This post continues the series of interviews I performed during my dissertation research on Agent-Based Modeling, intelligence analysis and policy-making. My interview with Leon Fuerth was the first in the series, and provided me several insights that shaped my research and subsequent interviews (most of which have since been posted already). Importantly, this is one of three final interviews where the subject was not able to review the transcript and provide any clarifying remarks. While I do believe that my write up accurately captured his comments, it is possible that interpretive errors do exist that have not been corrected.
On September 21, 2011 I was fortunate enough to talk to Leon Fuerth (LF) about his time in the White House, the role of models in policy, and intelligence analysis from a consumer’s perspective. LF served as the national security adviser to Vice President Al Gore, and served on the Principals’ Committee of the National Security Council, alongside the Secretary of State, the Secretary of Defense, and the President’s own national security adviser while in that position. Prior to that, he worked for Gore in the Congress, spent eleven years in the State Department’s Foreign Service, and served in other positions within the government over his more than thirty years in public service. He is the founder and the director of the Project on Forward Engagement, which seeks to address long-term, complex national security problems through the development of adaptive and anticipatory governance. In addition, LF is a Research Professor of International Affairs at the George Washington University’s Institute for Global and International Studies, and a Distinguished Research Fellow at the National Defense University’s Center for Technology and National Security Policy.
LF’s insights were quite refreshing for reasons that will be noted below. Importantly, a unifying theme bridged all of his responses to my questions, which largely weaved together a sense that the world is, and will always be, an uncertain place where free-will exists and frustrates policy-makers’ plans. What follows is a synopsis of our conversation, where answers have been combined and reordered for narrative clarity, rather than a transcript of our discussion.
The Roles of Models in Policymaking:
LF’s view of models and modeling was generally supportive, but different from many government sponsors or users that see models as opportunities to predict future outcomes. He was wary of the idea that human affairs could be predicted with high degrees of accuracy, and thought that efforts were better spent employing models to illuminate the kinds of unintended consequences or adaptive changes that might occur in a complex system given a particular policy choice. Additionally, he identified the value of working from models to discover particularly important levers or opportunities for influencing others.
Three challenges to the use and roles models were identified. The first challenge was practical—policy and modeling often move at different paces, and the time required developing and understanding a sufficiently rich and complex model that is credible with decision-makers may not be compatible with the speed of executive decision-making time-horizons. Likewise, increasingly sophisticated applications and explorations of models may generate more information than policy makers can digest, placing a heavy burden on new methods for distilling complex models and their behaviors into more familiar formats or new, highly efficient packages.
A second challenge philosophical—modelers and policy makers may become overly invested in particular models or worldviews that hinder their ability to develop durable or robust policies. This over vesting occurs when models and their results are confused with real-world findings. This often occurs at the intersection between abstract theories or concepts, supported by logical models or deductive proofs, which are then seen as sources of policy solutions to complex problems. LF immediately noted that many economic models that have driven policy, such as the Laffer Curve or the J-Curve. Indeed, any model that produces a permanently valid forecast has likely been misunderstood by those advocating its application. This definition of a model overlapped with that of a theory or abstraction, which could quickly transition from a helpful cognitive aid to a misleading and dangerous source of policy solutions if interpreted literally and applied uncritically.
A third and related challenge follows from the second—if there is not a permanently valid and predictive model of the world, then policy makers and modelers must both proceed with caution. The notion that models will continuously improve and predict with greater accuracy, rather than plateau or even decline in predictive power as situations and behaviors change, leads to an uncomfortable argument against free will. People are free to adapt, change their minds, behave unexpectedly, and so on. No model can capture all possible actions and options that the actors may consider. Moreover, the implications that a model has identified the drivers of future events implies that policy makers are powerless to alter the course of events—in which case, their actions have no consequences.
Given these challenges, models can provide a more supportive role by serving to illuminate alternative perspectives and trajectories of the international system, and focus more on understanding the sources of their dynamics than predicting outcomes. Uncertainty and risk cannot be eliminated, but policy makers can be made increasingly aware of their sources. From this point of view, models may produce more fruitful and useful results by focusing on the exploration and illumination of potential unexpected consequences, and the countermoves of others in the system that limit policymaker’s influence or power.
Being a Consumer of Intelligence Analysis:
LF’s perspective on intelligence support to policy provided a necessary counterpoint to the usual discussions of intelligence analysis that tend to reflect a producers viewpoint. An interesting and problematic fact of intelligence studies and efforts to reform the community in particular, has been the focus on producer’s side of the producer/consumer relationship. While reforms have introduced new technologies, organizations, guidelines for personnel matters, etc., there has been little examination of how policy makers actually use and interact with intelligence analysts. There has been a one-sided focus that shifts blame and responsibility onto collectors and analysts, while shielding policy makers from not acting whenever intelligence is not deemed actionable. LF’s perspective on intelligence from a consumer’s perspective provided an alternative revealed how intelligence can better serve the needs of their customers.
To start with, LF noted that his career in international politics preceded his time in the White House, and included time spent in the State Department’s Bureau of Intelligence and Research. He believed that this background gave him a considerable understanding how the intelligence community operated and allowed him to have a better working relationship with the intelligence community than other policy makers who have spent their formative years in other professions and have never dealt with the intelligence community before serving in the White House or other senior policy-making organizations.
What largely separated LF from other policy makers was his placement of the burden on policy makers for the success or failure of intelligence analysis. He emphasized that the quality and utility of analysis he received was directly correlated with the quality of the questions he asked. Moreover, he felt that once policy makers became angry upon the delivery of bad news or analysis that disagreed with their beliefs the relationship could be irreparably harmed. Instead, continued dialogs and subsequent questions were necessary to provide analysts with a context for understanding policy-makers’ interests, and revising collection and analysis in a way that could improve its timeliness and relevance.
This view was largely distinctive from other policy makers in that LF viewed interacting with the intelligence community as an opportunity, rather than a cost, and he noted that analysts tend to define their identity around the ability to provide policy makers with useful, relevant assessments on difficult, complex problems. Making analysts aware of the key challenges facing policy makers, allowed them to tailor their responses to his needs.
Another distinctive feature of his response related to the modeling questions as well, and that was the belief that intelligence serves to reduce uncertainty. This argument is often advanced in the intelligence studies literature, but may in fact be misleading. Indeed, LF expressed doubts as to extent the world could be rendered knowable or predictable, so there would always remain a cloud of risk and uncertainty. The existence of such epistemological limits meant that intelligence should not necessarily provide products focused on prediction, but rather be capable of remaining continuously engaged in the policy-making process, tracking changes in the international system and helping policy makers adjust their perspectives in fluid, dynamic situations. In a manner consistent with his views on asking questions, the character of producer/consumer relations is defined by continual interaction and personal relationships, not the delivery of products.
A final point that he made was on the speed of policy making and the ability of the intelligence community to keep pace with the needs of decision-makers. Again, reiterating the importance of personal and continued relations, LF noted that the multiple layers of management in the intelligence community could often work against the interests of consumers by slowing analytic production down to the point that ceased to be timely. For example, an official response to a question could take several days to get through the entire bureaucracy, while an unofficial response may be returned in hours. Understandably, these layers ensure that the community speaks with one voice, and that analytic products bring the widest possible set of information and expertise to bear on a problem, but LF suggested that skilled analysts that were fully engaged on a problem were consistently aware of what their peers knew and thought, and appropriately and responsibility caveated their unofficial assessments. In such cases, the layers of bureaucracy slowed down the delivery of analysis to consumers and largely served the interests of the producing organization, creating a track record and review that protected senior managers from criticism at the expense of timeliness. Other discussions with retired intelligence professionals indicate the direct relationship between analysts and policy makers is one of the most difficult and complex relationships to manage.
Importantly, his characterization of the consumer’s perspective, however unconventional, reflected the conclusion commonly reached by many intelligence scholars—that the relationships between producers and consumers may be the single most important factor in determining the quality and utility of intelligence analysis and that efforts to improve tradecraft, technology, organizational structure, etc. should be viewed through the lens of this relationship.