This article was downloaded by: [Princeton University]On: 17 September 2013, At: 11:03Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Studies in Conflict & TerrorismPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/uter20

Revisiting the “Problem From Hell”:Suicide Terror in AfghanistanHenry Romea

a Department of Politics, Princeton University, Princeton, NJ, USAAccepted author version posted online: 26 Aug 2013.Publishedonline: 26 Aug 2013.

To cite this article: Henry Rome (2013) Revisiting the “Problem From Hell”: Suicide Terror inAfghanistan, Studies in Conflict & Terrorism, 36:10, 819-838, DOI: 10.1080/1057610X.2013.823752

To link to this article: http://dx.doi.org/10.1080/1057610X.2013.823752

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Studies in Conflict & Terrorism, 36:819–838, 2013Copyright © Taylor & Francis Group, LLCISSN: 1057-610X print / 1521-0731 onlineDOI: 10.1080/1057610X.2013.823752

Revisiting the “Problem From Hell”: SuicideTerror in Afghanistan

HENRY ROME

Department of PoliticsPrinceton UniversityPrinceton, NJ, USA

A hallmark of the resurgence of antigovernment forces in Afghanistan was the massintroduction of suicide terrorist attacks. Between 2005 and 2006, the incidence of suicidebombings increased more than fivefold, marking a sea change in the Afghan conflict.Despite the initial jump in the volume of suicide attacks and the fear of more violence,the number of attacks actually flat-lined while the level of other attacks increased. Thisstudy will argue that it is the competency of the attackers employed in Afghanistan, notthe politics, technology, or targeting, that best explains the static level of suicide attacks.

On 3 August 2006, a Taliban fighter detonated a suicide car bomb in the middle of a crowdedmarket near Kandahar, Afghanistan, killing 21 Afghan civilians. Analysts later assessed thatthe fighter may have tried to target a North Atlantic Treaty Organization (NATO) convoythat was passing through the town. However, the convoy was some 400 meters away atthe time of the detonation. It is unclear whether the civilians were targeted, or whether theTaliban employed an undisciplined bomber who detonated the device prematurely.1 Thistype of suicide attack has repeated itself across Afghanistan over the past eight years—thesuicide attack, foreign to Afghanistan, has become a powerful tool for antigovernmentforces to kill civilians and military forces alike. And, similar to this 2006 attack, there isoften little clarity as to which entity was the intended target.

The study of suicide terrorism in Afghanistan, like the incidence2 of the attacks, peakedin 2007 with a landmark study by the United Nations Assistance Mission in Afghanistan(UNAMA). In the report, Tom Koenigs, Special Representative of the UN Secretary-General for Afghanistan, wrote: “I hope . . . that this study in itself is not the final word onthe matter, but the start of a wider exploration of what we can all do to protect Afghanistan,its neighbors, and the world, from this true problem from hell.”3

The destruction caused by suicide attacks, not only to the immediate victims but tothe fabric of Afghan society, should not be understated. Nevertheless, this study concernsitself with a surprising fact: attacks have flat-lined. While suicide attacks jumped between

Received 1 December 2012; accepted 28 April 2013.The author thanks Prof. Kosuke Imai, who provided invaluable advice on this article, which

began as a junior year independent research project at Princeton University. The author also thanksAlex Acs and two anonymous reviewers.

Replication data has been uploaded to the Harvard Dataverse Network.Address correspondence to Henry Rome, Department of Politics, Princeton University, 130

Corwin Hall, Princeton, NJ 08544-1012, USA. E-mail: hrome@alumni.princeton.edu

819

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

820 H. Rome

2005 and 2006, the sheer number of attacks remained essentially the same in 2007, 2008,2009, and 2010.4 The motivating question for this study, then, is not why suicide attacksin Afghanistan have increased so much, but why suicide attacks in Afghanistan have notincreased more.

An assumption undergirding this analysis is that suicide attacks have significant power.Over the past decade, terrorism scholars have recounted the lethality and effectiveness ofsuicide attacks. UNAMA and Hoffman present data reflecting the stunning kill counts ofsuicide attacks: UNAMA shows that attacks using suicide vests kill or injure 81 peopleon average, and suicide car bombs average 98 victims. Hoffman’s numbers, while moreconservative, still reflect the power of suicide attacks: suicide attacks kill four times as manypeople as other attacks and, in Israel, suicide attacks kill six times as many people and injure26 times as many people. Suicide attacks have been variously described as the “ultimatesmart bomb,” a “human weapons system” and a “poor man’s nuclear weapon.”5 Further,scholars have articulated the value of suicide terrorism in terms of strategic signaling aboutthe capabilities of a group or the impotence of the government. These theories bolster theassumption that if suicide attacks are as effective and unstoppable as scholars articulate,terrorist organizations should continue to order operatives to use such attacks.

This article, which reviews terrorist attacks against non-combatants in the years2005–2010, approaches this question from a variety of angles. First, this study exam-ines what, if any, relationship existed between the levels of international troops and thevolume of suicide attacks in Afghanistan. It has been reported that the deployment ofinternational troops to the south and east correlated with the peak of suicide attacks, andthis trend will be evaluated. Second, this study assesses whether antigovernment forces areskilled at suicide bombings; a static level of suicide attacks is little consolation if the attacksare getting deadlier.

This study concludes that there is little evidence to support the idea that suicideattacks in Afghanistan are the preferred tactic of antigovernment forces. To the contrary,in the presence of more coalition forces in a particular region, attackers have chosen touse Improvised Explosive Devices (IEDs) instead of suicide bombs, despite the purportedoverwhelming benefits of suicide missions. In addition, antigovernment forces have failedto improve the effectiveness of suicide bombs. At the same time, antigovernment forces inAfghanistan have continued killing non-combatants in record numbers using other tactics.

This study concludes that the lack of skill of individual attackers, more than politics,technology, or targeting, can explain this shift away from suicide bombings to other tactics.While this study is not the first to argue that Afghan suicide bombers are incompetent, thisstudy links attacker incompetence to the strategies of attacking organizations.

The study will proceed as follows. First, definitions and theories about suicide terrorismwill be reviewed. Second, the history of suicide terrorism in Afghanistan will be detailed.Third, the database used for this study, the U.S. National Counterterrorism Center (NCTC)’sWorldwide Incidents Tracking System (WITS), will be introduced. Fourth, the two casesintroduced above—troop levels and attacking skill—will be examined. Finally, the articleconcludes with a discussion and outlook.

Definitions

The 2006 attack in Kandahar embodies a fundamental question: What kind of attacks oughtto be considered “terrorism”? This article specifically studies suicide attacks that are usedas a terrorist tactic. In this study, “suicide terrorism” will be used quite intentionally. Thisstudy intentionally differentiates “terrorism” from “insurgency.”

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

Revisiting the “Problem From Hell” 821

Scholars have struggled to develop a uniform definition of terrorism, given the apho-rism that “one man’s ‘terrorist’ is another man’s ‘freedom fighter.”’6 Indeed, some scholarshave found more than a hundred definitions used throughout government and scholarship.7

In this article, the definition employed by NCTC will be used: “incidents in which sub-national or clandestine groups or individuals deliberately or recklessly attacked civiliansor noncombatants (including military personnel and assets outside war zones and war-likesettings).”8 Defining what constitutes a suicide attack is a controversial endeavor as well.What does it mean to be a suicide attacker—must one have the intent to die in an attack?NCTC’s definition seeks to avoid attempting to get into the head of the attacker:

The Suicide flag indicates that the perpetrator was either part of the device(e.g., wearing a bomb vest) or was commanding the device with the intentionof dying at the time of the attack (e.g., a car bomber driving a vehicle bombinto a checkpoint). The perpetrator must have died as a result of the attack.9

The WITS definition is consistent with those used by other databases and scholars (seeAppendix for full WITS methodology).10

The events in Afghanistan make applying these definitions challenging. A majority11

of suicide attacks in Afghanistan appear to target foreign or domestic military entities.Exclusively attacking the military, however, does not generally qualify as “terrorism.”Nevertheless, antigovernment elements have continued to use suicide tactics that mayostensibly target military units but instead kill far more civilians.12 This trend has causedobservers to determine that these attacks are acts of terrorism. For example, three of themost rigorous databases that catalog terrorist attacks—WITS, the University of Maryland’sGlobal Terrorism Database (GTD), and the RAND Database of Worldwide TerrorismIncidents—all have concluded that the 2006 attack on the Kandahar market should beconsidered an act of terrorism.13

This conclusion is reflected in how human rights groups categorize these attacks, too.While Human Rights Watch (HRW) does acknowledge attacker incompetence and doesaccount for the fact that antigovernment forces sometimes kill civilians as collateral damagein a military attack, HRW nevertheless argues that antigovernment leaders are fully awareof these realities. “The willingness of Taliban and other insurgent commanders to continueto deploy in highly populated areas a weapon—suicide bombers—that in practice is highlyindiscriminate amounts to a serious violation of international humanitarian law.”14 Theseinstances are included as terrorist attacks namely because, as the UNAMA report reiterates,the leadership of antigovernment elements is fully aware that the attacks will kill civilians,and leaders intend that outcome.15

This article also purposely includes all suicide attacks in Afghanistan, as opposedto attacks for which specific groups have claimed responsibility. While this approachruns the risk of falling into the “unitary foe” trap, this study argues that looking at allattacks can reveal a more accurate portrayal. UNAMA reports that extremist groups—namely, Al Qaeda, the Taliban, Hiz-i-Islami, among many others—are “eager” to claimresponsibility for suicide attacks.16 But, scholars and UNAMA17 itself acknowledge thatgroups are sometimes hesitant to claim responsibility for the deadliest of their attacks.18

Further, the fragmented nature of bands of fighters claiming to act on behalf of extremistgroups complicates efforts to attribute responsibility.19 The degree of fragmentation alsocould decrease the actual analytic value of attributing the attacks to a particular entity,when the group is not uniform and is not centrally directing attacks.20 Even if properacknowledgment is articulated, it may not even be acknowledged by the database. For

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

822 H. Rome

example, the Taliban publicly claimed responsibility for the 2006 attack in Kandahar butthis claim was not recorded by either RAND or GTD.21 WITS reported that while “[n]ogroup claimed responsibility . . . it was widely believed the Taliban was responsible.”22

In that vein, this study adopts the United Nations definition of “antigovernment ele-ments” (AGE):

These encompass all individuals and armed groups currently involved in armedconflict against the Government of Afghanistan and/or international militaryforces. They include those who identify as “Taliban” as well as individuals andgroups motivated by a range of objectives and assuming a variety of labelsincluding the Haqqani network, Hezb-e-Islami and al-Qaida affiliates such asthe Islamic Movement of Uzbekistan, Islamic Jihad Union, Lashkari Tayyibaand Jaysh Muhammad.23

This study is careful to not ascribe specific motivations to groups but to comment moregenerally on the actions of antigovernment elements.

Suicide Bombings in Afghanistan

In order to examine the current state of suicide attacks in Afghanistan, it is important tobriefly trace the development of the tactic. The first suicide attack in Afghanistan was the9 September 2001 assassination of Ahmed Shah Masood at the direction of Osama binLaden, in anticipation of a U.S. invasion.24 After 9/11, U.S. and Northern Alliance forcesdrove the remaining elements of the Taliban into Pakistan. By 2002, the Taliban began a“traditional insurgency”25 and were re-cast as the “Neo-Taliban.”26

Al Qaeda, not Taliban, forces are believed to be responsible for two attempted suicideattacks in Kabul in 2002.27 The following year, Al Qaeda operatives detonated two suicidebombs, including one in a car, killing police and International Security Assistance Force(ISAF) troops.28 Nevertheless, the attacks thus far, while surely relying on logistical supportin Afghanistan, were not carried out by Afghans.29 That changed in early 2004, when ayoung Afghan blew himself up near ISAF troops. While there were two additional attacksin 2004, the tactic was still not catching on among Afghans and was mainly perpetrated byforeigners.30

The suicide attacks, while few, caused tensions within the Taliban leadership. The AlQaeda tactic contradicted the Pashtunwali code that forbade suicide, and the leader of theTaliban, Mullah Omar, initially resisted executing suicide attacks.31 It is debated why theTaliban dropped its reservations. Williams, in several works, identifies two primary reasons:the persuasion of Al Qaeda and the embrace of technology from Iraqi insurgents.32 He writesthat Al Qaeda’s “greatest victory” was convincing the Taliban leadership to “embrace thecult of martyrdom,” while the influx of veterans from the Iraqi insurgency provided thetechnical expertise to build suicide bombs. Johnson notes that, similar to any other new andpowerful tactic, the Taliban was able to integrate it quickly into their ongoing operations.33

However, it is important to note that unlike the pre-2001 “Taliban,” the antigovernmentelements in Afghanistan, including the Taliban, are now very disparate. As Tarzi writes,“There did not appear to be an umbrella organization or a centralized body directingactivities, but instead several independent groups loosely linked by their drive to oustthe foreign forces in order to establish their own strongholds of power.”34 This accountchallenges the importance of Al Qaeda representatives’ persuasion of Mullah Omar; hisapproval or disapproval may have been irrelevant. While Omar was the near-mythic leader

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

Revisiting the “Problem From Hell” 823

of the Taliban in the 1990s, the extent to which fighters respect and obey his wishes todayis “questionable.”35

In any event, the number of suicide attacks grew rapidly and, increasingly, the attackswere carried out by Afghans themselves.36 By 2006, following a series of peace accordswith Pakistan, the Taliban secured operating space in the Federally Administered TribalAreas (FATA) in Pakistan for training and “large-scale indoctrination of suicide bombers.”37

Suicide attacks jumped rapidly from 16 in 2005 to 90 in 2006, then leveled off. There were106 suicide attacks in 2007, 103 in 2008, 99 in 2009 and 103 in 2010.38 These attacksprimarily appeared to target military or government locations, but instead predominantlykilled civilians.

Data

This study primarily uses data collected by NCTC, but also uses data from four othersources to back its conclusions.39 In this section, the NCTC database will be discussed indetail, and the other databases will be introduced.

NCTC maintained WITS as a comprehensive database of all terrorist attacksworldwide.40 It contained incident data from 1 January 2004 to 31 December 2011. Becausethe data from the fourth quarter of 2011 was in the process of being uploaded at the timeof this analysis, the data from all of 2011 was excluded. NCTC took down WITS in May2012,41 but the author maintains a full backup of the database.

The authors of WITS acknowledged that maintaining the database can be “more artthan science; information is often incomplete, fact patterns may be open to interpretation,and perpetrators’ intent is rarely clear.”42 The other publicly available databases, such asRAND and GTD, face the same challenges. WITS only includes data about attacks that itconsiders acts of terrorism. To qualify as an act of terrorism, the attack must involve a (1)subnational or clandestine group that (2) “deliberately or recklessly” attacks (3) civiliansor non-combatants.43

For the parts of the analysis that examine ISAF troop levels, the ISAF Troop Placemat isused. The Placemat shows the approximate strength of ISAF forces by Regional Command(RC). Unfortunately, additional granularity is not publicly available, which does limit thestrength of this analysis. Because troop levels are only available by RC, terrorist attacksfrom WITS are sorted into that breakdown. The RCs represent very large sections ofterritory, and it is potentially of concern that the RCs are geographically larger than theorganization of AGE. However, there is disagreement among analysts on this account. Forexample, Smith, writing in Giustozzi, recounts that the “Taliban’s leadership is reputed todivide the country into areas of operation, with the insurgents’ definition of the southernregion roughly mirroring the cluster of provinces in NATO’s Regional Command South.”He also writes that other experts have outlined theories of “parallel insurgencies” not unifiedin a single RC.44 Therefore, while this study is not afforded greater specificity because ofthe sporadic release of ISAF information, there is arguably still validity in examining theRC-level data.

ISAF data is only available by month from January 2007 to August 2010, limitingthe data points to 44 per Regional Command, because ISAF does not consistently releasemonthly troop levels to the public. Further, the data is released sporadically, so only 25months have data points. The months without specific troop values were coded “N/A.” Thefive RCs are: North, West, South, East and Capital (Kabul). In August 2010, a sixth RC(Southwest) was separated from RC South; because that only occurred on one data point,RC Southwest was combined with RC South in the data. In the WITS data, six attacks were

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

824 H. Rome

of an origin that overlapped between geographic areas. Those incidents were removed fromthe analysis.

Troops and Attacks

This section will evaluate the relationship between ISAF troops and suicide attacks, to betterunderstand if the stagnant levels of suicide attacks nationwide over time was sensitive tochanges in troop levels. Then, this section will posit several potential theories.

Antigovernment elements appear to have two primary goals with correspondingaudiences.45 First, they target international military forces in an effort to increase the costsof continued military presence, which could impact public opinion in the home country.Second, they aim to challenge the Afghan government’s law and order capabilities, raisingthe costs of centralized control from Kabul and signaling to local populations their power.46

Suicide attacks also serve to accomplish related goals of fund-raising and recruitment.The rapid rise in suicide attacks against civilians correlated with several conditions: a

technology transfer from Iraqi insurgents to their Afghan counterparts; a planned move byISAF to extend its control over the southern and eastern provinces; the expansion of ISAFinstead of U.S. forces; a planned Taliban resurgence in Kandahar.47 However, beyond thesecorrelates, scholars have not assessed the specific behavior or uses of suicide attacks.

There has been little scholarly research48 about the relationship between troop levelsand suicide attacks, and the research that most famously argues for such a relationship,the work by Robert Pape, is fundamentally flawed.49 The simplest relationship could bethat the increase in presence of foreign troops decreased the capability of antigovernmentelements to conduct any kind of terrorist attack. On the other hand, it stands to reason thatan increase in the presence of foreign troops could drive suicide attacks against civilianshigher, as antigovernment elements seek to demonstrate to the civilian population and toKabul that non-combatants will still die in large numbers (using fearsome suicide tactics)no matter how many well-trained and well-equipped troops are sent in.50

If the former relationship is true, and the capabilities of AGE were degraded, the levelsof both suicide and non-suicide attacks should decline. Alternatively, a decline in suicideattacks relative to other tactics could reflect a decision to use the suicide attacks less whenAGE are faced with increased military presence.

Results

The most obvious conclusion from Figure 1 is that, across Afghanistan, higher troop levelsare correlated with higher levels of suicide attacks against non-combatants (effect size1254.8, std. error 160.4). But this overview provides little insight into whether the increasein suicide attacks was simply a reflection of overall trends of increased violence. As overallviolence increases, are suicide attacks used at the expense of other tactics?

Using the WITS data, this study calculates a “suicide proportion.” The suicide propor-tion is the percentage of all terrorist attacks that involve suicide tactics. When examining therelationship between the suicide proportion and troop levels, the result is quite different. Thelarger the level of ISAF troops nationwide, the lower the suicide proportion. On the nationallevel, AGE use suicide attacks less as a proportion of overall attacks when more troops aredeployed. The relationship, while negative, is weak (effect size –18.86; std. error 18.22).

In order to ensure that the nationwide data are not confounded by time or RC, thenationwide data was analyzed after controlling for month and RC. In addition, the meansuicide proportion was subtracted from each suicide proportion, and the mean troop level

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

Revisiting the “Problem From Hell” 825

Figure 1. Suicide attacks in Afghanistan. The total number of suicide attacks is closely correlatedwith the total number of troops, reflecting the increase in attacks during the same time period (leftpanel). But a more useful statistic is how suicide attack usage compared to other tactics (right panel).In fact, as troop levels increased, the proportion of attacks that were suicide trended down. ∗Trooplevels presented on logarithmic scale. Sources: NCTC WITS; NATO/ISAF Placemat.

was subtracted from each troop level. This calculation, known as “de-meaning,” is standardprocedure for the fixed effects model, which allows for the analysis of the effects specific toeach data point. When controlling for these potential confounding variables, the effect size,while a small –0.002268, is still negative. It is important to note that, on the nationwidelevel, the p value after controlling for both month and RC is not statistically significant(0.27021; std. error 2.046e — 03; see Table 1).

However, analyzing the aggregate data on the nationwide level can mask importantdifferences in each RC. When examining the relationship by RC, the negative relationshipis stronger (see Figure 2). The effect size in the South is –75.40 (std. error 22.07), and theeffect size in the East is –64.54 (std. error 24.00). Nationwide, not a single RC experiencedan increase in suicide proportion vis-a-vis troop levels.

What type of attacks are AGE using instead of suicide attacks? The two other mostpopular tactics are attacks involving Improvised Explosive Devices (IED) and attacksinvolving firearms.51 In order to analyze the non-suicide attacks, attacks involving a suicideelement must be removed from the IED and firearm dataset. This is necessary for two

Table 1Controlling for possible confounding variables nationwide

Control forControl for month Control for RC month & RC

Coefficient −1.434e − 03 −3.181e − 03 −2.268e − 03Std. Error 1.344e − 03 1.806e − 03 2.046e − 03P value 0.2881 0.0806 0.27021

When controlling separately for Month and Regional Command, and when controlling for bothMonth and Regional Command, there is no statistically significant relationship between troop levelsand the proportion of attacks that are suicide.

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

826 H. Rome

Figure 2. Suicide attacks by region. The relationship between troop levels and suicide proportion isnegative among all of Afghanistan’s five Regional Commands. Troop levels presented on logarithmicscale. Sources: NCTC WITS; NATO/ISAF Placemat.

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

Revisiting the “Problem From Hell” 827

reasons. First, while attacks involving firearms could also involve a suicide attack, firearmattacks accounted for less than seven percent of suicide attacks.52 Second, every singlesuicide attack was also coded as an IED attack, so excluding “suicide” was necessary.

While all non-suicide firearm attacks increased in the presence of more troops, the“firearm proportion” decreased in the presence of more troops (effect size –93.18; std.error 22.43), much like the pattern seen in suicide attacks. On the other hand, the “IEDproportion” increased in the presence of more troops (effect size 85.86; std. error 15.28).Therefore, it can be concluded that the presence of more troops, nationwide, correlated witha substantial increase in the proportion of IED attacks, a substantial decrease in firearmattacks and, depending on the region, a decrease of varying strength in the use of suicideattacks.

It is not surprising that the South and East, the focus of a major ISAF campaign startingin 2005–2006, saw the largest numbers of suicide attacks against non-combatants. What issurprising, however, is that more troops coincided with an AGE move to increase use ofother tactics more than the suicide attack.

Tactical Skill

While the aggregate number of suicide attacks has been steady, it is little consolation to theAfghan people if the attackers are more skilled and are killing more people in each attack.This section will evaluate how effective suicide attacks are. Determining whether operativesare “effective” at suicide attacks is complicated. A bomb that destroys an empty roadsidefruit stand could be deemed both a resounding success (despite not killing anyone, theattack sends a powerful message) or an embarrassing failure (it must have been a mistakendetonation). In order to assess the tactical success of suicide attack, analysts have usedtwo measures: (1) the victim yield,53 which is the number of victims killed per attack orper attacker and (2) the count of how many suicide attacks actually failed to kill anyone,considering those to be mistakes. This section applies both measures to Afghanistan.

Previous Measures

UNAMA’s analysis of victim yields is limited to all of 2006 and half of 2007. For 2006, itreports that an average of 2.48 people (1.93 civilians) were killed per suicide attack. From1 January 2007 to 30 June 2007, 2.51 people (1.57 civilians) were killed per attack.54

As suicide attacks escalated, the actual efficacy of the attacks was called into question.Williams found that, in 2006, 43 percent of suicide attackers killed only themselves. Therewere many bombs that did not detonate; Afghan police have reported finding suicide vestsaround Kabul, apparently left behind by deserters.55 Williams recounts one terrorist whotried to push his explosive-packed car into the target. The car had run out of gas.56

New Measures

Have fighters gotten better? The short answer is “no” (see Table 2). Using the NCTC data,it is possible to isolate the average number of non-combatants killed or wounded per yearper suicide attack. The number of civilians, as opposed to police or government officials,57

can also be isolated.In 2005, a suicide attack on average killed 1.06 people and wounded 3.75. In 2008, a

suicide attack on average killed 5.48 people and wounded 10.70. By 2010, the numbers fellslightly, with 3.53 killed per attack and 10.40 wounded. Looking at the entire time span of

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

828 H. Rome

Table 2Victim yield in Afghanistan terrorist attacks

Total suicide Total non- Killed/suicide Wounded/suicide Killed/non- Wounded/non-Year attacks suicide attacks attack attack suicide attack suicide attack

2005 16 479 1.06 3.75 1.39 1.452006 90 872 3.40 9.24 1.08 1.462007 106 1016 4.94 9.81 1.41 1.342008 103 1116 5.48 10.70 1.28 1.602009 99 2025 4.73 12.97 1.14 1.442010 103 3243 3.53 10.40 0.88 1.17

Mean 3.86 9.48 1.20 1.41Mean (2006–2010) 4.42 10.62 1.16 1.40

Source: NCTC WITS.

2005 to 2010, each attack killed on average 3.86 people and wounded 9.48. The trends aremirrored when isolating only civilians: 3.50 civilians are killed and 8.76 are wounded inan average attack.

It is most striking to compare these kill rates to the average number of non-combatantskilled in all other types of attacks: on average, a non-suicide attack killed 1.20 people andwounded 1.41. When examining the subset of only civilians, only 0.58 civilians are killedand 0.72 wounded. It is important to note that the absolute level of casualties, compared toIraq, has some relationship to the types of targets chosen by Afghan organizations.58

It is also useful to examine rate of suicide attacks that have failed—attacks in which nonon-combatants are killed. Williams wrote that, in 43 percent of bombings in 2006, onlythe attacker died. He found similar numbers in 2007. The WITS data from 2006 and 2007confirms this observation; 36 percent of suicide attacks in 2006 and 29 percent of suicideattacks in 2007 resulted in zero non-combatant fatalities.59

In the following years, did the attackers get better? It is not unreasonable to assumethat they would—with more experience in conducting attacks, terrorist groups are knownto adapt and learn. When examining the data from 2008, 2009 and 2010, there is littleevidence that improvement has taken place. While 24 percent of attacks in 2008 had zerovictim fatalities, the numbers increased to 27 percent in 2009 and 30 percent in 2010.

Discussion

This study has shown that AGE have maintained the level of suicide attacks at a constantlevel while also decreasing the use of suicide attacks relative to other tactics. Also, thisstudy has illustrated that common indicators for attack effectiveness show little increase inthe effectiveness of suicide attacks. These findings stand in sharp contrast to the reputationthat suicide attacks have earned. What can explain this phenomena? Several potentialexplanations are evaluated in this section. Ultimately, this study will conclude that repeatedattacker incompetence, more than any other factor, can explain these trends.

Concern About Civilian Casualties

Scholars and practitioners of suicide attacks argue that suicide attacks, while having thepotential to be highly effective, are often indiscriminate. Therefore, a potential explanationfor the lack of expansion of suicide attacks in Afghanistan is a concern for civilian casualties.Williams argues that the low victim yields, combined with the lack of increase of suicide

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

Revisiting the “Problem From Hell” 829

attacks, reflects a deliberate desire by AGE to avoid civilian casualties due to the largelyPashtun roots of the attackers:60

This trend seems to fit the Pashtun warrior code (the Afghan bombers areexclusively Pashtuns, with only a few foreigners) with its emphasis on acts ofmartial valor. While Arab suicide bombers in Iraq seem to have no compunctionabout killing unarmed women and children, Afghan-Taliban bombers wereclearly reluctant to do so.61

Consistent with this explanation, the Taliban has both externally and internally sought tominimize civilian casualties from suicide attacks. In one case, following a particularlydeadly attack in Spin Boldak that prompted a powerful community backlash, a Talibanspokesperson said, “We have a problem with making sure they attack the right targets,avoiding killing civilians.”62 The Taliban has also issued instructions to its cadre, including a2010 “code of conduct” that imposes restrictions on the use of suicide attacks: (1) the bombermust be well trained; (2) attacks should focus on “high value targets”; and (3) attackers mustreceive permission. The Taliban further implores attackers to not target civilians.63 The UN,in response to the document, reported: “AIHRC [Afghanistan Independent Human RightsCommission] and UNAMA Human Rights did not observe any concerted effort by theTaliban to implement these orders or to take action against those commanders or memberswho disobeyed them.”64

It is conceivable that AGE have not aggressively pursued suicide attacks because theyfear that too many civilians would be killed. If this explanation is true, it is logical toexpect that forces loyal to the Taliban, as well as other AGE, would also try to minimizecivilian casualties in other attacks, not just suicide attacks. But this is not the case. BothUNAMA and HRW have documented skyrocketing civilian casualties caused by antigov-ernment elements, only a small proportion of which can be attributed to suicide attacks. In2007, UNAMA recorded 700 civilians65 killed by antigovernment forces.66 In 2008, 1,160civilians were killed; in 2009, 1,630 were killed; in 2010, 2,038 were killed. In 2009 and2010, UNAMA disaggregated civilian deaths caused by AGE suicide attacks. In 2009, 17percent of civilian deaths were attributed to suicide attacks, while in 2010 only 11 percentof deaths were attributed to suicide attacks.67

As illustrated with the discussion of the “suicide proportion,” AGE are increasinglyusing IEDs instead of suicide attacks. This could indicate that IEDs are more discriminate,meaning less civilians are killed. But this is also not the case. UNAMA explains that IEDsplaced on roadsides are often detonated by civilian buses, and IEDs are used in populatedcivilian centers, such as bazaars.68

In light of the indiscriminate killing of civilians in attacks that do not involve suicide,the argument that AGE are reducing their use of suicide attacks in order to avoid civiliancasualties is not viable.

Dissipation of the Iraq Effect?

A common element in the discussion over Afghan suicide bombings are the lessons learnedfrom Iraqi insurgents, dubbed the “Iraq effect.” However, the evidence that Iraqi insurgentlessons had an enduring role in the Afghan theater is mixed.

Suicide attacks, some observers have argued, are the most complex tactical operationto execute. For example, Wilner writes: “Suicide bombings are also the most sophisticated

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

830 H. Rome

type of violence to plan, the most difficult to organize effectively, and take a considerableamount of time, energy, and expertise to mount successfully.”69

There is a substantial body of evidence that AGE have received training from interna-tional insurgents. Scholars have linked the initial rise in Afghan suicide bombings to thetransfer of technology and personnel from Iraq—Iraqi insurgents educated their Afghancounterparts about constructing bombs for suicide missions.70 Taliban leader Mullah Dadul-lah acknowledged as much in an interview: “We have ‘give and take’ relations with themujahideen of Iraq. We co-operate and help each other.”71 In 2005, Iraqi insurgents traveledto Pakistan to meet with leadership of the Afghan Taliban.72 In addition to the rise of suicidebombings, influence from Iraq has been documented in a number of other contexts, includ-ing the adoption of sophisticated IEDs and new propaganda73 tools, such as videotapedbeheadings, which were common in Iraq.74 In contrast, other scholars have questionedwhether the technology transfer can be singularly linked to insurgents from Iraq, as fightersfrom Kashmir, Saudi Arabia, and Palestine are often collocated with Afghan AGE.75

Nevertheless, whatever initial lessons were learned, observers have noted that theseskills have been lost. UNAMA stated dryly that AGE demonstrated “little sustained inno-vation and concomitant refinement in technique,”76 which it said challenges the assumptionthat Iraq tactics were particularly helpful. In addition, the capacity for centralized trainingdoes not exist in Afghanistan; low-level commanders often take their own initiative.77

Can this lack of innovation following the transfer of technology from Iraq explainwhy attackers have not used the suicide attack more? This argument about suicide attacktechnology does not provide an effective explanation. AGE in Afghanistan have the tech-nical know-how to create suicide bombs. A lack of creativity in the actual building ofbombs, or of information about tactics from Iraq, is not convincing; the largest differencebetween the Iraqi tactics and Afghan tactics is the kinds of people carrying the bombs, notnecessarily the quality of the explosive device itself. In other words, there is little evidencethat technical failure is to blame when suicide attacks go wrong.

Attacker Competency

Instead of politics, technology or targeting, this study argues that the attackers are mostto blame for the lack of expansion of suicide attacks. Terrorist leaders continue to selectagainst suicide attacks because the attackers are not able to carry them out effectively.In particular, this section outlines recruiting challenges, incompetence and tribal codes aspotential keys for explaining this phenomena.

Reports from international nongovernmental organizations (NGOs) and internationalpress have recounted, horrifically, the increased use of children as perpetrators of suicideattacks in Afghanistan. UNAMA explains how children are kidnapped from schools, indoc-trinated and strapped with explosives. “The use of children, in particular, suggests that thegroups responsible for their ‘recruitment’ are seeing a need to employ increasing extremesof barbarity,” UNAMA wrote. But the use of children also could also indicate increasingdesperation. While children are a relatively unlimited resource, the increasing use of chil-dren could also indicate a failure to recruit traditional attackers—the so-called “militaryage male.” The use of children reflects a broader trend, supported by anecdotal evidencefrom Afghanistan and Pakistan, that antigovernment elements use attackers who are notsufficiently skilled to properly conduct suicide attacks.

Williams outlined anecdotal evidence that suicide attackers have been found to be“deranged, retarded, mentally unstable or on drugs”78 and others who had been bribed orcoerced into executing an attack. Another study shows that the attackers are “poor” and

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

Revisiting the “Problem From Hell” 831

“under-educated or uneducated.”79 This assessment is consistent with Giustozzi, who writes,on an insurgency-wide level, “The insurgents who operate in Afghanistan at the beginningof 2009 are not, by any standard of comparison, a particularly effective, disciplined orcapable lot.”80 The actual insanity or drug use should be perceived with caution; it wouldnot be surprising that Afghan authorities, who are the source for some of this information,seek to portray their foes as deranged for propaganda purposes.81 But a lack of competenceappears to be widespread.

Conventional Wisdom. The poor, uneducated, and incompetent bomber is consistent withthe popular wisdom about terrorists. However, the personal traits of Afghan suicide bombersis important precisely because it reflects the opposite of the prevailing research on terrorists.

Indeed, a great deal of literature has attempted to debunk the conventional wisdomthat suicide attackers are poor, deranged or religious zealots. Comprehensive quantitativestudies of the backgrounds of terrorists, for example, argue that the middle class is morelikely to participate in terrorism and that wealthier members of society are more likely tosupport suicide terrorism.82 Other studies83 question the assumption that bombers succumbto “all-consuming religious devotion” (but others argue that it is a mistake to fully discountthe role of Islam).84 Another thread of research claims that attackers are mentally ill orinsane85 but studies have consistently failed to find a common psychological affliction.86

In varying degrees, the opposite of these traits appears true in Afghanistan.How is the intelligence of a bomber and the success of an attack linked? An innovative

study by Benmelech and Berrebi applied principles of labor economics to a database ofPalestinian suicide bombers. The study found that suicide bombers who are older and moreeducated are more likely to be tasked to attack larger, more difficult targets, while youngerand less-intelligent attackers are tasked to smaller targets.87 They also found that “moreeducated and older Palestinian suicide bombers are less likely to fail or to be caught.”88

The general incompetence of suicide bombers in Afghanistan can explain why AGEleaders are choosing to employ suicide attacks less.

Criticism. There are at least two potential criticisms of this conclusion: First, it is dangerousand naıve to assume that one’s enemy is irrational, insane, or otherwise incompetent.Second, the lack of skilled attackers does not explain why IED attacks continue to beeffective.

There is no question that many terrorists are highly skilled, disciplined, and innovativeactors, and that many of the terrorist attacks over the past decade around the world werehighly sophisticated and effective operations. It is important to note that this study doesnot argue that terrorist leaders are incompetent or irrational. Actually, quite the contrary:this study argues that leaders are rational enough to limit the use of suicide bombings. Thedistinction between leader and operative is an important one in political science analysesof terrorism; Victoroff, in his famous review of the psychological traits of terrorists, arguesthat “[l]eaders and followers tend to be psychologically distinct. Because leadership tendsto require at least moderate cognitive capacity, assumptions of rationality possibly applybetter to leaders than to followers.”89

Second, the primary difference between IEDs and suicide bombs is the obvious one:the operative does not intend to die when planting or controlling an IED. Those who plantIEDs can develop a certain competency in repeatedly conducting such attacks, which isimpossible for suicide attacks. It stands to reason that the type of person who plants an IED,therefore, is different than the type of person who is chosen to kill themselves. Unfortunately,there is little empirical research on the demographics of non-suicide bombers.

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

832 H. Rome

Conclusion

This article used several quantitative measures to show that suicide attacks in Afghanistanhave flat-lined, and AGE have chosen to use other tactics, namely IEDs, instead. Theprimary explanation for this trend, this article argues, is the general incompetence of theattackers.

Further research could compare these findings to trends in attacks exclusively againstISAF troops. Specific attack data is not available publicly, and examining this informationcould provide valuable insight into whether tactics are different in direct engagement withISAF forces.

These findings have a number of potential implications. As already discussed, the useof uneducated and potentially drugged bombers is a pointed exception to the scholarlyview that suicide attackers are rational actors. These findings also have implications on theresearch about the “purpose” of suicide terrorism. A number of scholars have argued thatsuicide terrorism is the most potent form of strategic signaling, illustrating the commitmentand power of a terrorist group.90 That AGE in Afghanistan are choosing to forego suicideattacks could indicate that suicide attacks are primarily tactical, not strategic, choices.

Notes

1. Human Rights Watch, “The Human Cost: The Consequences of Insurgent Attacks inAfghanistan,” Human Rights Watch 19(6) (2007); RAND Corporation, “Incident Detail Aug 3,2006.” Available at http://smapp.rand.org/rwtid/incident detail.php?id=32029 (accessed 8 March2013); Global Terrorism Database, “Incident 200608030001.” Available at http://www.start.umd.edu/gtd/search/IncidentSummary.aspx?gtdid=200608030001 (accessed 8 March 2013); Na-tional Counterterrorism Center, Incident 200693848.

2. This peak is documented by WITS. See National Counterterrorism Center, “WorldwideIncidents Tracking System.” Available at wits.nctc.gov (accessed 24 April 2012). For additional con-firmation, see Thomas H. Johnson, “Taliban Adaptations and Innovations,” Small Wars & Insurgencies24(1) (2013), p. 3; Human Rights Watch, “The Human Cost.”

3. UNAMA, “Suicide Attacks in Afghanistan (2001–2007),” United Nations Assistance Mis-sion in Afghanistan (2007).

4. NCTC reports the following data: Year (Number of suicide attacks). 2004 (2); 2005 (16);2006 (90); 2007 (106); 2008 (103); 2009 (99); 2010 (103).

5. Paul Goldman, “Looking Inside the Mind of a Suicide Bomber”; Bruce Hoffman, “TheLogic of Suicide Terrorism,” The Atlantic Monthly 291(5) (2003), p. 40.

6. Richard L. Clutterbuck, Guerrillas and Terrorists (London: Faber and Faber, 1977); Fordiscussion of terrorism’s “pejorative” nature, see Bruce Hoffman, Inside Terrorism (New York:Columbia University Press, 2006); RAND Corporation, The Study of Terrorism: Definitional Problems(1980).

7. Alex Peter Schmid and Albert J. Jongman, Political Terrorism: A New Guide to Actors,Authors, Concepts, Data Bases, Theories and Literature (New York: Transaction Books, 1988);Walter Laqueur, The New Terrorism: Fanaticism and the Arms of Mass Destruction (New York:Oxford University Press, 1999); Henry Morgenstern and Ophir Falk, Suicide Terror: Understandingand Confronting the Threat (Hoboken, NJ: Wiley, 1999).

8. National Counterterrorism Center, “Methodology Utilized to Compile NCTC’s Database ofTerrorist Incidents.” Available at http://www.nctc.gov/witsbanner/wits subpage criteria (accessed 28April 2012).

9. Ibid. This definition is consistent with Assaf Moghadam, “Suicide Terrorism, Occupation,and the Globalization of Martyrdom: A Critique of Dying to Win,” Studies in Conflict & Terrorism29(8) (2006), pp. 707–729.

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

Revisiting the “Problem From Hell” 833

10. The definition for suicide attacks provided by the United Nations Assistance Mission inAfghanistan is similar: “In a suicide mission, the attacker deliberately and with premeditation useshis or her body to carry and deliver explosives with the explicit intent to attack, kill and maimothers, with the supreme aim of dying in that attack.” UNAMA, “Suicide Attacks in Afghanistan(2001–2007).” GTD uses the following definition: “those cases where there is evidence that theperpetrator did not intend to escape from the attack alive.” Global Terrorism Database, “Code-book: Inclusion Criteria and Variables” (2012). RAND judges only whether the attackers use “‘sui-cide’ tactics as part of the attack.” RAND Terrorism Incidents Database, “Database Definitions.”Available at http://www.rand.org/nsrd/projects/terrorism-incidents/about/definitions.html (accessed8 March 2013).

11. UNAMA in 2007 argued that the use of suicide attacks is “almost exclusively” limited totargeting military units, but by 2010, as fighters began intentionally targeting civilians, UNAMAbacked off this claim. See UNAMA, “Suicide Attacks in Afghanistan (2001–2007)”; UNAMA,“Annual Report 2010 Protection of Civilians in Armed Conflict,” United Nations Assistance Missionin Afghanistan (2011).

12. UNAMA articulates that “irrespective of the insurgents’ intended targets, the victims of thesuicide attacks have been largely civilian bystanders.” UNAMA, “Suicide Attacks in Afghanistan(2001–2007).”

13. RAND Corporation, Incident Detail Aug 3, 2006; Global Terrorism Database, Incident200608030001; National Counterterrorism Center, Incident 200693848.

14. Human Rights Watch, The Human Cost.15. UNAMA, “Suicide Attacks in Afghanistan (2001–2007).”16. Ibid.17. In its 2010 report, UNAMA writes: “In 2010, a particular armed opposition group sometimes

admitted responsibility for a specific attack. All the groups however used tactics that included stand-off attacks and ground engagement as well as asymmetric tactics, such as IEDs and suicide attacks,assassination and execution, abduction and intimidation. In this context, it was often difficult to attachresponsibility for specific attacks to an identifiable armed opposition group where no group admittingresponsibility.” UNAMA, “Annual Report 2010 Protection of Civilians in Armed Conflict.”

18. It is notable that several databases are unsure of the perpetrators of the very first suicideterrorist attack in the history of Afghanistan, the 2001 killing of Northern Alliance leader AhmedShah Massoud. There is considerable evidence that the attack was carried out by Osama bin Laden’sAl Qaeda (see Neamatollah Nojumi, “The Rise and Fall of the Taliban,” in Amin Tarzi and RobertCrews, eds., The Taliban and the Crisis of Afghanistan (Cambridge, MA: Harvard University Press,2008), p. 90); Steve Coll, Ghost Wars: The Secret History of the CIA, Afghanistan, and Bin Laden,from the Soviet Invasion to September 10, 2001 (New York: Penguin, 2004); In contrast, RANDreports the perpetrators as “unknown.” RAND Corporation, “Incident Detail Sep 9, 2001.” Availableat http://smapp.rand.org/rwtid/incident detail.php?id=8625 (accessed 8 March 2013); GTD reportsthat the Taliban were suspected of the attack. Global Terrorism Database, “Incident 200109100001.”Available at http://www.start.umd.edu/gtd/search/IncidentSummary.aspx?gtdid=200109100001 (ac-cessed 8 March 2013).

19. Tarzi recounts the proliferation of “official” Taliban spokesmen, who often contradict eachother and sometimes themselves. Amin Tarzi, “The Neo-Taliban,” in Amin Tarzi and Robert Crews,eds., The Taliban and the Crisis of Afghanistan (Cambridge, MA: Harvard University Press, 2008),pp. 294–296.

20. For discussion of splinter groups, see ibid., pp. 301–304.21. Human Rights Watch, “The Human Cost”; Global Terrorism Database, Incident

200608030001; RAND Corporation, Incident Detail Aug 3, 2006.22. NCTC, Incident 200693848.23. UNAMA, “Annual Report 2010 Protection of Civilians in Armed Conflict.”24. UNAMA partly attributes the attack’s success with his security detail’s unfamiliarity with

the idea of a suicide operation. UNAMA, “Suicide Attacks in Afghanistan (2001–2007).”

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

834 H. Rome

25. Williams notes that the Taliban “launch[ed] swarm attacks against exposed coalition out-posts in an attempt to seize Kandahar and Helmand.” Brian Glyn Williams, “Suicide Bombings inAfghanistan,” Jane’s Islamic Affairs Analyst (2007).

26. Tarzi, The Neo-Taliban.27. However, see Tarzi, who recounts ISAF’s skepticism about the Afghan government’s asser-

tion that Al Qaeda was responsible. Ibid., p. 283.28. In addition, Afghan police thwarted another suicide bombing in Kabul involving two Pak-

istanis and one Afghan. Williams, Suicide Bombings in Afghanistan.29. This claim is not universally accepted. For example, see Tarzi, who recounts that while

Karzai claimed a June 2003 suicide bombing was carried out by a foreigner, the Afghan interiorminister said that it was carried out by an Afghan. The German defense minister said the attack wasthe work of Al Qaeda. Tarzi, The Neo-Taliban, p. 284.

30. Williams, Suicide Bombings in Afghanistan; Omid Marzban, “The Foreign Makeup ofAfghan Suicide Bombers,” Terrorism Focus (Jamestown Foundation) 3(7) (2006).

31. Johnson, Taliban Adaptations and Innovations, p. 3; Brian Glyn Williams, “Mullah Omar’sMissiles: A Field Report on Suicide Bombers in Afghanistan,” Middle East Policy 15(4) (2008),pp. 26–46.

32. Williams, Suicide Bombings in Afghanistan; Brian Glyn Williams, “Talibanistan: Historyof a Transnational Terrorist Sanctuary,” Civil Wars 10(1) (2008), pp. 40–59.

33. Johnson, Taliban Adaptations and Innovations, p. 3.34. Tarzi, The Neo-Taliban, p. 291.35. Ibid.36. But see UNAMA, which skeptically recounts arguments made by Afghan officials, including

President Hamid Karzai, that Afghans do not commit suicide attacks. Nevertheless, “it is undeniablethat Afghans provide necessary support to suicide cells within Afghanistan.” UNAMA, “SuicideAttacks in Afghanistan (2001–2007).” Similarly, Williams notes that the Taliban was “clearly facili-tating” early suicide attacks, before executing its own starting in 2004. Williams, Suicide Bombingsin Afghanistan.

37. Robert Canfield, “Fraternity, Power and Time in Central Asia,” in Amin Tarzi and RobertCrews, eds., The Taliban and the Crisis of Afghanistan (Cambridge, MA: Harvard University Press,2008), p. 212.

38. National Counterterrorism Center, Worldwide Incidents Tracking System.39. Data from the UN Assistance Mission in Afghanistan, Human Rights Watch, University

of Maryland’s Global Terrorism Database and RAND’s Database of Worldwide Terrorism Incidentswere consulted for this analysis. UNAMA and HRW data consistently include more cases than doesWITS, and both HRW and UNAMA do not restrict attacks to those defined as “terrorism.” The GTDand RAND databases consistently report suicide terrorist attacks at roughly half the number of WITS.GTD, RAND, and WITS all employ rigorous definitions of terrorism and suicide attacks; the reasonfor the disparity is not clear.

40. National Counterterrorism Center, Methodology Utilized to Compile NCTC’s Database ofTerrorist Incidents.

41. Ibid.42. Ibid.43. The full WITS methodology is included in Appendix A. Ibid.44. Graeme Smith, “What Kandahar’s Taliban Say,” in Antonio Giustozzi, ed., Decoding

the New Taliban: Insights from the Afghan Field (New York: Columbia University Press, 2009),pp. 191–210.

45. UNAMA, “Suicide Attacks in Afghanistan (2001–2007).”46. Ibid.47. Thomas Barfield, “The Roots of Failure in Afghanistan,” Current History 107

(2008), pp. 410–417; Alec Barker, “Improvised Explosive Devices in Southern Afghanistanand Western Pakistan, 2002–2009,” Studies in Conflict & Terrorism 34(8) (2011), pp. 600–620.

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

Revisiting the “Problem From Hell” 835

48. For one of the only papers on the topic, see B. Peter Rosendorff and Todd Sandler, “SuicideTerrorism and the Backlash Effect,” Defence and Peace Economics 21(5–6) (2010), pp. 443–457.

49. See Moghadam, “Suicide Terrorism, Occupation, and the Globalization of Martyrdom: ACritique of Dying to Win”; Scott Ashworth, Joshua Clinton, Adam Meirowitz, and Kristopher Ramsay,“Design, Inference and the Strategic Logic of Suicide Terrorism,” American Political Science Review102(2) (2008), pp. 1–5.

50. See UNAMA, which argues that “anti-government elements effectively employ this terrify-ing tactic to increase the level of insecurity among Afghans; to diminish their confidence in the Afghangovernment; and to deepen resistance to ISAF operations, Coalition Forces and other internationalentities in Afghanistan.” UNAMA, “Suicide Attacks in Afghanistan (2001–2007).”

51. NCTC definition: “Handguns, shotguns, machine guns, military guns. Includes any weaponclassifiable as ‘small arms.’ Excludes mortars, artillery, and rocket-propelled grenades (RPGs).”National Counterterrorism Center, Methodology Utilized to Compile NCTC’s Database of TerroristIncidents.

52. These incidents probably refer to the so-called “fedeyeen attacks” in which an attacker maybe armed with multiple types of weapons in a suicide mission. For discussion, see Johnson, TalibanAdaptations and Innovations, p. 3; Barker, Improvised Explosive Devices in Southern Afghanistanand Western Pakistan, 2002–2009, pp. 600–620.

53. Aside from anecdotal evidence, UNAMA assesses the efficacy of attacks using what itcalls the “victim yield.” That number articulates the ratio between victims killed and attackerskilled in the suicide attack. UNAMA also measures the ratio between victims killed and the uniquenumber of attacks (as opposed to the number of attackers involved). In reality, though, these numbersare very similar, and the measures are used interchangeably throughout the UNAMA report. Thisstudy will use the victims killed per attack measure. Ibid. Williams uses a similar measure. BrianGlyn Williams, “The Taliban Fedayeen: The World’s Worst Suicide Bombers?” Terrorism Monitor(Jamestown Foundation) 5(14) (2007).

54. UNAMA, “Suicide Attacks in Afghanistan (2001–2007).”55. Williams, The Taliban Fedayeen: The World’s Worst Suicide Bombers?56. Ibid.57. All of which are considered non-combatants.58. Afghan bombers typically attack “hard” targets.” A “hard” target, like a passing military

convoy or a fortified government installation, can theoretically be effectively targeted by a suicidebomber wearing concealed explosives. The targeting of hardened locations, as opposed to less-protected “soft” targets, could explain the low victim yields, as well. But suicide attacks are notonly used against hard targets. In 2010, UNAMA recorded an increasing number of suicide attacks“conducted in areas with large numbers of civilians, for example outside hotels, on bridges and inmarket places and also against civilians” (UNAMA, “Annual Report 2010 Protection of Civilians inArmed Conflict”). Further, UNAMA records that six “major suicide attacks occurred during [2010]that caused large number of civilian casualties and took place in busy civilian areas often with nodiscernible military target.”

59. While the attacks kill no non-combatants, they are still considered suicide terrorist attacks,per the WITS database. The decision to consider these attacks as terrorist incidents is supported byUNAMA, which states that, “Even if the attacker manages to kill only himself or herself, the missionis indeed a success by the perpetrators as the assailant has attained the status of martyr for his or hercause.” UNAMA, “Suicide Attacks in Afghanistan (2001–2007).”

60. Magnus and Naby outline how the Taliban is made up of mostly Pashtun, with a “tokensprinkling of Tajiks.” Ralph Magnus and Eden Naby, Afghanistan: Mullah, Marx and Mujahid(Boulder, CO: Westview Press, 2002), p. 219.

61. Brian Glyn Williams, Afghanistan Declassified: A Guide to America’s Longest War(Philadelphia: University of Pennsylvania Press, 2012).

62. The spokesman also stated that “we do our best in our suicide attacks to avoid civiliancasualties” and included that dubious claim that “fighting with explosives is out of the control ofhuman beings.” Williams, The Taliban Fedayeen.

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

836 H. Rome

63. Muhammad Munir, “The Layha for the Mujahideen: An Analysis of the Code of Conductfor the Taliban Fighters Under Islamic Law,” International Review of the Red Cross 93(881) (2011),pp. 81–102.

64. UNAMA, “Annual Report 2010 Protection of Civilians in Armed Conflict.”65. This is a more specific categorization than NCTC WITS’ designation of “non-combatants.”66. UNAMA, “Annual Report on Protection of Civilians in Armed Conflict, 2008,” United

Nations Assistance Mission in Afghanistan (2009).67. UNAMA, “Annual Report 2010 Protection of Civilians in Armed Conflict”; UNAMA,

“Annual Report on Protection of Civilians in Armed Conflict, 2009,” United Nations AssistanceMission in Afghanistan (2010).

68. UNAMA, “Annual Report 2010 Protection of Civilians in Armed Conflict.”69. Alex S. Wilner, “Targeted Killings in Afghanistan: Measuring Coercion and Deterrence in

Counterterrorism and Counterinsurgency,” Studies in Conflict & Terrorism 33(4) (2010), pp. 307–329.70. Barnett Rubin, “Saving Afghanistan,” Foreign Affairs 86(1) (2007), pp. 57–74, 76–78;

UNAMA, “Suicide Attacks in Afghanistan (2001–2007).”71. Williams, Suicide Bombings in Afghanistan.72. Johnson, Taliban Adaptations and Innovations.73. The Taliban began idolizing suicide bombers, a sharp contrast to their traditional “iconopho-

bic” history. Atiq Sarwari and Robert Crews, “Afghanistan and Pax Americana,” in Amin Tarzi andRobert Crews, eds., The Taliban and the Crisis of Afghanistan (Cambridge, MA: Harvard UniversityPress, 2008), p. 311.

74. Williams, Talibanistan: History of a Transnational Terrorist Sanctuary, pp. 40–59.75. Canfield, Fraternity, Power and Time in Central Asia, p. 212; Barker, Improvised Explosive

Devices in Southern Afghanistan and Western Pakistan, 2002–2009, pp. 600–620.76. UNAMA, “Suicide Attacks in Afghanistan (2001–2007).”77. UNAMA cites the seemingly random geographic and temporal aspect of suicide attacks as

evidence that different cells work concurrently and separately. Ibid.; Antonio Giustozzi, “Afghanistan:Now You See Me?: The Pygmy Who Turned into a Giant: The Afghan Taliban in 2009.” LSE IDEAS,London School of Economics and Political Science (2009). But see Canfield’s discussion of morecentralized training. Canfield, Fraternity, Power and Time in Central Asia, p. 212.

78. UNAMA reported that some analysts in Pakistan believe that some attackers are addicted toheroine. UNAMA, “Suicide Attacks in Afghanistan (2001–2007)”; Williams, The Taliban Fedayeen.

79. UNAMA, “Suicide Attacks in Afghanistan (2001–2007).”80. Giustozzi, “Afghanistan: Now You See Me?”81. Williams, The Taliban Fedayeen.82. See Alan B. Krueger, What Makes a Terrorist?: Economics and the Roots of Terrorism

(Princeton, NJ: Princeton University Press, 2007); M. Najeeb Shafiq and Abdulkader Sinno, “Edu-cation, Income, and Support for Suicide Bombings: Evidence from Six Muslim Countries,” Journalof Conflict Resolution 54(1) (2010), pp. 146–178; For a critique of Krueger, see Efraim Benmelech,Claude Berrebi, and Esteban F. Klor, “Economic Conditions and the Quality of Suicide Terrorism,”The Journal of Politics 74(1) (2012), pp. 113–128.

83. For overview, see Jeremy Ginges et al., “Psychology Out of the Laboratory: The Challengeof Violent Extremism,” The American Psychologist 66(6) (2011), pp. 507–519; Robert A. Pape,“The Strategic Logic of Suicide Terrorism,” The American Political Science Review 97(3) (2003), pp.343–361; Laurence Iannaccone, “The Market for Martyrs,” Interdisciplinary Journal of Research onReligion 2 (2005).

84. Moghadam, “Suicide Terrorism, Occupation, and the Globalization of Martyrdom,” pp.707–729.

85. Clutterbuck, Guerrillas and Terrorists; Russell Razzaque, Human Being to Human Bomb:Inside the Mind of a Terrorist (Thriplow, Cambridge, UK: Icon, 2008); Arie W. Kruglanski et al.,“Fully Committed: Suicide Bombers’ Motivation and the Quest for Personal Significance,” PoliticalPsychology 30(3) (2009), pp. 331–357. For criticism, see Alex Mintz and David Brule, “Method-ological Issues in Studying Suicide Terrorism,” Political Psychology 30(3) (2009), pp. 365–371.

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

Revisiting the “Problem From Hell” 837

86. Ami Pedahzur, Suicide Terrorism (Cambridge, MA: Polity, 2004).87. Benmelech and Berrebi argue that the educated bomber “kills roughly four to six more

people when he attacks a large city target compared to an uneducated suicide bomber.” Benmelech,Berrebi, and Klor, “Economic Conditions and the Quality of Suicide Terrorism.”

88. Ibid.89. Jeff Victoroff, “The Mind of the Terrorist: A Review and Critique of Psychological Ap-

proaches,” Journal of Conflict Resolution 49(1) (2005), pp. 3–42.90. Bruce Hoffman and G. H. McCormick, “Terrorism, Signaling, and Suicide Attack,” Studies

in Conflict and Terrorism 27(4) (2004), pp. 243–281.

Appendix

WITS Methodology

“The data provided in WITS consists of incidents in which subnational or clandestine groupsor individuals deliberately or recklessly attacked civilians or noncombatants (includingmilitary personnel and assets outside war zones and war-like settings). Determination ofwhat constitutes a terrorist act, however, can be more art than science; information is oftenincomplete, fact patterns may be open to interpretation, and perpetrators’ intent is rarelyclear. Moreover, information may become available over time, changing initial judgmentsabout attacks. Users of this database should therefore recognize that reasonable people maydiffer on whether a particular attack actually constitutes terrorism or some other form ofpolitical violence. NCTC has made every effort to limit the degree of subjectivity involvedin the judgments and, in the interests of transparency, has adopted a set of counting rulesthat are delineated below.

Terrorists must have initiated and executed the attack for it to be included in thedatabase; failed or foiled attacks, as well as hoaxes, are not included in the database.Spontaneous hate crimes without intent to cause mass casualties are excluded, thoughit should be understood that often there is insufficient information to judge whether anattack was planned or spontaneous. While genocidal events can be interpreted as the mostextreme form of politically motivated violence against civilians, attacks in this categoryare excluded, in part because of the inherent difficulty in counting such events and becausethe inevitable undercount does not do justice to the scope and depth of such atrocities.Moreover, the question of whether or not acts of genocide should be included in the WITSdatabase was posed to a panel of academics at the 2008 Brain Trust on Terrorism Metrics.The panel concluded that acts that meet the criteria for genocide should not be included inthe database.

Determining when perpetrators have targeted noncombatants can also be difficult.Military personnel and assets outside war zones and war-like settings pose one challengeto the noncombatant provision of the definition, while police under military command andcontrol, and organized groups of armed civilians inside war zones and war-like settingspose another challenge. With the approval of the 2007 Brain Trust on Terrorism Metrics,NCTC developed a combatant matrix which details the various areas of war-like settings,and the common actors such as military police, militias, soldiers and other combatant-likeactors. The analysts use the matrix in complex cases to determine when an act targetingcombatant-like actors should be included in WITS. The combatant matrix is adjusted as thecircumstances in world conflicts change or evolve. The distinction between terrorism andinsurgency in Iraq was especially challenging in previous years, as Iraqis participated in boththe Sunni terrorist networks as well as the former-regime-elements insurgency, targetingboth civilians and combatants and often affecting both populaces. Terrorist attacks against

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013

838 H. Rome

combatants count as reckless and indiscriminate when terrorists could have reasonablyforeseen that their attack would result in civilian casualties. Therefore, combatants may beincluded as victims in some attacks when their presence was incidental to an attack aimedat noncombatants, and some attacks may be deemed terrorism when it recklessly affectscombatants.

The WITS database contains a field that allows analysts to categorize an incidentby event type. Event types are coded in the database as the following: armed attack,arson/firebombing, assassination, assault, barricade/hostage, bombing, CBRN, crime, fire-bombing, hijacking, hoax, kidnapping, near miss/non-attack, other, theft, unknown, andvandalism. While some incidents can clearly be coded using this taxonomy, other kindsof attacks are more difficult to define. When it can be determined, incidents that involvemultiple types of attacks are coded with multiple event types. Incidents involving mortars,rocket-propelled grenades, and missiles generally fall under armed attack, although impro-vised explosive devices (IED) fall under bombing including vehicle-borne IEDs (VBIED).VBIEDs include any IED built into or made a part of a vehicle including cars, trucks,bicycles, and motorcycles. Suicide events are also captured, but the perpetrator must havedied in the attack for the event type suicide to be included.”

Source: From National Counterterrorism Center, “Methodology Utilized to CompileNCTC’s Database of Terrorist Incidents.” Available at http://www.nctc.gov/witsbanner/wits subpage criteria (accessed 28 April 2012).

Dow

nloa

ded

by [

Prin

ceto

n U

nive

rsity

] at

11:

03 1

7 Se

ptem

ber

2013