Could a citywide network of sensors prevent terrorist use of bombs?

© 2003-2004 Devabhaktuni Srikrishna


Abstract: Sensors have been developed by R&D labs enabling remote (non-intrusive) detection of conventional, nuclear, biological, chemical bombs. They have not seen widespread deployment outside of airports and some transportation hubs for detection of terrorist use of bombs by law enforcement. To understand how the effective these sensors would be if deployed across cities, we study the next best proxy: just how effective have screening techniques at airports been at preventing bombings aboard commercial aircraft? Analysis of the RAND-MIPT terrorism incident database (1968-2003) reveals (1) government authorities have been unable to contain the accelerated rise in fatalities caused by of terrorist use of conventional explosives overall, and (2) by comparison, even imperfect sensor-based screening (passengers, luggage) instituted by civil aviation authorities has virtually eliminated bombings aboard commercial aircraft over the past decade. Therefore, a credible defensive program based on metro-area sensors networks is likely to be effective in detecting and deterring terrorist use of conventional explosives and WMD, even if the sensor coverage is far from perfect.


Could a citywide network of sensors prevent terrorist use of bombs?. 1

Introduction. 1

Underestimating the threat of WMD.. 2

Bombings continue to rise, but aircraft bombings have become increasingly rare. 3

Despite imperfections, airport security has been surprisingly effective. 5

Sensors for conventional explosives and WMD.. 7

Networks of sensors. 8

A cost-model for city-wide sensor networks. 9

Conclusion. 10

References. 10

Appendix A: How do we track trends in worldwide terrorism?. 12

Appendix B: Terrorist Incidents. 14


If you find this research useful, please consider linking to this URL.

or sending comments to


For a shorter version of this paper, see



Widespread use of sensor technology throughout metropolitan areas is not a novel concept. Street lamps in the United States and other countries incorporate photo-sensors that are used to switch them on at night and off during the day time, to save electricity. In this paper, we seek to understand whether widespread deployment of sensors to detect threatening materials, such as explosives, can be effective in aiding public safety authorities in their mission to stop terrorism.


To answer this question, we draw upon the experience of airports where passenger and luggage screens for conventional explosives have been widely deployed in order to see how effective screening is at preventing and deterring terrorism. We charted the number of fatalities due to terrorist attacks both before and after these measures were put in place by using data from the RAND-MIPT terrorist incident database [MIPT, 2002] kept continuously since 1968.[1] We discovered that overall fatalities resulting from terrorist use of conventional explosives have been consistently rising at an accelerated pace from 1968 to 2003. Since 1968, fatalities resulting from use of explosives aboard airplanes followed a similar trend, but have dropped dramatically only in the past decade or so. The timing of this reduction is correlated with specific aviation security acts by the United States Government and specific measures consequently taken by airports.


The risks posed by WMD tend to be grossly underestimated simply because use by terrorists has been historically infrequent. Generalizing from the experience of civil aviation authorities with explosives aboard airplanes, we argue that a credible framework for detection for both conventional explosives and WMD is likely to end up preventing and deterring these forms of terrorism across entire metropolitan areas and beyond. Such a system would need to enable public safety authorities to detect when terrorists are transporting bombs well before they reach their intended target where the maximum harm could be inflicted. This detection system could be created based on distributing remote sensors throughout wide areas that non-intrusively detect the existence of threatening materials. The sensors would need to be tied back to a monitoring facility using a data network.


In Appendix A, we survey the publicly available databases that track terrorism incidents including the RAND-MIPT database from which the data in this study was drawn. In particular, we point out that in contradiction to the US State Department data, the RAND-MIPT terrorism database shows that terrorism rose between 2001 and 2002. While charts of fatalities were employed in this body of the paper, Appendix B shows the corresponding charts involving number of incidents.

Underestimating the threat of WMD

Terrorist use of weapons of mass destruction (WMD), also known as nuclear, biological, and chemical (NBC) threats pose a particularly challenging problem for nations seeking to prevent such forms of terrorism. Shockingly few hurdles stand in the way of a committed and organized terrorist group seeking to assemble the materials and create such weapons. [p. 97, Falkernrath, 1998]. Not only are they are easier to build than military grade weapons, but they can be delivered by modes of civilian transport such as road or sea which militaries are not equipped to defend [p. 100, Falkernrath, 1998]. Evaluation of counter-measures designed to prevent such WMD attacks is inevitably complicated by three facts,

  1. Use of WMD has been historically rare and consequently there is little experience from which to judge the likelihood of future usage patterns by terrorists.[2]
  2. A lack of prior history is not a justification for complacency since several changing factors increase the likelihood of WMD may be used by terrorists in the future.[3]
  3. The attack modes are likely to be far more lethal than any terrorist incidents witnessed to date. A single WMD incident could have devastating humanitarian, economic, and environmental consequences.[4]


Therefore, WMD use by terrorists falls into a class of high-severity, low-probability events for which the human bias is (generally) to dismiss as not worth investing resources in [p. 60, Levenson, 1995]. The paradox of such a bias is that the likelihood of observing such events grows proportional to elapsed time, regardless of how unlikely such an event may be on a given day. Hence a nation-state’s reluctance to investing its resources to prevent such low-likelihood events is tantamount to playing Russian roulette with the lives of its citizens.


Human tendency is also (generally) to invest in preventative measures for the rarest calamities only after they have been personally experienced [p. 61, Levenson, 1996]. As a recent example, we note that it took the events of 9/11 before the cockpit doors of commercial airliners were secured to prevent unauthorized access. Such a measure could easily have been undertaken much earlier, and it was already in place on all airplanes of one airline [Walt, 2001].


Given the unprecedented devastation that could result from WMD attacks and in light of such biases, nation-states must recognize that they cannot depend solely on past statistics to identify and implement the appropriate terrorist counter-measures for WMD use by terrorists. Rather, nation-states have no choice but to anticipate and implement effective techniques to prevent the most lethal and destructive forms of terrorism within the resources available to them. Then the question becomes how can nation-states devise an effective counter-terrorism strategy that encompasses WMD? Success with airport screening again suggests that emulation and adaptation of the screening approach to WMD can offer an effective solution.

Bombings continue to rise, but aircraft bombings have become increasingly rare

Commercial passenger airplanes have historically been a high-visibility target for terrorists, yet statistics indicate that the incidence of aircraft bombings have all but ceased over the past decade. Lessons learnt from this observation are all the more valuable in light of the fact that terrorist incidents involving alternate modes of attack (without explosives) that were not fully appreciated by authorities have increasingly been responsible for much greater fatalities, as had happened with the attacks of September 11, 2001. We compare four time series of cumulative fatalities from terrorist incidents resulting from the following four categories[5],

  1. ALL: all terrorist incidents
  2. EXPLOSIVES: explosives were used or detonated
  3. AIRPORT/PLANE: explosives in an airport or onboard an airplane
  4. AIRPLANE: explosives onboard an airplane.


Figure 1

In Figure 1, we chart the total number of fatalities in these categories by date. For each category, this shows us how the fatality statistic changes over time. From 1989 and onwards, ALL fatalities and fatalities from EXPLOSIVES not only increase, but their rate is increasing over time as evidenced by the increasing steepness of the slope[6]. By contrast we observe that fatalities for AIRPORT/PLANE and AIRPLANE remain nearly constant after 1989, with AIRPLANE fatalities rising by only 14 over the course of 14 years.


Upon a closer examination of the data for categories AIRPORT/PLANE and AIRPLANE, we find there to be two extended periods of comparatively low fatalities preceded by periods of much more intense fatalities. In Table 2 below, we show these periods along with the fatalities for each category. Periods II and IV have much smaller fatalities for AIRPORT/PLANE and AIRPLANE when compared to the corresponding statistics during periods I and III. However, during all periods, including I and II, the number of fatalities in EXPLOSIVES or ALL showed no signs of slowing down. We conclude that during periods II and IV, there was a displacement[7] from airplanes as terrorist bombing targets into other environments.












09/02/1974 to








03/28/1977 to








08/07/1984 to




















Table 1

Despite imperfections, airport security has been surprisingly effective

Out of necessity, civil aviation authorities have pioneered techniques to prevent weapons and explosives from being used aboard aircraft through the use of screening of various forms. Critiques of these counter-measures have focused primarily on less than perfect detection probability of a weapon or explosive as it passes through the screening process at the airport. For instance, [p. 52, Szyliowicz, 2004] remarks on the “porousness” of airport screening by citing the United States GAO finding (July 2002) that fake weapons and explosives had passed through airport screeners a quarter of the time at 32 major airports [p. 2, Dillingham, 2002].  However, if the screening had no effect on terrorist decisions of whether or not to use conventional explosives at airports, the 25% detection failure rate observed by the GAO would have permitted many more fatalities than had actually been observed due to explosives aboard airplanes.


The fundamental fact of economics that resources are limited applies to both terrorists[8] as well as their target nation-states, albeit asymmetrically. When terrorists consider a particular mode of attack, accessibility of the required resources and likelihood of a successful attack in light of counter-measures will be primary considerations. Therefore, it is important to take into account not only the probability of detection resulting from a given counter-measure, but also how well the existence of the counter-measure dissuades terrorists’ from selecting that attack mode. Ultimately what matters is how much terrorism was prevented as a result of a counter-measure rather than specific metrics that describe the measure’s performance.


From the results of the previous section, it becomes evident that airport security and its role in the prevention of bombings aboard aircraft has become a model for prevention of terrorism. Despite rapid increase in fatalities for ALL and EXPLOSIVES, the presence of two periods of accelerated fatalities followed by extremely low fatalities for AIRPORT/PLANE and AIRPORT is telling. A number of social, political, and technological changes taking place during these periods also gave rise to the trends, and it is impossible to completely disentangle all of what is going on to explain them. A plausible explanation for these periods of low fatalities must involve factors that are different between airports and other environments. For instance, improvements in intelligence gathering capabilities would be expected to apply uniformly towards all incidents, rather than airports per se.


The thing that sets airports apart from all other environments is the coincidence with eras during these periods when airport screening programs were expanded. First, following frequent hijackings beginning in the late 1960s, a variety of security measures including metal-detectors and X-Ray machines to screen carry-on baggage were introduced in the United States as part of the Air Transportation Security Act of 1974 [Malotky, 1998]. Indeed, in a study using econometric techniques to analyze data from the ITERATE database, a reduction in skyjackings in the United States could be attributed to the introduction of metal detectors in 1973, while also increasing incidents not protected by the detectors such as those involving hostages [Enders, 1993]. In light of the trends in Figure 2 and Table 2, it is likely that introduction of airport security measures worldwide contributed primarily to the reduced number of fatalities in period II.


Second, following an intense wave of fatalities during period III primarily due to use of explosives in airports/airplanes culminating in the Pan Am airliner explosion over Scotland[9], increased vigilance and better procedures to sniff and screen for explosives in airports that followed appears to be directly responsible for the period of low fatalities beginning in 1989 and continuing on into the present era. This end of this period was also marked by the United States Aviation Security Improvement Act of 1990 [Bush, 1990]. Over the 1990s, increasingly better detection techniques and a greater degree of automation to screen for explosives were introduced at airports worldwide [Malotky, 1998] thereby continuing to raise the bar for terrorists who sought to employ explosives aboard airplanes.


During periods II and IV, there was also a significant reduction in terrorist attempts to deploy explosives in airports and airplanes, even though overall terrorist incidents using explosives continued to rise at an accelerated pace[10]. However, this reduction in incidents was not commensurate with the sharp reduction in fatalities. Hence, we conclude that although fatalities from airplane/airport bombings were successfully prevented despite significant ongoing attempts by terrorists, the screening measures most likely have been successful in deterring a much greater number of attempts that would have taken place had the screening not been present.


If airport security measures have led to an unparalleled reduction of risk from terrorism, then the lack of screening has permitted terrorist use of explosives in heavily populated environments where fatalities have been growing rapidly according to Figure 2. Could it be possible to replicate the comparative success of civil aviation in preventing terrorism to broader contexts?

Sensors for conventional explosives and WMD

Detection systems employed in many of today’s airports are built on technology that requires physical contact (swabbing) to wipe off traces of explosive which can then be analyzed by techniques such as ion mobility spectrometry [Malotky, 1998 and p. 114 Committee on Science and Technology for Countering Terrorism, 2002] within specialized machines. Recent advances in remote sensor techniques enable miniature trace-detection sensors for explosives [Pinnaduwage, 2003] which sniff for explosive particulate matter in the air and recognize specific explosives from a distance, analogous to how a dog does. Other passive and non-intrusive techniques are being investigated involving bulk detection of explosive compounds using low levels of neutrons, RF waves, and vapor-detection [Malotky, 1998 and p. 114, Committee on Science and Technology for Countering Terrorism, 2002]. For an overview of today’s sensor capabilities and other homeland security technologies, please refer to [Committee on Science and Technology for Countering Terrorism, 2002].


Chemical, biological, radiological, and nuclear bombs pose a danger much like explosives, but pose far greater potential for devastation [U.S. Congress, Office of Technology Assessment, 1993 and Levi, 2003]. Much attention has been paid to the doomsday arrival of a nuclear bomb in a container on a ship at a sea-port [Medalia, 2003]. With the tools at the disposal of today’s public safety officials, it’s not difficult to imagine scenarios for undetected transport of threatening materials or weapons of mass destruction (WMD) across and within the national boundaries that have nothing to do with sea-ports, but rather using streets and other entry/egress points. While sea-port security is an important component of a more comprehensive protection plan, it is not sufficient. Security is not achieved until the probability of detection is sufficiently high to deter attacks that involve these threatening materials.


In the case of civil aviation, the early use of dogs to detect bombs ultimately gave way to more efficient techniques and procedures once the need was clear, such as use of scanners for explosives in luggage in the 1990s [Malotky, 1998]. Research institutions whose mission is airport/airplane security have been created that continue to improve existing techniques and introduce better ones, such as the Transportation Security Laboratory [Transportation Security Laboratory, 2004] that was established by the Federal Aviation Authority (FAA) and is now part of the Transportation Security Authority (TSA). By comparison to the use of detection technologies in civil aviation and other industries, the National Academies point out that sensor research funded by the United States federal government has not resulted in significant improvements in counterterrorism capabilities or emergency preparedness for want of an overall development of systems designed for specific purposes, for use by specific authorities, available at specific costs[11]. Once such a framework exists and is in use by governments, only then may new sensor techniques may be improved to meet the most desired characteristics of the system and further sensing capabilities be introduced.


Networks of sensors

One possible approach is to distribute sensors like light bulbs, and along streets on street lamps at as high a density as needed to ensure sufficiently high probability of detection. This would make it increasingly unlikely that a terrorist would decide to transport a bomb. The resulting sensor “coverage” would encompass public and private areas, both indoors and outdoors, such that they can passively detect and locate the presence of bombs. In conjunction with metro-area wireless or power-line data communication networks enabling the sensors to relay their best estimate of an explosive’s location, law enforcement can use that information to non-intrusively locate the bomb, then track it down, and finally defuse it. Work on correlated sensors to gain more accurate location estimate from multiple weak signals is becoming available [Parker, 2001], and it should be possible to deploy systems for entire urban areas with increasing degrees of resolution and accuracy in the coming years.


An approach to deploying sensors throughout populated areas to localize WMD would be necessary to enable local authorities to react in a timely fashion well before they reach their terrorist’s intended target. Steps in this direction are being researched for their feasibility, and the initial results of experiments to leverage cell towers for the placement of such sensors appear promising [Kulesz, 2002]. The cost of deploying these systems for 120 major cities in the United States was estimated at roughly $2 billion [Munger, 2002].


Apart from aiding public safety authorities in doing their job, a sensor network for detecting threatening materials (bombs) could create problems if it had a high rate of false positives. First, false positives increase the inefficiency of operation of the monitoring authority, causing them to chase down bum leads. Second, false positives also impact personal privacy to the degree that authorities conduct searches that eventually turn up nothing. Experience with trace detection systems employed in airports has shown that they have a false-positive (nuisance) rate as low as 0.25%, with a majority of these false positives attributable to explosive residues [Malotky, 1998]. Making the false probability percentage acceptably low would need to be part of the system engineering goals of the sensor network.

A cost-model for city-wide sensor networks

We acknowledge that much work remains to be done in assessing the feasibility of a network of sensors. In this section we seek to establish the basic cost model into which specific inputs based on state of the art sensor technology may be plugged in to estimate the total cost of a metro-area sensor network. Given a fixed budget, this cost model determines the performance and cost targets that the individual sensors must meet for the system to be viable[12]. From this standpoint, range of detection is a critical parameter that describes the capability of any sensor—longer is obviously lower cost since it means fewer sensors. Advances in sensor technology are continually pushing the limits on what’s possible for each type of sensor, including detection distances among other characteristics. Conversely, if requirements on detection distance are known, they can be used to drive sensor research to meet those goals.


Strictly speaking, the maximum distance from which a sensor can detect the presence of its targeted material is highly dependent on its environment as well. However, the effects of the environment can be statistically averaged to produce a distance at which a given probability of detection is ensured. Therefore, we can simply model the distance as a constant number for the purposes of arriving at cost estimates. Considering the spectrum of possible threatening materials ranging from explosives, chemicals, biological agents, radioactive materials, etc., sensors targeted at different materials will be able to detect at distances that vary as well. When considered a system with multiple types of sensors, we would need to use the minimum detection distance across each of the sensors.


If sensors were to be deployed along streets[13] at a regular distance, they could detect the transport of threatening materials with sufficiently high probability to deter attack. An upper bound (conservative estimate) on the number of sensors needed can be obtained from the centerline street miles[14] of the area/region of interest. If the centerline miles in the area are C miles, and the sensors are placed evenly at a distance of at least d miles, then it follows from elementary geometry that the total number of sensors that would be required are approximately C/d.


To gain a more concrete sense for what costs would be involved, we consider the example of San Francisco County and the entire San Francisco Bay Area consisting of nine counties[15]. The total number of sensors needed assuming various values of spacing are given in Table 3.



(d miles)

# of sensors for

SF County

(C=891 miles)

# of sensors for

SF Bay Area

(C=21, 218 miles)

0.017 (100 feet)


1.24 million










Table 2 Number of sensors required at various spacing


To tie this back to costs incurred to operate the sensor network, even at a the smallest spacing we considered of 100 feet and a conservatively high cost of $1000 per sensor (per year) that could detect a variety of threatening materials, it would involve a cost of roughly $51 million to cover and operate for the entire city of San Francisco (per year). The cost of sensors manufactured en masse could drop to $100s of dollars or even much less per year, making the annual total cost of ownership in the range of a single-digit millions. By spacing the sensors further apart, say at each intersection or by random dispersal, the cost could be brought down to several hundred thousand dollars. Since these calculations are rather conservative, actual costs could work out to be even much less.


Based on the rates of terrorist incidents and the resulting fatalities over multiple decades, the case of civil aviation shows that terrorism can be effectively prevented through expanding and constantly improving defensive measures such as screening for explosives at airports. We conclude that extending distributed sensor networks into populated areas is technologically and financially within the realm of possibility. Serious governmental programs to incorporate them are, however, lacking.




  1. Associated Press. 2002. “Buffett offers scary prediction,” USA Today, 6 May. <>
  2. BBC News. 2001. “Indian parliament attack kills 12,” BBC News, 13 December. <>
  3. Bush, G. 1990. “Statement on Signing the Aviation Security Improvement Act of 1990,” 16 November. <>
  4. Committee on Science and Technology for Countering Terrorism. 2002. Making the Nation Safer, Washington. D.C.: The National Academies Press. <>
  5. Coughlin, C. C., et. al. 2002. “Aviation Securiy and Terrorism: A Review of the Economic Issues,” Federal Reserve Bank of St. Louis, September/October. <>
  6. Dillingham, G. L. 2002. “AVIATION SECURITY: Transportation Security Administration Faces Immediate and Long- Term Challenges,” United States General Accounting Office, GAO-02-971T, 25 July. <>
  7. Enders, W., T. Sandler. 1993. “The effectiveness of antiterrorism policies: a vector-autoregression-intervention analysis,” American Political Science Review, 87(4): 829-844.
  8. Falkernrath, R.A., et. al. 1998. America’s Achilles Heel. Cambridge, MA: The MIT Press.
  9. Kulesz, J. 2002. “SensorNet Factsheet,” Oak Ridge National Laboratory, 30 July. <>
  10. Levenson, N. G. 1995. Safeware: System Safety and Computers.  Boston, MA: Addison-Wesley.
  11. Levi, M. A. 2003. “Preventing Nuclear and Radiological Terrorism,” The Brookings Institution, 24 October. <>
  12. Malotky, L., S. Hyland. 1998. “Preventing Aircraft Bombings,” The Bridge, 28 (3), Fall, 1998. <>
  13. Medalia, J. 2003. “Terrorist Nuclear Attacks on Seaports: Threat and Response,” Congressional Research Service, Order Code RS21293, 13 August. <>
  14. Metropolitan Transportation Commission. 2001. “2001 Bay Area Road Miles.” <>
  15. MIPT. 2002. “MIPT Terrorism Database System,” 27 August. <>
  16. Munger, F. 2002. “Sensing a threat: Cell-phone towers could be armed to detect chemical, biological or nuclear hazards,” Knox News, 25 October.
  17. NHTSA. 2003. “National Overview of Recent Highway Safety Data”, 17 November. <>
  18. O’Hanlon, M. E., et. al. 2002. Protecting the American Homeland: A Preliminary Analysis, Washington D.C.: Brookings Institution Press, an updated version (March 2003) is available online at <>
  19. Parker, A. 2001. “Sensing For Danger,” Science and Technology Review, July/August: 11-17. Lawrence Livermore National Laboratory. <>
  20. Pinnaduwage, L. A., et. al. 2003. “Explosives: A microsensor for trinitrotoluene vapour,”  Nature. 425: 474, October 2, 2003.
  21. Szyliowicz, J. 2004. “Aviation Security: Promise or Reality?” Studies in Conflict and Terrorism, 27:47-63.
  22. RAND. 2003. "The RAND-MIPT Terrorism Incident Database," 16 September. <>
  23. Transportation Security Laboratory. 2004. “Homepage.” <>
  24. U.S. Congress, Office of Technology Assessment. 1993. Proliferation of Weapons of Mass Destruction: Assessing the Risks. OTA-ISC-559. Washington, DC: U.S. Government Printing Office. <>
  25. US Department of State.“Patterns of Global Terrorism.” <>
  26. US Department of State. 2002. “Patterns of Global Terrorism 2002.” <>
  27. Vinyard Software. 2003. “International Terrorism: Attributes of Terrorist Events,” 25 November. <>
  28. Walt, V. 2001. “Unfriendly skies are no match for El Al,” USA TODAY, 1 October. <>

Appendix A: How do we track trends in worldwide terrorism?

The United States State Department chronicles terrorist incidents annually in its publication, Patterns of Global Terrorism [US Department of State]. According to this chronology, international terrorists conducted 199 incidents in 2002, a drop of 44% from the previous year. However, analysis based on the RAND-MIPT terrorism incident database [MIPT, 2002] shows the total number of incidents in year 2001 as 1532 and year 2002 as 2631, thus representing an increase of over 70%. In each instance, they employ their chosen criteria decide what incidents are recorded as terrorism as shown in Table 1.


United States State Department

(Incident Review Panel’s definition)


(Research Team’s definition)

“An International Terrorist Incident is judged significant if it results in loss of life or serious injury to persons, abduction or kidnapping of persons, major property damage, and/or is an act or attempted act that could reasonably be expected to create the conditions noted.” [p. 83, US Department of State, 2002]

“For the purpose of this database, terrorism is defined by the nature of the act, not by the identity of the perpetrators. Terrorism is violence calculated to create an atmosphere of fear and alarm to coerce others into actions they would not otherwise undertake, or refrain from actions they desired to take. Acts of terrorism are generally directed against civilian targets. The motives of all terrorists are political, and terrorist actions are generally carried out in a way that will achieve maximum publicity. [RAND, 2003]”

Table 3 Definitions of what constitutes a terrorist incident

It appears that an order of magnitude more incidents were tracked by RAND when compared to the State Department in 2001/2002. These definitions alone offer sparse insight into the cause of this discrepancy between the two databases. In addition to RAND and the US State Department, other public databases for terrorist incidents exist within government and academia. One such database is ITERATE [Vinyard Software, 2003] that chronicles terrorism incidents from 1978 onwards[16]. However, we found that the RAND-MIPT database represents the most comprehensive, longest running, publicly available database of worldwide terrorist incidents, and provides a detailed summary of the incident in a format that is uniform across all incidents including description, fatalities, injuries, location, type of weapon used, terrorist organization responsible, to name a few.


Hence, in this paper, we employ[17] data from the database of terrorism incidents that has been kept by RAND and later MIPT continuously since January 1, 1968[18]. Current until March 12, 2003, the contents of the RAND database have been made available by the MIPT on their website [MIPT, 2002]. In Figure 1, we chart the growth of terrorism incidents according to the RAND-MIPT database.


Figure 2



At the time of writing, this site was still a beta version. With over 15,000 incidents recorded, the database system is not yet absolutely perfect. For instance, at least two well-known incidents appeared to be missing in the RAND database that can be attributed either to bugs in the web-based software system[19] or simply delayed in their process of tracking and following up on incidents[20]. These bugs are being fixed. Nevertheless, the RAND-MIPT database of incidents remains a useful chronology upon which we can draw reasonably accurate conclusions.


Terrorism is frightening for several reasons, perhaps in large part due to the uncertainty inherent in the terms we use to describe it such as its causes, its perpetrators, the methods they employ, their targets, etc. To place terrorism in context, consider that the total number of terrorism fatalities worldwide from January 1, 1968 through March 12, 2003 as calculated from the RAND Terrorism database [RAND, 2003] were less than half of total automotive accident-related fatalities that occur in the United States every year [NHTSA, 2003]. However, in contrast to automotive accidents, terrorism is rising rapidly and there are no actuarial tables upon which uncertainty can be managed or mitigated. Furthermore, when gaming out plausible scenarios for terrorist incidents, the fatalities and cost could potentially skyrocket beyond anything witnessed to date [Associated Press, 2002].


Appendix B: Terrorist Incidents



Figure 3

[1] For an overview of the RAND terrorist incident database used in this paper, see Appendix A

[2] For an illuminating discussion of the history of use of WMD by non-state actors (terrorists) and possible reasons why non-state actors have historically not sought or used WMD as much as one might have expected, see [p. 29-62, ACHILLEES].

[3] For a discussion of what changes are increasing the likelihood of WMD use by non-state actors, see [Chapter 3, ACHILLEES]. The reasons can be summarized as growing ease of proliferation and the rising interest of non-state actors to inflict mass casualties [p. 169, ACHILLEES].

[4] For a good discussion of the likely casualties from terrorist use of chemical, biological, and nuclear see  [p. 150, 152, 157, ACHILLEES]. For the official United States best estimates of the casualties from such an attack, see [p. 52, U.S. Congress, Office of Technology Assessment, 1993].

[5] Note that each successive category is contained by the previous category.

[6] The apparent discontinuity on the ALL graph represents the death of nearly 3000 victims on September 11, 2001

[7] for more on the use of the term “displacement” in the terrorism context, see [p. 2, O’Hanlon, 2002]

[8] See [Enders, 1993] for a seminal analysis of terrorists as rational actors in the economic sense.

[9] The bombing killed 259 people on board and 11 people on the ground [MIPT, 2002]

[10] We show the total number of incidents as a function of date in Figure 3 in Appendix B.

[11] As pointed out by the National Academies [p. 115-116, Committee on Science and Technology for Countering Terrorism, 2002].

[12] In what follows, we model the cost expended per sensor as the dominant cost in the system, with information systems and other centralized systems being secondary costs.

[13] most likely on top of street lamps.

[14] these are the sum total length of streets in the given region or area

[15] Centerline miles for these regions drawn from [Metropolitan Transportation Commission, 2001]

[16] Available for a fee

[17] With the aid of purpose-written automated (Perl) scripts, the entire contents of the RAND-MIPT database was downloaded, parsed, and arranged in a tabular spreadsheet (MS Excel). Having the incident data in this format enabled generation of charts and tables shown in this paper.

[18] Through 1997, only international terrorist incidents were recorded by RAND. From 1998 onwards, both domestic and international incidents were recorded.

[19] bombing of the USS Cole on October 12, 2000 [RAND, 2003]

[20] bombing of Indian Parliament on December 12, 2001 [BBC News, 2001]