The analytics of evil


The role of analytics in combating terrorism

Earlier this spring, I found myself walking through a quiet and peaceful grove of spruce trees south of the small hamlet of Foy outside of Bastogne, Belgium.  On travel in Europe, I happened to have some extra time before heading to London.  I had rented a car earlier that morning, and made the drive south from Brussels to the Ardennes forest.

101st Airborne "Easy" Company foxholes outside of Foy, Belgium
101st Airborne Easy Company foxholes outside of Foy, Belgium - dug December, 1944 are still there today

As I picked my way between the 4-foot-wide holes that dotted the woods, it was hard to imagine what that forest must have looked like 70 years ago.  In these woods, in December of 1944, Europe waged a very real battle for the fate of the world.

These foxholes – still there today – were dug during that brutal winter by 101st Airborne’s 506th Parachute Infantry Regiment, Easy Company as they dug in to hold the critical town of Bastogne for the Allied forces.  Lose Bastogne, and they lose the port of Antwerp.  Lose Antwerp, and the Allies lose their supply lines, likely forcing a political settlement with the Nazis to end the war.  Anyone who knows their history, or has read Stephen Ambrose’s book, Band of Brothers, knows that the 101st did in fact hold Bastogne, despite withering artillery fire from German forces and unbearable winter conditions – turning back the Nazis in the Battle of the Bulge.

As I walked through these woods, it struck me that there was no confusion as to the type of enemy the Allies were fighting here in these woods.  The European story of the 20th century turned around a fight against evil, and its very real threat to the world.  Just the day before my time in those woods, I was speaking in Brussels about analytics at a Europol counterterrorism conference commemorating the 1-year anniversary of the horrific Brussels attacks of March 22, 2016. While I spoke, 52-year-old Khalid Masood used his car as a weapon on Westminster Bridge in London, killing four people and injuring 50.  Another similar attack was attempted in Antwerp at the same time, but thankfully failed. 

As I reflected amongst the spruce trees about Europe’s struggle against evil in the 20th century, there is a new 21st-century existential evil that Europe – and the world – must confront.  And the stories continue: Brussels, Paris, Nice, Berlin, Westminster Bridge, and now Manchester and London Bridge (with thankfully unsuccessful recent attacks in Brussels and Paris, as well as a horrifying attack against Muslims as they left worship in London).

But what can we do about it?  Can Europe use some of the same approaches they used in the 20th century to defeat the Nazis, in their battle against domestic radicalization and terrorism?

Using analytics to keep people safe

When I worked for U.S. Homeland Security after 9/11, my passion was to use analytics to help government make better decisions to keep people safe; here at SAS, that’s still my passion today.  And as a student of history and an analytics practitioner, I know that analytics traces its history back to the great struggles against that 20th-century evil in World War II.  British physicist Patrick Blackett pioneered modern operations research as a way to optimize Allied effectiveness in the war.  From how to best attack submarines and what colors to paint Allied aircraft, to how close B-17 bombers should fly together, Blackett and his team used quantitative analysis and data observations to help military leaders make better strategic, operational and tactical decisions which gave them an advantage over the enemy.  After the Normandy invasion, operations research teams followed the allied advance through central Europe analyzing and optimizing the effectiveness of artillery, bombing and anti-tank operations against the Nazis – and analytics was born.

Today, as Europe and the rest of the world face a new, 21st-century evil in radical Islamist terrorism, the capabilities we have at our fingertips to do what Blackett and his team did – to apply data to our most difficult and challenging security decisions – provide unprecedented opportunities to make real progress in the fight against terrorism and the socioeconomic and demographic challenges that give rise to it.

Can analytics help reduce the unemployment found in many communities in which radicalization has been observed to take place?  Can we use predictive analytics and machine learning to better determine what kinds of conditions make radicalization more or less likely in neighborhoods and communities?  Can analytics reform how prison communities are handled to reduce the likelihood of radicalization?  Can analytics identify the best strategies in our educational institutions for promoting integration and unity starting with our children?  Can we use advanced anomaly detection to better screen for dangerous individuals and cargo at our borders?  Can we apply real-time text analytics to web and social media data to detect radicalization or remote terrorism cell activation?

I think the answer to these and many other related questions is yes.  Much can be gained just from analysis of fully public data, and in a way that respects data privacy in today’s Europe, reflected in the GPDR initiative.  In the same way that the 20th-century threat from the third Reich was met with creative solutions to turn data and observations into insights for decision making, so too can new 21st-century security challenges be met with the same creativity.

From data management to visualization; anomaly detection to optimization; text analytics to machine learning – the new disciplines of analytics present government leadership with a multilayered and multifaceted set of tools to use in the fight.  While we must be prepared to dig in, just as we did in those foxholes 70 years ago, today’s fight requires the flexibility of new 21st-century approaches.


About Author

Steve Bennett

Director of the Global Government Practice

Steve Bennett, Ph.D. is the Director of the Global Government Practice at SAS. With twelve years of experience in the U.S. Department of Homeland Security, he has a passion for improving the speed and quality of public sector decision making through analytics. While he enjoys discussing machine learning, data visualization, and decision theory, he also loves talking faith, family, and woodworking!


  1. Dr. Bennett -

    I am very interested in your work in this area, having written a paper or two on #DataForGood in the past. I am presenting next month at the Society for Terrorism's conference in New York. Will you be there? Perhaps we could talk sometime. My presentation addresses the issue of Reproducibility in statistical research, the subject of recent efforts and a new set of recommendations by the American Statistical Association. The example I am using for the talk - of a necessity something hard to prove, neither certain nor disproven - is an investigation of the hypothetical link between hate speech and hate crimes: the question of whether speech can be clearly identified as a driver of domestic terrorism.

    I hope we can talk more.


    David J Corliss, PhD
    Founder and Director, Peace-Work

    • Steve Bennett
      Steve Bennett on

      Hey David! (and please call me Steve).

      Thanks for your note - would love to connect further about the use of analytics in effective detection, prevention, and response to terrorism. I am unfortunately not going to be at the New York conference next month, but would love to see your presentation, or read any papers you might be able to share. The specific topic of your talk is particularly interesting, since particular speech can be what gets people onto law enforcement 'lists' of potential persons of interest. These lists become quite large (e.g. I believe after the London Bridge attack, the UK indicated they had tens of thousands of people on their list of persons of interest), and it becomes a signal-to-noise problem in which you have trouble figuring out which people are real threats to monitor and which are not. Ways to screen people off these lists (or prioritize them by risk) would give intelligence and law enforcement enhanced tools to focus on the real threats.

      Feel free to ping me on LinkedIn to connect further, more directly: Thanks again David for your comment!


Back to Top