligence’ as means-end activity, and ‘politics’ as a separate sphere, is deeply problematic. Yet intelligence studies often reproduces that very distinction. Shedding light on this literature and the instrumentalist language that underpins it, the purpose of this article was not to provide better explanations of the causes behind failures, or more accurate definitions of what a failure objectively ‘is’. Rather, the aim was to question the very notion of intelligence failures and the role it plays in reproducing ideas about what intelligence is or should be.

Key to moving away from the instrumentalist language of intelligence is seeing any form of power/rationality symmetry as predicated on a prior asymmetry, rather than the other way around. There is in this way no symmetry to begin with, hence no ideal intelligence process to return to, which means that power is not an aberration that deviates from the symmetrical relation as a prior or superior norm. It also means that there is no failure ‘as such’ or ‘in itself’. There is, only, the politics of intelligence failures. Stressing the role of power is to become aware of how this politics plays out both in discourses on specific ‘failures’, such as the Iraq NIE, as well as with respect to other aspects of intelligence work and practice, including the academic study of intelligence.

Another important implication of my analysis is to focus less on idealized understandings of the intelligence process and the subjects acting (at least potentially) rationally in accordance with it. More focus should be placed both on the social practices of intelligence,⁶² and on the life experiences of its subjects. Biographical accounts of real life experiences say much less about an idealized process and more about the unique life-world of its subjects.⁶³ Rather than relying on an objective notion of rational thinking and decision-making, which reduces these experiences to generalized statements applicable to different contexts, the emphasis here is on what makes them unique in the specific context(s) of intelligence work and practice.

Going from the conventional symmetrical view of the relation between power and rationality to the asymmetrical, it is possible to think of how power permeates different parts of intelligence work. In none of those parts, however, should power be seen as something that is later added on to the intelligence process; power is always already there to begin with. For the same reason, intelligence does not get politicized, but already is. Thinking about intelligence as potentially void of power and as purely rational and instrumental is, therefore, problematic since it obscures the workings of power and gives the impression that there is indeed an intelligence process free from power.

Going beyond the instrumentalist language means taking a step behind knowledge and focusing on power as that which conditions the production of knowledge. It means thinking of any claims to symmetry of power/rationality as predicated on a prior asymmetry. Intelligence failures remain an important subject to study in this way, but not for the same reasons they have traditionally been studied. With a renewed focus on the politics of intelligence failures, part of the added value relates also to the learning process that is often said to follow in the footsteps of failures. So what, then, are the ‘kernels of wisdom’ that can be generated from studying intelligence failures along these lines? If explaining the causes of failures in order to propose process modifications has traditionally been the main task, the approach suggested here questions the very possibility of making such modifications without at the same time reproducing the illusory distinction between power and rationality. Instead of reifying this distinction, studying failures has the potential to involve a more thoroughgoing examination of how power shapes the intelligence process. Rather than suggesting minor adjustments, this approach opens up the possibility of thinking about more far-reaching changes, which have the potential to re-structure intelligence work in ways that are more sensitive to the contexts of power in which this work takes place.

Acknowledgements

For his many insightful comments on several drafts of this article, I first of all want to thank Petter Narby. I am also grateful to Stefan Borg, Robert Frisk, Dan Hansén, and Dan Öberg for giving me vital feedback at various stages in the process of writing it. Finally, I wish to thank the anonymous reviewers for their valuable criticism.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

Research for this article was made possible by a Swedish Armed Forces’ FoT (“Forskning och Teknik”, 2022) grant.

Notes on contributors

Tom Lundborg

Tom Lundborg is Associate Professor of Political Science at the Swedish Defence University. He is currently exploring the relationship between power and rationality as it plays out in attempts to grapple with the purpose and meaning of the intelligence process. He is the author of numerous articles on international relations and security, published in journals such as International Political Sociology, Review of International Studies, Security Dialogue, and European Journal of International Relations

Notes

1. Marrin, “Evaluating Intelligence Theories,” 481.

2. Marrin, “Preventing Intelligence Failures.”

3. Handel, “The Politics of Intelligence,” 5.

4. Ben Jaffel et al., “Collective Discussion,” 326.

5. Flyvbjerg, Rationality and Power.

6. In Jervis, Why Intelligence Fails, 1.

7. Jensen, “Intelligence Failures,” 270.

8. Kent, Strategic Intelligence, 180.

9. Ibid., 155.

10. Ibid., 206.

11. Lowenthal, Intelligence, 6–7.

12. Jervis, Why Intelligence Fails, 152, 154–5.

13. Ibid., 153.

14. Ibid., 155.

15. “Fact Book on Intelligence,” Central Intelligence Agency (1983), quoted in Johnson, “Making the intelligence ‘cycle’ work,” 1.

16. Gill, “Explaining Intelligence Failure,” 54.

17. Johnson, “Making the Intelligence ‘Cycle’ Work,” 20.

18. Johnson, “Analysis for a New Age,” 670.

19. Jensen, “Intelligence Failures,” 278.

20. Marrin, “Preventing Intelligence Failures,” 669.

21. Jensen, “Intelligence Failures,” 273.

22. Fitzgerald and Ned Lebow, “Iraq: The Mother of All Intelligence Failures,” 895.

23. Betts, Enemies of Intelligence, 98.

24. Rovner, Fixing the Facts, 29.

25. Gill, “The Way Ahead,” 577.

26. Ibid., 578.

27. Jensen, “Intelligence Failures,” 265.

28. Gates, “Guarding Against Politicization,” 9. See also Westerfield, “Inside Ivory Bunkers, Part I,” 409.

29. Woodard, “Tasting the Forbidden Fruit.”

30. Flyvbjerg, Rationality and Power, 2.

31. Ibid., 2.

32. Ibid., 226.

33. Ibid., 226.

34. Foucault, “The Subject and Power,” 341.

35. Hamilton, de Werd, and Ivan, “Critical Intelligence Studies.”

36. Bean, “Rhetorical and Critical/Cultural Intelligence Studies,” 499.

37. Ibid., 504.

38. de Werd, “Reflexive Intelligence,” 514.

39. Ibid., 514.

40. Bean, “Rhetorical and Critical/Cultural Intelligence Studies,” 505.

41. Foucault, “Truth and Power,” 120.

42. Foucault, “The Subject and Power,” 343.

43. Ibid., 340.

44. Drumheller, On the Brink, 4.

45. Ibid., 91.

46. Drogin, Curveball, 251.

47. Ibid., 252.

48. Feldman, Formations of Violence, 9.

49. Drogin, Curveball, 256.

50. Ibid., 256.

51. Tenet, At the Center of the Storm, 511.

52. Ibid., 504.

53. Jervis, Why Intelligence Fails, 146.

54. Tenet, At the Center of the Storm, 503–4.

55. Ibid., 511.

56. Jervis, Why Intelligence Fails, 135.

57. Glaister, “Bush Ignored Intelligence on Iraqi Weapons, Says Ex-CIA Officer.”

58. Lucas, “Recognising Politicization,” 225.

59. Ibid., 227.

60. Drumheller, On the Brink, 2.

61. Ibid., 2.

62. See Ben Jaffel et al., “Collective Discussion.”

63. Bean, “Rhetorical and Critical/Cultural Intelligence Studies,” 514.