Geography Lessons From the 9/11 Terrorist Network
Mapping the travel geography of terrorist networks can help expose how they operate internationally. Olivier Walther, Joseph Padron, and Jason Scheuer of the University of Florida and Rafael Prieto Curiel of the Complexity Science Hub in Vienna take a close look at the 9/11 plot and find that terrorists who belonged to the same operational cell did not necessarily live in the same place at the same time. However, their itineraries closely matched their organizational structure. Distinct travel patterns and strong social ties not only made the 9/11 travel network resilient but also essentially allowed the 19 hijackers to hide in plain sight while being very mobile.
On June 3, 2000, Mohamed Atta, the ringleader of the Sept. 11 attacks, cleared Immigration and Customs at Newark Liberty International Airport after arriving from Prague, Czech Republic. Over the course of the next year and a half, Atta and 18 other terrorists embarked on a series of trips within the United States, from the suburbs of Phoenix to the ethnic neighborhoods of Paterson, New Jersey, and from the ritzy beaches of Ft. Lauderdale, Florida, to Portland, Maine.
The 9/11 hijackers also traveled extensively internationally, visiting more than a dozen countries and crossing international borders at least 45 times. From the moment they entered the United States until the morning that they killed 2,977 people, Atta and his accomplices each traveled, on average, more than half of the circumference of Earth.
This flurry of travel can help inform an understanding of terrorist networks. Our analysis of the travel patterns of the 9/11 hijackers suggests that mapping the travel geography of members of violent extremist organizations can yield important insights into the group’s overall structure.
Which Terrorists Flock Together?
To find out whether geography really matters, we collected detailed information about the known location of each hijacker using publicly available sources from the 9/11 Commission, the FBI, and the U.S. Congress. In total, we documented 231 trips between 48 metropolitan areas across the world, totaling more than 445,000 miles.
Our study shows that terrorists who worked closely together as part of the same operational cell during preparation for the 9/11 attacks tended to cluster in a few regions of the United States, including the Boston-Washington, D.C., corridor, southern Florida, and the triangle between Los Angeles, Las Vegas, and Phoenix (Figure 1). The entire set of hijackers of the two planes that crashed into the World Trade Center, for example, lived in or visited Fort Lauderdale at some point in time. There, they seemed to have lived relatively “normal” lives—opening bank accounts, visiting Lion Country Safari, and shopping at local supermarkets.
However, the 9/11 hijackers did not necessarily live in or visit the same places at the same time. Our analysis reveals that co-presence may provide crucial information about some aspects of the network but does not really identify the cell-based operational structure that enabled the coordinated, simultaneous attacks aboard different planes.
To demonstrate this, we calculated the number of days shared by each pair of hijackers in the same location. Individuals who spent many days together in the same city are represented with large squares in Figure 2. Our analysis shows that the largest number of shared days is not always found within each cell, colored according to their flight.
For example, several members of the cell that boarded American Airlines Flight 11, which crashed into the North Tower of the World Trade Center, spent as much or more time with some of those who hijacked United Airlines Flight 175, which crashed into the South Tower, and United Airlines Flight 93, which crashed in Pennsylvania.
Space Can Inform Social Structure
The hijackers’ pre-9/11 travel patterns reveal that the itineraries and destinations of the al-Qaeda operatives more closely matched their organizational structure than did co-presence. To demonstrate this, we represented each hijacker as a node in a network connecting different places in the United States and abroad. Hijackers who moved numerous times between the same cities are connected.
We then compared this spatial network with a reconstruction of the hijackers’ social network based on who they trained with, lived with, or had other known contact with before the attacks. As shown in Figure 3, the similarity between the two networks is striking: The spatial structure of the network based on similar trajectories mimics how hijackers were socially connected. In other words, the destinations of their travel are a rather good match for what we now know about the cell structure of the network.
The co-destination of cell members, or their traveling to the same places at different times, was both inevitable given the task and more revealing after the fact. This spatial aspect of the hijacker network was a function of the way in which they were socially organized as four discrete cells with identical tasks to be carried out simultaneously at separate locations. Because of this parallel structure, for example, pilots had to travel to certain flight schools to finish up their training, while “muscle” hijackers, who arrived in the United States less than five months prior to the attacks, needed to be moved to rented apartments to settle rapidly in the country.
This spatial organization also allowed the 19 hijackers to hide in plain sight while simultaneously being very mobile. Some of the early travel patterns of the network were successfully detected by intelligence agencies, such as those of Khalid al-Mihdhar and Nawaf al-Hazmi in Southeast Asia in 2000. However, the intensity and complexity of the domestic patterns followed by the four al-Qaeda cells became apparent only after the attacks, when intelligence about their prior contacts and locations was finally pieced together.
Spatializing Social Networks
Mapping how terrorists travel from place to place provides a much more comprehensive picture of their social organization than simply monitoring their location. The scale and sophistication of the 9/11 attacks did not require the routine co-presence of cell members, but within-cell travel patterns were detectably similar. In other words, hijackers tended to follow the same itineraries across the United States and the world, without necessarily always visiting the same destinations at the same time.
Thanks to recent developments in network and spatial science, the structure, geography, and temporal evolution of terrorist networks can now be modeled with a level of complexity that would have been unimaginable 21 years ago. Geolocalized data, for example, can be extracted from social media to monitor the spatial diffusion of violent events and identify the social structure of the most active accounts, as during the 2012 attack on the U.S. diplomatic compound in Benghazi, Libya.
In years to come, the growing availability of geospatial data and the use of artificial intelligence should make it increasingly easier to detect patterns in terrorist activities. “Knowledge discovery” techniques, for example, are already being developed for the intelligence and law enforcement community to predict the outcome of a specific event, identify hotspots where violence could be concentrated, understand connections between the actors involved, and more generally “make sense” of large-scale data in real time.
One thing hasn’t changed, though. While new technologies have facilitated the exchange of easily codified information, such as dates and names, space continues to exert a considerable constraint on connectivity. Then as now, the preparation of terrorist attacks still requires close linkages that can happen only in certain places at a certain time, if only briefly. Similarities in travel patterns, such as those observed in the 9/11 network, should remain difficult, if not impossible, for international terrorist organizations to hide.