Under the Hood – Learning and Innovation in the Islamic State’s Suicide Vehicle Industry

Abstract

This article explores how the Islamic State learned from its suicide bombings, a tactic that in theory undermines the in-group stability associated with organizational learning. Using interviews from Mosul and Baghdad together with internal documents from the Division of Soldiers it provides a case study of the learning and production process underlying innovations in suicide vehicle designs between 2014 and 2017. It examines how the group acquired, distributed, interpreted, implemented, and stored information from suicide vehicle operations in its military-industrial complex, and details how the group learned by maximizing continuity among personnel in support, coordinating, and production roles.

In the summer of 2014, the Islamic State launched an offensive in Iraq that enabled the group to establish a territorial presence in several of the country’s provinces, most notably Nineveh and its urban capital, Mosul.¹ This was the beginning of the group’s three-year war against the Iraqi government where it sought to expand and defend its newly consolidated Islamic State. During the conflict, the group began to employ battlefield tactics in open confrontation with military forces to a greater extent than its Iraqi predecessors, al-Qaeda in Iraq (AQI) and the Islamic State in Iraq (ISI), had done against the U.S. between 2003 and 2010.² At the same, time the Islamic State also kept tactics that had become signatures in these groups’ violent repertoires, most notably suicide bombings. The Islamic State used suicide bombings more in Iraq in its first year than AQI and ISI combined between 2003 and 2010.³ It also made greater use of suicide vehicles, introducing some of the most novel changes to the devices since Hezbollah put drivers in truck bombs in its attack against the U.S. and French marine barracks in Beirut in 1983.⁴ Its main suicide bombing innovation was both low-tech and simple – it began adding armor on the outside of vehicles to prevent premature detonation in direct confrontation with military forces, in essence adapting an inherited technique to new operational realities.⁵

Innovation is usually a concept reserved for radical departures from current methods, but it is also used to capture incremental modifications of existing technologies and procedures.⁶ While Hezbollah’s first suicide vehicle bomb in 1983 can be characterized as a radical innovation, the Islamic State’s suicide vehicle modifications are better described as incremental, aimed at producing gradual improvements rather than novelty-induced shock.⁷ Changes were made in response to new operational circumstances and evolving military countermeasures, and were the outcome of a learning process wherein the group acquired, analyzed, and applied knowledge to adjust equipment, aiming to gain the upper hand from government forces.⁸

But how did the Islamic State learn from its suicide bombings? After all, a tactic where the death of the perpetrator is a necessary condition for the success of the mission is not exactly conducive to learning.⁹ Although several studies have outlined why the Islamic State developed these vehicles, only a few have sought to explore how they did it.¹⁰ This reflects a trend in the terrorism literature more broadly, namely that “while many studies on terrorism acknowledge that successful terrorist organizations must learn, very few show how they learn.”¹¹ Using interviews from Mosul and Baghdad and documents from the military branch of the Islamic State, this case study explores the learning process and structures underlying incremental innovations in suicide vehicle designs in Iraq between 2014 and 2017.¹² It is structured around Huber’s model of organizational learning, examining how the group aquired, distributed, interpreted, implemented, and stored knowledge relevant to suicide vehicle development.

Whereas source availability makes it difficult to trace a particular learning process to a particular innovation outcome, it is nevertheless possible to outline general processes that were likely involved in a range of adjustments of both vehicles and related infrastructure. By so doing, the article finds that the Islamic State learned by ensuring continuity in support, coordinating, and production roles while making suicide bombers as peripheral as possible to everything but pressing the detonator. While this is commensurate with literature on learning in terrorist and insurgency groups, the sources used in this study grant a unique level of detail on how this division of labor was designed, achieved, and maintained. By outlining this in-depth, it increases our understanding of learning in the Islamic State, especially of how the group was able to innovate under conditions of high membership attrition.

The first reports of up-armored suicide vehicles in Iraq emerged around Baghdad in October 2014, with the Islamic State employing Humvees left behind by the Iraqi army following the fall of Mosul in June that same year.¹³ Because they were a limited resource and had utility in other aspects of battle, these cars proved impractical to use for suicide bombings in high numbers. Combining the benefits of the Humvee with the requirement for mass production, the group gradually developed a reproducible and bulletproof design based on civilian vehicles.¹⁴ Personnel would cover all or parts of cars with thick iron plates, slanting it in front to increase the effective thickness of the metal and heightening the odds of small arms and heavy munitions ricocheting off.¹⁵ They also added metal grids to increase the distance between exploding munitions and the car.¹⁶ In rural operations, the group would paint the armor beige to blend in with desert terrain and make the discovery by reconnaissance units more difficult.¹⁷ In urban operation the armor would be painted in more radiant colors to mimic civilian vehicles.¹⁸ The added armor initially caught the Iraqi military off-guard, and to effectively stop some of these new contraptions they had to procure Kornet missiles at around $250 000 apiece.¹⁹

While a relatively minor change to design, it facilitated a radical departure from how suicide vehicles had been used in Iraq and abroad before 2014. Following the U.S. invasion in 2003, the asymmetry in both technology and manpower between Islamist insurgents and U.S. forces made suicide vehicles an attractive instrument for inflicting damage without having more than one person face the enemy directly. The vehicles used by both AQI and ISI were therefore intended to be as inconspicuous as possible to maximize the likelihood of reaching the desired target undetected.²⁰ This emphasis on stealth characterized the majority of suicide bombing campaigns from the 1980s onwards, regardless of delivery mechanism.²¹ It was characteristic of Palestinian groups’ use of explosive belts in Israel, the Kurdistan Worker Party’s use of similar contraptions in Turkey, the Tamil Tigers’ use of belts, vehicles, and boats in assassinations and guerilla assaults in Sri Lanka, the Taliban’s use of suicide vehicles in Afghanistan, and the use of suicide bombing by al-Qaeda cells internationally.²²

While the Islamic State kept using inconspicuous suicide vehicles in clandestine operations, it also designed a new type with an emphasis on repelling, rather than avoiding, incoming fire. This expanded the tactical functions of suicide vehicles. In addition to their utility in guerilla assaults and as symbolic acts of terrorism, they could now be integrated into battlefield operations to weaken enemy defense positions in advance of ground-troop assaults, in some ways paralleling the use of traditional artillery.²³ This capability became particularly pertinent following the fall of Mosul in June 2014, when the Islamic State obtained enough arms to equip nearly 50,000 soldiers from Iraqi military stocks and sought to further expand the state.²⁴ As the tide of battle turned against the Islamic State around 2016, the vehicles also proved useful as a stalling mechanism to delay government forces. Historically, the closest functional equivalents among non-state actors are perhaps the Tamil Tigers’ integration of suicide attackers in some ground offensives and the Viet Cong’s occasional use of sapper units in advance of ground forces.²⁵ However, neither used suicide attackers for this purpose as systematically as the Islamic State, nor modified the associated equipment to the same extent. Whereas the ultimate success of the Islamic State’s suicide vehicle innovations can be debated considering the fall of Mosul in 2017, they reflect both adaptability and an aptitude for learning worth exploring in depth.

Learning in Terrorist and Insurgent Organizations

Regardless of whether an organization is legal or illegal, learning is important for both short-term performance and long-term success.²⁶ Researchers have defined organizational learning in several ways, but most articulate it as experience-driven knowledge change.²⁷ By seeing knowledge as expressed in behavior, learning is observed as improvements in either practices or performance.²⁸ However, experience can be partially internalized, poorly interpreted, and is subject to a range of cognitive biases that can lead organizations to either fail to make changes or to make the wrong ones. As noted by Michael Kenney, “experience is often dimly perceived. Participants may not understand what happened, why it happened, and whether what happened is beneficial or harmful to the organization.”²⁹ To take this into account, scholars have adopted a process view, defining learning as organizations broadening their range of potential behaviors.³⁰ Whether and to what extent the potential is actualized depends on how groups process experience-based information; how it is acquired, distributed, interpreted, and stored in the organization.

Terrorist and insurgent groups’ ability to learn is often indicated by their ability to adapt to changing operational circumstances.³¹ Adaptability can manifest in a range of behaviors, with the most studied being innovation and contagion. The literature on contagion has found that both violent and nonviolent political movements have a propensity to mimic tactics that are perceived to have been successful elsewhere.³² The likelihood of occurrence increases with the strength of the bond between the original user and aspiring adopter, with direct organizational ties being most conducive for exchange.³³ The literature on innovation has been more focused on identifying motives, conditions, and group characteristics that differentiate innovative from conservative groups.³⁴ Jackson identified eleven factors intended to predict the occurrence of innovative behavior, encompassing group characteristics, the nature of the technology, and the modalities of the struggle.³⁵ Similarly, seeing innovation as a physical manifestation of creativity, Gill and others explored factors that can predict the transition from one to another.³⁶

While indicators of adaptability are a focus in the literature, the underlying learning- and creative processes have been less explored. This might be due to the fact that empirical insight on this subject is rare, or, as suggested by Gill et al., that studies on the topic are less interesting for counter-terrorism efforts.³⁷ A notable exception is Kenney, who examined the learning process underlying competitive adaption in al-Qaeda.³⁸ One of his key contributions was to highlight that a group’s ability to adapt depends on whether the knowledge required to do so is codified or tacit. Codified knowledge has an agreed-upon language that lends itself to oral and written communication.³⁹ In this context learning is a rather straightforward process: know-what, like technical instructions, can be transferred via manuals, while know-how, how to carry out instructions, might require the addition of time and practice.⁴⁰ Tacit knowledge is unarticulated and manifests in a sense of effortlessness when executing a complex task, social traits like charm or trustworthiness, or – in an organizational setting – in ease of cooperation and coordination.⁴¹ This cannot be acquired from papers but requires interpersonal interaction and time to develop, with common transfer mechanisms being mentorships and apprenticeships.⁴²

For insurgent and terrorist groups, acquiring both types of knowledge is restricted by the organizational imperative for secrecy, especially in contexts where the state has high levels of control.⁴³ Moreover, while groups have developed several ways to acquire codified knowledge clandestinely, utilizing it to develop a capability or improve performance often requires knowledge on coordination.⁴⁴ Being a tacit skill, not only does this require time to develop in groups, but it also takes a degree of membership stability, neither of which are typically associated with organizations of the terrorist or insurgent variety.⁴⁵ This is arguably particularly acute in organizations that opt to employ suicide bombings. While it might have advantages from a tactical point of view, from an organizational perspective the intentional blowing up of operatives risks undermining the organizational stability required to improve performance.

Despite the level of membership turnover suggesting that systematic learning in rebel and terrorist organizations is difficult, it sometimes has the opposite or little effect.⁴⁶ The main reason for this is that insurgency groups are not homogenous decision-making actors, but separate between leadership and foot-soldiers in a way that allows the former to persist and learn from the demise of the latter.⁴⁷ While this division is well known, the finer details of how groups organize and process information to learn and innovate under high levels of membership attrition are not. The following section provides a detailed account of how this was done in the case of the Islamic State’s suicide vehicle industry.

Gaining Insight into the Islamic State’s Suicide Vehicle Industry

Learning and innovation in the Islamic State’s military effort took place within the Division of Soldiers (Diwan al-jund), the Islamic State equivalent of a Ministry of Defense. Its main responsibilities were managing wars, training and deploying soldiers, protecting the borders of the caliphate, and planning the overall conflict.⁴⁸ To this end the Division was comprized of a number of sub-departments and committees, with bureaus responsible for leadership, administration, logistics, management of special skills, and human resources.⁴⁹ The Division was a trans-national body with a particular presence in Syria and Iraq. While this article focuses on Iraq, the place where the group used suicide bombings most prolifically, the involvement of the Division in Syria and evidence of exchange across the border, suggests that the learning process was similar in both places and informed by knowledge gained in both countries.⁵⁰

To obtain insights into the procedures and activities of the Division, the study uses internal Islamic State documents and propaganda material in combination with information from thirteen interviews conducted in Mosul and Baghdad between March and August 2021. Due to the ethical and security challenges associated with interviewing Islamic State members inside Iraq, interviews were conducted with members of the Iraqi military and civilians who have all directly engaged with aspects of the group’s operations. These were carried out with the help of a local research assistant due to travel restrictions imposed by COVID-19.⁵¹ Participants were selected during a three-month preparation period where we worked to identify people with relevant expertise and more than a decade of experience in their fields. Consent was acquired in this period, which gave participants several weeks to determine whether they wished to take part in the study.⁵² Those who agreed to take part included engineers tasked with raiding Islamic State explosives facilities, army personnel tasked with countering and dismantling up-armored suicide vehicles, and intelligence officials tracking Islamic State networks. It also included people involved in a metal works facility incorporated by the Islamic State and doctors in and outside of areas under Islamic State control.⁵³ All requested full anonymity due to the security situation in Iraq. While participants had in-depth knowledge on aspects of the Islamic State’s organization and capabilities, they were not members. A key challenge in interpreting the interviews was therefore to separate hearsay from reliable information and empirically grounded knowledge from prejudice. This was done by matching the information given by a participant against his experience and knowledge in that field and by triangulating the information against documentary sources and interviews conducted on similar topics in the same environment.

The study also uses primary sources from the Islamic State collected and archived by scholars and analysts in the last decade. Work led by Aaron Zelin and Aymenn Tamimi have made accessible several hundred Islamic State administrative documents, publications, and propaganda material. Moreover, documents from the Division of Soldiers hosted at George Washington University (GWU) have recently enabled scholars to outline the group’s military organization in more detail.⁵⁴ Much of the GWU archive is restricted, but the author received access while working on this article. While these documents provide insights into the group’s activities, they have to be interpreted with caution. Documents from armed groups in conflict environments are likely to be imperfectly sampled, selectively preserved, and in an imperfect state. Islamic State documents are no exception, and archives constitute a nonrandom collection of papers made available to the public after several rounds of selection. Similarly, material from the Islamic State news services and propaganda apparatus are the result of the group seeking to project a particular image and is prone to suffer from both misrepresentation and omissions as a result.⁵⁵

This combination of sources nevertheless gives valuable and unique insights into both the structure of the organization and the nature of its activities. That said, the below analysis should be regarded as an outline of the learning process in the Division of Soldiers under ideal conditions by actors looking from the outside in and documents displaying the group at its most organized. Findings likely coexisted with a degree of improvization, miscommunication, and system breakdowns that are not captured in the sources used for this paper.⁵⁶

The Learning and Production Process Underlying Innovations in Suicide Vehicles in the Islamic State

How did the Division of Soldiers overcome the learning impediment associated with the systematic use of suicide bombings to produce incremental innovations in suicide vehicle designs? The process can be usefully analyzed in terms of Huber’s model of organizational learning.⁵⁷ Huber divides learning into four constructs based on recurring themes in the literatures on organizational learning and knowledge management: (1) knowledge acquisition, (2) information distribution, (3) information interpretation, and (4) integrating lessons in organizational memory. Learning can be defined as the sum of these activities, but as Huber was primarily interested in understanding processes and the range of new potential behaviors they aspire, his model does not include a construct for observable outcomes. Outcomes are therefore analyzed in an implementation section. Implementation is defined as the organization using interpreted information to realize one or several new potential behaviors, and is placed between information interpretation and embedding lessons in organizational memory.

The breadth of Huber’s model and its focus on knowledge management makes it apt for examining how the Islamic State learnt under conditions of membership turnover, but its broad scope comes at the cost of examining its constitutive constructs in detail.⁵⁸ For example, the information interpretation section assumes the pursuit of SVBIED innovation was grounded in instrumental considerations. While this was a important aspect, it does not consider the historical, social, cultural, and emotive antecedents that made SVBIEDs appear in the Islamic State repertoire in the first place. This has inspired books of its own but is outside the scope of this article as it does not examine why SVBIEDs became an attractive option, but rather how SVBIED innovation was achieved once it already featured on the “menu of tactics”.⁵⁹ With that caveat in mind, the below contains a step-by-step analysis of how the Islamic State developed and produced their suicide vehicles while blowing up their members with remarkable zeal.

Knowledge Acquisition

Most activities undertaken by the Islamic State were grounds for acquiring new knowledge, and suicide attacks were no exception.⁶⁰ However, suicide bombings were particularly challenging in this regard as few bombers survived long enough to learn from their mission, much less to author post-action reports to inform future deliberations of superior officers. On an individual level, perpetrators could only gain knowledge and communicate lessons if they failed. According to doctors in both Baghdad and Mosul, failure did happen, and if the perpetrator was lucky enough to return to a hospital under Islamic State control, the organization often granted them the opportunity to try again.⁶¹ While some had to be persuaded, others appeared much keener on a second attempt and tried until they got it right, despite – or perhaps due to – debilitating injuries in the first attempt.⁶² In any case, returning suicide bombers were the exception, most were either successful, killed, or arrested, rendering them unavailable for debriefs.⁶³

To ensure the group could conduct a series of suicide bombings while also acquiring feedback, the organization built a division of labor during operations that separated operative, supporting, and coordinating functions, leaving the bomber to play a bit part in his own demise. This created continuity in supporting and coordinating roles, enabling the group to communicate and make improvements to different aspects of its operations over time, despite experiencing continuous losses in the execution phase.

Suicide operatives were the purview of the special skills bureau, with prospective bombers organized in a section called the Martyrdom Operatives Battalion (Katibat al-Istishadiin).⁶⁴ Members of the suicide battalion would normally arrive at the area of operation shortly before the execution of an attack and spend the preparation period in isolation with clerics to build the mental fortitude required to execute this type of mission.⁶⁵ The battalion had no shortage of volunteers and, following the group’s acquisition of territory, its size far outgrew the tactical demand for suicide operations.⁶⁶ The Islamic State was not the first group to find itself in this situation. Around 2006, an ISI document shows that it too had more suicide bombers than the group was able to deploy.⁶⁷ However, while ISI was unable to utilize aspiring bombers for logistical reasons, the Islamic State was arguably unwilling to waste them. Winter’s analysis of suicide attack use in 2015-16, suggests that the group appeared reluctant to use bombers in campaigns where it did not feel it could win.⁶⁸ While this changed during the Battle of Mosul in 2016–17, it corresponded with the impression of several interviewees, one of whom remarked, “The Islamic State does not use suicide bombers unnecessarily, they are cheap, but they are also indispensable.”⁶⁹

The most important support functions for suicide operations were filled by intelligence and logistics personnel, responsible for providing mission-critical information and supplying equipment. Like the suicide bomber, the suicide vehicle and explosives would arrive shortly before the start of an operation, typically around one day in advance.⁷⁰ These were manufactured in a decentralized production chain overseen by a sub-division of the logistics bureau called the Committee for Development and Manufacturing (Hay’at al-tatwir wa al-tasnia’). This process is outlined in detail in the implementation section. The suicide vehicles could either be assembled near the attack site or completed in a factory and transported to areas of operation, depending on the type of vehicle and extent of Islamic State territorial control.⁷¹ Transportation of components was usually overseen by the intelligence department.⁷² If planning a series of attacks in one location, the leadership could order factories to be opened near the area of operation to reduce the risk associated with moving equipment around.⁷³

In addition to transportation, intelligence personnel were charged with collecting the information required for the suicide bomber to reach the target. Depending on the nature of the target, type of government presence, and nature of the operation, these reports could be prepared for hours or months with the use of anything from drones, to civilians, to intelligence operatives themselves scoping the area of interest.⁷⁴ For example, a plan to drive a covert suicide vehicle near a government installation inside Baghdad would likely be based on long-term surveillance by members of the intelligence department looking for routines, opportunities, and obstacles.⁷⁵ In active battlefields, reports would more likely stem from army battalion personnel on the ground or live drone imagery.⁷⁶ Operatives in these settings could also be guided to the target by personnel with strong local knowledge, senior members in radio contact, maps on iPads, and – eventually – drones.⁷⁷ With both the car and intelligence in hand, the suicide bomber would depart and attempt to carry out the mission.

The task of acquiring attack performance information was given to a person described as an attack coordinator.⁷⁸ The coordinator – or someone reporting to this function – would ideally be in direct contact with the suicide bomber from ignition to detonation. Coordinators were described as relatively senior Islamic State members that were tied to an operation room. The configuration and function of operation rooms varied, with some being mission and campaign specific, comprised of a small group of leaders, including from the intelligence section and relevant battalions in the province of operation.⁷⁹ Some had more overarching strategic functions, but these were likely inhabited by leaders with responsibilities on the Division level.⁸⁰

Information Distribution

Following the execution of one or several suicide operations, information on performance and outcome would be communicated by the attack coordinator up the chain of command, where it would be assessed and used for planning future operations in conjunction with other sources of information.⁸¹ The distribution of some feedback would stop at the provincial level and inform leadership deliberations in a local battalion but information considered to have wider strategic implications would likely reach the higher levels of Division leadership.⁸²

Interviewees noted that operational feedback would most commonly be circulated in chats on social media, using apps like WhatsApp and Telegram.⁸³ While sensible from a time-saving perspective and in clandestine operations where leaving a paper trail is particularly disadvantageous, these informal means of exchange are somewhat surprising knowing what we do about the Islamic State’s penchant for paperwork.⁸⁴ Documents from the drone unit known as the Al-Bara’ bin Malik Brigade, for example, indicate that evaluations of drone training operations were communicated in written reports.⁸⁵ While none of the archives consulted for this paper had similar documents for suicide missions, it is not impossible that they existed.

Information Interpretation

Upon receiving feedback from the attack, the leadership at the relevant level would use the information to imagine potential device changes and select which modifications the group should attempt to realize in light of operational needs. Because of the decentralized nature of the Islamic State’s military and production complex, systematic changes to technology, equipment, and infrastructure based on operational feedback would likely be decided by Division-level leadership.⁸⁶ Here, responsibilities included planning operations and coordinating the group’s five main armies.⁸⁷ This encompassed deciding what targets to pursue, instructing intelligence operatives to monitor particular areas, and – relevant to innovation in suicide vehicle equipment – ordering the manufacturing of necessary materials.⁸⁸

As this article does not have sources privy to the discussions in Division leadership, it is difficult to pin-point suicide vehicle outcomes from decisions made at that level with any degree of certainty. However, two decisions strongly indicate top-level involvement due to the scale and systematic nature of their implications. The first was the decision to develop up-armored suicide vehicle production infrastructure following experimentation with Humvees against military targets around Baghdad in 2014.⁸⁹ The second was the expansion of up-armored suicide vehicle production infrastructure in Mosul around 2016, after it had proven useful to stall military onslaught in other Iraqi cities.⁹⁰ Because these crossed provincial boundaries in scope and in personnel implications, it is unlikely that they could have been made without involvement by leaders responsible for coordinating between armies in different localities.

Although the efficacy of both decisions can be debated, interview participants tasked with countering the consequences considered Islamic State leaders highly competent at interpreting incoming information. As noted by one intelligence officer, “I think they learned something from every operation. They learned from their mistakes very quickly, both in terms of how they made the explosives, where they made it, and how they execute the operation.”⁹¹ Whereas international interest in the conflict gave interviewees incentives to exaggerate the ability of their opponents, the group did have a distinct advantage when it came to extracting relevant and actionable lessons from attack information. The individuals who came to inhabit leadership positions in the Islamic State after 2013 all had several years of fighting experience combined with periods of apprenticeships where they learned from senior fighters in U.S.-run prisons.⁹² Beyond the top echelon, this also had a favorable effect on the types of people that came to inhabit positions of mid-level leadership. As argued by Vera Mironova about jihadi groups in Iraq and Syria,

Three groups that had, from the very beginning, enjoyed the luxury of having experienced leaders were ISIS, Jabhat al-Nusra, and Ahrar al-Sham. (…) Because of their own previous fighting experience, they knew the exact qualities a fighter and leader needed. In other words, they knew what to look for when making a promotion or recruitment decision.⁹³

In this way, experience at the top positively impacted the quality of people lower in the hierarchy. This arguably improved the overall analytic ability of the group and enhanced the odds of interpreting information in a way that enhanced, rather than deteriorated, performance and capabilities. However, this quality was not a static feature of the leadership throughout the conflict but worsened over time due to in-group dynamics as the tide of battle turned against the group.⁹⁴

Implementation

While the leadership decided which changes to equipment the group should attempt to realize in line with operational needs, the actual modifications were developed and produced elsewhere. Following the interpretation of operational feedback, leadership at either the central or provincial level would issue orders about desired equipment modifications to sub-divisions in the Division of Soldier’s logistics section.⁹⁵ These sub-divisions would oversee a production process drawing on civilian and military expertise to meet requirements. As already mentioned in the discussion on knowledge acquisition, the most involved sub-division in suicide vehicle development was the Committee for Development and Manufacturing (CDM). The CDM was responsible for production infrastructure, development and production of weapons, as well as quality control across Islamic State-run workshops.⁹⁶ By 2015, documents indicate that the CDM was divided based on divisions and sectors, meaning they were present in different provinces and had a degree of subject specialization.⁹⁷

Interviews suggest some equipment orders would be aspirational, prompting a relatively centralized experimental and development process within the CDM, while others would be more easily achievable and integrated into a preexisting production chain. Most changes to suicide vehicle design fell in the latter category, with the CDM’s role being coordinating a network of evolving production infrastructure. Experimental work, on the other hand, centered around particular facilities. In Iraq, much of the experimental work took place in the laboratories of the University of Mosul, described by one engineer as an “ISIS research center”.⁹⁸ Before it was bombed in 2016, the group made particular use of the departments of engineering and chemistry, as well as the workshops of the technical institute.⁹⁹ They also incorporated some of the staff from these departments in their work.¹⁰⁰ While some capabilities were realized, like the group’s weaponization of drones, use of chlorine gas, and new IED linking methods, others never left the drawing table.¹⁰¹ An example of the latter was experimentation with remote-controlled vehicle bombs in Iraq, Libya, and Syria.¹⁰² While it appears in one Islamic State video from Mosul, and a document from the Research and Development Section of the Aleppo Province outlines rudimentary lessons from the manufacturing process, the contraption was hardly used and never put into mass-production.¹⁰³

As a rule, members with responsibilities in the CDM or its associate facilities were not involved in planning or executing operations but received orders from leadership in their province and supplied equipment to battalions in the field.¹⁰⁴ Like the division between operative, support, and coordinating functions during operations, this created continuity in key roles and enabled the group to make improvements to equipment and infrastructure over time despite experiencing continuous losses in other parts of the organization. The group’s ability to maintain the separation varied based on context. While it was rarely an issue in the production of up-armored suicide vehicles, as the factories required to prepare and assemble these were too large to establish outside of areas of Islamic State control, the operative and production roles would sometimes be merged in government-controlled areas.¹⁰⁵ This was likely because the tradeoff between cell size and detection risk was more pressing in that context. That said, production staff in Islamic State-controlled areas were not outside of government reach, with both factories and research facilities being targets of airstrikes.¹⁰⁶

Production Infrastructure and Expertise

So, what types of facilities and people were used in the production of suicide vehicles under the CDM? In areas of Islamic State control, suicide vehicle manufacturing was typically divided based on the component being produced, with different types of workers involved in each stage and the CDM engaged in a coordinating capacity.¹⁰⁷ Because ironwork required special machinery that was neither widely available nor easily moved, the decision to add armor increased the decentralization of production. In Mosul, for example, this led the group to make particular use of facilities and technical equipment in the eastern industrial zone.¹⁰⁸ The CDM would also take-over local metalwork businesses, with some owners being incorporated into the production chain and others dispossessed and replaced.¹⁰⁹ These locations could be staffed by civilians, some of whom gained employment through family members and similar informal avenues, while others responded to more traditional work-adds.¹¹⁰ One ironworker described his employment in the following way:

One day, one of my Islamic State relatives offered me to work with them because I didn’t have a job. I accepted to work with them, but I didn’t belong to the organization. I was in charge of 10 workers, the business was booming and never stopped. There were delegates from within the Islamic State in the Military Development and Manufacturing Authority who visited the site and supplied us with iron. (…) It was a very normal factory. We received raw materials, work orders, and measurements, and we told them what we needed.¹¹¹

The production of explosives was also overseen by the CDM but appears to have been a separate process confined to separate locations, usually staffed by Islamic State members belonging to explosives manufacturing battalions.¹¹² Military engineers tasked with raiding explosives factories described encountering two main types, one run by personnel focusing on vehicle-based bombs (Tafkhikh) and another focused on smaller IEDs (Tafjir).¹¹³ While the former required space and was usually located on the outskirts of cities, the latter could be established inside government-controlled territory. The composition of the explosives used in Tafkhikh factories varied based on vehicle type and available material, with common components being TNT, ammonium nitrate, gasoline, oxygen bottles, and rocket propellants.¹¹⁴ Engineers estimated that the ammonium nitrate, gasoline, TNT combination would typically cost between 15-20 dollars per kilogram to manufacture.¹¹⁵ In some cases, they would also use C4, but this was more common in suicide belts, vests, and motorcycles, instruments that would be the purview of Tafjir personnel focusing on smaller IEDs.¹¹⁶ Doctors in both Mosul and Baghdad noted that the damage of suicide bombs gradually increased after 2013, with the Islamic State making greater use of shrapnel and producing explosions that were more powerful than both ISI and al-Qaeda.¹¹⁷

Ensuring that factories under CDM supervision had the necessary material for production was the responsibility of another logistics sub-division called the General Supply Committee (Hay’at tajhiz al-’am). The Supply Committee oversaw warehouse management and procured materials for the group’s military industry, ranging from explosive precursors to electronic components. Soon after taking over Mosul, the group leveraged cross-border networks, partially built before 2014, to import materials en masse.¹¹⁸ In line with findings by conflict Armament Research (CAR), Islamic State documents and interviewees indicated that much of the material found in the group’s warehouses came from Turkey, through Syria.¹¹⁹ Army engineers noted that while explosives like TNT could be manufactured domestically but required importing raw materials from abroad, there were no signs of domestic C4 production, meaning this would have to be imported ready-made.¹²⁰ Some materials used in suicide vehicles, such as rocket propellants, could be found in the equipment left behind by the Iraqi army in 2014. Yet other components, like oxygen bottles, iron plating, and the vehicle itself, would be available in and around the Islamic State-controlled provinces.

Implementation of leadership orders in explosives battalions drew on technical expertise obtained from a range of sources, with the most important being people with experience in the field obtained prior to joining the organization. Having been in a state of insurgency for more than ten years, Iraq had much to offer on this front. All military interviewees mentioned former Ba’athist army engineers as the most competent explosive experts in the Islamic State. “In some cases, like the drone manufacturing process, the group would hire foreign experts,” stated one interviewee, “However, when it came to experts on booby traps and mines, Iraqis, especially those who worked in the former Iraqi Army Military Engineering Division, were the best.”¹²¹ That being said, researchers have suggested that the influence of Ba’athists on the Islamic State organization is exaggerated.¹²² It is entirely possible that these played a less important role than the interviewees indicated, and that other experts mentioned, especially foreign fighters from Azerbaijan, Afghanistan, and Europe, were more significant.¹²³ In either case, the geographical scope of production between 2014 and 2017 required distributing technical know-how to members in different areas to implement leadership orders. Engineers reported encountering suicide vehicle factories in Mosul, Ramadi, Fallujah, Tal Afar, and Tikrit.¹²⁴ Their list was not exhaustive, but these were all areas where the group faced military onslaught and had a sufficient level of control to set up the type of factory required for production.

Theoretical and practical knowledge was distributed in several ways, both in-person and remotely. Most members received lessons on explosive manufacturing during basic training, but only a few would receive more advanced instructions.¹²⁵ Candidates for the latter would often be identified based on exam results in the subject.¹²⁶ These individuals generally had some prior education, mainly in the sciences, although this was described as non-essential.¹²⁷ More specialized teaching was described as structured lessons with both theoretical and practical exercises, which corresponds with examination papers on the construction of improvised explosive devices found by CAR throughout the eastern sector of Mosul.¹²⁸

While some instruction took place in person, instruction videos and manuals were also shared online. Interview participants involved in raiding explosive factories reported finding both electronic training material and handwritten notes, and intelligence officials described amassing repositories of manuals in both physical and e-book formats in Arabic, Russian, English, and French.¹²⁹ Several participants noted that trainers would often avoid travel due to security risks, so in-person training conducted by explosive experts in one province would be recorded and distributed to personnel in other provinces.¹³⁰ This was primarily shared through closed channels on social media, although some instructions were also made publicly available on channels like YouTube.¹³¹ Training also happened “on the job” with factories described as hierarchical systems where more senior members guided less experienced and skilled workers.¹³²

Evaluating the Islamic State’s explosives competence, statements by interviewees were characterized by the same tension found in discussions on terrorist group expertise more generally – alternating between describing them as highly competent and completely clueless.¹³³ The contradictory assertions can in part be accounted for by the fact that different phases of the manufacturing process required different levels of technical knowledge. Some participants noted that the group forbade unqualified personnel from participating in production, but engineers described that what constituted necessary qualifications varied greatly based on task.¹³⁴ Assembling explosive devices did not require particularly high levels of skill, and members of the group with basic training – even children – would be capable of doing it given instructions.¹³⁵ Working with electric components was among the more difficult tasks which required advanced training, although this depended on the sophistication of the method.¹³⁶ This range in technical knowledge combined with on-the-job training likely contributed to a fair share of factory accidents described by doctors interviewed for this paper.¹³⁷

Integrating Lessons in Organizational Memory

Organizational memory captures how knowledge is stored for future use in both documentary form and the memories of individual group members.¹³⁸ As just discussed, the Islamic State made use of both documents and videos to store knowledge on production and manufacturing related to suicide vehicles and explosives manufacturing. The group also developed standard production procedures to ensure interoperability between provinces. In the CDM, standardization was the responsibility of a committee called the Central Organization for Standardization and Quality Control (Al-jihaz al-markazi la al-taqaiis wa al-saitara al-nua’ia) (COSQC). While it is unclear when the COSQC started operating, a document found by CAR, dated 31 August 2016, instructed all production facilities to produce only to the standards specified by the branch.¹³⁹ Another COSQC document contained technical standards, measurements, and acceptable ranges for mortar manufacturing.¹⁴⁰ Mortar components measured by CAR and compared to standard procedures in production facilities indicate that instructions were followed.

To what extent suicide vehicle manufacturing was codified in standard procedures is not entirely clear. On the one hand, the range of car types used in operations, which varied from SUVs to 4 × 4 vehicles and tankers, required a degree of improvization that made standardization impractical.¹⁴¹ On the other, the group did on occasion mass-produce designs based on particular vehicle types. The most striking example was during the Battle of Mosul when the group used so many Kia SUVs, Kaaman noted it was “as if IS had seized all the vehicles in a car dealership within the city.”¹⁴² This is not unlikely, with interviewees stating that the group had seized control of businesses relevant to its military industry and that it faced no shortage of civilian cars locally.¹⁴³ That said, the ironworker interviewed in Mosul reported regularly receiving measurements for iron plates, suggesting that these generally varied.¹⁴⁴ Some intelligence officials and engineers reported having seen standardized production schemas for the initial stage of explosives manufacturing, but others said it was largely random.¹⁴⁵ Detonation mechanisms appear to have been the most uniformly produced component, with Kaaman observing a higher degree of standardization between 2014–2017 compared to what groups used in Iraq between 2007–2012.¹⁴⁶

While suicide vehicle production might have seen some standardization at the component level, it did not reach the level seen in the group’s mortar production. Moreover, the standards the group did have faltered when it was under high levels of pressure. Toward the end of the Battle of Mosul in 2017 when the Islamic State lost territorial hold of the city, army personnel and doctors reported observing a wider variety of devices and a higher proportion of injuries consummate with low-tech improvization.¹⁴⁷ This local deterioration in ability was likely due to the destruction of production facilities, supply chains being cut, and high levels of membership attrition. The latter is considered the biggest threat to the human aspect of organizational memory, with high personnel turnover potentially leading to the loss of both know-how and more tacit forms of knowledge.¹⁴⁸ The group’s division between coordination, production and high-risk operative roles prevented this for as long as the organization could maintain territorial safe havens. However, the containment of personnel loss to expendable functions became harder when areas under Islamic State control came under direct military onslaught. When this happened, the group literally imploded locally, with scores of people blowing themselves up with a starkly diminishing rate of return to the battle effort. That said, lessons related to the use and production of suicide vehicles from the battle in Mosul were implemented in Syria, suggesting that the know-how was processed and stored in the organization but that the provinces in Iraq were increasingly unable to act on it.¹⁴⁹

Conclusion

How did the Islamic State learn from its suicide bombings? Using Huber’s model to analyze the learning process and structures underlying incremental innovations in its suicide vehicle industry, this article shows it did so by minimizing membership attrition in roles critical to processing information and implementing changes to vehicle components. This not only required a separation between leadership and foot-soldiers, but between each function relevant to planning, preparing, and executing operations. Given their certain death, suicide bombers were made as peripheral as possible to everything but driving the car and pressing the detonator. Personnel in supporting, production, and coordinating functions, in contrast, were specialized and shielded from military confrontations to ensure continuity in roles needed for information acquirement, distribution, interpretation, and implementation – in short, those necessary for learning and production.

Exploring the structures and learning processes underlying incremental innovations in suicide vehicle design gives rise to several questions that would be interesting to explore in more depth. While the certainty of losing the suicide bomber made the separation described above paramount for learning from suicide operations, the ever-present possibility of losing operatives in conventional attacks arguably makes it sensible in most situations. Do organizational demands associated with systematic use of suicide bombing shape organizational structures in a way that makes groups that employ them better at learning from other types of operations as well? Or is it the case that groups that opt to employ suicide bombings systematically already have the appropriate divisions in place because operating in a high-repression environment or executing other types of high-risk operations have sparked a similar division of labor?

The Islamic State grew out of a movement that had structures separating bombers from other functions and had members that were already socialized into seeing suicide bombers as a public good.¹⁵⁰ Suicide bombings come at a greater cost to young and small groups, as well as groups whose members are not socialized to accept the tactic, as they are less equipped to contain the effect of personnel loss to expendable individuals and sections of the group.¹⁵¹ The Islamic State was never forced to contend with this issue because the social and ideological controversies had been overcome by al-Qaeda throughout the 1990s and 2000s and the structures to absorb martyrs and support mass-use had been developed by AQI and ISI in Iraq between 2003 and 2010.¹⁵² Knowing how to execute and learn from such operations without creating an organizational imbalance, and with an experienced leadership to translate lessons into modifications based on operational needs, the group’s suicide bombing complex hit the ground running, resulting in the most novel changes to suicide vehicles since they were first employed.

The complexity of the structures underlying these innovations is cause for reflection about how learning might manifest differently in territorial and clandestine groups. Groups that operate under high levels of government pressure often divide roles in a way conducive to acquiring, distributing, and interpreting knowledge in the organization much like the Islamic State.¹⁵³ However, in lieu of a territory, it seems more likely that innovations in clandestine context would be confined to changes in practice rather than technology and equipment. To a certain extent, this would be a practical matter related to infrastructure and availability of materials, but not all innovations require the military-industrial complex underlying suicide vehicle development. Equally important to the Islamic State was continuity among its group of experts and a stable level of communication between organizational branches, neither of which are typically associated with a clandestine mode of existence.¹⁵⁴

That said, there might also be a difference between a group’s ability to conduct one-off innovations rather than systematic improvements over time. The past holds many examples of clandestine groups that have conducted technological experiments and developed proofs of concepts in areas of government control.¹⁵⁵ However, systematic changes based on operational feedback in the manner seen in the Islamic State are increasingly rare. This is arguably because the required continuity in key production functions is unfeasible given government countermeasures in modern high repressions states.¹⁵⁶ Should we, therefore, expect groups in high-repression environments to employ technological innovations developed in environments more hospitable to violent learning? This is not the current tendency, at least not in Europe and the U.S., but whether this is a short- or long-term trend remains to be seen.¹⁵⁷

Although these and other questions abound, the paper at hand had a much more limited scope, seeking simply to unpack the learning processes and structures underlying Islamic State innovations in suicide vehicle design and to detail how the group learned under conditions of high membership attrition. High levels of compartmentalization and specialization in the Islamic State’s military-industrial complex was key to this. While the paper is too narrow to contribute to broader theorizing on learning and innovation, it might prompt further exploration into whether these are necessary conditions for terrorist group learning to manifest in incremental technological innovations over time.

Ethics statement

This article includes human research participants and was approved by the Social Sciences and Humanities Inter-Divisional Research Ethics Committee (IDREC) at the University of Oxford, Ref No: R68407.

Acknowledgements

I would like to thank Dara T. A., as well as Hugo Kaaman and Vera Mironova, for their invaluable support in different stages of the data collection process. I am very grateful to Stathis Kalyvas and the participants of the T. E. Lawrence Program on the Study of Conflict at the University of Oxford, who all provided much needed feedback on an early draft of this article. I also benefitted greatly from the comments by Truls Hallberg Tønnessen, Petter Nesser, Mikael Hiberg Naghizadeh, Clara Voyvodic Casabo, Giuseppe Spatafora, and the two anonymous reviewers. Your constructive and helpful comments made this article significantly better. Finally, I would like to thank Sean Patrick Hughes and Anthony Long, whose coffee in Derry made the writing process thoroughly enjoyable.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This study was funded by The Norwegian Defence Research Establishment.

Notes

1 Suadad Al-Salhy and Tim Arango, ‘Sunni Militants Drive Iraqi Army Out of Mosul’, 10 June 2014, https://www.nytimes.com/2014/06/11/world/middleeast/militants-in-mosul.html.

2 Craig Whiteside, Ian Rice, and Daniele Raineri, ‘Black Ops: Islamic State and Innovation in Irregular Warfare’, Studies in Conflict and Terrorism, 2019, 1–28, https://doi.org/10.1080/1057610X.2019.1628623; Craig Whiteside et al., ‘The ISIS Files The Department of Soldiers’ (Washington D.C.: George Washington University, April 2021), https://isisfiles.gwu.edu/downloads/q237hr95t?locale=en.

3 According to the Global Terrorism Database (GTD), AQI and ISI conducted 108 suicide bombings between 2003 and 2010, while the Islamic State conducted 392 between 2014 and 2015 alone. If we include all groups in Iraq, the GTD registered 776 instances of suicide bombings between 2003 and 2010. In contrast, the Islamic State alone conducted 1023 attacks between 2014 and 2017. These estimates should be considered approximations. As noted by Charlie Winter, the limited presence of reporters, and military rather than terrorist nature of some of the group’s attacks could mean that the GTD could be undercounting. Global Terrorism Database, ‘National Consortium for the Study of Terrorism and Responses to Terrorism (START)’, [Data File]. Retrieved from https://www.Start.Umd.Edu/Gtd, 2018; Charlie Winter, ‘War by Suicide: A Statistical Analysis of the Islamic State’s Martyrdom Industry’, ICCT Research Paper (Online) 8, no. 3 (2017): 1–34, https://doi.org/10.19165/2017.1.03.

4 Mike Davis, Buda’s Wagon. (Place of publication not identified: Verso, 2017), 78–89.

5 Hugo Kaaman, ‘The History and Adaptability of the Islamic State Car Bomb’, Hugo Kaaman - Open Source Research on SVBIEDs (blog), 14 February 2017, https://hugokaaman.com/2017/02/14/the-history-and-adaptability-of-the-islamic-state-car-bomb/.

6 Dolnik, Understanding Terrorist Innovation: Technology, Tactics and Global Trends; Gill et al., ‘Malevolent Creativity in Terrorist Organizations’, 1 June 2013; Gill, ‘Tactical Innovation and the Provisional Irish Republican Army’, 2017; Bloom, ‘Constructing Expertise: Terrorist Recruitment and “Talent Spotting” in the PIRA, Al Qaeda, and ISIS’; DeVore, Stähli, and Franke, ‘Dynamics of Insurgent Innovation: How Hezbollah and Other Non-State Actors Develop New Capabilities’.

7 Paul Gill et al., ‘Malevolent Creativity in Terrorist Organizations’, The Journal of Creative Behavior 47, no. 2 (1 June 2013): 125–51, https://doi.org/10.1002/jocb.28.

8 Michael Kenney, From Pablo to Osama: Trafficking and Terrorist Networks, Government Bureaucracies, and Competitive Adaptation (University Park, Pa.: Pennsylvania State University Press, 2007), 13.

9 Diego Gambetta, Making Sense of Suicide Missions (Oxford: Oxford University Press, 2005).

10 Craig Whiteside and Vera Mironova, ‘Adaptation and Innovation with an Urban Twist: Changes to Suicide Tactics in the Battle for Mosul’, Military Review 97, no. 6 (2017): 78; Hugo Kaaman, ‘Car Bombs as Weapons of War: ISIS’S Development of SVBIEDS, 2014-19’, Middle East Institute, 2019.

11 Gill et al., ‘Malevolent Creativity in Terrorist Organizations’, 1 June 2013, 128.

12 The timeframe was selected because the Islamic State had a relatively stable level of organisation in Iraq. Before 2014 it operated clandestinely and following the fall of Mosul in 2017 the Iraq provinces underwent a re-organisation aimed at clandestine survival. Whiteside et al., ‘The ISIS Files The Department of Soldiers’, 8; Daniel Milton, ‘Structure of a State - Captured Documents and the Islamic State’s Organizational Structure’ (West Point: Combating Terrorism Center, June 2021), 33, https://ctc.usma.edu/wp-content/uploads/2021/06/Structure-of-a-State.pdf.

13 The Islamic State did something similar in Syria a year earlier, using a BMP-1 reenforced with steel pipes in a coordinated attack on the Menagh air base. In that case the Free Syrian Army (FSA) faction called the Northern Storm Brigade modified the vehicle and the Islamic State supplied a driver. Kaaman, Hugo, ‘The First SVBIED Attack Using a Modified BMP-1 Armored Personnel Carrier in the Syrian Civil War. Jabhat al-Nusra Used It to Attack a Loyalist Checkpoint in Daraa al-Balad on June 27, 2013 – h/t @E_of_Justice’, Twitter (blog), 18 September 2020, https://twitter.com/HKaaman/status/1306984555434438657?s=20. Max Fisher, ‘ISIS Just Pulled off Its First Carbombing with a Stolen American Humvee’, Vox, 27 October 2014, https://www.vox.com/2014/10/27/7078635/isis-carbomb-stolen-american-humvee-iraq.

14 For more comprehensible overviews of the different modifications, see: Kaaman, ‘Car Bombs as Weapons of War: ISIS’S Development of SVBIEDS, 2014-19’; Kaaman, ‘The History and Adaptability of the Islamic State Car Bomb’.

15 Kaaman, ‘The History and Adaptability of the Islamic State Car Bomb’.

16 Kaaman.

17 Wilayat Slah al-Din, Who Never Escape From War, 2016, https://jihadology.net/2016/06/08/new-video-message-from-the-islamic-state-but-god-will-perfect-his-light-wilayat-%e1%b9%a3ala%e1%b8%a5-al-din/; Wilāyat al-Ānbār, Determination of the Brave #3 (Anbar, 2017), https://jihadology.net/2017/12/24/new-video-message-from-the-islamic-state-determination-of-the-brave-3-wilayat-al-anbar/; Wilayat Slah al-Din, Holding Wounds of Their Nation #9, 2015, https://jihadology.net/2015/12/08/new-video-message-from-the-islamic-state-holding-wounds-of-their-nation-9-wilayat-%e1%b9%a3ala%e1%b8%a5-al-din/.

18 Wilayat Slah al-Din, Who Never Escape From War; Wilāyat al-Ānbār, Determination of the Brave #3; Wilayat Slah al-Din, Holding Wounds of Their Nation #9.

19 Interview with BTU1, August 2021; Tamer El-Ghobashy and Ali A Nabhan, ‘Iraq’s Vital Weapon Against ISIS in Mosul; $250,000 Kornet Missiles Target Toughest Islamic State Truck Bombs’, Wall Street Journal (Online), 4 November 2016, 1836035993, ABI/INFORM Global; Global Newsstream, https://www.proquest.com/newspapers/iraqs-vital-weapon-against-isis-mosul-250-000/docview/1836035993/se-2?accountid=13042.

20 Gambetta, Making Sense of Suicide Missions.

21 A history of the use and spread of suicide bombing is beyond the scope of this article, but can be found in the following works: Mia Bloom, Dying to Kill (New York: Columbia University Press, 2005); Gambetta, Making Sense of Suicide Missions; Robert A. Pape, Dying to Win, Epub (New York: Random House, 2006); Assaf Moghadam, The Globalization of Martyrdom: Al Qaeda, Salafi Jihad, and the Diffusion of Suicide Attacks (Baltimore: Johns Hopkins University Press, 2008).

22 Ami Pedahzur, Root Causes of Suicide Terrorism: Globalization of Martyrdom, Cass Series on Political Violence (London: Routledge, 2006), 1–12.

23 Kaaman, ‘The History and Adaptability of the Islamic State Car Bomb’.

24 UNSC, ‘The Islamic State in Iraq and the Levant and the Al-Nusrah Front for the People of the Levant: Report and Recommendations Submitted Pursuant to Resolution 2170 (2014)’ (The United Nation, 14 November 2014), 15.

25 C. Christine Fair, ‘Sri Lanka’, in Urban Battle Fields of South Asia, 1st ed., Lessons Learned from Sri Lanka, India, and Pakistan (RAND Corporation, 2004), 11–68, http://www.jstor.org/stable/10.7249/mg210a.10; Stephen Hopgood, ‘Tamil Tigers, 1987–2002’, in Making Sense of Suicide Missions (Oxford: Oxford University Press, 2005), 46; Pedahzur, Root Causes of Suicide Terrorism: Globalization of Martyrdom, 115–18.

26 Linda Argote and Ella Miron-Spektor, ‘Organizational Learning: From Experience to Knowledge’, Organization Science (Providence, R.I.) 22, no. 5 (2011): 1123, https://doi.org/10.1287/orsc.1100.0621.

27 Argote and Miron-Spektor, 1124.

28 George P Huber, ‘Organizational Learning: The Contributing Processes and the Literatures’, Organization Science (Providence, R.I.) 2, no. 1 (1991): 89, https://doi.org/10.1287/orsc.2.1.88.

29 Kenney, From Pablo to Osama: Trafficking and Terrorist Networks, Government Bureaucracies, and Competitive Adaptation, 5.

30 Huber, ‘Organizational Learning: The Contributing Processes and the Literatures’.

31 Brian A. Jackson et al., Aptitude for Destruction, Volume 1, 1st ed. (RAND Corporation, 2005), http://www.jstor.org/stable/10.7249/mg331nij; Brian A. Jackson et al., Aptitude for Destruction, Volume 2, 1st ed. (RAND Corporation, 2005), http://www.jstor.org/stable/10.7249/mg332nij; Kenney, From Pablo to Osama: Trafficking and Terrorist Networks, Government Bureaucracies, and Competitive Adaptation; Paul Gill, ‘Tactical Innovation and the Provisional Irish Republican Army’, Studies in Conflict & Terrorism 40, no. 7 (3 July 2017): 573–85, https://doi.org/10.1080/1057610X.2016.1237221.

32 Charles Tilly, From Mobilization to Revolution (Reading, Mass.: Addison-Wesley, 1978); Manus Midlarsky, Martha Crenshaw, and Fumihiko Yoshida, ‘Why Violence Spreads: The Contagion of International Terrorism’, International Studies Quarterly 24, no. 2 (1980): 262, https://doi.org/10.2307/2600202; Sarah A. Soule, ‘The Diffusion of an Unsuccessful Innovation’, The Annals of the American Academy of Political and Social Science 566 (1999): 120–31; Sidney Tarrow, The New Transnational Activism, Cambridge Studies in Contentious Politics (Cambridge: Cambridge University Press, 2005); Michael Horowitz, ‘Nonstate Actors and the Diffusion of Innovations: The Case of Suicide Terrorism’, International Organization 64, no. 1 (January 2010): 33, https://doi.org/10.1017/S0020818309990233.

33 Sidney Tarrow, ‘Dynamics of Diffusion: Mechanisms, Institutions, and Scale Shift’, in The Diffusion of Social Movements: Actors, Mechanisms, and Political Effects, ed. Kenneth M. Roberts, Rebecca Kolins Givan, and Sarah A. Soule (Cambridge: Cambridge University Press, 2010), 204–20, https://doi.org/10.1017/CBO9780511761638.012; Kristin M. Bakke, ‘Copying and Learning from Outsiders? Assessing Diffusion from Transnational Insurgents in the Chechen Wars’, in Transnational Dynamics of Civil War, ed. Jeffrey T. Checkel (Cambridge: Cambridge University Press, 2013), 31–62, https://doi.org/10.1017/CBO9781139179089.005.

34 Dolnik, Understanding Terrorist Innovation: Technology, Tactics and Global Trends; Mauro Lubrano, ‘Navigating Terrorist Innovation: A Proposal for a Conceptual Framework on How Terrorists Innovate’, Terrorism and Political Violence, 5 April 2021, 1–16, https://doi.org/10.1080/09546553.2021.1903440.

35 Brian A Jackson, ‘Technology Acquisition by Terrorist Groups: Threat Assessment Informed by Lessons from Private Sector Technology Adoption’, Studies in Conflict and Terrorism 24, no. 3 (2001): 183–213, https://doi.org/10.1080/10576100151130270.

36 Gill et al., ‘Malevolent Creativity in Terrorist Organizations’, 1 June 2013.

37 Gill et al.

38 Kenney, From Pablo to Osama: Trafficking and Terrorist Networks, Government Bureaucracies, and Competitive Adaptation.

39 J.-C. Spender, ‘Organizational Learning and Competitive Advantage’, (London: SAGE Publications Ltd, 2021), 56–73, https://doi.org/10.4135/9781446250228.

40 Amy C Edmondson et al., ‘Learning How and Learning What: Effects of Tacit and Codified Knowledge on Performance Improvement Following Technology Adoption’, Decision Sciences 34, no. 2 (2003): 197–224, https://doi.org/10.1111/1540-5915.02316.

41 Michael Polanyi, Personal Knowledge: Towards a Post-Critical Philosophy (London: Routledge & Kegan Paul, 1962); Edmondson et al., ‘Learning How and Learning What: Effects of Tacit and Codified Knowledge on Performance Improvement Following Technology Adoption’.

42 J.-C. Spender, ‘Organizational Learning and Competitive Advantage’ (London: SAGE Publications Ltd, 2021), 56–73, https://doi.org/10.4135/9781446250228.

43 Kenney, From Pablo to Osama: Trafficking and Terrorist Networks, Government Bureaucracies, and Competitive Adaptation; Jacob N. Shapiro, The Terrorist’s Dilemma: Managing Violent Covert Organizations (Princeton: Princeton University Press, 2015); Thomas Hegghammer, ‘Resistance Is Futile’, Foreign Affairs (New York, N.Y.) 100, no. 5 (2021): 44–53.

44 J. Bowyer Bell, ‘Aspects of the Dragonworld: Covert Communications and the Rebel Ecosystem’, International Journal of Intelligence and CounterIntelligence 3, no. 1 (January 1989): 15–43, https://doi.org/10.1080/08850608908435089; Anne Stenersen, ‘The Internet: A Virtual Training Camp?’, Terrorism and Political Violence 20, no. 2 (9 April 2008): 215–33, https://doi.org/10.1080/09546550801920790; René M. Bakker, Jörg Raab, and H. Brinton Milward, ‘A Preliminary Theory of Dark Network Resilience’, Journal of Policy Analysis and Management 31, no. 1 (December 2012): 33–62, https://doi.org/10.1002/pam.20619.

45 Edmondson et al., ‘Learning How and Learning What: Effects of Tacit and Codified Knowledge on Performance Improvement Following Technology Adoption’.

46 Dominic Johnson, ‘Darwinian Selection in Asymmetric Warfare: The Natural Advantage of Insurgents And Terrorists’, Journal of the Washington Academy of Sciences 95, no. 3 (2009): 89–112.

47 Dominic Johnson.

48 Milton, ‘Structure of a State - Captured Documents and the Islamic State’s Organizational Structure’.

49 Whiteside et al., ‘The ISIS Files The Department of Soldiers’.

50 Indications of this can be found in Islamic State media content, like in a 2017 issue of the Islamic State Magazine Rumiyah, where a commander in Raqqa noted “the brothers’ experiences [in Mosul] have been passed on to all the wilayat so they could benefit from them, both militarily and in terms of iman (faith).” Interviewees also mentioned seeing evidence of exchange to Algeria, Egypt, Libya, and Tunisia, but this is likely not an exhaustive list. Interview with BEE1, August 2021; Interview with BTU1; Interview with BI1, August 2021; ‘It Will Be a Fire That Burns the Cross and Its People in Raqqah’, Rumiyah, 2017, 34.

51 Each interview typically lasted around one hour.

52 Consent was obtained by the research assistant, who received training in research ethics and consent procedures prior to the start of the project. Participants were given the opportunity to reach out to the research assistant directly via telephone to pull out of the study or withdraw information as this was both safer and more accessible than calling internationally or emailing the author.

53 We did not use a tape recorder for the majority of interviews due to participant safety concerns. This required the research assistant to take detailed notes in Arabic. Upon completing an interview, the notes would be sent to and translated by the author, who would use the replies to inform the next interview with the same participant type. This was an iterative process aimed at acquiring as detailed a picture as possible and to corroborate information across sources without being able to conduct follow up interviews. Around halfway through the interviews, participants who had spoken together expressed being uncomfortable with questions being updated. To avoid causing undue stress, we stopped updating questions for the second half. This was not a big issue because by this time we had been able to identify questions that gave sufficiently in-depth responses.

54 Whiteside et al., ‘The ISIS Files The Department of Soldiers’; Milton, ‘Structure of a State - Captured Documents and the Islamic State’s Organizational Structure’.

55 The group’s portrayal of its suicide bombers is an apt example. While suicide operatives and vehicles feature heavily in videos of the group’s offensives, the operatives are portrayed as excited and there are no signs of coercion. An interviewee involved in dismantling old suicide vehicles, however, reported often finding metal bars intended to restrain drivers in case they tried to back out in the last minute. Moreover, Whiteside and Miranova found that suicide attack ‘volunteers’ could be a misnomer in some cases, as there were several endogenous group processes determining who would end up carrying out attacks. Interview with MTU1, April 2021; Whiteside and Mironova, ‘Adaptation and Innovation with an Urban Twist: Changes to Suicide Tactics in the Battle for Mosul’.

56 These aspects are better explored by examining accounts by group members, like those used to inform Mironova’s study of jihadi human resources, see: Mironova, From Freedom Fighters to Jihadists.

57 Huber, ‘Organizational Learning: The Contributing Processes and the Literatures’.

58 Mark Easterby-Smith and Marjorie A Lyles, Handbook of Organizational Learning and Knowledge Management, 2nd ed. (Chichester, West Sussex: Wiley, 2011), 183–349; Richard Michael Cyert and James G March, A Behavioral Theory of the Firm, Prentice-Hall International Series in Management (Englewood Cliffs: Prentice-Hall, 1963).

59 Several works address these questions in the context of Salafi Jihadi groups, among them: Moghadam, The Globalization of Martyrdom: Al Qaeda, Salafi Jihad, and the Diffusion of Suicide Attacks; Jeffrey William Lewis, The Business of Martyrdom: A History of Suicide Bombing (Annapolis: Naval Institute Press, 2012).

60 Huber, ‘Organizational Learning: The Contributing Processes and the Literatures’, 91.

61 Interview with MD1, April 2021; Interview with MD2, April 2021; Interview with MD3, May 2021.

62 Interview with MD1; Interview with MD2; Interview with MD3.

63 Interview with MD1; Interview with BD2, August 2021; Interview with BD1, August 2021.

64 ISIS Files, ‘16_001042’ (George Washington University, n.d.), 001, accessed 20 February 2021; ISIS Files, ‘16_001043’ (George Washington University, n.d.), accessed 20 February 2021; ISIS Files, ‘16_001044’ (George Washington University, n.d.), 00, accessed 20 February 2021.

65 Interview with MEE1, April 2021; Interview with MTU1.

66 An anecdote by a foreign fighter interviewed by Vera Mironova captured one of the situations that followed: “Once the group had a suicide mission volunteer detonate a car filled with explosives under a bridge. But there was no enemy near the bridge, and we could have just gone there at night, quietly positioned the explosives, and detonated them remotely.” Vera Mironova, From Freedom Fighters to Jihadists, Causes and Consequences of Terrorism (New York: New York: Oxford University Press, 2019), 169, https://doi.org/10.1093/oso/9780190939755.001.0001.

67 Document Islamic State in Iraq, ‘Analysis of the State of ISI’ (Combating Terrorism Center at West Point, 2006).

68 Charlie Winter, ‘War by Suicide: A Statistical Analysis of the Islamic State’s Martyrdom Industry’, 15.

69 Interview with BI1.

70 Interview with MEE1.

71 Interview with MBS1, June 2021; Interview with MEE2, May 2021.

72 Interview with BI1.

73 Interview with MTU1.

74 Interview with BEE1.

75 Interview with BEE1.

76 Interview with MI1, March 2021.

77 Kaaman, Hugo, ‘Through the Desert & down the Euphrates – Islamic State SVBIED Use & Innovation’, Hugo Kaaman – Open Source Research on SVBIEDs (blog), 29 May 2018, https://hugokaaman.com/2020/09/21/the-unique-svbieds-used-in-the-2013-capture-of-menagh-airbase/; Kaaman, ‘Car Bombs as Weapons of War: ISIS’S Development of SVBIEDS, 2014–19’.

78 Interview with BI1. Interview with MTU1; Interview with BEE2, August 2021.

79 Operation rooms could be established for particular operations and was considered an early indicator that an attack was underway. Interview with BI1.

80 Interview with BI1.

81 Interview with MTU1; Interview with BEE2.

82 Huber, ‘Organizational Learning: The Contributing Processes and the Literatures’, 102.

83 Interview with MI1; Interview with MEE1.

84 Aymenn al-Tamimi, ‘The Evolution in Islamic State Administration: The Documentary Evidence’, Perspectives on Terrorism (Lowell) 9, no. 4 (2015): 117–29.

85 According to Whiteside et al., the Al-Bara’ bin Malik Brigade was a part of the Committee for Development and Manufacturing, the Brigade likely fell under the Committee’s aviation unit. ‘Specimen 30C’, 2015, http://www.aymennjawad.org/2016/09/archive-of-islamic-state-administrative-documents-2; ‘Specimen 30D’, 2015, http://www.aymennjawad.org/2016/09/archive-of-islamic-state-administrative-documents-2; ‘Specimen 30E’, 2015, http://www.aymennjawad.org/2016/09/archive-of-islamic-state-administrative-documents-2; ‘Specimen 30F’, 2015, http://www.aymennjawad.org/2016/09/archive-of-islamic-state-administrative-documents-2; Whiteside et al., ‘The ISIS Files The Department of Soldiers’; Don Rassler, Muhammad Al-Ubaydi, and Vera Mironova, ‘The Islamic State’s Drone Documents: Management, Acquisitions, and DIY Tradecraft’, CTC Sentinel, 31 January 2017, https://www.ctc.usma.edu/ctc-perspectives-the-islamic-states-drone-documents-management-acquisitions-and-diy-tradecraft/.

86 Textbook examples and templates for military orders from the ISIS Files indicates how communication along the hierarchy and between the departments might have looked like. That being said, interviewees said most communication happened over social media apps, from messaging services to video games. ISIS Files, ‘13_000972’ (George Washington University, n.d.), accessed 20 February 2021; Interview with BEE2.

87 Whiteside et al., ‘The ISIS Files The Department of Soldiers’.

88 Interview with BEE2. Interview with BI1.

89 Fisher, ‘ISIS Just Pulled off Its First Carbombing with a Stolen American Humvee’.

90 Charlie Winter, ‘War by Suicide: A Statistical Analysis of the Islamic State’s Martyrdom Industry’.

91 Interview with MI1.

92 Mironova, From Freedom Fighters to Jihadists, 256–60.

93 Mironova, From Freedom Fighters to Jihadists, 261.

94 Mironova.

95 Interview with MTU1; Interview with MEE2.

96 Interview with BI1.

97 The geographical reach of the CDM is difficult to discern. That being said, it had a physical presence in multiple cities, with interviewees mentioning Mosul, Baghdad, Kirkuk, Fallujah, Tel Afar, and the Ramadi dessert, but this should be viewed as indicative only. Interview with MEE2; ‘Specimen 30H’, 2015, http://www.aymennjawad.org/2016/09/archive-of-islamic-state-administrative-documents-2; ‘Specimen 30I’, 2015, http://www.aymennjawad.org/2016/09/archive-of-islamic-state-administrative-documents-2; ‘Specimen 30K’, 2015, http://www.aymennjawad.org/2016/09/archive-of-islamic-state-administrative-documents-2; Interview with BTU1; Interview with BEE2; Conflict Armament Research, ‘Standardisation and Quality Control in Islamic State’s Military Production’ (London: Conflict Armament Research, December 2016), 6, https://www.conflictarm.com/dispatches/standardisation-and-quality-control-in-islamic-states-military-production/.

98 Interview with MEE1.

99 Interview with MEE2; Interview with BEE2; Interview with MTU1.

100 Interview with MEE2; Interview with BEE2; Avi Asher-Schapiro, ‘The US-Led Coalition Bombed the University of Mosul for Being an Islamic State Headquarters’, Vice, 22 March 2016, https://www.vice.com/en/article/mbnkba/the-us-led-coalition-bombed-the-university-of-mosul-for-being-an-islamic-state-headquarters; Interview with MTU1.

101 AFP, ‘ISIS Using Hobby Drones to Bomb Forces in Mosul’, Al Arabiya News, 12 January 2017, https://english.alarabiya.net/News/middle-east/2017/01/12/ISIS-using-hobby-drones-to-bomb-Iraqi-forces-in-Mosul; Interview with MEE2; Interview with MEE1.

102 Hugo Kaaman, ‘The Myth of the Remote Controlled Car Bomb’ (European Eye on Radicatization, 15 September 2019), 7.

103 ‘Aleppo_RCVBIED_outcomes’ (Aleppo Province of the Islamic State, n.d.), Document obtained by author; Kaaman, ‘The Myth of the Remote Controlled Car Bomb’.

104 Interview with MBS1; Interview with BEE2; Interview with BTU1; Interview with BI1.

105 Interview with BEE1.

106 Interview with MD3; Interview with MI1; Christopher Woody, ‘Watch a US-Led Airstrike Level an ISIS Bomb Factory Days before the Assault on Mosul’, Business Insider, 18 October 2016, https://www.businessinsider.com/us-coalition-airstrike-isis-bomb-factory-mosul-2016-10?r=US&IR=T; Asher-Schapiro, ‘The US-Led Coalition Bombed the University of Mosul for Being an Islamic State Headquarters’.

107 Interview with MEE2; Interview with MBS1. Statements in the interviews resonate with the observations by CAR, whose researchers in Mosul noted, “Although production facilities employ a range of non-standard materials and chemical explosive precursors, the degree of organisation, quality control, and inventory management, indicates a complex, centrally controlled industrial production system” Conflict Armament Research, ‘Standardisation and Quality Control in Islamic State’s Military Production’, 4.

108 Interview with BEE1; Interview with MEE2; Interview with BEE2.

109 Interview with MEE2; Interview with MTU1.

110 ‘Specimen 17F’, 2015, http://www.aymennjawad.org/2016/01/archive-of-islamic-state-administrative-documents-1; ‘Specimen 39U’, 2015, 39, http://www.aymennjawad.org/2016/09/archive-of-islamic-state-administrative-documents-3; ‘Specimen 33W’, 2015, http://www.aymennjawad.org/2016/09/archive-of-islamic-state-administrative-documents-2; Interview with BTU1; Interview with BI1.

111 Interview with MBS1.

112 ‘Specimen 17F’; ‘Specimen 33W’; Interview with BTU1; Interview with BI1.

113 Interview with MEE1; Interview with MEE2; Interview with BTU1.

114 Interview with MEE1.

115 Interview with MEE1.

116 Interview with MEE2; Interview with BEE1; Interview with MEE1.

117 Interview with MD1; Interview with MD2; Interview with BD1; Interview with BD2.

118 Interview with MEE2.

119 Islamic State documents showing the connection to Turkey, most importantly Specimen 21Z, titled “Report on fertilizer entering from Turkey through Tel Abyad”, was made public by the People’s Defense Units and the Syrian Democratic Forces (YPG and SDF) in 2015. According to Tamimi, this was part of an effort on the part of the YPG “to push the line of active Turkish state support for the Islamic State that can be traced to the highest levels of government.” What the document really shows, he argues, “is bribery and corruption above all at the local administrative level.” Aymenn Jawad Al Tamimi, ‘Archive of Islamic State Administrative Documents (Cont.)’, Aymenn Jawad al Tamimi (blog), 11 January 2016, http://www.aymennjawad.org/2016/01/archive-of-islamic-state-administrative-documents-1; Conflict Armament Research, ‘Standardisation and Quality Control in Islamic State’s Military Production’; Conflict Armament Research, ‘Weapons of the Islamic State’ (London: Conflict Armament Research, December 2017), https://www.conflictarm.com/dispatches/standardisation-and-quality-control-in-islamic-states-military-production/. Interview with MI1.

120 Interview with MEE2; Interview with MEE1; Interview with BTU1.

121 The interviewee did not mention where the foreign experts were from, but in their examination of drone manufacturing documents Rassler et.al. found a prevalence of fighters from Bangladesh. Interview with MEE2; Rassler, Al-Ubaydi, and Mironova, ‘The Islamic State’s Drone Documents: Management, Acquisitions, and DIY Tradecraft’.

122 Craig Whiteside, ‘A Pedigree of Terror’, Perspectives on Terrorism 11, no. 3 (2017): 2–18.

123 Interview with BEE1; Interview with MI1.

124 Interview with MEE2; Interview with BEE1; Interview with BEE2.

125 Interview with BTU1.

126 Interview with BEE2.

127 Interview with BI1; Interview with BEE2; Interview with MI1.

128 Conflict Armament Research, ‘Standardisation and Quality Control in Islamic State’s Military Production’, 6.

129 Interview with MEE1; Interview with BEE1. Interview with MTU1; Interview with BI1.

130 Interview with MI1; Interview with BI1; Interview with BEE2.

131 Interview with BEE2; Interview with BEE1.

132 Interview with BEE2.

133 John Mueller and Mark G Stewart, ‘The Terrorism Delusion: America’s Overwrought Response to September 11’, International Security 37, no. 1 (2012): 81–110, https://doi.org/10.1162/ISEC_a_00089; Bloom, ‘Constructing Expertise: Terrorist Recruitment and “Talent Spotting” in the PIRA, Al Qaeda, and ISIS’.

134 Interview with BI1; Interview with MEE2; Interview with BEE1.

135 Interview with BEE1. Interview with MEE2.

136 Interview with MEE2; Interview with MTU1.

137 Interview with MD3.

138 Huber, ‘Organizational Learning: The Contributing Processes and the Literatures’, 90.

139 Conflict Armament Research, ‘Standardisation and Quality Control in Islamic State’s Military Production’, 32.

140 Conflict Armament Research, 31.

141 Interview with BEE1; Interview with MEE1.

142 Kaaman, ‘The History and Adaptability of the Islamic State Car Bomb’.

143 Interview with BEE2; Interview with BI1; Interview with BEE1.

144 Interview with MBS1.

145 Interview with BI1.

146 Hugo Kaaman, ‘Islamic State of Iraq – A Snapshot of SVBIED Design & Use (2007-2012)’, Hugo Kaaman - Open Source Research on SVBIEDs (blog), 11 January 2019, https://hugokaaman.com/2019/01/11/islamic-state-of-iraq-a-snapshot-of-svbied-design-use-2007-2012/.

147 Interview with MEE1; Interview with MD3.

148 Huber, ‘Organizational Learning: The Contributing Processes and the Literatures’, 105.

149 Kaaman, Hugo, ‘Through the Desert & down the Euphrates – Islamic State SVBIED Use & Innovation’.

150 Eli Berman and David D. Laitin, ‘Religion, Terrorism and Public Goods: Testing the Club Model’, Journal of Public Economics 92, no. 10–11 (October 2008): 1942–67, https://doi.org/10.1016/j.jpubeco.2008.03.007.

151 The advantage is captured by comparing the Martyrdom Battalion to the beginning of the Kamikaze Corps in Japan. Despite a longstanding culture of heroic sacrifice in Japan, none of the professional Japanese soldiers volunteered when the corps was instituted, and leaders had to order officers of the military academies to fill these roles. One of the officers who received this order, Captain Seki Yukio, expressed that, “There is no more hope for Japan, if it has to kill such a skillful pilot like myself. I can hit an aircraft carrier with a 1,102 lb. bomb and return alive, without having to make a suicidal plunge.” Ivan Morris, The Nobility of Failure : Tragic Heroes in the History of Japan (London: Secker & Warburg, 1975); Emiko Ohnuki-Tierney, Kamikaze, Cherry Blossoms, and Nationalisms : The Militarization of Aesthetics in Japanese History (Chicago, Ill.: University of Chicago Press, 2002), 166.

152 Assaf Moghadam, ‘Motives for Martyrdom: Al-Qaida, Salafi Jihad, and the Spread of Suicide Attacks’, International Security 33, no. 3 (1 January 2009): 46–78, https://doi.org/10.1162/isec.2009.33.3.46.

153 Assaf Moghadam, ‘Palestinian Suicide Terrorism in the Second Intifada: Motivations and Organizational Aspects’, Studies in Conflict and Terrorism 26, no. 2 (2003): 65–92, https://doi.org/10.1080/10576100390145215.

154 The importance of territory is also evident in long-term innovative work conducted by other territorial groups, such as the FARC in Columbia. Michelle Jacome Jaramillo, ‘The Revolutionary Armed Forces of Colombia (FARC) and the Development of Narco-Submarines’, Journal of Strategic Security 9, no. 1 (2016): 49–69, https://doi.org/10.5038/1944-0472.9.1.1509.

155 Jackson et al., Aptitude for Destruction, Volume 2.

156 A possible exception here is the Provisional Irish Republican Army, which managed to continue mortar innovation despite increasing levels of government repression. However, technological advances in state capabilities in the last twenty years have arguably made activities of that ilk unfeasible. Hegghammer, ‘Resistance Is Futile’; Bloom, ‘Constructing Expertise: Terrorist Recruitment and “Talent Spotting” in the PIRA, Al Qaeda, and ISIS’; Tony Geraghty, The Irish War - The Military History of a Domestic Conflict (London: Harper Collins Publishers, 2000).

157 Hegghammer, ‘Resistance Is Futile’.