recommendations for improving analysis by the Intelligence Community,
A Blueprint For Survival
The Coming Intelligence Failure
The year is 2001. The Intelligence Community (IC) budget has remained under pressure and manpower cuts have continued, but bureaucratic politics and legislative prerogatives have perpetuated about a dozen national-level agencies and forced a further division of analytic labor. By the turn of the century, analysis had become dangerously fragmented. The Community could still collect "facts," but analysts had long ago been overwhelmed by the volume of available information and were no longer able to distinguish consistently between significant facts and background noise. The quality of analysis had become increasingly suspect. And, as had been true of virtually all previous intelligence failures, collection was not the issue. The data were there, but we had failed to recognize fully their significance and put them in context. At a time when the interrelationship among political, economic, military, social, and cultural factors had become increasingly complex, no agency was postured to conduct truly integrated analysis. From the vantage point of 2001, intelligence failure is inevitable.
Part I: The Path to Failure
Despite our best intentions, the system is sufficiently dysfunctional that intelligence failure is guaranteed. Though the form is less important than the fact, the variations are endless. Failure may be of the traditional variety: we fail to predict the fall of a friendly government; we do not provide sufficient warning of a surprise attack against one of our allies or interests; we are completely surprised by a state-sponsored terrorist attack; or we fail to detect an unexpected country acquiring a weapon of mass destruction. Or it may take a more nontraditional form: we overstate numerous threats leading to tens of billions of dollars of unnecessary expenditures; database errors lead to a politically unacceptable number of casualties in a peace-enforcement operation; or an operation does not go well because the IC is not able to provide the incredibly specific data necessary to support a new generation of weapons. In the end, we may not suffer a Pearl Harbor, but simply succumb to a series of mistakes that raises questions about an intelligence budget that dwarfs the entire defense budget of most countries.(1)
The Community will try to explain the failure(s) away, and it will legitimately point to extenuating circumstances. But we are going to begin making more and bigger mistakes more often. It is only a matter of time before the results rise to the level of acknowledged intelligence failure. It will get so severe that the IC's relevance will be seriously questioned--far more than has been the case to date. The reasons will be simple: we have gotten away from basics--the collection and unbiased analysis of facts. When we do the postmortems and try to reconstruct the broader institutional causes for the failure, we will find them spread throughout the national security apparatus--some a function of this period of history, others a function of mistakes:
Executive Branch. US national security policy will continue to evolve as we adjust to the end of the Cold War. As a result, the formation of security policy will continue to be done on something of an ad hoc basis. This presents the IC with a dilemma: as specific issues come to the fore and as interests and their priorities change, how does the IC focus its limited resources? Any attempt to program resources according to consumer needs is a recipe for getting whipsawed from crisis to crisis and cannot be sustained. Answering consumer questions presupposes a level of knowledge, the maintenance of which must be considered a cost of doing business. We have to come to grips with the fact that the entire "needs" process and associated tier strategy were ill-conceived for this period of history and need to be fundamentally rethought.
Legislative Branch. Congress will bear some responsibility for our forthcoming intelligence failure. It has pressed the Community to end duplicative analysis and achieve a division of labor. This push by Congress has significantly diminished competitive analysis within the Community and should, therefore, be seen as an acceptance of increased risk. There is, however, a more pernicious aspect to this division of labor. By operating under the premise that we can divide intelligence analysis into military, economic, and political subcomponents and then parcel out discrete responsibilities to various agencies, we are sowing the seeds for inevitable mistakes. This artificial distinction has never existed in history, but the IC is going to be expected to operate under such a regimen and do high-quality analysis. We are setting ourselves up to do bad political, economic, and military analysis; by implication, support to all our consumers is going to get worse.
Intelligence Community. Finally, a combination of bureaucratic politics and self-inflicted wounds within the IC will prove to be critical factors responsible for our failure. Many midlevel managers' priorities are misplaced, and loyalty to one's agency too often has primacy. As a Community, we have largely lost track of the view of intelligence articulated by former Deputy Director of Central Intelligence Dick Kerr: "All we bring to the table are facts and analysis of those facts." Management of intelligence is valued more than collecting and analyzing intelligence, and we thus have fewer and fewer good analysts. In the mid-1990s, the IC finds itself filled with individuals who have a tremendous equity in the retention of the current structure. Somewhere in this process, the corporate needs of the country have gotten lost in the shuffle.
[Top of page]
Adding It All Up
Any huge bureaucracy has problems in various aspects of its operation, but, in this case, the most serious is the diminished ability to get the facts straight and to use them as building blocks for high-quality analysis. We are far ahead of any other institution in the world in terms of the ability to collect sensitive information. Corporately, however, the IC is getting to the point where in many instances we do not even know what we do not know. Generally speaking, fewer analysts have less time to read more traffic and still fewer can keep up with their part of an increasingly complicated world; analysts have little opportunity for reflection, much less longer term research (2). Consequently, they stand little chance of putting whatever analysis they do into context--a recipe for irrelevance, if not outright failure. Within our overall analytic effort, a lack of fusion and a lack of objectivity will be principally responsible for the IC failing the nation.
Lack of Fusion. Organizationally, we are not set up to cull critical facts and fuse them into analytic products that respond directly to our consumers' needs. This stems from a glut of information, substantial personnel cutbacks that occurred at the end of the Cold War, the retention of Cold War structure in the face of those cuts, and, most important, the division of labor that occurred partly in response to these factors and to Congressional pressure.
This division of labor sounded good in theory, but it has virtually balkanized the Community. We tried to split economic, political, and defense analysis among the various Agencies and to divide defense analysis into discrete elements among DIA, the Service intelligence organizations, and the Commands. (3) This approach presupposes that the various issues falling under the rubric of intelligence analysis are unrelated, can be examined in a vacuum at the various organizations, and can then be added together to produce "the" answer. In reality, analysis is all about context, and the notion of dividing the labor represents the destruction of that context. Now, no agency has either the critical mass of analysts or, in most cases, the charter to look in depth at the political, military, social, economic, and cultural aspects of a problem. In the end, the lack of fusion and integration capability means that the IC "whole" is substantially less than the sum of its parts.
Lack of Objectivity. The second and related problem has to do with our decreasing ability to ensure objectivity. We have inadvertently built a high potential for bias into the system, striking at our integrity and at the core of the IC. By dividing the labor within Defense Intelligence, we have given an increased voice to Command Joint Intelligence Centers (JICs/JAC) and allowed the Service intelligence production organizations to speak virtually for the country on many matters of interest to their particular Service.
There are many brilliant analysts in these organizations, but corporate objectivity can come under severe pressure. First, in each instance they respond directly to a higher authority that has an agenda, one that may involve a competition for forces in the case of a Command, or funding for weapon systems and force structure in the case of the Services. Second, the analysts' narrow charter can result in an excessively conservative risk equation that stems from their perception of what is "really" important. Taken in their entirety, these forces can lead to an overly robust threat portrayal.(4) When each of these organizations is potentially gilding its own particular lily and the division of labor argument has been used to preclude any capacity for quality control, there is no basis upon which to assess risk objectively.
[Top of page]
In addition to the problems with fusion and objectivity, a host of other shortcomings confront the analytic end of the business and will contribute to our forthcoming failure. They are perhaps best illustrated by responding to a series of common myths:
- There are thousands of analysts "out there." In reality, the analytic base is dangerously thin, and we have far fewer people staying abreast of raw data, adding to the corporate knowledge, than is generally believed. We are underinvested in analysts in favor of the non-substantive functionaries necessary to run the IC's multiple agencies.
- Analysts are fungible. The belief that we can meet crises by moving analysts between disciplines has distinct limitations. We can always throw people at a problem, but it is the analysts' training and expertise that will determine our ability to support our consumers. The training time required for wider fungibility is not consistent with a world of rapidly developing crises.
- Technology is our panacea. Technology can help sort and rapidly move information, but finding the right piece of data, assimilating the information, and putting it in context is never going to be the job of a machine. Although Intelink is a powerful tool, it is only as good as the information that is loaded on it. And the notion that we can "simply" use technology to fuse work being done at disparate locations is an idea held by those who have never been analysts.
- We just need to manage the IC a little better. For the ever-growing number of functionaries in the Community, intelligence is about management. Apparently assuming that analysis just "happens," a disturbing number have little knowledge of or interest in the substantive end of the business. They believe that they can quantify everything, and they are intent on studying the IC to death. At a time when we should be reducing overhead, we are increasing such investment in the mistaken belief that we can manage ourselves out of this mess.
- We are on the right track. This myth is based on the fact that we have not yet failed. Nonetheless, we are operating on borrowed time, living off residual expertise, and not recapitalizing. Electronic databases and our overall command of the facts are falling into disrepair. As we fine-tune our structures and marginally change our programs, we are, in essence, getting the deck chairs on the Titanic nice and neat.
[Top of page]
Part II: Avoiding Failure
The very real constraints on IC management in the early 1990s left it with few choices, none of them good. Now, however, it is time to stop pretending that the current structure can work and start acknowledging the full extent of the problem. The system is built on fallacious assumptions about what intelligence analysis is and how it is carried out; as such, the system has to fail. To fix it, we will need to refocus on the analytic process and establish a structure that actually facilitates analysis rather than impedes it. (5) The following questions and answers provide some insight into the necessary fixes.
What Does Intelligence Do?
We need to accept the fact that this country does not have an actionable national security policy that the IC can use to program analytic and collection resources. This is not an indictment--it is simply a fact. We are just beginning a major debate about who we are and how we relate to the rest of the world; that debate could easily continue for a decade without anything close to national consensus being achieved. This reality implies that a "needs" process that presupposes the consumer community knows what it wants (and, more important, will want), in priority order, is inherently flawed and has to be substantially revised.
The lack of an actionable national security policy means that to varying degrees we have to "do the world." Despite the practice of the last several years, the IC does not have the luxury of deciding it just will not do things. We will always be concerned about the ways other countries can threaten us and our interests militarily. At the same time, it is apparent that in many parts of the world there is an ongoing shift in the definition of state power away from military strength, and this will cause high-level consumers to become increasingly interested in an entire range of nonmilitary issues.
Certainly, there need to be some threshold decisions, including the extent of our role in economic or environmental intelligence. Beyond such basic issues, there are few factual matters associated with a broad interpretation of US security interests that should be beyond our purview. To meet such a wide variety of needs, our role should be that of an information clearinghouse capable of addressing all the security issues of the early 21st century; as such, we have to maintain worldwide expertise or know enough to know where to get it.
With the exponential increase in information, the number of politically motivated pundits and opinions is also increasing. Who is a purveyor of accurate information and who is simply repeating uninformed platitudes? Ultimately, the policy community is going to look for someone in the government who will provide an unbiased assessment of the varied and multifaceted aspects of security--from the nitty-gritty details of databases to the facts and assessments underlying an evaluation of a foreign country's national security policy. Much of this will be unclassified, but much will also be available only from our unique sources. We have to be able to fuse these varied sources of information into a coherent story. If the IC does not do it, who will?
To fulfill this information-clearinghouse function will require us to be far better attuned to the work being done in the academic community, other governmental institutions, industry, and the myriad of other entities that collect and analyze data. While we are making strides in this area, most would agree that we could do much better. For example, we could introduce rotationals from academia into the midlevel ranks of the IC. Moreover, while it will be controversial, I believe we should have a much closer relationship with responsible journalists, extending beyond the "backgrounder" process to a more routine give-and-take among professionals interested in accurate information.(6) Such interaction will need to be thought through and carefully controlled, but we need to accept the fact that the press often has better access and insights than we do.
[Top of page]
What Threats Will Confront US Interests?
Before discussing structure, it is necessary to review briefly the nature of the future security threats to US interests. This will ultimately say a great deal about the IC's future, and particularly about the amount spent on intelligence. A vastly exaggerated version of those threats only confuses the discussion and could lead to a false sense of security about the future size of the IC.
This excessively worst-case approach has both regional and technical components. Regionally, we will allegedly be confronted with two Major Regional Contingencies (MRCs), a regenerated Iraq and a North Korea that somehow survives into the 21st century; Iran will be the center of Islamic extremism and capable of sustained military operations around the Persian Gulf; and a potential peer competitor--either Russia or China--looms on the horizon. Technologically, we will be confronted with a rogue state with an ICBM; R&D on most fourth- and fifth-generation systems will be completed, and procurement will occur in militarily significant numbers, thereby confronting us with the risk of technological inferiority; a revolution in military affairs (RMA) and information warfare (IW) will be new challenges to the nation's security. And, beyond these challenges, we will also be confronted with all the transnational "ism" threats and low-intensity problems so prevalent throughout the world.
But such a forbidding future will not come to pass. Instead, the backward view from 2010 is likely to be one when the United States enjoyed an extended period in which the major military challenges were substantially diminished from those of the early- to mid-1990s.(7)
Strategically, Russian nuclear forces will drop regardless of START II, and Chinese forces will remain at second-strike, deterrent levels. Regionally, there will be, at most, one MRC. Russia will continue to be a basket case for much of this period, though still assertive and nationalistic. Iran will suffer self-inflicted socio-economic woes, while continuing to foment subversion around the Persian Gulf and pursuing weapons of mass destruction. China's economic growth will continue to benefit the military, but not at a breakneck pace. Technologically, because North Korea will be long gone, no Third World country will have an ICBM capable of reaching any part of the United States. Many of the foreign systems in R&D will either be drawn out extensively or die on the vine from lack of procurement dollars. RMA will be a total bust. And, other than some legitimate concerns with the continental United States's vulnerabilities, the mid-1990s fascination with IW will be tempered substantially.
Despite these optimistic trends, other threats will prove to be very real in the first decade of the 21st century. Terrorism will have come increasingly to our shores. Proliferation of technologies associated with weapons of mass destruction will have continued. In the lesser developed world, crises will erupt routinely. Eventually, there will be spillover problems affecting US interests, particularly if the world has not started addressing the severe pressures on the Third World before the turn of the century. As for classic military threats, regional actors will have limited capabilities that could impact on US concerns: Iran in the Persian Gulf, China on its periphery, perhaps Russia against an expanded NATO, and so forth.
[Top of page]
How Should We be Structured?
The defense budget is going to suffer additional cuts as the nation begins to realize that it is running out of major enemies, is confronted with a less challenging technological future, and is faced with real problems that may not always have military solutions.(8) Once this happens, the IC budget is living on borrowed time. Because intelligence is a force multiplier and provides early warning in a time of military force structure cuts, it is a safe bet that we will not have to absorb a pro rata share of these cuts. Nonetheless, more budget cuts are inevitable. The IC needs to start thinking smaller--perhaps substantially smaller.
At the turn of the century, we are going to find ourselves with an IC that is dysfunctional because of a misguided effort to decentralize missions that cannot be decentralized (fused, integrated analysis) and with a system that will be too big to sustain in light of the security environment confronting the United States. And the obvious questions follow: how does the Community ensure it has sufficient analysts to "do the world"; how do we reverse the fragmentation of analysis; how do we ensure fusion and objectivity; and how do we do all this when we are already underinvested in analysis and facing a further decline in manpower?
[Top of page]
The answer lies in increased consolidation and centralization of responsibilities: first, because it is the only way the system can hope to work, and second, because, as in the private sector, "mergers" are the way the IC can slim down and shed overhead in a period of increased resource constraints.
All Source Production
In examining proposed future structures, I will again limit discussion to the major all-source production elements,(9) with the key question being the degree of consolidation. As previously indicated, the reality is such that there are good analysts and bad analysts in all the agencies, and, consequently, I believe that the cost of a Community is a worthwhile safeguard. But, if we are going to invest in all of the overhead associated with having an IC, then make use of it. Having a Community and then effecting a division of labor among its members fails the common sense test. If we have decided to eliminate redundancy (competitive analysis), we should also decide to eliminate the Community; if we are not going to get the benefits of a Community of disparate, reasoned views on the same subject, why pay the huge price of all the management and overhead that go with all the individual agencies? Instead, put one person in charge and save a lot of money and the efforts of thousands of people. We would be much more effective at a much lower cost to the taxpayer.
If we do opt to maintain a Community, it will mean that we have accepted significant amounts of overlap. CIA needs to retain sufficient military and technical expertise to ensure that it can analyze Country X's national security policy in some detail; CIA analysts should no more be a hostage to DIA's military analysis than should DIA be forced to rely on CIA's political judgments. That said, CIA will not do the kind of detailed order-of-battle analysis that should be the province of the Defense intelligence community.
Similarly, DIA will not do microeconomic or pure political analysis, but it has to retain enough political (security-issues-related), and economic (principally defense-resource-related) analytic capability to provide a holistic view of Country X's national security policy. To some degree, DIA also has to retain the kinds of cultural, ethnic, and religious expertise that are becoming increasingly important in understanding the reasons for, and the nature of, post-Cold War conflict.
Such a regimen would re-establish a true Community in which agencies would have a legitimate basis upon which to debate analytic positions. In the same vein, should the country decide it wants one super-intelligence agency, then some mechanism would have to be established to institutionalize alternative analysis. A substantially strengthened National Intelligence Council could fill this role, but there would have to be some entity within the national production elements that avoids the risk of being trapped by groupthink. Formally chartering a "Team B" concept to play devil's advocate would also be a worthwhile investment.
[Top of page]
Consolidating Defense Intelligence
In some ways, the political versus military (CIA versus DIA) issue masks a more obscure but far more acute problem within the Defense intelligence community. Whether we ultimately decide to retain a national-level community or to compress everything, one change should be implemented immediately. Independent military service intelligence production organizations at the national level should be eliminated--merged with DIA--and a single organization subordinate to Office of the Secretary of Defense should be established.(10)
This change would address the lack of fusion and the lack of objectivity that otherwise will lead to intelligence failure. By putting one person in charge and getting rid of all the problems associated with the division of labor, we would create an organization that could do true fusion analysis. By getting out from under potential Service biases, there would be substantially less concern about lack of objectivity. Moreover, this entity would take back responsibility for all databases from the Command JICs/JAC (making a virtue of necessity because it is readily evident that most will never be able to fulfill their database responsibilities); this would leave them free to do exactly what they were always going to do anyway--whatever the CINC or Command J2 wanted. The Command Intelligence Centers would no longer have responsibilities for national-level production. They would be welcome to argue their cases, and even include their views in NIEs. Because the Command is hardly an unbiased observer of the threat in its area of responsibility, however, it would not be unduly influential in shaping the position of an NIE. National concerns over politicization would be allayed, further fixing the objectivity problem.
Finally, as a result of a consolidation of Defense intelligence, tremendous savings would accrue in diminished overhead as the number of functionary positions would drop precipitously. It would allow us to strengthen our analytic capacity to "do the world" and to provide the necessary surplus to accommodate the inevitable mandated reductions in the future.
For a long time, we have been misleading ourselves with regard to our abysmal tooth-to-tail ratio. We are smothering analysis with the huge amounts of administrative overhead associated with a multiplicity of agencies. By eliminating the hidden unemployment and returning those billets to actual analytic responsibilities, we could demonstrate that the number of people is not--and never has been--the problem. These people are not doing the right things, and consolidation would be the remedy.
[Top of page]
A Tough Transition
In the transition period for any major IC consolidation, we would lose effectiveness for a substantial time and would be operating for years at less than optimal performance, potentially. This alone suggests that we should concentrate first on Defense intelligence production consolidation--and only then go on to further reorganization. If we try to tackle an entire Community restructuring in one step, it would entail substantial risk, leaving the United States Government without a fully functioning intelligence apparatus. While we effect consolidation within Defense intelligence, it would fall to the CIA to ensure that the government has its finger on the "big-picture" pulse during the lengthy transition. Once Defense intelligence is consolidated, further steps, perhaps some consolidation of various collection functions, could then be made. At any rate, restructuring the Community would be complicated and would need to be phased in carefully to reduce the risk of the very intelligence failure we seek to avoid.
[Top of page]
Back to the Future
As we look ahead to 2001, the United States will have the luxury of at least several more years in which we will be in pretty good shape internationally. The world will certainly continue to be messy, and there will be innumerable instances where the United States could choose to engage militarily. With the possible--and increasingly less likely--exception of North Korea, however, the direct major military threats to United States national security interests will be virtually nonexistent for the rest of the decade. How long into the 21st century that will continue is anybody's guess. Accordingly, if we are going to try to fix an IC that is ill-equipped to analyze the complexities of today's world, we should start soon.
The kind of restructuring that is required will take a huge short-term toll on our effectiveness, and it will take a substantial time for the dust to settle. That in itself may encourage those ultimately responsible for restructuring the Community to leave the existing structure largely intact. But waiting is not a viable alternative--at least not one that will work over the long term. Either we fix it now in hopes of being in a position to support America's intelligence needs at the beginning of the new millennium, or we fix it later--under the cloud of failure.
[Top of page]
(1) The combined total annual defense budgets of North Korea, Iran, Iraq, Syria, Libya, and Cuba approximate half that of the US Intelligence Community.
(2) The disparity between what we collect and what we process gives rise to the concern that we are overinvested in technical collection at the expense of processing and analysis. Archiving is often suggested as a response, but, given the lack of in-depth research now being done, the utility of this approach is at least open to question.
(3) An example from Defense intelligence might help with this abstract notion. DIA retains responsibility for analysis of Country X's threat perceptions, national security outlook, and infrastructure, and has a limited capability to analyze forces. The Services retain most scientific and technical responsibilities and analyze the ground, air, and naval forces of Country X, including future systems that X might deploy--even though they have no capacity to compare Country X's interservice priorities or to incorporate defense economic constraint analysis. The responsibility for Country X's database may reside at a Command, but the Command has little expertise in other areas such as whether and how (in terms of logistics and C3, for example) Country X will fight a war. In other words, all organizations have a piece of the puzzle, but none can put the puzzle together.
(4) This in turn is picked up by those with a political agenda and leads to shopping among the intelligence agencies in search of the appropriate level of threat for an avowed political goal.
(5) The various plans being considered to revamp the IC do little more than make passing references to analytic issues. While aspects of these plans undoubtedly have merit, unless and until we tackle our analytic shortcomings none would substantially diminish the likelihood of intelligence failure.
(6) This should not be confused with the recent debate over whether to use journalistic cover for clandestine agents.
(7) By implication, IC efforts focused on our ability to support two MRCs are misguided. To be looking at a requirement for post-2000 intelligence capabilities based on two MRCs is a recipe for making the wrong investment decisions.
(8) Those who choose to believe that the "sum of all our fears" future is more likely should take no solace from the possibility that the defense and intelligence budgets might be larger; all the same problems with the IC will pertain. Whether we retain something like the current size or are forced to downsize much further than we already have, the failure is inevitable unless we begin to analyze the underlying problems.
(9) I will not actually address State/INR, but this organization shows what can be done with a small number of high-quality people. Its influence is far greater than its size would suggest.
(10) This would require a legislative change to Title 10, which the Services have used to justify a production element as part of their man, train, and equip functions. Under the optimal approach, the former Service subordinate centers outside Washington would be slimmed down and would revert to basic S&T intelligence. A large fusion center in Washington would consist of regional, transnational, infrastructure, and technical components, with separate regional elements that address current (J2), basic, and longer term estimative work (a robust staffing of regional offices would ensure proper warfighter support by maintaining a ready pool of experts available for crisis task forces). Overall, this approach would provide for a natural progression of analysts who begin in more basic database analysis or in areas which focus on the simple accumulation of facts. During their careers, they would gradually work up a pyramid of complexity, reflecting the fact that portfolios range substantially in difficulty.