Jon Askonas | University of Oxford

Closed Loop: Structural Sources of Epistemic Closure in Large Bureaucracies

It took the US Army 4+ years to adapt itself to the realities of fighting the war in Iraq. While we are used to bureaucracies moving slow, we ought to be more surprised by failures of learning, not least because it did not take individual officers (junior and field-grade) four years to adapt to Iraq. If a wide spread of individuals throughout an organization demonstrate significant learning behavior and growth in understanding in substantially shorter timespans, why does this knowledge fail to become embedded in the organization, while other kinds of knowledge succeed? I am interested in investigating three phenomena which may hamper learning in large bureaucracies. First, not all modes of knowledge are equally easy to embed in organizational communications; certain modes of knowledge (chief among them, contextual local knowledge, or metis) are simply more difficult to communicate effectively in the media of bureaucratic organization – presentation, memo, report, and meeting (e.g. experiences, intuitions, skills). Second, according to cognitive psychology, different modes of knowledge are experienced, remembered, and interpreted differently. Even where information is effectively communicated, the receiving party may not have the cognitive frameworks to properly interpret it, and his memory of the information will be biased by the medium it is in. Third, there is the question of unknown unknowns; intelligence reporting procedures presuppose which kinds of intelligence will be important (through metrics, intelligence requirements, etc.). Where information falls outside of these built in data-structures, it will travel more slowly through the organization. While each of these biases may be overcome in particular instances, I am also interested in how they aggregate over the entire military hierarchy and over time through biases in knowledge transmission. The fundamental conviction behind this work is that we have not spent enough time thinking about the gap between modern organizational practices and human capacities, nor about what it means for warfighting to have outgrown human-scale organizational systems over the span of fifty or seventy years.

I am still looking to specify and validate an empirical puzzle by identifying the specific components of organizational underperformance that cannot be explained by IR’s existing theories of organizational behavior and designing a research question that gets at that puzzle specifically. Initial investigation has confirmed that the US Army’s adaption to the wars in Iraq and Afghanistan may be such a case. Initial pre-interviews have revealed a number of avenues where informal and bottom-up knowledge identified problems earlier than formal knowledge production processes. Potential empirical cuts at this project include examining intelligence-gathering and disseminating procedures and workflows, investigating the work of the Commanders Action Group and Asymmetric Warfare Group, using statistical models to infer learning in the SIGACTS database or a database of Rules of Engagement (ROEs) and Standard Operating Procedures (SOPs) from battlefield environments, process tracing information flows in the US Army through interviews and simulation modeling, and running statistical analyses of Vietnam War casualty data over deployment cycles looking for evidence of adaptation and learning at the individual and unit level.