William L. Mitchell, phd. studerende ved Forsvarsakademiet
Foreign and security policy (FSP) construction is mainly supported by strategic level intelligence analysis as it usually focuses on the national and supranational issues or threats. It is primarily collected in response to requirements by a government across a spectrum of national and international military, diplomatic, political, and economic issues. Strategic level intelligence analysis may be understood as the type of analysis that relates directly to achieving the overarching (that is, strategic) objectives of organisations, corporations, and certainly governments. It is a process that is not restricted to the formulation of policy and plans at agency, national or international level but is best understood in its simplest form as the development of an overall plan that encompasses all the details necessary to arrive at the main objective.
Sept. 11, 2001 constituted a strategic level threat for American FSP not because it was a devastating action that attracted world wide attention, but simply because of the fact that the action itself was only a small part of the overarching plan of those responsible for it.
Observations in this paper concerning the American Intelligence Community (AIC) and events around Sept. 11 are about assessing the analytical capability of an intelligence agency, and therefore focus on the strategic intelligence analysis process, its methodologies and structures and not the semantic labelling of violent behaviours. In the post-Sept. 11 tidal wave of “terrorism” literature, I hope that the approach adopted in this paper will be a positive one in that it will not contribute to the ‘mystification’ or ‘sensationalising’ of “terrorism” but rather to a better general understanding of related issues in intelligence studies. As far as this paper is concerned what is called “terrorism” could instead be labelled ‘organised crime,’ ‘crimes against humanity,’ ‘war crimes,’ or ‘alternative threat systems’ it would not affect the approach of this paper.
During the Cold War the AIC had produced a formidable intelligence bureaucracy that quickly came under a lot of scrutiny during the 1990’s for post-Cold War spending cuts and management restructuring. The AIC began to spend more time assuring them selves a new role without the Soviet Union than really coming to grips with the analytical challenges of the post-Cold War international environment. Arguably the AIC itself recognised that it was not adjusting to a globalisation process that was producing threats of a much wider variety than those produced by the former Warsaw Pact - and certainly of a more complicated character than within the context of a bi-polar ideological standoff. The problems did not lie so much in the physical ability of the AIC to collect intelligence to be used for threat identification, but rather in the quality of their strategic level analysis capability. Problems between the various AIC communication and management structures had weakened the analysis process by thinning out the contextual expertise for intelligence processing. Though recognised and responded to by some of the AIC structures to a limited degree – the whole of the AIC did not. To further exasperate the problem, the AIC focused almost exclusively on the management of the massive amounts of intelligence coming in, and neglected the need for better analysis.1 Sept.11 brought this state of contextually degenerative analysis within the AIC to an end.
This essay suggests that the primary affect of Sept.11 was to force the AIC to acknowledge and act upon the necessity for better intelligence analysis to deal with a wider range of threats of a more complicated character. In effect, ending a decade long period of uncertainty within the AIC over which direction to take for the future development of their analytical capability.
The structure of the article consists of a general overview of intelligence capabilities, structures, processes, and methodologies that support strategic level intelligence analysis. These will act as reference guides for later discussion on the problems before Sept.11 and proposed changes after. The article will close with an evaluation of the changes and expectations for improvement of intelligence analysis in the AIC. Before starting I would like to point out that assessing the AIC on the basis of the one event, Sept.11, could mislead people into thinking that there was widespread unprofessional attitude within the AIC. This would be a mistake. Therefore place the discussion in this paper within the following context; while many of us were drinking champagne during the millennium celebrations of 19992000, personnel from the AIC were responsible for thwarting plans for up to 15 large terrorist attacks world wide.2
A typical intelligence structure will have capabilities for clandestine collection, counter intelligence, analysis & estimates, and covert action. It should be mentioned that open source collection has an increasingly important role to play (Internet, NGOs, etc.) but usually is carried out within the realm of analysis & estimates as needed.3
Clandestine collection may be dealt up into intelligence provided by humans (HUMINT) and intelligence provided by technology (TECHINT).4 In relation to each other HUMINT is, by enlarge, the most traditional and most sensationalised aspect of the intelligence community. The advantages of HUMINT in terms of threat assessment are that any direct information obtained will go a long way in clarifying the intentions and capabilities of suspected threats. TECHINT where it concerns satellite images for example, provide very little to clarify the intentions of the target under focus. Therefore HUMINT is preferred as it directly enriches the context for intelligence analysis. Furthermore in situations where suspected groups counter the TECHINT by structuring them selves in much more complicated manner – HUMINT would be highly desirable. However HUMINT, depending on the circumstances may have severe limitations such as too few possibilities for infiltration because of a complete cultural ‘lock-out.’
As for TECHINT, it also has a range of limitations that vary in proportion with resources and capabilities of both the intelligence agency as well as those available to the target. For example, TECHINT may provide impressive satellite photos of what is apparently a strategically important geo/political area to their adversaries, but may lack a properly trained geologist to sort out what is man made and what is natural. Thus they are possibly unable to identify the significance of what has been observed or collected. Furthermore the availability of resources and capabilities in TECHINT also play an important role in terms of developing TECHINT countermeasures, or adapting to the counter measures of others. Also in today’s rapid advancing world of technological development, one can assume that TECHINT will consistently be open to countermeasures that may deceive or mislead intelligence analysis.
Certainly an important capability for any intelligence organisation, and a capability that will become more important as the level of co-operation between intelligence agencies in the international environment increases. Some typical activities include managing defectors, co-operative relationships with other services, infiltration of target groups or services, and managing special operations for uncovering enemy agents. A general rule of thumb is that the resources necessary for counter intelligence rise proportionally to the level of internal or external intelligence cooperation. New exchanges of information between intelligence agencies (foreign or internal) would require that more resources be delegated to counter intelligence activities.
The ability to undertake covert actions effectively requires a good deal of resources. Actions such as supporting insurgencies, or training other intelligence or security services, propaganda, confidential political support and counsel, and assassination, are all examples of covert activity. During the last 30 years in the West there has developed distinct political (and judicial) norms for limiting these activities in peacetime.
Analysis & Estimates
As stated in the introduction the focus of this paper is on the analysis & estimation capability of the AIC. Though probably the least sensational of the capabilities, it is the most important as it co-ordinates all other capabilities for a single purpose. Intelligence gathered is only as useful as it is interpreted in relation to the desired ends. It sets the parameters of questions to be answered that in turn determine resource delegation. It provides the synergy with purpose for the concerned intelligence community that focuses on answering the following three questions to varying degrees: What is happening? Why is it happening? What will happen next? The approach to how an intelligence organisation will try to answer these questions is reflected in its analysis structures, while the degree to which it attempts to answer these questions is found in its methodology.
Most intelligence organisations are reflected to some degree in one of three models. A confederated model, for example the departmental analysis of the Army, Navy, USAF, and others.5 The centralised model, that usually divide up responsibilities internally by department or desks such as the Canadian Security and Intelligence Service (CSIS). The competitive model where Team A vs. Team B in the same area of responsibility, arguably CIA/FBI/DEA in the US as it concerns the narcotic trade. These platforms are certainly susceptible to blurring. One could imagine a little competition between different departments within the centralised organisational platform, where crosscutting issues may exist due to special circumstances.
However differences in methodology are not so straightforward as the differences in organisational structures. At this point it is suffice to say that differences in methodology usually occur in the degree or analytical weight the intelligence apparatus gives to answering each question, the what, the why and how, or the what next?
A very simple generic process common to most types of strategic intelligence analysis is one that usually includes tasking, planning, data collection, analysis, and preparing the final report. The process begins with a task that is negotiated with the client, followed by the project planning stage where time limitations, command and control, scope of tasking, and support structures are planned. These include the initial resource delegation necessary to start the initial steps of the following stage data collection.
Once data is collected the analysis and estimation stage can begin, here data processing can further be divided into three distinct phases; the first is the integration phase where data is prepared for analysis. The second is the analysis or evaluation phase where data is sorted and valued according to the relevant preestablished framework (a series of related hypothesis or sub-hypotheses for example.) Finally the interpretation phase will take the results of the evaluation phase and give them meaning, and possibly place them back into the context of overall task at hand.
When examining the AIC and Sept.11 you will read several references to the importance of ‘context’, it is here in the evaluation phase of the analysis stage that context is important for extracting significant information that might not be so obvious.
The reporting stage is where the results of the process are presented to clients, it is important here that results are framed within the context of the tasking – or simply said they answer the questions asked to the best of ones’ ability. In relation to maintaining the integrity of the analysis, it is here that the objectivity of the process is most exposed, for example to the political influence of clients.
Fig. 2.0 also indicates, the generic process may be applied at three different levels: the strategic, the operational, and the tactical. Differences between the three levels of analysis are apparent at different stages throughout the process. For example the differences between the levels at the project planning stage concern the amount of flexibility in terms of forward planning. At the strategic level you will have more time to assure forward planning as the complexity of the analysis project unfolds. The time and associated flexibility, in terms of forward planning, decreases as you move down through the operational analysis level to the tactical level. (However the operational flexibility – or the ability to react to the immediate circumstances - should increase.)
Differences in data collection at the strategic level are found in the depth & range of the scope as compared with the other two levels. The strategic level will have the widest range and variety of sources and data across a plethora of categories such as politics, military, social, and economic. At the analysis stage the main differences between the threes levels is that at the strategic level you can expect to spend much more time working out just what types of analysis suit what categories of data. Subsequently one can also expect to spend more time trying to manage qualitative data in manner that gives it meaning - even where there is no substantive statistical proof to be found. This requires a contextual understanding for the data being analysed, in order to detect significant intelligence that might not be so obvious.
Finally, differences at the reporting stage between the levels usually reflect all the differences mentioned at the previous stages. Strategic level reports are usually written and provide extensive information, analysis, organisational information on the report itself, as well as an executive summary of some sort. As one moves down through the operational level to the tactical level one can expect the written reports reflect the diminishing scope and eventually the majority becoming verbal reports and briefings to commanders at the tactical level.
This paper focuses on strategic level intelligence analysis before and after Sept.11 within the AIC as many of the pre-Sept.11 problems noted, concern the analysis capabilities of AIC in relation to a weakened contextual understanding.
Intelligence Analysis Methodology
In a very general characterisation, just as in the left to right political ideological model of political science, one could also summarise the methodological approaches in the intelligence field by drawing a horizontal line with ends represented by ‘description’ and ‘prediction.’ The more an approach tends to focus on the technical collection of special information for the specific purpose of answering, “what is happening?” the more descriptive it tends to be. A good historical example would be intelligence analysis during WWII that concentrated on code breaking.
At the other end of the scale closer to prediction is what some call the Colby Model or the ‘academic’ approach.6 This model extensively uses social science techniques and methodologies applied to intelligence interpretation and collection not only to answer the what, but also the why, and most importantly what will happen next? An interesting historical example inside the AIC could be the research and analysis section of the Office of Strategic Services (OSS) during the World War II.7
In between these two extremes are two approaches both of which are similar to each other fundamentally, yet retain clear differences in terms of how much effort is put into prediction. The dominant approach since 1949 has been the Kent approach where strongest efforts are on intelligence collection for the purposes of description.8 This approach hopes to be predictive while being executed under limited policy guidance. A critic of Kents’ approach led to another option that lies a little further towards the academic side of the scale.9 This approach, the Kendall Model, has been called opportunity analysis or actionable analysis, both terms indicate the greater amount of interpretation and policy led tasking that is involved in collection planning. It is used to ID future threats and to purport specific opportunities to advance policy-makers objectives. The approach certainly retains the usefulness of technical collection and analysis related to description but focuses a little more on the prediction end through a richer contextual analysis (see Fig 3.0).
September 11 and the AOC
The intelligence problems identified before and after Sept.11 fall largely under the capability of ‘analysis and estimates’ and decidedly on the strategic level of the analytical process. There is no better example of a pre-Sept. 11 AIC self-critique of this capability than that of Russ Travers “The Coming Intelligence Failure,” written in 1997.10 The uncanny prediction of a serious intelligence failure in 2001 aside, Travers zeros in on many of the key problems with the post-Cold War AIC analysis structures and methodology. Travers writes that the “ Management of intelligence is more valued than collecting and analysing intelligence and thus we have fewer and fewer good analysts,” this observation underscores the fact that the significance of collected intelligence was already being missed.11
In effect the analysis process was transforming itself into being more operationally oriented in 1997 than strategic, thus reducing forward planning flexibility and the analytical capability to handle complexity. Specifically Travers pointed to the process of analysing problems separately under the traditional categories such as economics, military, and political, and that did not sit well with a globalisation process that blurred such distinctions. This subsequently weakened analysts’ ability to extrapolate what was significant from what was just ‘background noise,’ from the increasing mass of collected data. Resulting in a weakened contextual understanding within the analytical bodies.
The problem did receive some attention - at least within the CIA. On May 5th, 1998, CIA Director George J. Tenet suggested that the Directorate of Operations would work to improve two of its capabilities, all-source analysis and clandestine HUMINT collection to strengthen the contextual understanding of the international environment. However these efforts would be made primarily within the CIA alone so there would be little affect on the intelligence community as a whole. This problem would be exasperated further by the loose US confederated intelligence community of 13 agencies, and the inherent organisational competition on relative areas of responsibility. Not only were analysts loosing their analytical skills, outright communication problems between agencies would sabotage any structural support for analytical fusion at the strategic level.12 It could be argued that specifically where terrorism is concerned, in the 1990’s, some US agencies may have had better co-operation in terms of information sharing with foreign intelligence services than it did with some of its fellow American agencies. 13
In terms of methodology, the AIC took a step to the left towards the description end of the scale during the 1990’s pushed on by a mass of TECHINT, more data managers, but no analysts with a coherent contextual understanding complex enough with which to analyse it. The result was that the quality of contextual analysis related to the change in the international environment stalled, and the danger of missing opportunities to exploit the intelligence increased.
Problems with analysis would continue to be noted on occasion within the AIC up to 2001, with a few but largely indirect and insignificant improvements. For example, the passing of the Counter Intelligence Reform Act in 2000 that would improve communication of warrant requests from the Foreign Intelligence Surveillance Court. However in terms of the size and role of the AIC - it was a drop in the analytical bucket that could possibly improve secondary analysis by providing faster raw intelligence for sub-hypotheses testing, but again only in situations necessitated by a counter intelligence activity.14
Another example can be found in the US Commission on National Security in the 21st Centurys’ final draft report of Jan.31, 2001. The report stated that “US intelligence has at times, been unable to respond to the burgeoning requirements levied by more demanding consumers trying to cope with a more complex array of problems.”15 However their subsequent recommendations for executive actions produced very little in the way of improving the integrated strategic level analysis of the AIC, though it did stress the importance of HUMINT.
The problems noted before Sept. 11 within the AIC may be summarised as follows: The confederated structure was not conducive to the complex contextual analysis required by the sudden end of the Cold War and an international system where political, economic, ethnic, religion, technological, and military issues became intertwined.16 The traditional AIC methodology, was losing its capability to fuse together a sufficient contextual understanding for extracting the significance from collected intelligence that in turn reduced their capability to predict the ‘what next?’ Essentially the events of Sept.11 confirmed the theoretical problems within the analytical capability of the AIC identified before Sept.11. It further provided the impetus for the formal investigation/reviews of the AIC that quickly found the physical evidence of the problems. Internally within the confederated model of the AIC analysis apparatus, analytical fusion was being weakened by poor communication between confederated bodies. For example the CIA did not send pertinent information to the FBI on suspected terrorist al-Mihdhar.
Even if they had sent it, is not certain it would have received any attention by FBI analysts who were unable to place other similar tactical level intelligence (the possibility that flying schools were inadvertently training terrorists) provided by its own agents, within a strategic level context. It is a clear example of how both the analytical structures and methodology of the AIC could produce analytical lapses. Sept. 11 made it clear that the AIC was weakened by its own structural and methodological disjunction at a time when threats from the international environment where more complex and required greater analytical fusion. So what was to be done about it?
The AIC Reacts
The solution that presently appears to be on course for adoption partially follows the same philosophy that the CIA adopted in 1998 - except applied to the whole of the AIC and Law Enforcement community. That is more efforts to improve allsource analysis and more HUMINT. To accomplish this the President has proposed a new Department of Homeland Security (DHS) that will bring a single analytical focus to homeland security. It is to merge under a single department the capabilities to assess threats to the homeland, map threats against US vulnerabilities, and take the appropriate action to protect the US. It has further been suggested this AIC structure will provide the platform for a much more strategic level of intelligence analysis that will include all AIC sources, with further access to both private and public data.17
As it concerns terrorism the new department’s role will be to “translate assessments about evolving terrorists threats” including the targeting, training, and doctrines of terrorists overseas into an effective system of homeland protection. The three pillars of the new Departments strategy against terrorism include aggressive HUMINT collection, global offensive actions to deny sanctuaries to terrorists, and better all-source analyse on the home front with close co-ordination with the FBI.18
It appears to be a very good plan and here is why. The main problem that was facing the AIC in the 1990’s was that it was facing not only a wider range of possible threats but also threats of a more complex character than they were used to. (The Cold War world was a much slower moving target than the post-Cold War world.) Therefore on the methodological scale it was the external international environment that was developing faster than the analytical process forcing it towards the descriptive side of the scale, and down towards the operational level of intelligence analysis. It was a trend that reduced the AICs’ analytical capability to conduct more complex analysis and undermined the strategic level flexibility for forward planning.
In terms of re-establishing a methodologically sound approach, the new DHS should constitute a large swing back in the direction of the prediction end of the methodological scale, by bringing the whole of AIC all-source analysis to bear on threat identification. The dynamics behind this rejuvenation will lie in the consequent enriching of the analytical context for the processing of raw intelligence (Fig. 4.0). Thus as new intelligence arrives there is a lesser chance that the significance of data may be lost, creating more forward planning options in terms of focus and resource delegation, while strengthening the ability of the AIC to deal with the complexity of the international environment. In other words strengthening the strategic level analysis of the AIC.
This analytical improvement however is still very much reliant on the practical structural improvements within the AIC. The structural problem before Sept.11 was rather simple in that the loose confederated model of the AIC did not support analytical fusion enough relative to the complexity of the problems they were facing in the international environment after the Cold War. (In areas of duplicated interest such as terrorism, some may argue that AIC elements may have become competitive, turning the confederated model into a de facto competitive model at the cost of ‘significance’ identification.) The new DHS should correct that structural fallacy by reigning in the AIC agencies through a level two command and control authority for the purposes of analysis. The question is whether or not the new DHS will receive the co-operation expected of the entire AIC to provide the intelligence necessary to maintain the desired contextual benefits?
It appears that in the initial stages planners have thought of this challenge and have proposed several solutions by which to deal with it. The AIC itself has proposed to support the new DHS in three main areas: The first is information sharing where the intelligence community has promised to make available all HUMINT, TECHINT, and /or all- source intelligence assessments. Secondly, the AIC has promised to support the new DHS, by promoting ‘connectivity’ - physically this translates to larger numbers of inter-agency personnel exchanges. Thirdly, a fresh idea, in that the AIC will support the new DHS with analytical training and development. This is a good idea for both parties in that it will promote analytical fusion while allowing for third party opinions on institutionalised analytical processes. (Perhaps 3rd parties may see opportunities for improvements that those inside the institutionalised process could not see.)
To conclude, the affect of Sept.11 on the AIC was to finally remove from the analytical bodies the last of the methodological limitations and influences of an uncertain post-Cold War environment. In terms of threat identification it has brought the deeper issues surrounding the importance of ‘context’ to the analytical process, to the forefront. This should help with the identification of new threats - whether they are based on religion, money, politics, or something completely new, and do so earlier before they become too well organised. The new DHS should provide the structural support for these methodological improvements in the AIC. Therefore the affect of Sept.11 has been to insure the modernization of the AIC into a more effective intelligence analysis organisation. In many ways it represents the final chapter of the AICs’ own period of globalisation, a dramatic ending to a decade long period of adaptation, learning, and paradigm adjustment. It now looks like the AIC may be back on track.
1 For a thorough examination of many of these issues before Sept. 11 see Gregory F. Treverton, Reshaping National Intelligence for an Age of Information (Cambridge, Uk: Cambridge University Press, 2001.)
2 See testimony by Director of Central Intelligence George J. Tenet before the Senate Select Intelligence Committee Hearing on National Security Threats to the United States 6 February, 2002.
3 Cord Meyer, Facing Reality: From World Federalism to the CIA (Lanham, Md.: University Press of America, 1982): pp.360-388
4 This includes signals intelligence better known as SIGINT.
5 Also the relationship between FE and PET here in Denmark as there is a clear demarcation of areas of responsibility.
7 See Barry M. Katz, Foreign Intelligence (London England: Harvard University Press, 1989.)
8 Sherman Kent, Strategic Intelligence for American World Policy (Princeton, N.J.: Princeton University Press, 1949.)
9 Wilmoore Kendall, “The Function of Intelligence,” World Politics 1:4 (July 1949)
10 Russ Travers, “The Coming Intelligence Failure,” Studies in Intelligence Vol. 01 No. 1, 1997.
11 Ibid., pg.2.
12 Despite the creation of the Director of Central Intelligence’s Counter terrorist Centre in 1986, in terms of integrated analysis throughout the AIC it was not having much affect.
13 This could mainly be due to US action on the 25 recommendations from the Paris Terrorism Ministerial of July 30, 1996, where several actions were taken by the US to improve information exchange at the strategic, operational, and tactical levels.
14 See Senate Rpt. 106-352 – The Counter Intelligence Reform Act of 2000. (Printed July 20th, 2000.)
15 The United States Commission on National Security/21st Century. Road Map for National Security: Imperitive for Change. Final Draft Report, jan.31, 2001.
16 See “US intelligence efforts fractured” BBC, 18 May, 2002 for a light discussion on related issues.
17 See Testimony of the Director of Central Intelligence before the Government Affairs Subcommittee, 27 June 2002.
18 Testimony of the Director of Central Intelligence before the Government Affairs Subcommittee, 27 June 2002
PDF med orginialudgaven af Militært Tidsskrift, hvor denne artikel er fra: