Jump to ContentJump to Main Navigation
The Terrorist's DilemmaManaging Violent Covert Organizations$

Jacob N. Shapiro

Print publication date: 2013

Print ISBN-13: 9780691157214

Published to Princeton Scholarship Online: October 2017

DOI: 10.23943/princeton/9780691157214.001.0001

Show Summary Details
Page of

PRINTED FROM PRINCETON SCHOLARSHIP ONLINE (www.princeton.universitypressscholarship.com). (c) Copyright Princeton University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in HSO for personal use (for details see http://www.universitypressscholarship.com/page/privacy-policy). Subscriber: null; date: 17 August 2018

Discrimination and Control in Ireland

Discrimination and Control in Ireland

(p.169) Chapter 7 Discrimination and Control in Ireland
The Terrorist's Dilemma

Jacob N. Shapiro

Princeton University Press

Abstract and Keywords

This chapter studies the three most prominent terrorist groups operating in Northern Ireland from the 1960s through 2003: the Provisional IRA, the Ulster Defense Association (UDA), and the Ulster Volunteer Force (UVF). Because the history of terrorism in Northern Ireland is so well known, the case provides an excellent venue for testing hypotheses about the relationship between discrimination and control. The history of the groups fighting in Northern Ireland also provides a critical illustration of the policy importance of this kind of organizational analysis. From 1987 on, leaders on both sides of the Northern Ireland conflict understood the broad contours of a negotiated settlement, but it took them many years to work the internal politics of their organizations to the point at which ceasefire orders were obeyed.

Keywords:   Northern Ireland, Provisional IRA, Ulster Defense Association, Ulster Volunteer Force, negotiated settlement, internal politics, discrimination

7.1 Introduction

The Northern Ireland conflict provides an excellent place to further examine how terrorist groups are organized. It is almost certainly the best studied terrorist conflict of the twentieth century. Historians started working closely with prominent participants on all sides of the conflict from almost the first day of the Troubles in August 1969 and continue doing so to this day. The conflict has been one of the longest terrorist struggles in recent history. For almost forty years the country was home to a conflict between two camps of terrorist organizations: Republicans (Catholics) who sought union with the Republic of Ireland or, at a minimum, greater independence from the United Kingdom (UK); and Loyalists (Protestants) who sought to ensure that Northern Ireland remained a part of the UK, or to make sure it did not become part of the Republic of Ireland. During this period the main Republican organization was the Provisional Irish Republican Army (PIRA). On the Loyalist side, there were two prominent organizations—the Ulster Volunteer Force (UVF) and the Ulster Defense Association (UDA).

The rich secondary history of this conflict is complemented by a raft of memoirs written by former fighters and some government agents. Beyond these qualitative accounts, the records of who was attacked by whom and why are so good for the Northern Ireland conflict that we can directly assess how control was exercised, both in terms of the level of care different actors took to avoid killing their own civilians and in terms of how much internal violence existed among combatants on each side. Equally important, the three main terrorist groups in this conflict differed in clear ways in terms of the level of discrimination in the use of violence demanded by their political goals, but were quite comparable on most other dimensions. As we will see in greater detail, the PIRA’s goals required more discrimination than those of the Loyalist paramilitaries. Additionally, the level of discrimination required by the PIRA’s goals increased over time as the group’s leaders placed a greater emphasis on (p.170) politics and as the Catholic population became less tolerant of civilian casualties. Within the Loyalist side, the UVF placed a greater value on discrimination than the UDA, because its leaders envisioned themselves as a small elite outfit, but also because it became affiliated with a political party much earlier, which created greater incentives to avoid hurting Protestant civilians than the UDA faced.

Outside of these differences in discrimination, the PIRA, UDA, and UVF look quite similar on other dimensions, including preference divergence, uncertainty, and security pressure. All had their origin in the communal violence of 1966–1969. All recruited from similar working-class populations, largely by working through existing social ties. And all three were proscribed, or at least their militant wings were, for most of the conflict. In the early years of the conflict, the PIRA and UVF faced greater security pressure than the UDA, but by the mid-1980s, all faced concerted counterterrorism pressure. There was some difference in the basic quality of their recruits. The most able potential recruits on the Loyalist side joined the state security apparatus, which meant that, on average, the UVF and UDA recruited from a less talented pool than the PIRA.1 But this difference does not seem to have been critical, and the differences we will see between the UVF and UDA provide some confidence that the differences between the Republican and Loyalist groups were not driven solely by the relative quality of their operatives.

Finally, the Northern Ireland conflict presents a stark illustration of how terrorists’ internal organizational problems interact with the government’s strategic calculus. In particular, from 1990 onward, if not earlier, leaders of both the PIRA and the British government knew there would be a negotiated settlement. From that point until the Good Friday Accords were finally signed in 1999, the British faced a delicate challenge. They had to put sufficient pressure on the PIRA leadership that they would accept terms the British could live with, but they could not put too much pressure on the leaders or they would lose their ability to control their operatives and so would not be able to deliver on the deal.

This chapter starts by assessing whether the relationship between discrimination and control in terrorist organizations predicted in earlier chapters holds in this conflict. Given the differences among the PIRA, UDA, and UVF in terms of discrimination, we should expect that:

(p.171) H1: PIRA leaders will exercise greater control than UVF or UDA leaders.

H2: PIRA leaders will exercise more control over time.

H3: UVF leaders will exercise greater control than UDA leaders because they saw a higher political value to the careful application of violence.

Our previous discussion of terrorists’ organizational dilemmas also suggests that we should find plenty of evidence that:

H4: Leaders’ abilities to exercise control will be restricted by preference divergence.

To test the differences between groups, section 7.2 begins by examining the broad patterns of violence in this conflict, showing that they are consistent with hypotheses 1–3. In order to provide a sense of what to look for in the detailed case studies, section 7.3 then lays out a method for assessing differences in control across groups. Sections 7.4 and 7.5 analyze the histories of the PIRA and the two main Loyalist paramilitaries. Section 7.6 concludes by reviewing the evidence for and against the four hypotheses and discusses some of the main implications of this chapter.

7.2 Discrimination and Patterns of Violence

The political goals of the PIRA required great care in the application of violence. The group had to place sufficient military pressure on the British so as to compel them to give up Northern Ireland, while at the same time avoiding excessive violence that would cost the PIRA the support of moderate Catholics or make it politically impossible for British politicians to “abandon” the Protestant majority by leaving Northern Ireland. The demand for discrimination by Republican leaders increased throughout the 1980s as Sinn Féin, the political wing of the movement, took on an increasingly central role in Republican strategy.

By comparison, the main political goals of the Loyalist organizations were well-supported by indiscriminate violence. These goals were to: (1) maintain the political status quo; and (2) deter Republican violence against Protestant civilians.2 One way to maintain the political status quo was to convince British leaders that the Protestant majority opposed (p.172) any moves away from a union with Great Britain. The existence of violent organizations that opposed any concessions supported this goal, regardless of how those organizations used violence. Indiscriminate violence also served a useful deterrent purpose, at least in theory. Loyalist paramilitaries clearly saw essentially random attacks against Catholic civilians as an effective means of protecting Protestants.3 Indeed, from a theoretical standpoint, giving up some control over the response to an attack can enhance the power of a deterrent threat.4

There was also substantial variation in the need for discrimination between the Loyalist paramilitaries. The relatively successful Progressive Unionist Party (PUP) emerged out of the UVF in 1979 and remained closely tied to the UVF leadership.5 By comparison, the UDA did not form its political wing, the Ulster Democratic Party (UDP), until 1989 and it never achieved much in the way of electoral success.6 Simply put, the electoral success of their political wing mattered to the UVF’s leaders, which meant that hitting the wrong targets was more costly for the UVF than for the UDA.7

Given these differences, we should expect higher levels of control among Republicans to be reflected in the aggregated data on violence. Specifically, we should see that: (1) Republicans killed fewer civilians per attack and (2) there was more internal violence among Republicans. To test these predictions, table 7.1 summarizes the patterns of killings by the two sides in this conflict for the entire period from 1969 through 2001. To provide some context for the numbers, Northern Ireland, where the vast majority of the violence took place, was roughly 38 percent Catholic in 1991 and 40 percent Catholic in 2001.8 (p.173)

Table 7.1. Killings in Northern Ireland, 1969–2001









































Source: Malcom Sutton, An Index of Deaths from the Conflict in Northern Ireland, available at the CAIN (Conflict Archive on the Internet) website, http://cain.ulst.ac.uk/sutton/. Last accessed June 22, 2007.

As H1 predicts, Republican organizations killed fewer civilians than Loyalists, which was only half as many as a proportion of all their killings. Indeed, Republican attacks killed approximately one Catholic civilian for every five enemy combatants (227 to 1133), while Loyalists killed three Protestant civilians for every one enemy combatant (132 to 42). As we would expect given their increasing need for discrimination over time, PIRA killings became more selective later in that conflict.9

(p.174) As we also expected, Republican organizations engaged in proportionally more intra-community violence than the Loyalists. This fits with the theory outlined so far. As the value of discrimination increases, terrorist leaders should be more willing to actually sanction their operatives. If, after all, the value of discrimination is low, why would leaders incur the security costs of punishment? In chapter 5, we focused on equilibrium in which the threat of punishment precluded operatives from behaving against their leaders’ wishes. If we think about that same managerial problem in a probabilistic world, in which agents just mess up sometimes (or in which leaders occasionally recruit really bad operatives by mistake), then the probability of observing groups punishing operatives will be higher, all other things being equal, in groups that place a higher value on hitting the right target.

Much of this intra-community violence was driven by disciplinary actions against combatants and so provides evidence on the extent to which control was exercised through violent sanctioning. Of the 135 PIRA operatives killed by fellow Republicans, 132 (97.8%) were killed by their own organization. In general, when Republican groups killed combatants, they were four times more likely to be attacking their own personnel than the other side. By contrast, Loyalists were only twice as likely to be attacking their own personnel. That the rate of violence against one’s own combatants was higher on the Republican side is consistent with the notion that Republican leaders placed a greater value on controlling the use of violence and were therefore more likely to need to sanction their own members.

At first glance a similar comparison does not work clearly within the Loyalist side as there was not a great deal of variation in the proportion of civilians killed by different groups. Of the UVF victims, 84 percent were civilians compared to 80 percent of UDA victims. There was, however, a good deal of variation in the targeting of other militants between the UDA and UVF. The UVF killed twice as many Loyalists as Republicans, while the UDA and its flag of convenience organization, the Ulster Fighting Force (UFF), killed more than three times as many Loyalists as Republicans. Initially this pattern is hard to interpret. The UVF placed a greater emphasis on discrimination, meaning it should have been more likely to punish any given infraction.

But, if we account for group size, then disciplinary violence seems to have been much more prominent in the UVF, reflecting a greater effort (p.175) to control the organization. At its peak in the mid-1970s, the UDA had as many as 30,000 dues-paying members. By the end of the 1990s it was down to 2,000 members.10 By contrast, the UVF peaked at 2,000 members in the mid-1970s and had as few as 120 members by 1998.11 With this difference in mind we can break down the Loyalist-on-Loyalist killings into two categories: those that occurred during disputes between organizations and those that involved intra-organizational disputes. Of the forty-one Loyalists killed by the UVF, twenty-nine (70.7%) were UVF members. Of the thirty-seven Loyalists killed by the UDA, twenty-eight (75.6%) were in the UDA.12 Since the UVF has never been more than one tenth the size of the UDA, these numbers mean the UVF was killing its own members at a rate of more than ten times that of the UDA.13 That the two groups killed similar numbers of their own personnel, in other words, means that internal discipline was much more often enforced in the UVF. This is exactly what we should expect given the greater value the UVF placed on discrimination, and it is in line with H3.

7.3 Measuring Control

To measure the level of control exercised by political and ideological leaders we need a simple, relatively objective way to decide which group exercised more control. One way to do this is to focus on areas of organizational activity where we can generally find evidence of whether leaders were able to exercise control. In the Northern Ireland context (and the Palestinian-Israeli one studied in the next chapter), there are six such areas:

  1. 1. Target choices

  2. 2. Operational procedures

  3. 3. Personnel assignments

  4. 4. Resource allocation

  5. 5. Fund-raising methods

  6. 6. Recruiting practices

(p.176) By asking how many of these areas are controlled by the political and ideological leadership, we can generate an ordinal scale of control. A group in which five of six areas are centrally controlled can be said to exhibit a higher level of control than a group in which only three of the six are.

So, what exactly does it mean for leaders to exercise control over each of these areas? Control over target choices means that leaders either designate targets, have a chance to exercise an effective veto over targeting choices, or are sufficiently satisfied with the targeting choices of operatives that they choose not to get involved. When leaders express dissatisfaction over attacks but do not influence subsequent targeting choices, that indicates a lack of control. Control over operational procedures entails leaders being able to decide how operations are conducted or demonstrating an ability to control things like the number of connections between cells, or the process by which new members are trained. Control over personnel assignments simply means that leaders control who fills which positions, either directly or by manipulating the selection process. Control over resources requires that leaders either directly manage resources such as weapons and funds, or that they willingly relinquish control, but there is little evidence that members appropriate organizational resources for private use. Control over fund-raising requires that leaders either raise funds through a centralized unit whose procedures they control, or that they be able to limit the methods by which operatives fund themselves. Control over recruiting involves either making sure recruiting happens through an approved method, or by being directly involved in selecting and vetting new members.

With these definitions in mind, let us turn to brief analytical narratives of the main terrorist organizations in this conflict.

7.4 The Provisional IRA

From its formation in 1969 through the present, the Provisional Irish Republican Army has probably been the most studied “secret” army in the world.14 There is a wealth of secondary literature on the organization, from excellent histories, first-person memoirs and journalistic accounts, to scholarly analyses and the wealth of primary and secondary resources available from the Conflict Archive on the Internet (CAIN) website.15

(p.177) Drawing on this literature, we can divide the organization’s history into three periods:

  • the early period from its formation in late 1969 to the first major truce in February 1975;

  • the middle period from the ceasefire until 1989 when the leadership realized it was in a military stalemate with Britain; and

  • the long, slow movement toward peace as the leadership fought a series of internal political battles from 1989 through 1997 to bring a still-militant organization to the peace table.

By July 1997, the vast majority of Republican militants had accommodated themselves to the negotiated settlement that would be embodied in the April 1998 Good Friday Accords. From that point on, terrorist violence on the Republican side would be the sole province of radical splinter groups such as the Real IRA (RIRA) and the Irish National Liberation Army (INLA), which conducted three attacks after the Accords before declaring a ceasefire which was maintained until 2007 when the group claimed credit for shooting and injuring a policeman.16 Rather than retread a history that is told in great detail elsewhere, my account will focus on evaluating whether this case supports the hypotheses outlined above.

7.4.1 Formation to Truce: 1969–1975

The early history of the Provisional IRA is traditionally told along the outlines best described in Richard English’s Armed Struggle: The History of the IRA: failure of old IRA in 1956–62 border campaign; IRA moves left politically and military capabilities atrophy; non-nationalist civil rights movement becomes very active in 1967–68, triggering violent response from Loyalists in summer 1969; the PIRA emerges to defend Catholic communities.17 Descriptions of the movement of the PIRA into open warfare against British forces tend to follow a similar narrative: British originally welcomed by Catholics; ham-handed peacekeeping efforts by British Army alienate the Catholic population; the PIRA gains new recruits and begins attacking British forces due to groundswell of support for doing so among rank and file; and steady escalation of violence up to 1975 ceasefire.18

(p.178) The most important pattern for our purposes is that at this early stage the PIRA’s central leadership exercised little control over targeting choices, but the leaders of local IRA units did control targeting choices within their areas. This pattern of control was due largely to the fact that the PIRA began in 1969 when locally cohesive IRA units switched their allegiance from Cathal Goulding and the leftist members of the IRA leadership to Sean MacStiofain and his more nationalist compatriots.19 Those who remained loyal to the Goulding leadership formed the Official IRA (OIRA) which competed with the PIRA for influence and recruits until 1972, when the OIRA turned away from militancy.

Tensions within the IRA began to mount in the summer of 1969 over the unwillingness of the Dublin-based Goulding leadership to do more to defend Catholic areas from Loyalist violence.20 By August, a group of the Northern IRA leaders, mostly from Belfast, began making plans to break from the Southern leadership, which they saw as insufficiently committed both to the military defense of the Catholic community and to the idea of a united Ireland.21 After taking over the Belfast IRA in September, the militant faction led by Sean MacStiofain split off completely in December 1969 to form the Provisional IRA, creating its own political wing, Provisional Sinn Féin, the next month.22

By April 1970, the PIRA had grown so large that it had to reorganize, moving from sixteen or seventeen (depending on the source) quasi-independent areas under one full-time staff to three geographically defined battalion areas in Belfast, each of which had a set of locally recruited, full-time members, supported by part-time auxiliaries and local defense committees.23 Within these areas, leaders controlled their operatives’ tactical choices to support strategic objectives. For example, during intercommunal rioting in the summer of 1970, Gerry Adams reportedly (p.179) ordered the Ballymurphy IRA not to break out its firearms. This led to more intense rioting in that Belfast neighborhood than elsewhere in the city. The intense rioting helped Adams’ unit because it radicalized the population and created many more opportunities for the British Army to anger local youth, who would flock to join the PIRA.24 A similar pattern of very effective local control over operatives emerged outside of Belfast. Despite being renowned for their lack of attention to orders from the Dublin leadership, South Armagh PIRA operatives scrupulously followed local orders not to fire on the Republic of Ireland police force, the Gardaí, even if it meant being arrested.25 South Armagh was the most important border area for the PIRA and the local leadership presumably felt it was worth losing a certain number of operatives to avoid alienating the population on the Republic of Ireland side of the border.

As the violence between the PIRA and the British Army intensified in 1971, Billy McKee, the senior leader in Belfast, proved unable to control his Volunteers. The Dublin leadership’s inability to punish violations appears to have been the key problem. Those who broke the rules laid down by the PIRA leadership in Dublin faced few consequences.26 Operations were more tightly controlled in localities where discipline was easier. For example, the Derry Brigade leader, Martin McGuinness, made it a priority to avoid civilian casualties during his 1972 bombing campaign against commercial targets. The smaller population and very tightly-knit Catholic community in Derry made it relatively easy for McGuinness to impose internal discipline. As a result, his orders to abort missions because of civilian casualty risks were followed, and the campaign devastated central Derry without any such deaths.27

Where the leadership did exercise control during this early period was over targeting at the strategic level, specifically operations in England. The decision to begin bombings in England was made in late 1972 or early 1973 following a particularly damaging series of arrests of operatives in Northern Ireland.28 It is not clear whether this decision was driven by: (1) a desire to take the pressure off the group by getting the security forces to focus on England; (2) the fact that Seamus Twomey, who took over as PIRA Chief of Staff following MacStiofain’s arrest in November 1972, was more aggressive; (3) a strategic calculation that bombs in England could influence British public opinion against the presence of the British Army (p.180) in the North;29 or (4) a consensus on the PIRA Army Council that the attacks in Derry and Belfast were not sufficiently persuasive to the British government.30 What is clear is that the core PIRA leadership chose the personnel for the mission,31 set the boundaries for how careful the team had to be about civilian casualties,32 and could turn the bombings on and off to support their negotiating strategy in the lead up to the 1975 ceasefire.33

Despite their inability to tightly control violence in the North, the PIRA leaders in Dublin, Sean MacStiofain and Ruari O’Bradaigh, were able to control personnel movements at the senior levels and for major operations. When McKee was arrested in April 1971, Joe Cahill was chosen as his replacement and directed the promotion paths of those moving up as a result.34 Likewise, when Cahill held an ill-advised press conference following his escape from the British internment sweeps in August 1971, MacStiofain ordered him to Dublin, chose Seamus Twomey as his replacement, and assigned Cahill a new role running the centrally controlled portion of the group’s finances.35 MacStiofain also chose the new chain of command in Derry where the sweeps caught the group’s senior leaders.36 These personnel movements highlighted how much control the PIRA leadership exercised over personnel matters at this point.

Throughout this early period, important aspects of operational procedure were also controlled from the center, at least when operatives’ actions began to hurt the group. Following the August 1971 sweeps, many PIRA operatives fled to Dundalk in the Republic of Ireland, where their poor behavior—mainly drinking and publicly brawling among themselves—began to alienate the local population. This border area was vital for PIRA operations, meaning the group could not afford to lose local support. Eventually the problem became prevalent enough that the PIRA leaders (p.181) forbade men on the run to live further north than Dublin, so that it was easier for the leadership to monitor their activities.37

The PIRA central leadership also exercised a good deal of control over fund-raising during this period. The Army Council both prohibited actions they felt would hurt the cause and ensured that orders about raising funds were followed. Until May 1973, the leadership prohibited raising funds through armed robbery in the Republic of Ireland because doing so in the 1940s had brought the organization into conflict with local officials. For reasons that are unclear, the Dublin-based leaders changed their mind on this score and ordered Southern units (those operating in the Republic) to begin raising funds through robberies in May 1973. Local PIRA leaders in Kerry ignored the order, arguing that it would sour their relationship with the local civilians. Four senior leaders from Belfast went to Kerry in response, dismissed the old leadership, and replaced them with more compliant individuals.38 In this case, the central leadership exercised control by removing agents who were not following orders.

7.4.2 Ceasefire, Reorganization, and Stalemate: 1975–1990

When a PIRA bomb exploded in Birmingham, England, on November 21, 1974, killing 19 and injuring 184, it is unlikely that many expected it to catalyze the peace process and push the PIRA into its first longterm truce. However, the bombing had three effects that paved the way for exactly that. First, it triggered widespread revulsion toward the PIRA, which the group was unable to redirect even though it denied responsibility.39 Second, the bombing allowed for the swift passage of the Prevention of Terrorism Act (PTA) in England. The PTA made the IRA illegal, which made it easier to prosecute prominent Republicans, and also contained a number of other useful legal tools, including the ability to issue “Exclusion Orders,” which allowed the security services to detain and interrogate designated individuals entering the UK without (p.182) having to present detailed evidence supporting the detentions.40 Both of these effects put additional pressure on the PIRA. More positively for the group, the bombing made clear how bad it would be if the PIRA really began to engage in indiscriminate bombing, providing additional bargaining leverage in negotiations with the British. It was leverage that could be lost if the organization went too far with future attacks.41

These three factors combined to give Republicans favoring a truce the upper hand in internal deliberations. Secret talks between PIRA leaders and the British had been taking place since at least January 1974 and with the conditions created by the Birmingham bombing, these at last led to a temporary ceasefire in December 1974. Over the next two months the terms of a long-term truce were worked out. During the process, the PIRA briefly called off its ceasefire to conduct a series of six carefully calibrated bombings in London and Manchester. These bombings were designed to remind the British of just how bad things could get, without triggering a total political disaster all over again.42 The bombings apparently did the trick of wrestling some last concessions from the Northern Ireland Office of the British government, and on February 10, 1975, the PIRA announced a cessation of hostilities in Northern Ireland.43

The 1975 truce had its limits though. It did not apply to PIRA bombings in England. Nor did it apply to internecine fighting within the Republican movement or to Republican attacks on Loyalist civilians and paramilitaries. Much of the violence in 1975 was locally driven, a result of low-level tit-for-tat feuding between the PIRA and both the OIRA and the UVF.44 This violence was counterproductive for the PIRA: it increased the level of fear in Protestant communities, thereby enhancing support for the Loyalist paramilitaries;45 and it led to substantial increases in the British military presence, especially as the truce began to break down in late 1975.46

(p.183) As the truce wore on without significant political gains for the PIRA, recruiting began to suffer, morale dropped, and support for the ceasefire within the group withered.47 At the January 1976 Army Council meeting the PIRA called off the truce. In the wake of the failure, the PIRA leadership began a major reorganization intended to address three major issues: (1) the increasing level of security-force penetration of PIRA units due to the lengthy interrogations permitted under the 1974 PTA;48 (2) the desire of leaders like Ivor Bell and Gerry Adams to have greater control over military and political activities;49 and (3) the sense that frequent leadership turnover was hindering the organization.50 The reorganization plan was worked out by Ivor Bell while he was in prison, was approved by the PIRA Army Council in late 1976, and was largely implemented by Gerry Adams, who was appointed to the PIRA Army Council on his release from Long Kesh prison in February 1977.51 That the leadership could direct such a wholesale reorganization demonstrates that they had a great deal of control over operational procedures.

A more detailed look at the reorganization tells a very interesting story about the relationships among security, efficiency, and control over targeting. The reorganization plan had four key components, three of which were clearly intended to increase security. The first of these was to move to a cellular structure for military operations. The idea was to organize around small Active Service Units (ASU) of approximately four members, only one of whom was to have contact with the larger organization.52 The ASU were intended to specialize in particular types of operations—robberies, sniping, and so on—and were to have no control (p.184) over weapons or explosives, which were to be controlled by the Brigade commanders.53

In theory, such a structure should have increased both security—by reducing members’ knowledge of who was in the organization—and control, at least so long as local commanders did not provide weapons for operations the central leadership had not authorized, which for the most part they did not.54 In practice, the extent to which the cellular model achieved these goals is unclear. There was resistance to the changes in rural areas. In Newry, for example, members ostensibly in different cells continued to talk and discuss PIRA business, while drinking in some cases.55 Urban cells could not sever all ties with the old company structure because the companies were needed for logistical support; to hide guns, help people get cars, and the like.56 Even where the organization did compartmentalize, the desired increase in control did not happen, at least not for violence at the local level.57 Full compartmentalization, moreover, was never feasible because it would have made the group critically reliant on the few individuals who created connections between cells.58

The second security-oriented change was to create an internal security squad to track down infiltrators and punish those who provided information to government forces.59 The third security-oriented change was to require all Volunteers be trained in counter-interrogation techniques.60 Whether due to the move to a cellular structure, the introduction of an internal security service, or the requirements for counter-interrogation training, the reorganization successfully stemmed the tide of penetrations that had been so damaging.61

(p.185) The fourth element of the reorganization was the creation of a Northern command which stood between the Army Council in Dublin and the Northern units. Part of the motivation for this change was to achieve efficiency through functional specialization; Northern Command would fight the British while Southern Command, which included all units not under Northern Command, would focus on logistics. This change had the desired effect insofar as it allowed for the more rapid spread of operational technologies.62 It also increased control over some units as the leaders of the rural units in East Tyrone and South Armagh were brought into the Northern Command hierarchy.63

The reorganization’s overall impact on leaders’ ability to control operatives’ tactical choices is unclear. In general, the ASU had greater day-to-day autonomy, but access to weapons and technology were restricted and the central leadership could exercise control when they wanted to.64 Decisions about when and how to conduct high-profile operations such as major bombings or operations in England were still made at the top.65 On balance, the reorganization appears to have marginally increased leaders’ control over big operations but decreased control over responses to sectarian killings.

In some ways, the most interesting pattern from the 1976–1979 reorganization is that it led to a more secure, more efficient organization, without a huge drop in control. On the face of it, this appears to contradict the framework laid out in chapters 2 and 3. It is important to keep in mind, however, that the reorganization also entailed the introduction of new organizational technology. Just as new production technologies in industrial organizations can lead to greater output with fewer inputs, this organizational innovation led to more security without an obvious (p.186) decline in control.66 By 1979–1980 the PIRA had turned into a more stable organization, one that began looking to a longer struggle against the British, what those within the organization would term “the long war.”67 These changes point to the role of technical innovation and learning in determining how terrorists’ managerial dilemmas play out.68

Two additional important patterns can be discerned in the period from the end of the re-organization through late 1988 when the leadership finally recognized the military campaign that had stalled. The first is that the senior leadership maintained the control over fund raising and personnel which they had exercised in previous periods.69 The second pattern is that, as politics took on a larger role in the strategy of senior leaders, discrimination became more important to them, leading to efforts to exert more control. During the 1981 Hunger Strikes a number of the strikers ran successfully for parliament under the Sinn Féin banner. These victories marked the start of the “Armalite and Ballot Box” strategy, which emphasized the mutually reinforcing relationship between political and military action. Senior PIRA leaders first entered politics in a serious fashion in October of the following year when Gerry Adams and Martin McGuinness were chosen by Sinn Féin to contest assembly elections in Northern Ireland. Adams and McGuinness remained on the PIRA Army Council but were divorced from operational decisions.70 Their initial foray into politics was unsuccessful, but in June 1983, Sinn Féin won 13.4 percent of the vote in the Westminster General Election for Northern Ireland, becoming the fourth largest party in Northern Ireland.

This electoral victory was accompanied by a largely successful effort to rein in unsavory fund-raising activities such as robbery and extortion.71 (p.187) Yet control over violence remained elusive. Throughout 1983 and 1984, failed or poorly thought out operations continued to cost Sin Féin political support.72 The new-found prominence of normal politics within the PIRA was highlighted in 1985 when Ivor Bell and three fellow radicals were expelled from the organization for opposing the diversion of funds to Sinn Féin political activities.73 By mid-1985, PIRA leaders were seeing their political prospects fall and sought to limit attacks to military targets or to individuals who would be seen as legitimate high-visibility political targets.74 In spring 1986, Northern Command sought and received permission from the Army Council to vet most tactical operations in order to prevent electoral damage.75 Perhaps seeking to make it even easier for local units to know whom to target, the Army Council published a list of “legitimate targets” in August 1986.76

None of this worked. By 1987, the extent to which the PIRA was losing support, particularly because of the civilians who had been killed, brought into question the very feasibility of making political progress through war.77 In the February 1987 election in the Republic of Ireland, Sinn Féin polled only 1.9 percent of the vote, nicely summarizing its political fortunes. As the group’s political fortunes fell, and the need for discrimination became more apparent, Sinn Féin politicians’ criticisms of killing civilians became much more clear.78

The trend of counterproductive violence at a time when the leadership was increasingly focused on politics came to a head in November 1987 with the PIRA bombing at a Remembrance Day celebration in Enniskillen that killed eleven and injured sixty-three more.79 Coming toward the end of a year when the PIRA had already killed nine civilians in hard-to-justify circumstances, the bombing led to the collapse of support (p.188) for the PIRA in the Republic of Ireland.80 The PIRA responded by disbanding the unit responsible, and in January 1988, Adams publicly stated that the accidental killing of civilians would have to stop.81 By the fall of 1988, even the PIRA policy for how to respond to Protestant killings was changed in a way that reflected the increasing importance of discrimination to the leadership. Prior to that point the policy had been to respond quickly with an equally violent attack against Protestant civilians. After that, the policy was to only hit back if a named target could be shown to be involved in the killings. Moreover, local commanders had to vet their retaliation plans through the Northern Command leadership.82 In 1989, the leadership set a further example of how important discrimination had become by disbanding the IRA unit in Fermanagh because of its continued sectarian killings.83

7.4.3 Internal Politicking and the Move to Peace: 1990–1997

By 1990, the core leadership of the PIRA had realized that they were in a military stalemate and, by their accounts, opened backchannel communications with the British government.84 The PIRA couldn’t be destroyed, but neither could they inflict sufficient costs on the British and the Loyalists to get them to give in to a united Ireland.85 The outstanding feature of this period was that the PIRA continually struggled to find the right use of violence to meet a diverse set of concerns. First, the violence had to impose sufficient costs on Britain so that the British government would make some concessions. Targeting Britain’s economic infrastructure (p.189) was seen as the way to achieve this goal.86 Second, the violence had to be sufficiently controlled so that the group would not completely alienate other elements of the Nationalist movement. This meant concentrating on major targets in England while avoiding civilian casualties in Northern Ireland.87 Third, the violence had to be sufficiently close to what the rank and file wanted so that the leadership could maintain control of the organization.88 This meant maintaining a level of activity that was higher than was ideal for the first two goals.

By early 1993, the combination of increased control imposed on local operations and more aggressive action in England began to pay off. The April 1993 Bishopgate bombing in the City of London did tremendous damage to one of the engines of the British economy, leading to a brief suspension of secret talks by the British government, but also making clear the potential costs to Britain of maintaining the status quo.89 Talks soon resumed and the bombing coincided with a series of electoral gains by Sinn Féin. This coincidence was understood within the movement as validating the leadership’s Armalite and Ballot Box strategy.90

In December 1993, the British and Irish prime ministers issued the Downing Street Declaration, which offered the PIRA a chance to join multiparty negotiations if the violence ended.91 This led to a great deal of tension within the organization as the rank-and-file Volunteers were confused that the Army Council did not simply reject the declaration outright.92 In March 1994, the PIRA leadership allayed the Volunteers’ concerns with a series of mortar attacks against London’s Heathrow airport. These attacks solved the dilemma faced by the Adams leadership by reassuring the rank and file that the violent struggle would not be abandoned, while reminding the British what the PIRA could do if the peace process failed. Because the mortars were not primed to explode, the attack killed no one, avoiding the loss of life that would have hurt the group’s reputation with moderate Republicans. It is unlikely the mortars were left unprimed by mistake since the PIRA had conducted a series of very effective mortar attacks in the 1980s and the technology was well established.

(p.190) As negotiations moved on in the summer of 1994, the PIRA Army Council introduced the TUAS policy. The policy was never officially defined, but the grassroots militants were told TUAS stood for “Tactical Use of Armed Struggle.” Sinn Féin’s Nationalist political allies were told it stood for “Total Un-Armed Struggle.”93 This highly ambiguous policy illustrates the political dance the PIRA leaders faced as they tried to negotiate between their internal constituents and the larger Republican community. The ceasefire was accompanied by continued training and an increased focus on community policing by Volunteers—punishment beatings of criminals—as local commanders sought to keep militants from engaging in sectarian attacks or the kinds of criminality that cost the organization popular support in South Armagh during the ceasefire.94

In August 1994 the PIRA finally declared a ceasefire, one that was not popular with most of the organization. Over the next sixteen months the Army Council repeatedly told the Volunteers to prepare for new hostilities and then prolonged the ceasefire.95 By February 1995 the PIRA Executive Council, the internal body responsible for representing the interests of the average Volunteer, was becoming increasingly critical of the ceasefire and morale among the rank and file dropped throughout 1995.96 These members recognized that the longer the ceasefire went on, the more the network of supporters that made PIRA actions possible would degrade.97 In January 1996, facing a vote to end the ceasefire at the upcoming General Army Convention, the Army Council preemptively ended the truce.98 The need to maintain hearts and minds within the organization motivated violence that the rank and file wanted, but that leaders felt would be counterproductive.

Preparations began immediately for a major bombing in Britain. The Army Council decided on the Docklands area of London and assigned the bombing to the South Armagh IRA.99 The Docklands attack in February 1996 was the first in a five-month series of bombings that ended only with the July arrest of the PIRA ASU in England.100 The bombing campaign (p.191) did little to help the PIRA negotiating position, but had to be conducted for the PIRA to remain together as an organization.101 New teams were sent to Britain and the attacks resumed with a series of bombings and hoaxes in March and April 1997.

By mid-1997 the factions supporting an end to violence had sufficiently strengthened their political position within the PIRA to call a General Army Convention for October. When the September Sinn Féin convention endorsed the Mitchell Principles for settling the conflict, dissenting members of the PIRA Executive broke off to form the Real IRA (RIRA).102 At the October convention, the Army Council, which favored the ceasefire, rigged the transportation arrangements so that those dissenting from the peace process would arrive last. This gave the Council first crack at convincing the delegates, with the expected result: the Convention voted to accept the Mitchell Principles.103 After the vote, a number of more militant members, largely from rural areas, left to join the RIRA, which went on to conduct a number of bombings in 1998 and 2000.104

The PIRA ceasefire was almost completely maintained in the face of ongoing Loyalist violence in the late 1990s and first part of the new century. It’s impossible to know if this was because: (1) the changes of the late 1980s meant that leaders had sufficient control to prevent renewed sectarian violence; (2) Volunteers who were set on carrying out more violence could join splinter groups; or (3) the internal politics of the 1990s had persuaded the vast majority of Volunteers that an end to violence best served the Republican cause. For our purposes it is sufficient to know that October 1997 marked the end of the PIRA as an active terrorist organization.

7.5 The Loyalist Paramilitaries

As with the PIRA, the organizational history of the Loyalist paramilitaries can be divided into three distinct periods:

  • the early period from the formation of the UDA and UVF in the wake of the Civil Rights movement to the first PIRA ceasefire;

  • (p.192) the middle period, during which the UDA became mired in criminality while the UVF responded aggressively to increased PIRA activity; and

  • the final period, which saw a brief revitalization of the UDA as the PIRA moved to the 1994 ceasefire, before the UDA once again descended into criminality and internal feuding.

Throughout this history, a few now-familiar themes keep emerging. The most prominent of these is that the Loyalist leaders’ ability to control their organizations were limited when the rank and file did not share their preferences. Loyalist leaders limited what they asked of their organizations to a much greater extent than their Republican counterparts. Part of the difference in how much leaders demanded was that Loyalists’ goals simply did not require careful violence. Still, it is striking that when rank-and-file Loyalists wanted violence, they got it, often to the detriment of the movement’s political goals.105

7.5.1 Formation to PIRA Ceasefire: 1966–1975

From their very beginnings, the UVF and UDA followed different organizational paths. A small group of Loyalists calling themselves the UVF began killing Catholics years before the Civil Rights movement turned violent, committing a brief spate of attacks, in April and May of 1966. After the attacks the group’s founder, Gusty Spence, and the other core UVF leaders were arrested and the group essentially disappeared for eighteen months. It was rebuilt in late 1968 by a core group of militants from the Shankill Road area of Belfast.106 The increasingly violent demonstrations and counterdemonstrations that marked the latter stages of the Civil Rights movement provided an ideal recruiting backdrop for a group claiming to defend the Loyalist population. The nascent organization was able to pick and choose (p.193) its recruits from the many youth gangs involved in sectarian rioting.107 As the inter communal rioting grew worse over the summer of 1969, the UVF leadership organized militants to protect Orange Lodge marches and coordinated a series of sabotage attacks against public utilities.108 Despite its relatively centralized beginnings, the UVF did not develop its military-style hierarchy until the summer of 1972, when Spence escaped from prison and reorganized the group.109 By September 1972 the UVF had clear structures and a well-defined recruiting process in which the central leadership could veto undesirable recruits.110

While the UVF coalesced around a core group of militants, the UDA began as a coordinating body for the local defense associations that had grown up in response to the sectarian violence plaguing Northern Ireland from 1969 to 1971.111 After a series of coordinating meetings that began in the summer of 1970, the UDA was officially formed in August 1971 by the leaders of the Shankill Defense Association, Oldpark Defense Association, Woodvale Defense Association, and the East Belfast Defense Association.112 The process of developing some measure of control over these disparate militias was marked by frequent bloody feuds and continual tension over where control should reside and over the commitment of local organizations to the objectives of the central leadership.113 In the end, control over violence almost always resided with the local “brigadiers.”114

Interestingly, the variation in how the Loyalist groups were organized had a self-reinforcing quality through its influence on recruiting. During the early 1970s, the UDA and its component militias were in constant competition with the UVF for recruits and resources. The UVF pitched itself to potential militants as the “clean force,” the one not engaged in racketeering or wanton violence.115 While the UVF was more selective in its recruiting, it was no more organized early on than the UDA. A (p.194) UVF ‘company’ was often little more than a group of men who drank in the same bar.116 Whatever the reality of the situation, the image of greater coherence and discipline helped the UVF recruit young men more willing to follow orders and accept the restrictions necessary to operate covertly.117 This in turn eased internal control relative to the UDA.

Leaders of both Loyalist groups experienced similar problems of control that would plague the PIRA in later years. There were two reasons for this within the UDA. First, there was a notable unwillingness during this period to punish Loyalists conducting unsanctioned attacks on their own.118 Second, there was a real lack of consensus on what the political goals should be. Andy Tyrie became the UDA’s leader in late 1972 in part because of disagreements between the two most powerful “brigadiers,” Charles Harding Smith and Tommy Herron. Some accounts by UDA veterans suggest the core disagreement was about whether the goal of the group should be community defense or offensive attacks against Republicans.119 Others report that it was really about tactics, about whether the group should selectively target Republicans or simply seek to compel the Catholic community to stop supporting the PIRA.120 Either way, Tyrie was a compromise candidate who did not have a strong power base of his own and was never able to exert much control over any aspect of the group.121

Within the UVF, central control was constrained in part by the members’ ability to leave the group. When Spence took over again in July 1972, he sought to limit sectarian violence and focus the group on Republican activists. His efforts ran afoul of former UDA men who had joined the UVF out of dissatisfaction with the lack of action by the UDA. These men broke off to form a fringe group in August 1972.122 Pressure from below also limited the UVF leaderships’ options. By the summer of 1973, the rapid growth of the UDA led UVF operatives to push for more action, and the leadership simply could not say no.123

Some of their reticence to restrain their operatives’ sectarian aggression likely stemmed from the fact that the Loyalist leaders just did not have (p.195) strategic reasons to control their operatives during the early 1970s. In their own words, both the UVF and UDA sought to “terrorize” the nationalist (Catholic) community into demanding that the PIRA stop its attacks.124 As one UVF statement put it:

By bombing the heart of Provisional enclaves we attempted to terrorise the nationalist community into demanding the Provisinals either cease their campaign or move out of the ghetto area. By bombing business premises and other such properties which we had reason to believe were terrorist meeting places or sources of revenue, we believed we could force the Provisionals out of business or at least cause a drastic reduction in their operational activity.125

This goal was well served by random sectarian violence. To the extent the Loyalist leaders had to worry about controlling violence, it was to avoid alienating segments of the Protestant population whose support would be required for them to impact the political process.126

By late 1973 the UDA had settled into the pattern of quasi-racketeering that would mark it for the remainder of the conflict. Local leaders had divided up many of the Protestant areas of Belfast, and their extortion and protection rackets were well established.127 Moreover, the UDA itself was generating income for the senior leadership. UDA members during the 1970s all paid regular dues to the organization. In the early 1970s these dues were spent locally, but by the middle of the decade dues were collected at “company” meetings and passed through the UDA “Battalion HQ” on the Shankill Road. This created opportunities for corruption if the wrong people were handling the money. Unfortunately for the group, the UDA suffered a serious adverse selection problem according to one prominent UDA member.128 The problem was that UDA commanders were not paid. This meant that people with jobs were unwilling to give up their regular income to seek mid-level positions. These positions were thus filled by those who were unemployed or had criminal backgrounds.129 Quite predictably, these individuals took advantage of their positions to skim from the members’ contributions.

During the Ulster Workers’ Council (UWC) Strike in May 1974, the UVF and UDA worked well together to provide vital community services (p.196) and enforce the strike. Even though leaders of both groups wanted a ceasefire in the latter half of 1974, competition between them soon led to a steady increase in sectarian and internecine attacks.130 The UDA leadership could not have halted the killings even if doing so had been a priority. They simply did not have sufficient power, nor did they have the ability to use violence to punish deviations, over local units.131

The UVF faced similar problems of control, but for slightly different reasons. The key reason was that extreme sectarian views were continually being reinforced by the intense bonding that went on in the more covert UVF units,132 a process similar to that which plagued the combat organizations of early Russian groups. Because of this, UVF operatives generally wanted to implement more violence than their leaders. In late 1974, UVF leaders’ ability to restrain their operatives took a hit as the UVF-backed political party’s abysmal showing in the October 1974 elections destroyed the credibility of claims that restraint was the path to political influence.133

The UVF continued its sectarian attacks in 1975, despite the February PIRA ceasefire. By late 1975, however, the unpopular violence was taking an obvious toll on the UVF. Gruesome attacks, such as the execution-style killing of the popular Miami Showband and continued feuding with the UDA, began to strain relations with the community and reduced the quality of recruits coming in to the UVF.134 Moreover, as the violence increased, so too did the number of incarcerated operatives, creating a negative feedback loop. By late 1975 a large portion of UVF operators were busy with the robberies needed to raise funds to support the families of hundreds of prisoners. These robberies further damaged the group’s reputation.135 In October, a group of the longer-serving UVF men staged a coup intending to reduce sectarian violence and re-focus the group on attacking known Republicans.136 While the new leaders were able to control much of the group, they were either unwilling or unable to control their most extreme sectarian units.137

(p.197) 7.5.2 Ceasefire and Organizational Divergence: 1976–1987

The eleven-year period following the end of the first lengthy PIRA ceasefire would see the Loyalist organizations move in very different directions. But the divergence did not begin immediately. In the first two years after the PIRA ceasefire, four factors caused an across-the-board decline in Loyalist violence.138 First, as the security services became more aggressive against the PIRA, there was less demand for paramilitary activity in the Loyalist community. Second, PIRA violence dropped, further reducing community support for Loyalist paramilitary activity. Third, the security services became more aggressive against the UVF and UDA, penetrating both groups quite effectively.139 Finally, the UDA leadership became more focused on crime and its leaders didn’t want to give up the comfortable lifestyle this provided.140 From 1980 to 1985, Loyalist violence remained at its lowest levels since the formation of the UVF and UDA in the early 1970s.

In 1979 the UVF followed the PIRA example and shifted to a more cellular mode of organization. The core changes followed those made by the PIRA, shifting to small teams for operations, restricting communications between teams, and establishing an internal security arm.141 The change appears to have worked for UVF and the organization remained relatively stable over the next eight years. Major operations were centrally planned throughout the 1980s and were coherently connected to political events.142 UVF violence began to pick up only after the 1985 Anglo-Irish Agreement. By 1987 the UVF was quite effectively putting pressure on the PIRA by carefully targeting its operatives.143

For the UDA, the period after the ceasefire was marked by a steady slide into criminality.144 As the 1980s wore on, there was increasing displeasure (p.198) from the UDA membership and the Protestant community over UDA gangsterism.145 By the mid-1980s the UDA military capability essentially existed to serve the rackets by making it easy to threaten businessmen.146 As the PIRA began conducting bigger operations through the 1980s, the lower ranks of the UDA became increasingly frustrated with their own organization’s failure to respond in kind.147 Of course, some of the dissatisfaction may have been because little profit was passed down to the lower ranks.148

7.5.3 Revitalization of UDA and the Descent into Internal Feuding: 1988–2003

By 1988, internal resentment against the UDA leadership boiled over and Tyrie was deposed.149 The internal politics of the leadership change are quite revealing. There are essentially two versions of what happened. The first is that Tyrie was held accountable for the organization’s problems because he committed two sins: appointing unpopular “brigadiers” in violation of the UDA tradition of having local volunteers elect their leader; and failing to investigate his appointees when suspicions of corruption arose.150 The second version is that during the 1980s, Tyrie had managed to anger two major factions within the group, creating a large coalition favoring his removal.151 His efforts to centralize the organization angered the traditionalists who favored a good deal of independence for the brigades, while his efforts to reduce the sectarian killing angered the more (p.199) hardline members. Whichever explanation is accurate, both point to the centrality of internal politics in larger terrorist organizations and illustrate the limits imposed on their leaders by the need to maintain the allegiance of the membership.

This example raises a more general point, which deserves to be highlighted again; terrorist leaders are often constrained by their operatives’ preferences in subtle ways. In the models in chapter 5, this happened because leaders’ threats were outweighed by the value to operatives of doing what they wanted. In Tyrie’s case, it happened because his move to centralized authority threatened valued autonomy. Tyrie’s expulsion was followed by a purge of his allies.152 This coincided with a series of arrests that would change the character of the organization’s leadership. In September 1989, the Stevens Enquiry into collusion between the UDA and the Northern Ireland security services began. The Enquiry led to the arrest of many of the corrupt older leadership, especially those who were working with security services.153 These arrests had the twin effect of easing security pressures and clearing the way for a younger, more aggressive leadership.154 Just as the UVF’s reputation had helped it recruit in the 1970s, word of the new UDA leadership’s commitment to violence helped UDA recruiting in the late 1980s as well.155

In late 1991 the UDA began conducting more attacks. This led to a virtuous cycle: more operations led to more prestige, which meant better recruits and more disciplined violence, which further increased the group’s status.156 While the new UDA leadership was indeed more aggressive, this increase in violence was not driven from the top. UDA violence increased despite an agreement between the UDA and UVF to limit “offensive activity.”157

As the PIRA internal politics ground slowly on between 1992 and 1994, UVF and UDA violence continued apace. Major operations like the October 1993 Greysteel attack were centrally planned and coordinated.158 (p.200) But, central planning was the exception. Most Loyalist violence during this period, especially UDA violence, was generated at the local level, only sometimes in response to specific PIRA activities.159 By the summer of 1994, the Loyalists’ paramilitaries had tamped down their internal feuding and were putting effective pressure on the PIRA.160

In October 1994, one month after the PIRA announced their next-to-last ceasefire, the UVF responded in kind. The UDA joined in one month later. It was easier for the UVF, with its centralized structure and restricted membership, to make the decision about whether or not to follow the PIRA initiative.161 The UDA leadership had to convince a much larger population of militants who were not used to having policy decided at the top.162 Following the truce, both the UDA and UVF had problems finding peaceful roles that would satisfy their members.163 And, by some accounts, the senior UVF political leaders had an easier time controlling their organization because their political representatives in the PUP, men like David Ervine, were seen as credible hardliners.164

Following the end of the PIRA ceasefire in February 1996 there was a great deal of dissension among both Loyalist groups about how to respond, providing a useful illustration of how leaders’ ability to exercise control is constrained by their operatives’ preferences. The disagreements degenerated into violent feuding within the UDA.165 While there was no internal violence within the UVF, unsanctioned attacks against Catholics by the Portadown UVF in July 1996 led to a split in the organization.166 The UVF tried to force their leader in Portadown, Billy Wright, into exile. Wright refused, as his local power base was sufficiently strong that the UVF leadership could not effectively threaten him.167 Instead, Wright took his local supporters and formed the Loyalist Volunteer Force (LVF) which quickly began to engage in more sectarian killings and was to be the most active Loyalist group against Catholics for the remainder of the (p.201) conflict.168 Neither the UVF nor the UDA officially ended the ceasefires they had started in the fall of 1994, but Loyalist violence began to rise, mostly conducted by the LVF.

As 1997 progressed with few Republican attacks in the North, the Loyalist paramilitaries increasingly focused their energies on each other. Throughout 1997, conflicts between the UVF and UDA continued, driven by a wide variety of local grievances.169 In mid-January 1998, the UDA briefly broke its sectarian ceasefire and its political party, the Ulster Democratic Party (UDP), was expelled from the multi party talks.170 At this point the conflict was obviously going to be resolved politically and the costs of not being at the table were clear. By late January UDA leaders convinced their militants to reinstate the ceasefire so the UDP could rejoin the multiparty talks.171

After January 1998, Loyalist violence was almost entirely confined to internecine feuding, much of it revolving around control of drug territory and related revenues.172 As in previous periods, UDA Brigade commanders had good control over their local areas, but the political leaders held little sway.173 In fact, UDA-UVF violence reached its worst in the latter half of 2000 because of this independence. One UDA company commander in the Shankill area, Johnny Adair, decided to try to clear the UVF from his area, against the wishes of both his superiors and his fellow company commanders in the West Belfast Brigade.174 UVF leaders did not have such problems, in part because the core UVF leadership were able to violently punish their own. For example, the 2000 Shankill feud ended in part because the UVF shot one of their own operatives as punishment for his role in attacking Adair.175 By 2002, Loyalist feuding was almost entirely confined to fights within the UDA. The last of these was settled in February 2003 with an oddly formal surrender of the main renegade UDA unit.176

(p.202) 7.6 Conclusion

This brief study of the major terrorist organizations that fought in Northern Ireland provides strong support for three of the four hypotheses set forth in the beginning of this chapter. Table 7.2 summarizes the case study findings for each of the six areas in which we can find evidence of central control. Where there was no solid evidence, the table reports an area as not being under central control.

Table 7.2. Control in Northern Ireland: 1969–2001




Target choices




Operational procedures




Personnel assignments
















(1) The level of control exercised by PIRA leaders increased dramatically over time. By the late 1980s, the leadership effectively modulated violence to support negotiations.

(2) The level of control exercised by UVF leaders increased over time. Target choices for major operations were centrally controlled by the late 1980s.

Turning to our hypotheses, we see that hypothesis 1 is clearly supported. Throughout the conflict, the PIRA leadership exercised a greater degree of control over their organization than did the leaders of either Loyalist paramilitary. Operational procedures and personnel assignments were controlled by the PIRA Army Council for most of the conflict. On occasion, local units flouted orders about operational procedures. For example, in May 1971 a very effective four-man ASU in Newry disbanded rather than follow orders to operate with men from Belfast who were known to the security services.177 Such isolated incidents pale in comparison to the wholesale restructuring of the organization in the late 1970s and the control over promotions during the personnel changes of the 1980s.

(p.203) Hypothesis 2 is similarly supported. While the PIRA Army Council never had perfect control over target choices, they were able to direct surges and lulls in attacks to support negotiations in the mid-1970s and in the 1990s. Control over smaller-scale activities also increased over time, not the least due to the development of an internal disciplinary arm in the late 1970s. Importantly, control over tactical choices in the PIRA was increasing over time for exactly the reason embedded in hypothesis 2: the political damage from indiscriminate attacks began to matter more to the group’s leadership.178 It was not just control over attacks that changed over time. Part of the late 1970s reorganization involved bringing such resources as guns and explosives under the control of local quartermasters who reported directly to either Northern or Southern Command.179

The histories of all three groups provide support for hypothesis 4. Leaders in the PIRA, UVF, and UDA were restricted in what they could ask of their organizations. Moreover, we saw direct evidence of the causal process embedded in hypothesis 4. Again and again leaders refrained from making certain demands exactly because they knew they could not enforce certain orders.180 Indeed, one explanation for the prominence of internal politics in the secondary literature on the Northern Ireland groups may be that historians’ main primary sources—former militant leaders like Adams, McGuinness, Spence, and Ervine—spent a great deal of their time working to change their operatives’ preferences to more closely mirror their own.

The evidence in support of hypothesis 3 is somewhat more ambiguous, owing to the paucity of detailed information on the internal organization of the two Loyalist paramilitaries.181 The broad patterns of the conflict certainly support the notion that the UVF leadership exercised greater control. While UVF militants were carefully targeting PIRA activists in the second half of the 1980s, the UDA was still busy with sectarian attacks and internal feuding. Moreover, the UVF leadership clearly controlled (p.204) recruiting early on and successfully reorganized along more cellular lines shortly after the PIRA did. Later in the conflict, UVF leaders discussed the links among political goals, discrimination, and the need to control violence in exactly the terms we would expect given all we have seen so far. They understood the political costs of indiscriminate violence and sought to control it.182 Control was exercised mainly through persuasion, but the UVF leaders did use sanctioning when they had the power to do so. While the UDA leadership recognized the same constraints, they made fewer efforts to exert control because discrimination was simply less important to them.183

Overall, the history of how terrorist groups in Northern Ireland were managed is broadly consistent with the argument that terrorists face a consistent set of organizational dilemmas, regardless of what they are fighting over. Importantly, the Northern Ireland case highlights that both pro- and anti-government terrorist groups face similar challenges. While pro-government groups face less security pressure, their leaders’ ability to manage violence is constrained when operatives wield a credible threat against them, just as that of leaders in anti-government groups. Still, one may reasonably object that the first two case study chapters have focused on European groups, and that the managerial dilemmas faced by groups in other settings (e.g., the Middle East), or by groups that have an explicitly religious motivation, may be fundamentally different. To address that concern we turn next to a comparative analysis of secular and religious groups in the Israeli-Palestinian conflict. As we will see, both organizations face similar dilemmas, or at least dilemmas that differ as we should expect them to given differences in preference divergence between the groups.


(1) This dynamic is common to conflicts that see militants fighting both for and against the state and means pro-state terrorist organizations tend to be less effective than anti-state ones. For a good discussion of the differences between pro- and anti-state terrorists, see Steve Bruce, “The State and Pro-state Terrorism in Ireland,” in The State: Historical and Political Dimensions, ed. Richard English and Charles Townshend (London: Routledge, 1999), 184–209.

(2) The PIRA also claimed their attacks served a deterrent purpose against Loyalist violence, but this purpose was not as central in the group’s statements as it was on the Loyalist side.

(3) Sarah Nelson, Ulster’s Uncertain Defenders: Protestant Political, Paramilitary, and Community Groups in Northern Ireland (Belfast: Apple Tree Press, 1984), 120–121.

(4) There are a variety of ways in which creating uncertainty about the nature of the action that will follow an attack can enhance deterrence. For examples see Thomas C. Schelling, The Strategy of Conflict (Cambridge, MA: Harvard University Press, 1960), 183–203. In the Northern Ireland conflict, attacks intended to deter usually led to cycles of escalation. This dynamic was quite apparent, for example, in the violence leading up to and following the 1993 Shankill Road bombing. See Richard English, Armed Struggle: The History of the IRA (New York: Oxford University Press, 2003), 282–283.

(5) On the closeness between the UVF and PUP, see: Ian S. Wood, Crimes of Loyalty: A History of the UDA (Edinburgh: Edinburgh University Press, 2006), 179; Roy Garland, Gusty Spence (Belfast: Blackstaff Press, 2001), 265–267; and author interview with UVF leader David Ervine, March 8, 2006.

(7) Neither paramilitary-linked party has approached the success of the more mainstream parties, the Ulster Unionist Party and the Democratic Unionist Party.

(8) Northern Ireland Statistics and Research Agency, <http://www.nisra.gov.uk/Census/>, accessed July 17, 2012.

(9) For a thorough discussion of this trend see Luis de la Calle and Ignacio Sánchez-Cuenca, “The Production of Terrorist Violence: Analyzing Target Selection in the IRA and ETA,” (Manuscript, Juan March Institute 2007). The authors distinguish between “control” attacks, which must be selective and “attrition” attacks that can be selective or indiscriminate. From 1970 to 1989, only 7.7 percent of PIRA violence involved control attacks while 25.4 percent did so from 1990 to 2004. De la Calle and Sánchez-Cuenca make a different argument than I do about why the PIRA became more selective over time. They argue that terrorists use indiscriminate violence if their supporters favor it and they have sufficient resources to pursue an “attrition strategy” against government forces. Their “attrition strategy” is contrasted with a “control strategy” which involves using discriminate violence to compel the civilian population to provide support to the group. As the troubles in Northern Ireland wore on, the Catholic population’s support for sectarian violence dropped and British security forces learned more about the PIRA. The latter meant the PIRA had to devote ever more resources to security, reducing what it could use on producing violence. In the authors’ theory, both changes push the group to use violence more selectively.

(10) Figures based on CAIN organizational abstracts and U.S., Patterns of Global Terrorism 2001 (Department of State, 2002), 130.

(12) UDA killings count actions by the UDA and the UFF.

(14) For the best overview of the PIRA’s organizational structure, see Horgan and Taylor, “The Provisional Irish Republican Army: Command and Functional Structure.”

(15) The CAIN site contains basic background information on many aspects of the “Troubles,” detailed chronologies of key periods, links to scholarly articles, bibliographies on a variety of topics, and the Sutton database of killings during the conflict. See <http://cain.ulst.ac.uk/index.html>.

(16) Global Terrorism Database, accessed July 17, 2012.

(18) The slow degradation of relations between the Catholic community and the British Army is well described in Bishop, Patrick, and Eamonn Mallie, The Provisional IRA (London: William Heinemann, 1987), 54–137. On the bottom-up pressure within the PIRA to begin attacking British forces, see Peter Taylor, Provos: The IRA and Sinn Féin (Bloomsbury, 1997), 88–89; and Bishop and Mallie, The Provisional IRA, 137.

(19) English, Armed Struggle: The History of the IRA emphasizes that there is a great deal of continuity between the old IRA and the PIRA which emerged in late 1969. The split in the Republican movement was, in many ways, about the development of parallel leadership groups, one of which won the political struggle for the allegiance of local units.

(20) The failure to defend the civil rights marchers played a lesser role. See Bishop and Mallie, The Provisional IRA, 68–69, 89–90.

(21) English, Armed Struggle: The History of the IRA, 105–107. Robert W. White, Provisional Irish Republicans (Westport, CT: Greenwood Press, 1993), 52–55.

(22) Taylor, Provos: The IRA and Sinn Féin, 60. J. Bowyer Bell, The Secret Army: The IRA 3rd ed. (New Brunswick, NJ: Transaction Publishers, 1997), 367–368.

(23) Ed Moloney, A Secret History of the IRA (Westport, CT: Greenwood Press, 2002), 87.

(24) Ibid., 91.

(25) Bell, The Secret Army: The IRA, 406–407. Toby Harnden, ‘Bandit Country’: The IRA and South Armagh (London: Hodder & Stoughton, 1999), 63.

(27) Ibid., 154.

(29) Gary McGladdery, The Provisional IRA in England: The Bombing Campaign 1973–1997 (Dublin: Irish Academic Press, 2006), 58–60.

(31) For a wonderful description of the meeting where what would become known as the “Balcombe Street Gang” was chosen, see Moloney, A Secret History of the IRA, 125. Essentially the PIRA sought volunteers unknown to the security services, but the only people willing to risk the treason charges faced by bombers operating in England were those who were already on the run or who had been interned.

(32) Taylor, Provos: The IRA and Sinn Féin, 155. Tim Pat Coogan, I.R.A. 5th ed. (London: HarperCollins, 2000), 388.

(33) Ibid., 398–399.

(38) This incident is noted in several sources but is best described in Sean O’Callaghan, The Informer: The Real Story of One Man’s War Against Terrorism (London: Bantam Press, 1998), 45–46.

(39) The reaction was so negative that the PIRA’s competitor—the Official IRA (OIRA), which was made up of individuals who had remained loyal to the traditional southern leadership during the 1969 split—felt compelled to announce it had adopted an arms-for-defense-only policy. See Bell, The Secret Army: The IRA, 413–4. See also McGladdery, The Provisional IRA in England: The Bombing Campaign 1973–1997, 99.

(40) Ibid., 94–95, 97–98.

(41) Ibid., 90–92.

(42) In this case, “calibrating” the attacks meant using modest amounts of explosives and offering adequate warning for the police to evacuate the targeted area. See Bell, The Secret Army: The IRA, 419–420.

(43) For an excellent account of internal Republican politics between the initial December 1974 ceasefire and the February 1975 truce, see Taylor, Provos: The IRA and Sinn Féin, 176–186.

(44) The South Armagh PIRA in particular kept up attacks during the ceasefire. See Harnden, ‘Bandit Country’: The IRA and South Armagh, 20. See also Bishop and Mallie, The Provisional IRA, 256.

(46) For example, a particularly gruesome round of sectarian violence in South Armagh beginning in November 1975 led the British to send in Special Air Service (SAS) units to target PIRA operatives for the first time. See English, Armed Struggle: The History of the IRA, 171–172. SAS ambushes of PIRA volunteers in South Armagh were to take a heavy toll on the organization in the years ahead.

(48) Because the PTA allowed the British to interrogate subjects for up to a week before charging them, the proportion of arrestees who broke and shared information greatly increased. Horgan and Taylor, “The Provisional Irish Republican Army: Command and Functional Structure,” 21–23, Bishop and Mallie, The Provisional IRA, 255.

(51) Bishop and Mallie, The Provisional IRA, 275; Moloney, A Secret History of the IRA, 157–160. Moloney (164) provides a fascinating description of the machinations Seamus Twomey went through to get Adams appointed to the Army Council without violating the PIRA constitution.

(53) The 1976 ‘Staff Report’ detailing the reorganization was captured in December 1976 and is reproduced in Coogan, I.R.A., 465–467.

(54) In December 1978, for example, the operations proposed by the Newry IRA reportedly had to be vetted up the chain of command before weapons would be issued. See Eamon Collins, Killing Rage (London: Granta Books, 1997), 14–15.

(55) For a typical description of how compartmentalization worked, and often didn’t work, in Newry, see ibid., 67.

(58) For a nice analysis of the management of urban terrorism which highlights this point and others, see John B. Wolf, “Organization and Management Practices of Urban Terrorist Groups,” TERRORISM: An International Journal 1, no. 2 (1978): 169–186.

(59) Killings of suspected informers rose sharply after the introduction of the internal security department. See Moloney, A Secret History of the IRA, 155–156.

(60) The July 1979 killing of PIRA member Michael Kearny by the internal security unit was widely interpreted within the organization as a warning to others who might break during interrogation, as Kearny had. See ibid., 155.

(62) For an example of how Northern Command was able to spread innovations developed in South Armagh to other areas through secure channels rather than ad hoc coordination at the local level, see Moloney, A Secret History of the IRA, 160–161.

(63) Ibid., 157–161. Northern Command did not always follow Army Council directions, making several unauthorized attacks in February 1977, apparently without being punished for it. See Bishop and Mallie, The Provisional IRA, 259.

(64) For example, when the Army Council decided it wanted to conduct a symbolic punishment attack for the Loyalist paramilitaries’ behavior during the 1981 hunger strikes, it took proposals from the Brigades for targets and chose the one proposed by the Belfast Brigade. See Collins, Killing Rage, 94–95, 98; and Bishop and Mallie, The Provisional IRA, 309.

(65) O’Callaghan, The Informer: The Real Story of One Man’s War Against Terrorism, 141–144. There were limits to the central leadership’s control. Orders to avoid operating near Adams’ holiday cottage were ignored and unauthorized attacks against the Royal Ulster Constabulary (RUC) did occur. See Collins, Killing Rage, 154; and Moloney, A Secret History of the IRA, 243–244.

(66) Such innovations seem relatively rare in that few groups adapt in ways that simultaneously improve both security and control.

(68) Organizational adaptation in covert groups is studied in depth in Michael Kenney, From Pablo to Osama: Trafficking and Terrorist Networks, Government Bureaucracies, and Competitive Adaptation (University Park: Pennsylvania State University Press, 2008).

(69) On financing see O’Callaghan, The Informer: The Real Story of One Man’s War Against Terrorism, 118–120. He describes how a cell in Galway restricted their use of armed robberies to avoid violating the preferences of the Army Council and Chief of Staff but were later ordered to increase their use of armed robbery. On the personnel side, see the discussion of the assignment of personnel to bombings in England in ibid., 149–150. On changes made following disciplinary actions, see Collins, Killing Rage, 165, 216–217. On the choice of personnel to run a major weapons smuggling operation see Moloney, A Secret History of the IRA, 17–18.

(71) Sabine Wichert, Northern Ireland Since 1945, 2nd ed. (New York: Longman, 1999), 188–189. Indeed, political considerations are the reason the PIRA was never keen on straightforward protection rackets like the UDA. See Bishop and Mallie, The Provisional IRA, 312–313.

(79) Enniskillen is frequently cited by PIRA members as an incredibly damaging event that fundamentally changed public perceptions of the group. Richard English, private communication, August 7, 2007.

(80) Taylor, Provos: The IRA and Sinn Féin, 278. On deaths, see Sutton database. Many PIRA members cite the Enniskillen bombing as an exceptionally damaging event for the group. Richard English, private communication, August 7, 2007.

(84) The British government places the date of renewed contacts in 1993. See Eamonn Mallie and David McKittrick, The Fight for Peace: The Secret Story Behind the Irish Peace Process (London: Heinemann, 1996), 104–109. Mallie and McKittrick provide the most detailed history of the peace process from 1987 to 1994.

(85) English, Armed Struggle: The History of the IRA, 307–308. Mallie and McKittrick put the date a bit later, in 1992. See Mallie & McKittrick, The Fight for Peace: The Secret Story Behind the Irish Peace Process, 315. Bell puts the date of this realization slightly later still, in 1993. See Bell, The Secret Army: The IRA, 634.

(87) Ibid., 160.

(88) Ibid., 150 cites John Major making exactly this point in his memoirs.

(89) Ibid., 172–173.

(92) Moloney, A Secret History of the IRA, 415, 431. Indeed, a PIRA mini-convention in February 1994 came out against any possible ceasefire.

(93) See ibid., 423. The document describing the TUAS policy is available at http://cain.ulst.ac.uk/othelem/organ/ira/tuas94.htm.

(96) Ibid., 439.

(105) Throughout the conflict in Northern Ireland, the Loyalist paramilitaries were much less careful in their use of violence than the PIRA. One of the very first attacks by the newly formed UVF, in May 1966, nicely illustrates the almost-casual targeting process that was to mark both UVF and UDA attacks. On May 26, 1966, a group of UVF men set out to try to kill a prominent IRA operative, Leo Martin. Martin was not at home when the UVF men arrived, so they set fire to his home and proceeded to a bar for a drink. At the bar, the UVF team ran into four Catholic barmen, decided on no real evidence that the barmen were in the IRA, followed them outside, and killed them. See Peter Taylor, The Loyalists (London: Bloomsbury, 1999), 34; David Boulton, The UVF: 1966–73: An Anatomy of Loyalist Rebellion (Dublin: Gill & McMillan, 1973), 49–50; and Jim Cusack and Henry McDonald, The UVF (Dublin: Poolbeg, 1997), 12–13.

(111) Steve Bruce, “Loyalists in Northern Ireland: Further Thoughts on ‘Pro-State Terror,’” Terrorism and Political Violence 5, no. 4 (1993) 50. Taylor, The Loyalists, 83.

(112) Steve Bruce, “Terrorists and Politics: The Case of Northern Ireland’s Loyalist Paramilitaries,” Terrorism and Political Violence 13, no. 2 (2001), 31.

(114) Henry McDonald and Jim Cusack, UDA: Inside the Heart of Loyalist Terror (Dublin: Penguin Ireland, 2004), 32.

(116) Steve Bruce, The Red Hand: Protestant Paramilitaries in Northern Ireland (Oxford: Oxford University Press, 1992), 109, 112–114.

(118) Ibid., 21.

(120) Ibid., 27. Crawford is reporting different points of view expressed by UDA members during his interviews with them.

(124) Indiscriminate retaliation for specific attacks was part of this strategy. See Taylor, The Loyalists, 125–126.

(128) Sam Duddy described an adverse selection problem, he did not use that terminology.

(133) Ibid., 187–189.

(134) Ibid., 189–191.

(137) Specifically, the gang that would become known as the Shankill Butchers. See Taylor, The Loyalists, 153, 155.

(139) Jim Cusack and Max Taylor, “Resurgence of a Terrorist Organisation—Part 1: The UDA, A Case Study,” Terrorism and Political Violence 5, no. 3 (1993), 3.

(141) These changes were first discussed by UVF prisoners in 1977, but the details of what this change meant for the UVF in practice are not as well known as for the PIRA. See Garland, Gusty Spence, 218–219; Cusack and McDonald, The UVF, 193–194.

(142) See for example the planning to target civil servants implementing the 1985 Anglo-Irish accords. See ibid., 238.

(145) The gangsterism is exemplified by the UDA building rackets described in Taylor, The Loyalists, 169–171.

(148) For an excellent overview of Loyalist fund-raising, see Silke, “In Defense of the Realm: Financing Loyalist Terrorism in Northern Ireland—Part One: Extortion and Blackmail,” 334–335. Interestingly, the dissatisfaction in the UDA over the distribution of profits stands in distinct contrast to the attitude of South Armagh PIRA Volunteers who apparently felt little resentment over the economic privileges enjoyed by their senior leaders. Perhaps the rank and file’s tolerance for the South Armagh leaders’ lifestyles had to do with their obvious success in prosecuting the conflict with the British.

(156) Ibid., 120–121, interview with “Jackie.”

(159) For example, a long series of tit-for-tat attacks followed the PIRA attempt to kill Johnny Adair, a UDA company commander who would later cause major rifts in the organization. See Taylor, The Loyalists, 223–225.

(163) Steve Bruce, “Turf War and Peace: Loyalist Paramilitaries Since 1994,” Terrorism and Political Violence 16, no. 3 (2004) 505.

(167) Ibid., 347–348.

(168) Steve Bruce, “Turf War and Peace: Loyalist Paramilitaries Since 1994,” Terrorism and Political Violence 16, no. 3 (2004), 511.

(171) See ibid., 220–1. See also the CAIN chronology at http://cain.ulst.ac.uk/othelem/chron.htm.

(172) Neil Jarman, “From War to Peace? Changing Patterns of Violence in Northern Ireland, 1990–2003,” Terrorism and Political Violence 16, no. 3 (2004), 427.

(175) Ibid., 370–371.

(177) Collins, Killing Rage, 81–82; Kevin Fulton, Unsung Hero (London: John Blake Publishing, 2006), 32.

(178) In terms of the model presented in chapter 5, b, was increasing over time as the PIRA leadership became more invested in the political process and as the Catholic population became less willing to tolerate sectarian killings.

(179) At least they were supposed to. That they did even some of the time meant there was an increase in central control over resources compared to earlier periods.

(180) This is analogous to the situation in the models in chapter 5 in which the leader refrains from communicating because he knows it won’t do any good and the agent will just do what he wants as the motivational constraint is not met. Recall that this occurs when the additional utility the agent gets from taking preferred action instead of doing what the leader wants is greater than the costs the leader can impose.

(181) Indeed, the fact that even those who have done extensive interviews with UDA militants lack a clear picture of the organization is in itself evidence in support of hypothesis 1, in so far as a coherent organization is a necessary condition for leaders to control agents with heterogeneous preferences.

(182) UVF leader Gusty Spence even reached out to the PIRA leadership in prison in 1976 with the suggestion that both groups’ political goals would be better served if they could tamp down the sectarian killings. See Garland, Gusty Spence, 200–202.

(183) Recall that discrimination was less important to the UDA both because the group lacked a successful political affiliate and because for much of the conflict the UDA leaders were more interested in racketeering than politics.