A Critical Look at Britain’s Spy Machinery
Collection and Analysis on Iraq
The understated reaction in the UK to [Iraq WMD intelligence failures] has been the source of some curiosity amongst American observers.
Editor’s Note: British political scientist Philip Davies’ examination of UK intelligence reporting and analysis prior to Operation Iraqi Freedom in 2003 broadens understanding of US post-mortems on Iraq by setting them in a coalition context. This article is based on seminars that Dr. Davies led at Carleton University’s Norman Patterson School of International Affairs and George Washington University’s Institute for European and Eurasian Studies in the autumn of 2004, while he was in the United States researching his forthcoming comparative study of UK and US intelligence.
* * *
Requirements Sections are supposed to ensure that goats remain goats. — Former senior SIS officer.
Since the invasion of Iraq, the understated reaction in the UK to what one former British official has described as the “worst intelligence failure since 1945” has been the source of some surprise and curiosity amongst American observers of the intelligence process. In the United States, the Intelligence Community, Congress, and commentators alike have been swept up in the intelligence reform debate that culminated in the last-minute passage of the Intelligence Reform and Terrorism Prevention Act in the closing days of the 108th Congress. In Britain there have been a succession of furors to be sure, but these have been only minimally concerned with the actual failure of intelligence to discern the limited extent of Iraq’s nonconventional weapons programs and capabilities.
For the most part, UK debate has pivoted around three issues: the publication of national assessments of Iraqi capability under the acknowledged authorship of the Joint Intelligence Committee (JIC); the possibility that the government pressured the JIC drafting team to include intelligence reports or claims known or suspected to be unsound or unreliable; and how the debate about this latter possibility led ultimately to the suicide of Dr. David Kelly in July 2002.
An intensive review of intelligence on so-called “weapons of mass destruction” (WMD) led by Lord Butler of Brockwell (a former Cabinet Secretary) was published in the summer of 2004; however, it attracted far less interest—and far less informed or, at least, comprehending discussion—than the January report of Lord Hutton of Bresagh on the Kelly suicide. Above all, the publication of the Butler review was not heralded with the kind of demands for comprehensive review and reform that accompanied the US Senate Select Committee’s report on pre-war intelligence estimates on Iraq.
This may have been a consequence of both the scope of the Butler review and the language of its final report. Butler was tasked to review all major counterproliferation investigations, not just Iraq. As a result, the failure on Iraq was examined alongside at least four other, loosely comparable problems—Libya, Iran, North Korea, and the Pakistan-based, transnational Abdul Qadir Khan network—which had been handled successfully. In this context, the Iraq failure was not seen as a comprehensive breakdown of the intelligence process and systemic malaise, but rather as one failure against four successes. Hence, it was viewed as a failure due to Iraq-specific factors that somehow tripped up an otherwise effective system.
The language of the Butler report was likewise comparatively understated. It avoided the often hectoring and accusatory tone of the Senate Select Committee report on US prewar intelligence on Iraq, stayed away from personalizing blame, and examined the Iraq failure chiefly in terms of the “collective responsibility” ethos of Britain’s Cabinet government and the collegiality of the JIC system in its Cabinet Office. But it also has to be said that intelligence analysis (or assessment, in UK parlance) is generally the least appreciated and least addressed aspect of the intelligence process in the UK. On the one hand, this is because assessment is scholarly rather than sexy; on the other, as has been pointed out in a number of forums elsewhere, assessment is viewed in the UK as a government function and not specifically as an intelligence function.
The conclusions reached by Butler’s review team were also less hostile than those of the Senate Select Committee. To be sure, they found that a measure of groupthink had been at work— in looking for evidence to corroborate the suspicions that the JIC had insisted on sustaining despite a lack of hard evidence (a long-recognized, inherent risk of the JIC system’s collegial methods) and a tendency to overcompensate for the optimistic assessments of the limits of Iraqi nuclear developments discredited after the first Gulf War. But no damning appraisal of comprehensive groupthink, analytical “layering,” or “broken corporate culture” appeared in the report. It concluded that publishing intelligence for public persuasion in the so-called September Dossier (drawn from a classified 9 September 2002 JIC estimate) had been a “bad idea” and “should not be repeated” under any circumstances, but that, for the most part, the causes of failure had been Iraq-specific and not endemic.
This is not to say that no institutional or structural issues were raised in Lord Butler’s review of intelligence on WMD. Indeed, toward the end of the report, he expressed a number of concerns regarding the effectiveness of the intelligence validation components of the Secret Intelligence Service’s management structure—the “Requirements” side of SIS. One of the factors behind the failure of UK Iraq assessments was the practice of placing “greater weight” upon a number of human intelligence (HUMINT) reports “than they could reasonably bear,” in the words of the Review. Butler identified a structural weakness in SIS’s quality control system embodied in its Requirements machinery. According to the report, confronted with both urgent demands for assessments of Iraqi nonconventional weapons capabilities and limited operational resources as a consequence of post–Cold War “peace dividend” cutbacks during the early 1990s, the Requirements system was not equal to the task of a rigorous evaluation of SIS reporting on Iraq.
As the following analysis will show, the malaise in Requirements that led to the intelligence failure on Iraqi WMD represents an even deeper, longer-term trend in the management of SIS than the Butler review identified. During the 1990s, I undertook a detailed administrative history of the Service, in the process discovering how the Requirements component of SIS had been progressively scaled back over more than two decades. It is in this context that we have to understand the breakdown in validation at SIS in 2002—the catastrophic failure over Iraq was not just a result of a short-term breakdown beginning in the mid-1990s.
How the Machine Works
For most of its existence, Britain’s Secret Intelligence Service has been centered around a basic organizational architecture in which its “Production” side mounts operations in response to specific demands laid upon it by a tasking, validation, and dissemination apparatus referred to as its “Requirements” side. The Production side is divided regionally under area controllers, who oversee an assortment of operational “P Sections,” each of which handles several countries and manages the agency’s resident stations abroad. The Requirements side and its “R Sections” task the Production side, validate its product, and disseminate that product to SIS consumers at Whitehall and Downing Street. Prior to the 1980s, Production and Requirements were separate and had approximately equal representation on the SIS Board of Directors.
Requirements officers are roughly analogous to reports officers inside the CIA’s Directorate of Operations (DO). Partly because SIS is an operational organization with no analytical function (except as a participant in the collective assessment process in the Cabinet Office Joint Intelligence Committee) and partly because of the peculiar circumstances of history, R Sections traditionally have occupied a far more central role in SIS than do reports officers in the DO.
From an industrial management point of view, Requirements plays a dual role within SIS: It provides marketing (representing the agency to its consumers and vice versa) and quality control (scrutinizing SIS product to see that it meets consumers’ needs in terms of both relevance and reliability). In industry, combining marketing and quality control in a single body would seem counter-intuitive and potentially a conflict of interest between the priorities of selling a profitable volume (increasing revenues) and ensuring a potentially expensive high standard of output (increasing costs). In intelligence, unlike the commercial world, however, there are natural economies of scale in combining these dissimilar tasks into a common organizational entity.
Evolution of “Requirements”
The Requirements side began life as a cluster of consumer-liaison sections shortly after the First World War. Under this scheme, SIS’s largest and most powerful consumers—the War Office, Admiralty, Foreign Office, and, later, Royal Air Force—seconded sections of their own intelligence branches to SIS to lobby for their partisan interests.
During and after the Second World War, however, the composition and function of the Requirements Sections, as they were termed after 1946, began to shift. To start with, the sections dealing with economic and industrial intelligence and with scientific and technological intelligence found themselves representing a range of consumers with common intelligence needs. They were no longer single-customer departments. At the same time, the Foreign Office liaison section began to grow into a minor empire of its own, subdivided along the same geographical lines as the Production side’s regional divisions, the Area Controllerates. In all three cases, the sections began to be staffed by internal SIS appointees.
Another shift came as a consequence of the Joint Intelligence Committee moving from the Ministry of Defence to the Cabinet Office, in which capacity it was to “give higher direction to and keep under review the organisation and working of intelligence as a whole at home and overseas . . . .” This mandate included formal responsibility for the “national requirements cycle” in which producers and consumers agreed on intelligence requirements and priorities on an annual basis. Under this new system, the Requirements Sections became responsible for overseeing SIS implementation of tasks under what became the annual National Intelligence Requirements Paper. R Section heads also represented SIS on the various JIC subcommittees (which later became Current Intelligence Groups). Both trends shifted Requirements away from partisan representation to a broader tasking, validation, and dissemination role.
For much of SIS’s existence, Requirements constituted a significant part of the management structure. When the headquarters staff in the 1930s numbered “less than twenty officers,” perhaps half a dozen were members of the liaison sections. After the Second World War, the Requirements Sections were grouped together in a Directorate of Requirements whose director (D/R) sat on the SIS Board of Directors alongside the Director of Production (D/P).
From the mid-1970s onward, however, the institutional position of Requirements began to weaken. SIS Chief Sir Maurice Oldfield reorganized Requirements along geographical lines to match both the Controllerates and the Current Intelligence Groups, which drafted national assessments under the Joint Intelligence Committee. Although essentially redistributing the work of the existing sections, this reform also separated the armed service representatives from the Requirements Directorate— moving them out of the chain of command into an SIS Secretariat under the Chief of Service. This deprived D/R of the weight and authority of having the three armed services and the Ministry of Defence behind him on the Board of Directors.
SIS Headquarters at Vauxhall Cross.
(photo © Richard Aldrich)
The 1970s was a period of national financial retrenchment, and SIS suffered from the cutbacks in funds and personnel as much as any other part of the UK defense and security apparatus. In 1979, under SIS Chief Sir Arthur ‘Dickie’ Franks, it was concluded that Requirements was too small to warrant being a full Directorate. Production and Requirements were amalgamated under a Director of Production and Requirements (D/PR), with the Requirements Sections being managed by a Deputy Director Requirements (DD/R).
This development coincided with a trend during the second half of the decade toward collocating regional Requirements and Production sections within SIS’s main office. The rationale for collocation was to allow tasking and validation functions to be more directly factored into operational management. This development was controversial. One senior officer of the period recalls:
The closer relationship between Requirements and Production began to raise all sorts of questions, especially amongst old-school Requirements officers, about whether this arrangement retained the independence of the Requirements process. On the other hand, this arrangement helped the Requirements Section officers know the agent better, helping with the assessment of the product and increasing alertness to the possibility of fabrication. This is particularly important since the case officer develops something of a partnership with his agent, developing a bond of loyalty. This relationship tends to make sheep out of goats and Requirements Sections are supposed to ensure that goats remain goats.
Although numerically and institutionally diminished, the Requirements mechanism continued to operate reasonably effectively until Permanent Secretary of Defence Sir Michael Quinlan conducted a post–Cold War review of intelligence. As a result of the Quinlan Review and subsequent cuts to intelligence expenditure as part of the “peace dividend,” SIS experienced a 25 percent reduction in staffing and expenditure, part of which took the form of a 40 percent decrease in senior staff. The DD/R was abolished and the R Sections were placed under the direct authority of the area controllers. Under the new arrangement, the area controllers, rather than Requirements officers, came to represent SIS on the Current Intelligence Groups of the JIC’s Assessments Staff, marginalizing Requirements further by depriving it of its role in the JIC-SIS relationship. The only remnant of formal independence for Requirements was a Requirements Board, headed by a senior R officer, but without representation on the Board of Directors. Just as collocation seemed relatively nonthreatening to the independence of Requirements in 1979, so subordinating the function to the senior Production managers seemed benign to participants at the time. 
This realignment, moreover, appeared to be a natural development in line with the spread of information technology. “Modern communications technology and computers,” observed one former officer, “have made it easier for everyone to know the same thing at the same time.” In other words, the day had passed when disseminating intelligence involved officers carrying locked briefcases across St. James’s Park. The increased centrality of tasking and dissemination dominated perceptions of this change, both to participants and observers (myself included). The quality control implications were completely overlooked.
SIS Iraqi Sources
In contrast to the United States, where the Senate Select Committee was perturbed to find a complete absence of national estimates on Iraq before the 2003 invasion, the JIC had produced a steady stream of national assessments on aspects of the Iraqi problem throughout the decade or so prior to the second Gulf war. Lord Butler’s inquiry in 2004 systematically retraced the steps of these assessments, paying particular attention to the basis for the estimates on the Iraqi threat in the final year before the invasion. From the present perspective, however, the crucial interval was the period following operation DESERT FOX in 1998 and the associated withdrawal of UN inspectors from Iraq. At that point, covert human sources acquired a primacy in the assessment process that they did not have when a substantial body of overt information was available through UN auspices.
Butler reveals in his report that SIS had a stable of six human sources inside Iraq. He describes four of these as “main” sources and two of the four as “dominant sources,” producing some two-thirds of all intelligence reports on Iraq that were circulated. He cautions that “volume is not necessarily a measure of influence; even a single [report] can have a significant impact.” All four of the main sources were considered reliable prior to the invasion and, in most respects, all four emerged as being generally reliable after the war, but with some significant qualifications.
First source: The first dominant source “reported accurately and authoritatively on some issues” but “on production of stocks of chemical and biological agents, he could only report what he learned from others in his circle of high-level contacts in Baghdad.” In other words, the first dominant source may have had direct knowledge of a number of key issues but on nonconventional weapons he was reporting hearsay.
Second and third sources: The second dominant source likewise was judged overall to be “an established and reliable source” whose reporting “on other subjects had previously been corroborated.” However, this second source began to pass information received from one of his contacts who acted as a subsource reporting on chemical and biological programs and intentions. The subsource’s reporting served to underpin a number of JIC assessments on Iraqi WMD, even though reports based on his information “properly included a caution about the subsource’s links to opposition groups and the possibility that his reports would be affected by that.” During the post–war SIS validation exercise, “serious doubts” were raised about the reliability of his reports. As a consequence, the reporting from the second dominant source may have been sound where he was reporting his own knowledge, but the information from his subagent was unsound.
Fourth and fifth sources: The other two main sources continued to be judged as reliable under SIS post–war validation efforts, but, Lord Butler notes significantly, “reports from those sources tended to present a less worrying view of Iraqi chemical and biological weapons capability than [reporting] from the sources whose reporting is now subject to doubt.”
Thus, all four of SIS’s main sources prior to the war proved to be reliable overall; the problem with this stable of agents was not wholesale inaccuracy, but rather hearsay reporting by one and reporting on behalf of an unreliable subagent on the part of another. Viewed in terms of the quality assurance function of the Requirements mechanism of SIS, it should have been apparent that both the first and second dominant sources were reporting in part at second hand. But were the reports properly tagged?
There has been a reasonably clear picture of UK HUMINT reporting procedure in the public domain for more than a decade. In 1993, an intelligence furor flared up in the media, driven by the prosecution of senior managers in Matrix-Churchill, one of a number of firms engaged in the export of dual-use technologies to Ba’athist Iraq. During the trial, it transpired that the managing director, Paul Henderson, had been an information source for SIS (and earlier for MI5), reporting on Iraqi weapons development programs. As a consequence, SIS and MI5 agent-handling techniques came under public scrutiny as operational reports were identified by the defense and submitted as evidence. The material submitted included the contact notes made by individual case officers after meeting with agents and the subsequent source reports based on the contact notes or other agent communications. Reporting procedures were made clear, including the requirement to distinguish between firsthand factual reporting, secondhand and hearsay information, and information that expressed an opinion or interpretation on the part of the source.
Iraq reporting appears to have followed these practices. According to Butler, reports from the second dominant source’s subsource did indeed go out with a rider alerting recipients to a question mark about the objectivity and reliability of that subagent’s information.
Sixth source: Evaluating the sixth and final source is a somewhat complicated matter. At various points, the Butler report refers to an individual “source” and to an alleged “subsource” who appears to have been part of a larger subagent network. Butler’s description of the sixth source runs as follows:
Finally in mid-September 2002 SIS issued a report, described as being from a “new source on trial,” on Iraqi production of chemical and biological agents. Although this report was received too late for inclusion in the JIC assessment [on Iraq] of 9 September, it did provide significant assurance to those drafting the government’s dossier that active, current production of chemical and biological weapons was taking place. A second report from the new source, about the production of a particular chemical agent, was received later in September 2002. In July 2003, however, SIS withdrew the two reports because the sourcing chain had by then been discredited. SIS also interviewed the alleged subsource for the intelligence after the war, who denied having ever provided the information in the reports. We note, therefore, that the two reports from this source, including one which was important in the closing stages of production of the Government’s September Dossier, must now be treated as unsafe.
In sum, the stable of SIS sources in Iraq was hardly as strong as the JIC assessments and, more critically, the September Dossier suggested, but neither was it as catastrophically poor as has been suggested in the media. The fluctuation in the first dominant source’s reports between direct knowledge and hearsay is typical of human sources. For example, a Soviet source working for the CIA and SIS known for his thousands of photographed documents also provided political assessments and interpretations that were so idiosyncratic and colored by his hostility to the Soviet regime that both agencies disseminated the two kinds of products under different cryptonyms. Likewise, as far back as the 1930s, SIS agent runners like Leslie Nicholson, who did not even have the benefit of something like the contemporary Intelligence Officers New Entry Course, were acutely aware of the difficulties and uncertainties of dealing with networks of subagents. So there were good reasons behind SIS’s placing dominant source one’s hearsay information and source two’s subagent reporting in parentheses; this did not necessarily impugn the source’s direct reporting. And, of course, both other main sources have retained their credibility with SIS and the Butler review team.
As a result, SIS had two unqualified good sources (four and five), two bad sources (three and six), and two sources that were a bit of both, but good at least when sticking to first-hand knowledge (one and two). However, how these sources were factored into the national assessment process at the JIC level and represented to the public in the September Dossier is an entirely different matter.
Britain’s Analysis on Iraq
Lord Butler is quite specific about how the various sources in SIS’s Iraqi stable were factored into JIC deliberations. On Iraqi ballistic missile programs, a JIC assessment of 10 February 2001 asserted “We know that Iraq has retained key components of disassembled 640-km-range Al Hussein missiles. Recent intelligence suggests that they may have assembled some of these.” According to Butler, this estimate rested partly on prior (worst case) estimates that Iraq had concealed missile components, but also on:
. . . three pieces of human intelligence from three separate sources on Iraqi possession of Al Hussein missiles. One of those sources provided the actual number of “up to 20 missiles” being concealed, which was subsequently reflected in all future JIC estimates . . . . that source was, in our view in a position to comment authoritatively; and we have established that he reported reliably both before and after the report. But we note that he was passing on the comments of a subsource, who reported only once. SIS had not, by the time we finished our Review, been able to contact the subsource to validate the reliability of his reporting.
In a 10 May 2001 assessment of Iraqi nuclear, biological, chemical, and ballistic missile programs, the JIC cautioned in the body of the report that it had “no clear intelligence” on Iraqi capabilities; however, in its Key Judgement on the nuclear matter, it stated that there was evidence of an Iraqi program to acquire dual-use items potentially applicable to a nuclear weapons program and to conducting unspecified nuclear-related research that could contribute to a break-out production capability if sanctions were lifted. Commenting on this, Butler observes:
[The] judgement was based on two human intelligence reports, both from new sources and neither speaking from direct, current experience. Unusually in the nuclear field, we conclude that those reports were given more weight in the JIC assessment than they could reasonably bear.
In the same report, the assessment of Iraqi chemical weapons (CW) capability was that “we believe Iraq retains some production equipment, stocks of CW precursors, agents and weapons.” But the assessment simultaneously warned that “intelligence of other CW activity, including possible weaponisation, is less clear.” According to Lord Butler, the HUMINT sources behind this estimate consisted of “a single report from a new source who reported details of a project three years earlier to integrate the nerve agent VX into rocket artillery warheads and the subsequent filling of 60 warheads” and “a further single report from a new source, passing on the comments of a subsource that he had been part of a project to produce the nerve agent VX in the period to 1998, again three years earlier.” As Butler observes rather sharply, “the intelligence applied to mainly historical (as opposed to current) activity and, even so, was by no means conclusive.”
The JIC produced another assessment on 21 August 2002, titled “Saddam’s Diplomatic and Military Options.” The report, prepared in response to a requirement from the Ministry of Defence, warned that “although we have little intelligence on Iraq’s CBW doctrine, and know little about Iraq’s CBW work since late 1998, we judge it likely that Saddam would order the use of CBW against coalition forces at some point.” Lord Butler points out that, given the context of the requirement, “the Key Judgements of that assessment would rightly have been prepared on a precautionary basis . . . when set against intelligence on Iraqi programmes contained in advice for Ministers in March, the [August] JIC assessment reflected more firmly the premise that Iraq had chemical and biological weapons and would use them in war.” Significantly, the conclusions “were based in part on one human intelligence report from one source but mainly the JIC’s own judgements.” In other words, the 21 August assessment was not a predictive one, but a speculative one that necessarily had to employ a worst-case approach and err on the side of caution.
The 9 September Assessment
The next assessment, dated 9 September 2002, addressed “Iraqi Use of Chemical and Biological Weapons—Possible Scenarios.” This estimate, which served as the main source for the unclassified September Dossier, reflects caveats similar to those of the other estimates, but, significantly, reached much firmer conclusions than prior reports. It warned that “Intelligence remains limited and Saddam’s own unpredictability complicates judgements about Iraqi use of these weapons” and that “Much of the paper is necessarily based on judgement and assessment” rather than hard evidence. Despite this, it also asserted, in apparent self-contradiction, that “Recent intelligence casts light on Iraq’s holdings of weapons of mass destruction and its doctrine for using them” and, with unprecedented confidence, that “Iraq has chemical and biological weapons capability and Saddam is prepared to use it.” The assessment further claimed that “other recent intelligence indicates that production of chemical and biological weapons is taking place.” Again somewhat inconsistently, the supporting discussion held that Iraq could produce chemical agents “within weeks” and biological agents “within days,” and that Baghdad had retained “up to 20 al Husseins.” These points were consistent with the more tentative earlier reports, which had placed emphasis not on stockpiles in hand, but on research and development and a latent, but possibly growing, “break-out capability” to kick-start chemical and biological agent production and weaponization programs once sanctions were lifted.
The Butler Review details the sources underpinning the 9 September assessment, and hence the published Dossier, in considerable detail, and it is worth quoting the review at some length on this.
The more definite judgements inside the assessment were based on the receipt of significant new intelligence in August and September 2002, in response to the routine requirement on SIS to obtain information to support the drafting of JIC assessments . . . . Four reports were received in total, from three sources, which were influential in the JIC’s assessment.
The first provided material from a range of original informants via an intermediary source. We have noted, however, that the individual informants did not confirm directly that Iraq had chemical weapons. They came from senior Iraqi officials who were believed at the time to have direct knowledge of Iraq’s intentions, use, deployment and concealment of chemical weapons, but were based for most of the informants on an assumption (not direct knowledge) that Iraq had such weapons.
The second and third were from a source who had previously reported reliably and who continued to do so in the following months. This source, too, could not confirm from direct knowledge that Iraq had chemical weapons, resting upon “common knowledge” within his circle that chemical agent production was taking place. The second report from this source seems to us to duplicate much of the first.
The fourth was a single report, from a reliable and established source reporting a new subsource who did not subsequently provide any further reporting, which was described as “confirming” the intelligence on Iraqi mobile biological agent production facilities received from [CIA]. Contrary to the JIC view at the time, we believe that this report would have been more accurately described as “complementary” to, rather than “confirming,” it.
Unsurprisingly, the Butler team concluded: “We were struck by the relative thinness of the intelligence base supporting the greater firmness of the JIC’s judgements on Iraqi production and possession of biological weapons, especially the inferential nature of much of it.”
The Review identified one last source during the final interval before the war. This informant, who played a central role during much of the postwar debate about the quality of the raw intelligence fed into the estimates of Iraqi WMD, was the source of a report received after 9 September, during the drafting of the September Dossier. The report proved especially controversial because it was used to quash objections to the wording of the Dossier raised by Dr. Brian Jones and the analysts at the Directorate of Scientific and Technical Intelligence at the Defence Intelligence Service (DIS).
Just before the 9 September assessment was completed, an SIS source, viewed at the time as having a “proven and reliable” track record, provided information that unspecified Iraqi chemical weapons could be prepared for use in “45 minutes or less.” The source in question may indeed have been individually reliable, but he was a subagent in a rather long reporting chain. His “45-minute claim” played a relatively minor role in the 9 September JIC assessment. However, due partly to its inherently alarming nature, partly to vague wording that amplified its alarming nature, and partly to ill-considered repetition in the September Dossier, the 45-minute claim acquired a central role in the subsequent controversy over the “sexing up” of intelligence by the government, the accusations of BBC journalist Andrew Gilligan, and the suicide of Dr. David Kelly. During the postwar validation review, reports Butler, “SIS interviewed the alleged subsource . . . who denied ever having provided the information in the reports.” The claim evidently had been fabricated by an intervening member of the reporting chain, based on a Soviet-era military handbook specification.
Before the Iraq invasion, analysts at DIS had been dubious about the validity of some of the HUMINT reporting—including the 45-minute claim—and the strength of the conclusions that could be reached on the basis of that reporting. On seeing a draft of the September Dossier, Jones prepared a memorandum to his line manager that challenged the conclusion that Iraq had chemical agents in hand. He argued that “We have not seen intelligence which we believe ‘shows’ that Iraq has continued to produce CW agent in 1998–2002, although in our judgement it has probably done so.” Jones’s objections were overruled by the deputy chief of defence intelligence, in part because of the time pressure under which the Dossier was being drafted. Another factor working against Jones, however, was a report from the last-minute source that supposedly corroborated the 45-minute claim.
This brand new source “reported that production of biological and chemical agents had been accelerated by the Iraqi regime.” Jones had been denied access to this report on the grounds that the author was a “new source on trial.” In giving evidence to the Butler team, SIS Chief Sir Richard Dearlove explained that it was SIS practice to limit the circle of individuals indoctrinated into any new source during the agent’s “initial, very sensitive period of development.” As a consequence, dissemination and evaluation of the new agent’s report was confined to SIS’s own technical experts who “took a preliminary and provisional view that the report should be issued, as being from ‘a new source on trial.’”
The restricted dissemination in and of itself indicates that this last minute information was hardly from the sort of tried and proven line of reporting that would ordinarily carry enough weight to tip the analytical scales in one direction or another. There appears, therefore, to have been a mishandling of the new source’s information at two points: The first, as Lord Butler points out, was the unwillingness to make untested and uncertain materials available to “the few people in the UK intelligence community able to form all-round, professional technical judgements on its reliability and significance”; the second was the fact that, despite qualifications placed upon the source by SIS’s own technical validation personnel, this source was viewed by senior SIS and DIS managers at the JIC as being sufficient to negate Jones’s concerns. Rather than dogmatically following a source protection protocol, argues Butler, senior managers in the DIS and SIS should have “made arrangements for the intelligence to be shown to DIS experts instead of making their own judgements on its significance.” Significant fault therefore rested with senior SIS and DIS officials—evidently at the JIC level—in their representation of, and the relative weight given to, this source, as well as other aspects of HUMINT reporting on Iraqi nonconventional weapons.
What pressures, assumptions, or incentives may have propelled the top intelligence management and analytical team in the UK government to place such excessive weight on sources? Lord Butler reached a series of key conclusions about the validation and assessment of intelligence at both the agency and JIC levels. While judging that the use of émigré sources was not a significant problem, the Butler review raised concerns about the reliance on subagent networks with long reporting chains: “Even when there were sources who were shown to be reliable in some areas of reporting, they had in other areas of intelligence concern where they did not have direct knowledge to draw on subsources or sub-subsources.”
The inclusion of insufficiently validated subsource reporting was no doubt driven by the fact that “agents who were known to be reliable were asked to report on issues going well beyond their normal territory.” Also, “because of the scarcity of sources and the urgent requirement for intelligence, more credence was given to untried agents than would normally be the case.”
These problems put strains on the SIS collection and reporting system that should have been detectable by Requirements Sections performing their quality-control function effectively. But, as Butler makes evident, Requirements was so diminished, it could not do so.
Throughout its assessments after the first Gulf war, the JIC had sustained suspicions that there might be weapons, components, and precursors that were slipping beneath the horizon of UN inspections and the available hard intelligence. These preconceptions that the weapons were there led the JIC to an overly robust interpretation of current reporting. Instead of being derived primarily from the evidence at hand, JIC judgments in 2002 were formed with strong reference to Saddam Hussein’s prior history of WMD production, concealment, and use. Assessments, Lord Butler concludes, tended to be “coloured by over-reaction to previous errors,” and there was a definite process of what the US Senate report termed “layering”— whereby “over-cautious or worst-case estimates shorn of their caveats” were carried over from one assessment to another, becoming “prevailing wisdom.” In other words, the preconceptions contributed to a level of groupthink where the JIC was looking less for indications of what might be the case than for what they expected to be the case.
It is significant to note, however, that, even with the layering and groupthink, the JIC was scrupulous about caveating its estimates and acknowledging lacunae in its data set, confining itself to asserting nothing more than suspicion, at least until the 9 September assessment. But even the September assessment was qualified in terms of the available intelligence and the JIC’s reliance on judgment— inference and informed speculation—as the basis for its firmest and least defensible conclusion about Iraqi possession of WMD.
Fixing the Machine
It might be invidious to observe that SIS’s six sources in Iraq, for all their variable quality, were six more than certain key allied intelligence services had, but it is important to keep in mind that SIS was successfully recruiting sources in a very hard target state. Nonetheless, although reporting was strong in some areas, there can be no doubt that very serious problems existed with respect to the quality of raw human intelligence reporting on Iraqi nonconventional weapons programs.
One must be cautious about reaching overly strong or sweeping judgments on the basis of the limited information. While SIS certainly had more penetration operations running against China and Russia in the Cold War than against Iraq in absolute numbers, it would be instructive to see how the ratio between strong, variable, and weak sources compared with that in Iraq. Moreover, the Butler Review does not give a comparable stocktaking of the number/quality distribution of the sources in hand for those relative successes against Iran, Libya, North Korea, and the Pakistani arms network. Did having six sources—with reporting that was 33 percent solid, 33 percent variable, and 33 percent weak—constitute a reasonable, poor, or indifferent performance against a regional counter-proliferation requirement? This evaluation cannot be confidently made based on the volume and quality of information available through the Butler review and other inquiries that followed the Iraq invasion.
What is evident, however, is that the JIC made stronger judgments on Iraq than available sourcing could support. Therefore, the question that has to be asked is what went wrong on SIS’s Requirements side that led to the failure to adequately assert and sustain the distinction between sheep- and goat-quality reporting and monitor the use of that reporting in the national analytical process in the Cabinet Office.
What went wrong was not a lack of sources but a failure to adequately lift the intelligence signal out of the background noise and make sure that the signal reached consumers, analysts, and decisionmakers with the required clarity. The real failure of the SIS validation system was not the failure to provide reliable reporting on Iraq, but, rather, a failure to effectively separate the reliable reporting from the less so.
This failure has to be seen not as a short-term breakdown in the SIS validation machinery resulting from cutbacks in the 1990s, as Butler contends, but as the culmination of a steady weakening of the Requirements mechanism for handling tasking, dissemination, and validation, since 1974. The abolition of a separate identity for Requirements was accompanied by successive moves to push responsibility “further down the organisational pyramid,” as one officer put it. The question has to be asked how far down can one push a function in an organizational hierarchy before it is deprived of any influential voice at the decisionmaking levels. As the Butler team observed:
The quality assurance function of the SIS “Requirements” officer . . . became subjected to the operational imperative of the team leader [Controller] to produce results. At the same time, we were told [by one SIS official], “Requirements” posts were increasingly staffed by more junior officers as experienced staff were put into improving the operational teeth of the Service. Their ability to challenge the validity of cases and their reporting was correspondingly reduced.
These sentiments were echoed by a second interviewee, who expressed concern that the “staff effort overall, and the number of experienced case officers in particular” applied to the Middle East and Global and Functional Controllerates was “too thin to support SIS’ responsibilities” so that “source validation, especially on Iraq, had suffered as a consequence of both problems with what were in the witness’s view sources with dubious motivation being overgraded for reliability.”
An oft-heard refrain in US intelligence literature is that what is needed most is not more collection but, rather, more analysis. Much the same point seems to apply to validation in the British system: Validation is not highly manpower-intensive, but it makes a disproportionately important difference to the quality of finished intelligence. Because this is a counterintuitive conclusion, it is easy to sympathize with Dearlove’s observation during the Butler review:
It is very, very difficult, particularly when the pressure on the Service is to produce good intelligence, to put your officers who are the only ones who can do production as well into the Requirements tasks. I accept problems and the fact that in an ideal world you would only staff your Requirements desks with very experienced operational officers. In practice that is not possible.
The problem, of course, is that this trade-off of validation against operational capability had been made not once but at least three times between 1974 and 1994, leaving very little slack in the system when it came under increased pressure from consumers between the autumns of 2001 and 2002.
One of Lord Butler’s few explicit recommendations on reform was to “urge the Chief of SIS to ensure that this task [validation] is properly resourced and organised . . . and we think that it would be appropriate if the [Parliamentary] Intelligence and Security Committee were to monitor this.” In mid-January 2005, there was a flurry of UK press interest in a briefing from the prime minister’s spokesman to the effect that SIS had reinstituted the position of Deputy Director Requirements as part of the implementation of the Butler Review’s recommendations. What this reform does, however, is simply return Requirements to its previous, diminished status quo ante of 1993, under the equivalent of a Deputy Director subordinate to D/PR. Responsibility for representing SIS on the JIC’s Current Intelligence Groups also remains to be returned to the Requirements Sections. This is necessary not only to ensure the most objective possible representation of SIS product in the assessment drafting process, but also to restore to Requirements the authority of speaking on behalf of the Assessments Staff within SIS.
However much SIS’s own senior officials may believe that Requirements has remained undiminished, it is evident that it did not have the voice and the authority at the top level in SIS to prevent the agency’s product from being oversold in the JIC’s deliberations on Iraq. As a consequence, SIS may not remain truly effectively reformed until quality control and the Requirements side once again have their own independent presence on that agency’s Board of Directors.
Dr. Davies would like to acknowledge helpful comments from his colleague at the Brunel University Centre for Intelligence and Security Studies, Professor Anthony Glees, and from participants in the Carleton and GW conferences, most notably Prof. Martin Rudner, Prof. Richard Aldrich, and Prof. James Goldgeier. He is also indebted to Michael Herman who, he acknowledges, was probably the first person other than himself to realize that the analysis of Secret Intelligence Service (SIS) structure in Davies’ book, MI6 and the Machinery of Spying (London: Frank Cass, 2004), had exposed the weakening of Requirements. The research for this article was made possible in part by a Leverhulme Trust Research Fellowship.
Interview with former senior SIS officer, 27 February 1997.
Information under the Chatham House rule.
See, for example, Anthony Glees and Philip H. J. Davies Spinning the Spies: Intelligence, Open Government and the Hutton Inquiry (London: Social Affairs Unit, 2004).
Lord Butler of Brockwell, Review of Weapons of Mass Destruction (London: TSO, 2004), HC898 of 2004.
Lord Hutton of Bresagh, Report of the Inquiry into the Circumstances Surrounding the Death of Dr. David Kelly C.M.G. (London: TSO, 2004).
Senate Select Committee on Intelligence, Report on the US Intelligence Community’s Prewar Intelligence Assessments on Iraq (Washington: United States Congress, 2004), S108-301.
This brief parallels that of the Presidential Commission on the Capabilities of the United States Regarding Weapons of Mass Destruction, which reported in March 2005; however, the SSCI report was published at roughly the same time as the Butler review and set the tone and agenda for US intelligence reform discussions in a way that the Presidential Commission has not.
See, for example, Michael Herman Intelligence Power in Peace and War (Cambridge: Cambridge University Press, 1996), 265; Philip H. J. Davies, “Ideas of Intelligence: Divergent National Concepts and Institutions,” Harvard International Review 24, no. 3: 62–64; and “Intelligence Culture and Intelligence Failure in Britain and the United States,” Cambridge Review of International Affairs 17, no. 3: 496–520.
See, for example, John Hughes Wilson, Military Intelligence Blunders and Cover-Ups (London: Robinson, 2004), 260–308, 407–8.
Davies, MI6 and the Machinery of Spying.
The following narrative is summarized from the text of MI6 and the Machinery of Spying, with citations given for new sources of information acquired since publication or where there are particular points of contention. For a detailed discussion of Requirements in particular, also see my article, “MI6’s Requirements Directorate: Integrating Intelligence into the Machinery of Government,” Public Administration 78, no. 1 (January 2000).
Reports officers are something of a dog that has failed to bark in the history of US intelligence and the CIA. Although they occupy a significant nexus between clandestine service officers and Directorate of Intelligence analysts, they have tended to attract far less attention than either in the literature. Indeed, little more than a dozen documents refer to them in the CIA Records Search Tool holdings at the National Archive in College Park. Significantly, however, at one point they were referred to as Requirements/Reports Officers. One of the few published studies of the reports officer role is W. J. McKee’s “The Reports Officer: Issues of Quality,” reprinted from Studies in Intelligence in H. Bradford Westerfield, ed., Inside CIA’s Private World (London: Yale University Press, 1995), 108–17.
The official history of British intelligence in the Second World War attributes this scheme to the Secret Service Committees of 1919 and 1921 and calls it the “1921 arrangement.” See F. H. Hinsley, et al., British Intelligence in the Second World War (London: HMSO, 1979), 17. Alan Judd, however, argues that it should really be credited to Director of Military Intelligence George W. M. MacDonogh, who proposed it in 1917. See Alan Judd The Quest for C: Mansfield Cumming and the Founding of the Secret Service (London: HarperCollins, 1999), 392–93.
“Terms of Reference for the Joint Intelligence Committee,” JIC (57) 101, 1 October 1957, in CAB 158 30, UK National Archive (formerly the Public Record Office).
Christopher Andrew, Secret Service (London: Sceptre, 1987), 487.
This is also sometimes given as Director, Requirements and Production, or D/RP.
 Interview with former senior SIS officer, 27 February 1997.
Mark Urban, UK Eyes Alpha: The Inside Story of British Intelligence (London: Faber & Faber, 1996), 29.
Confidential conversation with former senior SIS officer, 19 January 2005.
Interview with former senior SIS officer, 27 February 1997.
Senate Select Committee on Intelligence, The Intelligence Community’s Pre-War Assessments on Iraq, 51. Senator Bob Graham—in his book Intelligence Matters (New York: Random House, 2004), 179–80—describes his reaction at the time as “shocked” that there had been no Intelligence Community estimate of as major a national security issue as Iraq.
See, for example, David Leigh, Betrayed: The Real Story of the Matrix-Churchill Trial (London: Bloomsbury, 1993), 133, 135. The contact notes/source report procedures elaborated during the Matrix-Churchill trial and subsequent inquiry were specifically those of the Security Service, but standards and procedures in the two agencies are necessarily similar in this regard. Matrix-Churchill was one of a series of such prosecutions that eventually prompted a judicial inquiry by Sir Richard Scott. The Scott investigation was the most transparent inquiry into intelligence activities in the UK until the Hutton inquiry a decade later. See Richard Scott, Report of the Inquiry into the Export of Defence Equipment and Dual-Use Goods to Iraq and Related Prosecutions (London: HMSO, 1996), 5 vols., plus index and CD ROM.
 Butler, 100–101, emphasis in the original.
Jerrold Shecter and Peter Deriabin, The Spy Who Saved the World: How a Soviet Colonel Changed the Course of the Cold War (London: Brassey’s, 1995), 333–35.
See Nicholson, writing as John Whitwell, British Agent (London: William Kimber, 1967). In particular, Nicholson provides an instructive account of his ALEX network, identified by appended numbers such as ALEX-1, ALEX-2, and so forth, 85–86.
 Butler, 61.
 Ibid., 56. It is hard to reconcile the “two new sources” given here with the stocktaking of SIS assets presented earlier. One might indeed be the subagent to the second dominant source, but the other last minute agent mentioned on a previous page did not report until September; therefore, the second new source given here must not have been in the Iraq stable. He may have been either an émigré or part of the reporting chain to the September new source.
Ibid., 73, emphasis added.
 Ibid., 74–75, emphasis added. Regarding the first reporting chain, the Review includes a footnote that refers to the summary of SIS sources, which notes that “We were told by SIS during the course of our Review that there is now doubt about the reliability of this reporting chain and hence of the reports derived from it.”
Richard Dearlove, “Evidence to Parliamentary Foreign Affairs Committee,” FAC/3/26 (2003).
Butler, 111, 127.
Quoted in Butler, 138. Jones’s concerns were first publicly addressed in an elliptical style in the Parliamentary Intelligence and Security Committee’s report Iraqi Weapons of Mass Destruction: Intelligence and Assessments (London: TSO, 2003), 29–30. They then became a significant feature in the evidence compiled during the Hutton inquiry when Jones himself was invited to testify—see oral evidence to Lord Hutton’s Inquiry, at: http://www.thehuttoninquiry.org.uk/content/transcripts/hearingtrans28.htm—and then in still more direct detail in his article “Hutton Report: The Aftermath,” The Independent, 4 February 2004.
The persistence of these suspicions is a recurrent theme throughout the Butler Review.
Confidential briefing by former senior SIS officer, 19 July 2005.
 Butler, 103.
Ibid., emphasis added.
 Quoted in Butler, 104.
For an example of the caliber of press response, see Richard Norton-Taylor, “MI6 acts to curb rows over spying,” The Guardian, 12 January 2005. For a detailed and well-informed account, see Gordon Corera, “UK Makes Changes to Secret Intelligence Service,” Jane’s Intelligence Review, February 2005: 48–51.
Confidential conversation with senior SIS official, 19 January 2005.
See, for example, Sir Richard Dearlove’s remarks to the review panel on this question, in Butler, 102.
Philip H. J. Davies is deputy director of the Brunel University Centre for Intelligence and Security Studies where he is also senior lecturer in politics. He is the author or co-author of several books on intelligence matters.