This is the final summary of formal interviews conducted during the course of my dissertation research. I spoke with Dr. Sims on August 20, 2012 shortly after she left her position at Georgetown University for a book project on intelligence and the international system. What follows is my summary of our discussion, which may contain uncorrected errors or misinterpretations of her points given that I was unable verify my summary with her. Nevertheless, I found the discussion very informative and valuable.
Background: Jennifer Sims (JS) is currently working on a grant to continue to develop her theory of adaptive realism, which incorporates intelligence activities into international relations theory. She was previously the Director of Intelligence Studies at Georgetown University’s Security Studies Program. JS also served as Deputy Assistant Secretary of State for Intelligence Coordination, the Department of State’s first Coordinator for Intelligence Resources and Planning, and on the staff of the Senate Select Committee on Intelligence.
Discussion: On August 20th, 2012 I interviewed JS to discuss her theory of adaptive realism and the intelligence community more generally. As with other interviews, our conversation was far reaching, open and discursive. Thus, the following summary presents several key points that came up in conversation rather than provide an exhaustive transcript of our exchange.
One of the first questions I asked JS was about producer/consumer relations. She noted that the community must adopt a wider perspective regarding its role in policy making and that without the authority and expertise to examine US actions, goals and capabilities it cannot adequately determine what developments in the international system are truly threatening or identify opportunities to be exploited. Thus, JS argued that the intelligence community must increasingly perform net assessments and broaden their analytic focus to include assessments of the relative strengths and weakness of the US (blue) and foreign actors (red). From her perspective, the context of how the US and others compete or interact is crucial for fulfilling the analytic mission of the community, and this context cannot be established without carefully studying and considering blue as well as red.
Her emphasis on interactive context is important from a methodological perspective, particularly when considering the employment of Agent-Based Models (ABMs). Whereas the traditional analytic paradigm within the intelligence community is to study capabilities and intentions, ABMs place actors into an interactive and often strategic context that has its own independent effects on the system. Factors such as the order in which agents act can affect outcomes, despite the fact that they are not properties of agents’ capabilities or intentions. Indeed, in many models, agents with homogeneous intentions and capabilities can nevertheless produce ranges of distinct outcomes, indicating that other features of the system are in play and shaping results. Thus, ABMs might play an important role in broadening intelligence analysis and collection by adding the interactive structure through which interactions occur to the list of actor specific capabilities and intentions already examined.
JS’s emphasis on net assessment also revealed the need to expand the reach of the intelligence community to include nontraditional, domestic consumers. She noted that one of the problems with strategic warning prior to 9/11 was that the National Security Advisor, Condoleezza Rice, had no domestic constituency and therefore was unable to alert parties outside of the foreign policy community, such as the FAA, about the threat posed by al Qaeda. JS noted that a richer set of intelligence community consumers, especially those now embedded in Department of Homeland Security, would have given intelligence analysts and policy makers greater flexibility to respond to the threat posed by al Qaeda.
JS’s view of intelligence is distinctive due to her emphasis on strategic interaction, secrecy, and deception. Both she and Jim Bruce (interviewed the week earlier) have given considerable attention to these topics, particularly secrecy, denial and deception, and the integration of intelligence collection and analysis, yet each has emphasized different aspects of analytic efforts. Whereas Bruce focused on epistemology, the scientific method and hypothesis testing in order to understand the quality of analytic judgments, forecasts, and insights, JS has focused on whether intelligence information provides decision makers with a competitive advantage relative to their rivals. The implications of their differing emphasis is stark—Bruce’s concerns are about how we know whether intelligence estimates are justified based on empirically verifiable truths, while JS’s concern is whether intelligence consumers possess a relative advantage over their rivals, accepting that both may possess inaccurate or illusory perspectives.
As we spoke, JS and I repeatedly returned to the subject of strategic intelligence, particularly its importance to consumers and difficultly to produce. She noted that many changes were required within the intelligence community in order to redress existing deficiencies.
First, JS noted that analytic methods should first and foremost concentrate on the relationships between producers and consumers. This means placing matters such as achieving the most accurate assessment secondary, behind building trust with consumers and getting them to accept analytic inputs. Importantly, this does not mean that matters such as truth or accuracy are unimportant, but rather that the development and use of strategic intelligence is a dynamic process whose sequence first requires producers to earn the trust of consumers before they will accept their inputs and give alternative, challenging perspectives consideration.
From JS’s perspective, the objective of strategic intelligence to get policy makers out of their existing paradigms and consider alternative perspectives on world events and potential futures. This requires the intelligence community to provide mechanisms for preparing consumers for think through the implications of potential events and warn of their prospective occurrence. This is achieved through the integration of collection and analysis, where each informs and guides the other. Thus, collection tests analytic hypotheses by searching for indicators that would confirm or refute expectations, while providing the information needed to specify, parameterize or calibrate analytic models (both formal and informal) in order to ensure that analysts work from the best available representations of reality. Alternatively, analysis is needed to specify and prioritize intelligence targets against which to focus collection and provide interpretive frameworks for making sense of the data that collectors provide. While strategic intelligence is often considered an expansive act that proliferates the number of scenarios and perspectives to present to consumers, JS noted that the effective integration of collection and operations can also limit the extent to which the intelligence community and policy makers must consider alternative perspectives or worldviews by continuously ruling out developments for which no combination of theory and data can be found. Thus, at the nexus of analysis and collection, intelligence producers continuously develop and suggest alternatives for consumers’ consideration while simultaneously eliminating previously identified possibilities for which there is no evidence to merit continued attention.
A second strategic intelligence issue identified by JS was the movement from estimative to anticipatory intelligence. This largely means transitioning from the delivery of forward-looking yet static assessments of foreign activities and developments, towards a dynamic and interactive evaluation that includes consumers gaming how they might respond if a particular scenario comes to pass. JS argued that successful anticipation is based on starting from speculation to derive the implications of potential scenarios and associated indicators of their occurrence. Once strategic threats (or opportunities) have been identified, getting consumers to buy into analytic assessments about their significance is essential. Afterwards, the intelligence community can search for indicators of different strategic developments and warn policy makers of their occurrence. By placing consumers between the development of scenarios and the search for their occurrence, JS ensured that warning would be accepted by consumers when delivered by intelligence producers because of their prior exposure to the framework against which warning is provided.
As JS and I continued our discussion, we talked about the potential for modeling and simulation to make significant contributions to intelligence analysis. Because of her views about strategic interaction and bounded rationality, already central to her theory of adaptive realism in international relations, she understood a great deal about how game theory could be applied yet was sensitive to the cognitive biases and limits of organizations and individuals. Because Agent-Based Modeling provided opportunities to simulate strategic interactions, including deceptive behavior and the use of private or secret information by decision makers, she acknowledged they could make a contribution, especially in cases where coevolution, learning, and adaptation were the drivers of competitive balances and advantages.
A final topic covered in our discussion concerned the application of models more generally in analytic processes. One way was to focus on gathering sufficient data to properly calibrate or tune models in order to provide projections of contemporary circumstances, or at least precisely specified initial conditions. A second way to employ models was to search broadly across the parameter space in order to identify what outcomes are correlated with what inputs, and whether there are features that are insensitive to parameter choices and structurally determined vs. results that are contingent on specific combinations of inputs or sequences of events. JS noted that both approaches had their merits, but believed that linking models to data would be essential to earn the trust of consumers.