OR Forum: Intelligence Operations Research - The 2010 Philip McCord Morse Lecture


In the November-December issue of Operations Research Ed Kaplan writes about the subject of his 2010 Philip McCord Morse Lecture, “Intelligence Operations Research” (full article available in articles-in-advance).  Here he discusses applications of operations research to intelligence problems in national security and counterterrorism.  As he illustrates in his review of the literature, this is a distinctive problem area to which he has made notable contributions but also offers many opportunities for new research with the potential to improve the security of our society.

Invited Comments 

The editors have invited comments on this article from three experts military and/or intelligence analysis.

Moshe Kress ( ezembed

KressComments -Password-protected content.

pdf KressComments ) Dr. Moshe Kress is Professor of Operations Research at the Naval Postgraduate School (NPS), where he teaches and conducts research in combat modeling and related areas. His current research interests are counter-insurgency modeling, sensor deployment and operations, homeland security problems, and UAV employment in IW situations. His research has been sponsored by DARPA, ONR, USSOCOM, JIEDDO and TRADOC. He is the Military and Homeland Security Editor of the OR flagship journal Operations Research. Dr. Kress has been twice awarded the Koopman Prize for military operations research (2005 and 2009) and the 2009 MOR Journal Award.

Mitchell Silber ( ezembed

SilberComments -Password-protected content.

pdf SilberComments ) Mr. Silber is Executive Managing Director at K2 Intelligence in New York. Before joining K2, Mitch served as Director of the Analytic and Cyber Units in the New York Police Department’s Intelligence Division, where he supervised the Department’s entire portfolio of ongoing terrorism-related investigations.  Before that, he reported to the NYPD’s Deputy and Assistant Commissioners of Intelligence, managing strategic threat assessments and the Department’s relationships with international police and intelligence agencies. He has a Masters Degree in International Affairs from Columbia University, and BA in European History and Economics from the University of Pennsylvania.

Lawrence D. Stone ( pdf StoneComments )  Dr. Stone is the Chief Scientist of Metron where in the past he has been Chief Operating Officerand Chief Executive Officer  His technical work has included modeling the operational Anti-Submarine Warfare (ASW) effectiveness of nonacoustic sensors and developing tactical decision aids for ASW search and localization.   He is a coauthor of the 1999 book,Bayesian Multiple Target Trackingand continues to work on a number of detection and tracking systems for the U. S. Navy and Coast Guard. Dr. Stone is the recipient of the 1975 ORSA Lanchester Prize for his book,Theory of Optimal Search.He isa member of the National Academy of Engineering, an INFORMS Fellow and in 2008 the recipient of the . J. Steinhardt Prize for outstanding contributions to Military Operations Research by the Military Applications Society.


One of the most exciting aspects of Operations Research is the ability to walk into almost any area of human endeavor with the confidence that you have a world view and methodological toolkit that give you a good chance at making it better, ergo “The Science of Better”. On the other hand, one of the most frustrating aspects of Operations Research is that often you are an outsider.  As an outsider one tends to struggle to understand the domain being studied and be sure that the problem you are focusing on is both relevant and significant.  But what if an accomplished OR modeler gives you a map of the main operational activities in this domain as well as a careful description of the organizational structure?  Surely having such a guide in hand will greatly increase your chances of making an impact in this domain.  In his OR Forum article based on his 2012 Philip McCord Morse Lecture, Ed Kaplan has created just such a guide for “Intelligence Operations Research”


Ed Kaplan has provided an excellent and useful overview of the intelligence process within the United States.  This overview provides operations researchers with a framework for understanding the intelligence process and for determining where they can provide useful assistance to the Intelligence Community (IC).  In the first four sections of his paper, Kaplan defines intelligence, tell us who produces it, and steps the reader through the intelligence cycle which involves, planning and direction, collection, processing and exploitation, analysis, and dissemination.

Lawrence Stone

While Kaplan has given a good mapping of the intelligence landscape that categorizes problem types both Mitch Silber and Moshe Kress highlight an additional important distinction in intelligence work, namely timeliness. 


Given the IC’s significant focus on terrorism since 2001, the collection and analysis of tactical intelligence has been the highest priority and timeliness has been considered mission critical.

Mitchell Silber




The article mostly addresses peacetime intelligence and focuses, on the one hand, on homeland security (e.g., “how many terror plots are in progress?”) and, on the other hand, on the national-strategic level (e.g., NIPF and NIEs). Intelligence during combat operations has somewhat different characteristics that may require different treatment. First, the time scale of the intelligence cycle during combat is much shorter; it may be measured in hours, and perhaps even in minutes.

Moshe Kress

The message here is that an Operations Researcher both can work on intelligence problems that are more strategic related to planning over relatively long time horizons or on tactical problems that require real time responses.  The same intelligence cycle may exist at both levels but different time scales lead to very different types of problems and methods.

Beyond the focus on tactical results Mitch Silber points out that there is an inherent lack of incentives to tackle organizational and protocol changes the IC.  Therefore I expect that the dearth of OR capability within the IC, that Kaplan has observed, is not likely to change by fiat from above.   This suggests that the best path for OR to make inroads into the IC is by proving itself on the tactical and real-time level.  This can lead to hiring of OR trained staff who can eventually become advocates for and producers of the strategic analyses.  In many ways this situation is not much different than that seen in other industries.  In manufacturing and logistics OR has had a long history of providing solutions to fundamental issues and has been the basis for competitive advantage. This has been particularly true in the military sphere.  Intelligence is akin to the service sector where the role of OR has been less obvious and less of a tradition.  However, as the scale of the information processing that the IC has to do has exploded and exactly because it has developed a more tactical urgency, with terrorism, the IC is ripe for application of OR making Kaplan’s article very timely.

The commenters also identify some additional directions for research picking up on some of Kaplan’s themes.  Lawrence Stone identifies Bayesian analysis as a method that can readily be applied in intelligence. 


A starting point could be the Analysis of Competing Hypotheses (ACH) method in which an analyst develops a matrix.  On the rows are observations or “facts.”  The columns represent potential explanations or hypothesis.  For each observation and hypothesis, the analyst indicates in the matrix whether the observation is consistent with the hypothesis or not.  The analyst then scans the columns to see which hypotheses are most consistent with the data.  This process is a short step from a Bayesian analysis where the hypotheses are given prior probabilities and the rows corresponding to the observations are converted to likelihood functions where the entries in the matrix give the probability of obtaining the observation given the hypothesis is true.  The matrix combined with the prior will produce a Bayesian posterior on the hypotheses.  The usual objections to Bayesian inference because it requires subjective judgments are moot here.  The present ACH process requires subjective judgments.  The Bayesian process quantifies these judgments and combines them in a principled and consistent manner.


Moshe Kress identifies management of the “information glut” in a timely way as a critical problem in intelligence operations. 


One of the critical processing tasks is screening the collected intelligence items and determine those items that are relevant for analysis. More specifically, given an intelligence query--a concrete question posed to the collector by an intelligence analyst or a decision maker--each item is either relevant to the query or irrelevant (in general, there may be multiple levels of relevance). The collector's task is to pass on to analysts as many relevant items as possible during a limited time period, which is insufficient for screening all the items collected. The objective is to determine the screening sequence of a given length that maximizes the expected number of identified relevant items. This problem can be looked at as a multi-armed bandit problem with two significant distinctions from the classical model: First, the “arms” (items) are not independent, and second, the time horizon is finite. Bayesian updating techniques and machine-learning algorithms may help in this formidable task.