Speaker: Koven, B. (University of Maryland START)
Date: 11 April 2018
SMA hosted a speaker session presented by Dr. Barnett Koven (University of Maryland START) as a part of its SMA CENTCOM Speaker Series. Dr. Koven began the session by presenting some of the problems evident in data analyses. He argued that removing erroneous data is difficult, automated analyses are still devoid of theory and human-like cognition, and we often apply the wrong theory when performing data analyses. Dr. Koven proceeded to explore this last problem further. He spoke about the Loss-Exchange Ratio (LER) and how it often generates misleading results. He also discussed an instance in which applying the “correct” theory can give you a better understanding of data. He used post-ISAF Afghanistan as an example and applied the rationale presented by Stathis Kalyvas in his book The Logic of Violence in Civil Wars. Dr. Koven noted that competing theories should be evaluated and considered when looking at data. Finally, he presented a few possible extensions of his research, including creating a predictive attack model to predict the timing and/or the location of violence during Taliban GIRoA negotiations.
Dr. Koven’s relevant 2017 Small Wars Journal article entitled “The End of Afghanistan’s Spring Fighting Seasons and the Demise of the Afghan National Security Forces?” can be found at: http://smallwarsjournal.com/jrnl/art/the-end-of-afghanistan%E2%80%99s-spring-fighting-seasons-and-the-demise-of-the-afghan-national-secu