Can Operations Research Help Find Terrorists?

The debate over how to respond to the threat of domestic terrorism is shaped by the fear that an unknown number of terror cells is hidden among the civilian population. Yale SOM’s Edward H. Kaplan uses queuing algorithms to estimate how many cells exist and determine how to efficiently combat them.

The terrorist attacks late last year in Paris and San Bernardino, California, put a fresh spotlight on the threat posed by terror cells living, and plotting, among civilian populations.

The debate over how to prevent domestic terror is reignited every time there’s a new attack. In January 2016 alone, the New York Times reports, there were seven attacks directly linked to, or inspired by, ISIS, in locations as diverse as Indonesia, France, Egypt, and Pennsylvania.

The fear of such attacks is amplified by the fact that much about the threat is unknown—since by their nature, active plotters try to avoid detection. Edward H. Kaplan, the William N. and Marie A. Beach Professor of Operations Research at the Yale School of Management, has used innovative and surprising methods from operations research to try to estimate how many undetected plots may exist at any given time.

Kaplan’s work applies the principles of operations research to this question by framing the issue as a “queuing problem,” employing algorithms more typically used to determine how many call center agents are needed to efficiently handle incoming calls, or how many beds are needed in a busy hospital.

“The actual risks are very, very low in this country,” Kaplan says. “People seem to think that there could be hundreds or even thousands of terrorists just waiting to do something in the country. My analysis suggests that on average the number of terror plots might be around three undetected plots at any point in time. Maybe as an upper limit, five to seven, but not hundreds and not thousands.”

Using data from court proceedings, undercover agent reports, and plot timeframes, Kaplan’s research looks for the optimal approach to applying counterterror agents in balance with other needs.

“The real question becomes, how do you keep the country safe by mounting an appropriate response,” he says, “as opposed to going overboard and investing way too many resources into this area at the expense of other things that government and public agencies are supposed to do for us?”

Stressing that his research was conducted before the rise of ISIS, and that the factors impacting attack rates are fluid, Kaplan says his model can help policymakers as they make resource-allocation decisions.

“Obviously, you can’t catch everything, and we saw the terrible results in Paris recently, and unfortunately I’m sure we’ll see more of those,” Kaplan says. “But it’s all a question of keeping it in balance and having a system where you can efficiently detect and act upon detection once the information is there.”

Socially Efficient Detection of Terror Plots” by Edward H. Kaplan appeared in Oxford Economic Papers in January 2015.

Optimal Control of a Terror Queue,” by Andrea Seidl (Institute of Statistics and Mathematical Methods in Economics, Vienna University of Technology), Edward H. Kaplan (Yale School of Management, Yale School of Public Health, and Yale School of Engineering and Applied Science), Jonathan P. Caulkins (H. John Heinz III College, Carnegie University), Stefan Wrzaczek (Institute of Statistics and Mathematical Methods in Economics, Vienna University of Technology), and Gustav Feichtinger (Institute of Statistics and Mathematical Methods in Economics, Vienna University of Technology) appeared in the European Journal of Operational Research in January 2016.

William N. and Marie A. Beach Professor of Operations Research, Professor of Public Health & Professor of Engineering