4352 002 The-Roots-Of-Restraint WEB

THE ROOTS
OF RESTRAINT
IN WAR
REFERENCE
This report is based on research carried out by independent researchers commissioned
by the International Committee of the Red Cross (ICRC) to advance the organization’s
understanding of the workings of armed forces and armed groups. It does not include
information that was obtained in a confidential manner during ICRC operational
activities.
The icons used in Chapter 6 of this publication have been taken from the OCHA Icons Library, adapted for use by the ICRC, and icons
made by Freepik from www.flaticon.com
THE ROOTS
OF RESTRAINT
IN WAR
THE ROOTS OF RESTRAINT IN WAR 3
ACKNOWLEDGEMENTS
This publication was written by Fiona Terry and Brian McQuinn based on empirical
research led by Andrew Bell, Francisco Gutiérrez Sanín, Yvan Guichaoua, Ferdaous
Bouhlel, Oliver Kaplan and Naomi Pendle. Benjamin Eckstein, Brian McQuinn and Fiona
Terry coordinated the field research.
The project was overseen by a steering committee at the ICRC chaired by Helen Durham,
ICRC director of law and policy. The committee comprised: Knut Dörmann, Michael
Dynes, Luigi Fratini, Pierre Gentile, Irénée Herbet, Dorothea Krimitsas and Hugo Slim.
Research protocols were approved by an ethical review board composed of Gilles

Carbonnier, Claudia Seymour and Hugo Slim. Ethical clearance for the survey of
Australian soldiers was provided by the Australian Department of Defence and
Veterans’ Affairs Human Research Ethics Committee. Special thanks go to Lieutenant
General Angus Campbell for approving the research.
We are grateful to the following people for their help, input and inspiration:
Elisabeth Jean Wood, Helen Kinsella, Eva Svoboda, Lindsey Cameron, Olivier Bangerter,
Tina Bouffet, Sarah Grey, Sarah Gale, Kevin Meister, Kars Aznavour, Erica Potts, Paul
Baker, Pete Evans, Frederico Almendra, Ken Hume, Evecar Cruz-Ferrer, Pascal Porchet,
Leonard Blazeby, Stephanie Riddell, Natalya Wells, Christoph Luedi, Christoph
Harnisch, François Stamm, Béatrice Oechsli, Dorsa Nazemi-Salman, Ahmed
Al-Dawoody, Jean-Nicolas Marti, Abbas Daiyar, Ihcène Kiamouche, Nicole Van Rooi-
jen, AZM, Sarah Roxas, Nicole Martins-Maag, Vincent Bernard, Thomas Saint-Maurice,
Antoine Grand, Christina Grisewood, Cordula Droege and all the others who supported
the project along the way.
In addition, we would like to thank the UK Department for International Development

for its contribution to funding this study.
4 THE ROOTS OF RESTRAINT IN WAR
A. Hofford/EPA
CONTENTS
Foreword................................................................................................................. 6
Executive summary................................................................................................. 8
Introduction............................................................................................................11
Chapter 1: Norms of restraint, organizational structure and socialization.............. 17
Chapter 2: Integrated State armed forces................................................................ 27
Chapter 3: Centralized non-State armed groups..................................................... 37
Chapter 4: Decentralized non-State armed groups ................................................45
Chapter 5: Community-embedded armed groups................................................... 53
Chapter 6: Implications and approaches.................................................................63
Further reading...................................................................................................... 72
FOREWORD
It is a pleasure to introduce this new study on restraint in war. It is the fruit of an interdis-
ciplinary, cross-sector partnership between the International Committee of the Red Cross
(ICRC) and distinguished scholars from around the world. For us, it marks an important
step forward in understanding the sources of influence on soldiers and fighters when it
comes to respect for the principles and norms of international humanitarian law (IHL).
As the title suggests, the study aims to better identify the roots of restraint – the factors
that induce weapon bearers across the spectrum to observe certain limits when
engaging in armed violence and to preserve a minimum of humanity even in the heat
of battle. The researchers’ insights into the culture and practices of two State mili
taries and several types of non-State armed group reveal various political, ethical and
socio-economic reasons why different parties to conflict behave as they do. The key
is socialization – the process by which norms and rules become socially accepted and
then fulfilled on the battlefield.
The ICRC has a long history of working with State armed forces and non-State armed
groups in a constant effort to foster respect for the rules of war at the strategic, oper
ational and tactical levels. To this end, we work with senior commanders, policy-makers
and front-line fighters across the world to promote the integration of humanitarian
norms and IHL into their training and decision-making.
Accordingly, the ICRC’s approach has traditionally relied on established systems of

command and control, so that IHL is valued and respected throughout the chain of
command, and every fighting unit has a basic knowledge of the legal norms. This goes
some way towards socializing those norms, but not the whole way. The study shows
that we can do better by understanding how a culture of restraint is socialized, not
only formally and vertically, from the top down, but also informally and horizontally.
It opens up new avenues for ensuring that the basic principles of IHL are embedded in
the DNA of all members of armed forces and armed groups.
Today’s conflicts are characterized by a plethora of armed actors with differing goals
and ideologies. There is also a growing tendency for conflicts to be fought in coalitions,
with a number of States joining forces or State militaries partnering with non-State
armed groups to achieve a given, shared purpose. This makes it all the more import‑
ant that we gain a better idea of the many and varied ways in which these forces
inculcate respect for humanitarian norms, and thus restraint, in their members, and
what external influences there may be. By improving our understanding of these pro-
cesses and influences, we can work more effectively with all parties to armed conflict
to ensure civilians, detainees, wounded people and others protected by the rules of war
are treated humanely in accordance with IHL.
I thank everyone involved in the research, coordination and funding of the various
studies. I commend their important findings to all those who are committed to bringing
about greater restraint in the exercise of armed violence. The novel insights from this
research are of direct relevance to the ICRC and military authorities. They will help us
further strengthen the “roots of restraint” and thus, we hope, make a greater impact
on the conduct of many of today’s protracted armed conflicts.
Professor Gilles Carbonnier

Vice-President
International Committee of the Red Cross
EXECUTIVE SUMMARY
As the reference organization on international humanitarian law (IHL), the Inter
national Committee of the Red Cross (ICRC) seeks to ensure that the rules and norms
aimed at restraining the destructive forces of armed conflict are known and respected by
soldiers and fighters around the world. This report is a contribution to that endeavour.
The report, based on two years of research by a group of distinguished scholars, sets
out to identify the various sources of influence on the behaviour of those bearing arms
in different types of armed forces and armed groups. To date, the bulk of the ICRC’s
work in this domain has centred on State armed forces and on ensuring that IHL is
incorporated into their doctrine and directives, into the regular training of soldiers and
into the disciplinary mechanisms designed to enforce compliance with the rules. As
such, it has focused predominantly on the formal norms prescribed by IHL.
The ICRC has also engaged with many non-State armed groups, encouraging them to
adopt codes of conduct to align the behaviour of their fighters with the norms of IHL.
But the nature of armed conflict has changed over the last decade, particularly in the
proliferation of non-State armed groups that do not have a central hierarchical struc-
ture through which to transmit, and train members in, the rules of IHL. This has neces-
sitated new research into how both formal and informal norms condition behaviour in
the wide array of armed groups encountered in the ICRC’s work, and how ICRC staff
might promote restraint within their ranks.
This report draws on a rich body of empirical studies seeking to explain armed-group
behaviour. Two constants stand out: first, there is considerable variation in the patterns
of violence and restraint between and within armed organizations, and in the beliefs,
mechanisms, resources and people that influence their behaviour; second, those vari
ations may also change over time. Therefore, rather than formulating new directives for
the ICRC to adopt in its dealings with armed forces and armed groups, the report offers
a framework of analysis to assist its staff in situating armed groups on a spectrum
according to their organizational structure. It further explains how the transmission
and adoption of norms might occur in these groups depending on where they fall on
the spectrum. The report then suggests approaches that might be adapted effectively
to specific contexts.
MAJOR FINDINGS
1. Integrating the law into doctrine, training and compliance mechanisms in
centrally structured armed forces and armed groups increases restraint on
the battlefield. The intensity of training and how norms are taught make a
difference, and adherence is best tested under duress.
2. An exclusive focus on the law is not as effective at influencing behaviour as a
combination of the law and the values underpinning it. Linking the law to local
norms and values gives it greater traction. The role of law is vital in setting
standards, but encouraging individuals to internalize the values it represents
through socialization is a more durable way of promoting restraint.
3. Understanding the structure of armed groups is a first step in identifying
potential sources of influence over their behaviour. The more decentralized the
armed group, the more the sources of influence are external to the group.
4. By focusing on restraint as well as violence, we broaden our understanding of
who or what influences behaviour. Analysing patterns of violence can help to
pinpoint instances where restraint has been exercised.
5. Youth make up the bulk of present and future fighters. Finding innovative and
locally adapted ways to reinforce norms of humanity among them, including via
digital media, is essential.
6. External entities are able to influence the behaviour of armed forces and armed
groups. Making it a criminal offence for humanitarian organizations and local
communities to interact with armed groups is counterproductive and hampers
efforts to promote respect for humanitarian norms.
S. Pessolano
INTRODUCTION
This publication, based on two years of research by a group of distinguished scholars,

is an update of the ICRC’s 2004 study, The Roots of Behaviour in War.1 The original
study explored the social and psychological processes that condition the behaviour of
soldiers and fighters during armed conflict, and sought to identify ways in which the
ICRC might persuade them of the need to comply with the rules of war contained in the
four Geneva Conventions of 1949 and other instruments of international humanitarian
law (IHL).
The study’s findings led to significant policy changes at the ICRC. The organization
expanded its focus from making the law better known to pursuing more robust efforts
to ensure that the legal framework was incorporated into the inner workings of armed
forces and armed groups. Recognizing the importance of group conformity2 and obedi
ence to authority3 in conditioning behaviour, the study recommended that the ICRC
encourage armed forces to integrate IHL into their doctrine and training and to apply
sanctions for breaches of the law. In other words, the ICRC’s role was to persuade
armed forces to make respect for IHL a clear command within their ranks and to advise
and assist them in developing related training programmes and compliance mech
anisms. The original study also recommended that the ICRC approach the teaching of
IHL as a legal and political issue and not as a moral one – to give precedence to legal
norms over the values underpinning them, since the latter were seen to shift according
to the reasons for, and way in which, conflicts were fought. The policies that ensued
were dubbed the “integration approach” and have since guided the ICRC’s efforts to
promote adherence to IHL among a wide variety of armed forces and armed groups.
Just over a decade later, this report revisits some of the findings of that study in light of
the changes in the nature of conflict and of armed groups that have taken place in the
intervening years. It also benefits from the burgeoning academic and policy attention
to the issue of armed-group behaviour.4
PURPOSE OF THE STUDY

The principal objectives of this new study are twofold. First, the research sets out to
deepen the ICRC’s understanding of some of the processes and mechanisms influencing
behaviour that were identified in the earlier study, seeking evidence of what works
best. Of particular interest are the impact of IHL training on behaviour and the peda
gogic methods judged to be most effective by soldiers themselves. The “integration
approach” is unpacked to assess which aspect of the model – knowledge of the law,
training in the law, or threat of punishment under the law – has a greater influence
on behaviour, and how it compares with the influence of informal norms, particularly
peer-group conformity.
1 Daniel Muñoz-Rojas and Jean-Jacques Frésard, The Roots of Behaviour in War: Understanding
and Preventing IHL Violations, ICRC, Geneva, 2004.
2 Dave Grossman, On Killing: The Psychological Cost of Learning to Kill in War and Society, Little,
Brown and Company, New York, 1995, updated 2009; Christopher R. Browning, Ordinary Men:
Reserve Police Battalion 101 and the Final Solution in Poland, HarperCollins, New York, 1992.
3 The seminal work cited being Stanley Milgram’s Obedience to Authority: An Experimental View,
Harper & Row, New York, 1974.
4 See, for example, Hyeran Jo, Compliant Rebels: Rebel Groups and International Law in World
Politics, Cambridge University Press, Cambridge, 2015; Humanitarian Exchange, No. 58,
July 2013, special issue on humanitarian negotiations; Geneva Academy of International
Humanitarian Law and Human Rights, Reaction to Norms: Armed Groups and the Protection
of Civilians, Policy Briefing No. 1, Geneva, January 2014; and Geneva Call, “Exploring Criteria
& Conditions for Engaging Armed Non-State Actors to Respect Humanitarian Law & Human
Rights Law”, Conference Report, Geneva, 4-5 June 2007.
Introduction 13
Second, the study encompasses the growing number of non-State armed groups that
have emerged in recent years and that do not have a vertical, hierarchical structure
suited to the “integration approach”. How can we encourage members of these hori-
zontally organized armed groups to adopt norms of restraint? Are they as fragmented
and unstructured as often depicted, or do they have clear sources of influence over their
behaviour? If so, how can we identify and seek to influence the influencers?
The report begins here by contextualizing the issue of armed-group behaviour within
today’s broader political environment, highlighting some of the major emerging
challenges to ensuring the safety of civilians in armed conflict. It then briefly explains
the study’s methodological approach. Chapter 1 provides an overview of what we
mean by norms of restraint, why organizational structure is relevant, and how norms
are instilled in members of armed forces and armed groups through socialization.
Chapters 2 to 5 present the findings of empirical research conducted with four types
of armed group, exploring in each case the sources of influence over the development
of norms of restraint. Chapter 6 draws together the findings most applicable to the
work of humanitarian organizations and proposes a framework to guide analysis and
reflection on factors to take into account when seeking to influence the behaviour of
soldiers and fighters.
MORE ARMED CONFLICTS,

MORE ARMED GROUPS
Important new trends over the last decade have raised profound challenges for humani
tarian organizations. To begin with, the number of armed conflicts around the world
has risen significantly over the last decade and a half. According to the ICRC’s legal
classification, the number of non-international armed conflicts has more than doubled
between 2001 and 2016, from fewer than 30 to more than 70.
The number of non-international armed conflicts has more than

doubled between 2001 and 2016.
The number of parties fighting in these conflicts has likewise grown exponentially.
ICRC data show that only one-third of conflicts today are between two belligerent
parties: 44 per cent have between three and nine opposing forces, and 22 per cent
have more than ten. Some conflicts have hundreds: by the end of the war in Libya in
October 2011, 236 separate armed groups were registered in the city of Misrata alone,5
and the Carter Center counted over 1,000 armed groups fighting in Syria in 2014.6 The
sheer numbers complicate efforts to understand and engage with these armed groups.
5 Brian McQuinn, After the Fall: Libya’s Evolving Armed Groups, Small Arms Survey, Geneva, 2012,
p. 13.
6 Carter Center, Syria: Countrywide Conflict Report No. 5, Carter Center, Atlanta, February 2015.
Exceptional forms of violence are, moreover, jeopardizing the progress made in incul-
cating restraint in war. Humiliation and perceived injustices and corruption have driven
many new recruits into the arms of self-proclaimed jihadist groups.7 These groups
have proliferated and spread through the Middle East, Africa and Asia, aided, amongst
other things, by the rapid expansion of low-cost telecommunications technology and
social media platforms. In 2017, some 40 per cent of States experiencing armed conflict
were confronting jihadi groups, and the vast majority of all foreign interventions are
currently against armed groups with a jihadist agenda. The extreme and sometimes
indiscriminate violence practised by some jihadi groups has prompted many States
to enact harsh counterterrorism laws that risk eroding the very freedoms that these
States profess to be protecting. Pronouncements by several leading politicians of the
desire to see fighters from Islamic State group killed rather than detained or prosecuted
radically departs from longstanding international law on the treatment of captured or
surrendered fighters.8
Armed conflicts are increasingly fought in cities, leading to heavy

civilian casualties and the destruction of vital infrastructure.
Armed conflicts are increasingly fought in cities, leading to heavy civilian casualties
and the destruction of vital infrastructure such as power grids and water-treatment
plants. The interconnectivity of infrastructure means that the loss of one such service
will have a knock-on effect on other services and hamper efforts to repair damage.9
Explosive weapons with a wide impact area are more likely to be indiscriminate when
used in densely populated zones, with devastating consequences for the civilian popu-
lation, as was seen in Aleppo, Mosul and Raqqa.
Lastly, several powerful States are increasingly outsourcing warfare to human and
technological “surrogates”10 in order to maintain a geographical distance from the
battlefield and ease the domestic costs of direct involvement. This may take the form
of logistical, training, intelligence, advisory, air or other support to belligerent parties.
While such support generally goes to State military forces, it may also be directed to
private security companies, non-State armed groups, militias and community vigi
lantes working at the behest of the domestic State. The increasing use of cyber warfare,
remote technologies such as surveillance and combat drones, and the development
of autonomous weapon systems create further distance. Taken together, both human
and technological “outsourcing” can be seen as an attempt to dilute responsibility
for battlefield conduct, as State sponsors eschew accountability for the actions of
their partners (despite their legal obligation to ensure respect for IHL), while the use
of such surrogates and the detachment of drone operators from their targets make it
7 See the work of Scott Atran, who has extensively studied the factors that drive recruits to
these groups. See also the debate between Olivier Roy and Gilles Kepel on whether the growth
of jihadism is about the Islamicization of radicalism or the radicalization of Islam. For a brief
overview, see Adam Nossiter, “‘That Ignoramus’: 2 French Scholars of Radical Islam Turn
Bitter Rivals”, The New York Times, 12 July 2016.
8 See Anne Barnard, “Red Cross warns of ‘dehumanizing’ rhetoric in ISIS fight”, New York
Times, 26 October 2017. Unfortunately such statements persist; see Jessica Elgot: “British
Isis fighters should be hunted down and killed, says defence secretary”, The Guardian,
8 December 2017.
9 ICRC, Urban Services during Protracted Armed Conflict: A Call for a Better Approach to Assisting
Affected People, ICRC, Geneva, 2015.
10 Andreas Krieg and Jean-Marc Rickli, “Surrogate warfare: The art of war in the 21st century?”,
Defence Studies, January 2018, pp. 113–130.
Introduction 15
easier to dehumanize the enemy. As Alex Leveringhaus said in a recent blog, “Far from
ushering in an age of more humane warfare, the introduction of new, distance-enhancing
combat technologies may, in reality, undermine more informal violence-restricting
norms.”11
RESEARCH APPROACH
The present study sought to identify sources of influence over the development of
norms of restraint in four types of armed group, two with centralized organizational
structures and two with flatter, more horizontal structures. Within each category,
research focused on two armed forces or groups, for a total of eight subjects. The groups
identified were not necessarily parties to armed conflict according to the criterion for
legal classification under IHL.
Chapter 2 summarizes quantitative and qualitative research conducted by Andrew Bell

with, respectively, 409 and 1,030 members of the Australian and Philippine armies.
Through interviews and experimental surveys posing hypothetical scenarios, the
research teased out the relative importance of different sources of influence on com-
batant views. This research constitutes the first-known survey of active-duty combat-
ants in State armed forces on issues regarding IHL, combat ethics and conduct towards
civilians during conflict, providing unique insights into socialization mechanisms in
State armed forces.
Chapter 3 presents the findings of research in Colombia led by Francisco Gutiérrez

Sanín on patterns of violence and restraint by the Fuerzas Armadas Revolucionarias de
Colombia-Ejército del Pueblo (FARC-EP) and the Ejército de Liberación Nacional (ELN).
Based on extensive data from the Registro Único de Víctimas (RUV) database of violence
against civilians,12 and supplemented by 115 interviews with former and active combat-
ants, victims and social entities, the research sheds light on the role of armed-group
structure and ideology on behaviour.
Chapter 4 explores sources of influence on two jihadi groups in northern Mali,

Ansar Dine, active in the Kidal region, and the Movement for Unicity and Jihad in
West Africa (MUJAO) in Gao. Yvan Guichaoua and Ferdaous Bouhlel looked at changes
in the behaviour and sources of influence of the two groups between when they con-
tested power and when they governed in their respective regions. Data sources included
speeches and radio broadcasts, official documents and statements made by the groups,
along with interviews with people close to the groups and those who observed their
behaviour first-hand.
Lastly, research for Chapter 5 focused on gelweng, titweng and gojam armed cattle-
keeping groups of the south-western Dinka communities (former Lakes State),
north-western Dinka communities (former Warrap State) and western Nuer commu
nities (former Unity State) respectively in South Sudan. The research, led by Naomi Pendle
between July 2016 and August 2017, built on ethnographic studies she and her team
undertook in the country between 2009 and 2015. Data sources included interviews
with cattle keepers and their families, community leaders, politicians, spiritual leaders
and community members, and was supplemented by analysis of press articles and the
content of bull songs sung by the cattle keepers.
11 Alex Leveringhaus, “Autonomous weapons mini-series: Distance, weapons technology and

humanity in armed conflict”, Humanitarian Law & Policy (blog), 6 October 2017.
12 https://www.unidadvictimas.gov.co/
S. Olson/Getty Images
CHAPTER 1
NORMS OF
RESTRAINT,
ORGANIZATIONAL
STRUCTURE AND
SOCIALIZATION
This chapter provides an overview of what we mean by norms of restraint, why organ-
izational structure is relevant, and how norms are instilled in armed forces and armed
groups through socialization processes.
1. NORMS OF RESTRAINT
This report explores restraint, defined here as behaviour that indicates deliberate
actions to limit the use of violence. Most scholars of armed conflict, and organizations
such as the ICRC, tend to focus on documenting and understanding the use of excessive
violence by those bearing weapons, especially that which violates IHL.13 The more com-
mon violations witnessed today include attacks on non-combatants, disproportionate
attacks, the use of indiscriminate weaponry, forced displacement, sexual violence, and
attacks on health-care infrastructure and personnel. But work over the last decade by
several scholars, mentioned below, demonstrates the utility of broadening the scope
to identify instances or patterns of restraint. A reduction in the frequency or type of
violence used against civilians in a given conflict, or respect for a previously disre-
garded symbol, structure or person protected under IHL, can shed light on the sources
of this restraint and on previously unknown influences on behaviour. It can also indi-
cate the extent of commanders’ control should restraint be shown by their fighters in
one instance and not in another. The sparing of Kodok hospital during a violent attack
on the town in 2017 (see box) is one such case.
Sparing the Kodok hospital: An example of restraint

During a three-way clash between armed actors in Kodok, South Sudan, in April 2017,
the entire town was looted except for the hospital. This instance of restraint stands out
markedly from the looting and destruction of health facilities that have characterized
the violence affecting the country for the past four years. Having been informed by all
sides that fighting was likely to reach the town, the ICRC evacuated hospital staff and
patients and padlocked the wooden doors. To their surprise, they returned to find that
only items in the outer hospital compound had been looted. The padlocks remained
in place.
This show of restraint enabled the ICRC to reflect on possible contributing factors.
The organization had sent real-time messages to contacts at various levels of each
command structure, stipulating that “the medical personnel and health facilities in
Kodok must be respected at all costs”. The specific request to spare Kodok hospital
capped a broader, longer-standing dialogue with all levels of the hierarchy of the
armed forces and armed groups in the country on the importance of adhering to
humanitarian norms.
The receipt by the ICRC of an advance warning to evacuate the town attests to
successful trust-building with all sides. Even so, the ICRC compound in Kodok was
looted and the same approach failed to prevent attacks on hospitals elsewhere,
suggesting that other factors were at play. Further analysis led to the conclusion
that the desire to retain the population in place was a major factor in the ordering
of the hospital’s protection.
13 See, for example, Stathis N. Kalyvas, The Logic of Violence in Civil War, Cambridge University
Press, Cambridge, 2006; Hugo Slim, Killing Civilians: Method, Madness and Morality in War,
Columbia University Press, New York, 2008, reprinted 2010; Benjamin A. Valentino, “Why
we kill: The political science of political violence against civilians”, Annual Review of Political
Science, Vol. 17, May 2014, pp. 89–103.
Norms of restraint, organizational structure and socialization 19
This case demonstrates that commanders in South Sudan have more control over the
intensity and targeting of violence than is sometimes claimed, and can potentially play
a greater role in preventing violations.
Restraint can be conceptualized in different ways.14 In his analysis of why violence

escalated to genocide in Rwanda but not in Côte d’Ivoire, Scott Straus identifies
sources of de-escalation and restraint.15 Comparing historical patterns of violence in
the Israeli-Palestinian conflict, James Ron advances the notion of what he calls “savage
restraint” – the worst that, given past experience, could have been done, but was not.16
And Elisabeth Wood explores why fighters from Sri Lanka’s Liberation Tigers of Tamil
Eelam (LTTE) engaged in many forms of extreme violence against the civilian popu-
lation but rarely in sexual violence.17 In each study, restraint was measured against a
different benchmark: comparison with a similar case; comparison with past behaviour;
and comparison with other forms of violence.
For this report, restraint is measured against the standards set by IHL. Comparisons
between armed groups operating in the same context, and in relation to historical pat-
terns of violence, also help to monitor improvements in the way civilians are treated.
Genuine restraint should not be confused with restraint caused by mechanical factors:
for example, attacks might decrease because of the mass desertion of fighters, the
disruption of weapon supplies, or seasonal weather. Furthermore, comparing violence
and restraint between armed groups requires that differences in size, capabilities and
terrain be taken into account.
Although this report uses the norms of IHL as a benchmark for measuring restraint,
there is a tighter version of restraint in counter-insurgency warfare known as “cour
ageous restraint”. It recognizes that while some civilian casualties might be lawful
under IHL, incurring such casualties can undermine the very purpose of the mili-
tary operation and increase civilian support for insurgents. Empirical evidence from
Afghanistan confirms this assumption, finding that harm inflicted on civilians by the
International Security Assistance Force (ISAF) increased support for the Taliban, while
Taliban-inflicted harm on civilians did not increase support for ISAF.18
Identifying genuine restraint is challenging since it is essentially demonstrating a

counter-factual: something that could have happened but did not. Francisco Gutiérrez
Sanín and Elisabeth Wood have developed a framework to track patterns of violence
that could be helpful in this endeavour.19
14 We are grateful to Francisco Gutiérrez Sanín for a rich discussion on this issue.
15 Scott Straus, “Retreating from the brink: Theorizing mass violence and the dynamics
of restraint”, Perspectives on Politics, Vol. 10, No. 2, June 2012, pp. 343–362.
16 James Ron, “Savage restraint: Israel, Palestine and the dialectics of legal repression”,
Social Problems, Vol. 47, No. 4, November 2000, pp. 445–472.
17 Elisabeth J. Wood, “Armed groups and sexual violence: When is wartime rape rare?”,
Politics & Society, Vol. 37, No. 1, March 2009, pp. 131–161.
18 Jason Lyall, Graeme Blair and Kosuke Imai, “Explaining support for combatants during
wartime: A survey experiment in Afghanistan”, American Political Science Review, Vol. 107,
No. 4, November 2013, pp. 679–705.
19 Francisco Gutiérrez Sanín and Elisabeth J. Wood, “What should we mean by ‘pattern of
political violence’? Repertoire, targeting, frequency, and technique”, Perspectives on Politics,
Vol. 15, No. 1, March 2017, pp. 20–41.
PATTERNS OF VIOLENCE
Documenting and analysing patterns of violence in an armed conflict can help to better
understand that violence and to identify examples of restraint. A pattern of violence
consists of the “repertoire” or type of violence regularly used (such as killing or forced
displacement), and for each type observed, the target (who), the frequency (how often)
and the technique or method used (for example, killing by machete or burning houses
to evict residents). After the Islamic State group occupied parts of northern Iraq, for
example, its repertoire of violence included sexual violence (type), against Yazidi
women and girls (target), many times (frequency), using sexual slavery (technique).20
Overall patterns can also be broken into sub-patterns that combine two variables, such
as frequency and target, or type and technique, providing clues as to the fine print of a
group’s ideology or rationale for perpetrating violence.
Analysing patterns of violence is useful for the ICRC’s field work in three ways. First, it
can indicate changes in an armed group’s behaviour over time. This can help pinpoint
periods of restraint when the repertoire or target of the violence has narrowed or the
overall use of violence has decreased. This in turn allows us to retroactively investigate
possible reasons for, and thus possible influences on, the change of behaviour. At one
point in the Colombian conflict, for example, the ELN ceased blowing up oil pipelines
owing to pressure from conservationists – a group that the ICRC would not normally
consider when mapping “actors of influence”.
Tracking an armed group’s pattern of violence over time can also help to predict vio-
lence and restraint. By documenting violence perpetrated by the Lord’s Resistance Army
between 1994 and 2003, Jessica Stanton demonstrated that the group tempered its vio-
lence during peace negotiations with the Ugandan government, but resumed it with
renewed vigour each time one of the seven mediation attempts broke down.21 The ability
to predict periods of violence and restraint could guide preventive and remedial responses.
Having a solid record of an armed organization’s pattern of violence also increases the
persuasiveness of the ICRC’s arguments in dialogue with the leadership (assuming that
such dialogue is possible). For example, in discussions with Jamā'at Ahl as-Sunnah
lid-Da'wah wa'l-Jihād (JASDJ – aka Boko Haram), the ICRC would be able to demon-
strate that the group’s tactics changed between 2009 and 2016 from exclusively attack-
ing military targets to increasingly targeting those not belonging to the security forces
(see Figure 1).
Figure 1: Targets of JASDJ violence 2009–2016.22

100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
2009 2010 2011 2012 2013 2014 2015 2016
 Political assassinations  Attacks on security forces  Attacks on civilians
20 Mara Revkin and Elisabeth J. Wood, “The Islamic State’s pattern of violence: Ideology and
institutions, policies and practices”, paper presented at the annual meeting of the American
Political Science Association, San Francisco, 31 August–3 September 2017.
21 Jessica A. Stanton, Violence and Restraint in Civil War: Civilian Targeting in the Shadow of
International Law, Cambridge University Press, Cambridge, 2016, p. 261.
22 We are grateful to Nathaniel Allen for providing this analysis based on data from the Nigeria
Social Violence Project at Johns Hopkins University School of Advanced International Studies.
Second, analysing patterns of violence allows us to draw comparisons between groups

operating in the same context, furthering our understanding of the mechanisms and
factors that influence armed-group behaviour. Comparing patterns of violence between
two Islamist groups controlling territory in Mali, for example (see Chapter 4), demon-
strates that despite espousing the same Islamist doctrine, one engaged in a much
broader repertoire of violence than the other, including suicide bombings, kidnapping
of foreigners, and attacks against other non-State groups and the State.
Third, analysing patterns of violence at the unit level within an armed force or group
allows us to make a distinction between violence as an ordered “policy” and violence
which is tolerated but not ordered, which Wood terms “practice”.23 This important dis-
tinction can steer the ICRC’s intervention to the appropriate level of responsibility. As
Chapter 3 demonstrates, where sub-units of armed groups are responsible for oppor-
tunistic violence not sanctioned by the leadership, interventions would best target the
sub-commander and larger force levels.
Measuring influence via social media

For this report, we tested whether big data from social media could help to identify
people with influence, specifically through the content of their posts and the extent
to which they were shared.
The research focused on Twitter, using key words during fixed periods to catch
references to the “restraint fatwa” issued by Ayatollah al-Sistani in early 2015 in
response to allegations of abuses by the Popular Mobilization Units fighting the Islamic
State group in Mosul. Al-Sistani was instrumental in mobilizing the units. His “Advice
and Guidance to the Fighters on the Battlefields” specified acts that are not permitted
under Islamic law and tradition, most of which accord with important elements of
IHL, including respect for non-combatants and for the sanctity of the dead.
Ayatollah al-Sistani is the most influential Shia cleric in Iraq, yet the importance of his
“restraint fatwa” was not reflected in social media content accessible to us. This finding
suggests the limitations of using social media to measure influence in this case.
2. W
HY ORGANIZATIONAL STRUCTURE
MATTERS
An armed group’s organizational structure is an important determinant of its behav-
iour. This structure is shaped by several factors, including: ideology and doctrine;
leadership preferences; recruitment strategies; funding sources; group history; and
pre-existing social networks. Structure is also shaped by external factors, such as the
opposing force’s strength and effectiveness, the topography of the group’s operating
terrain and, most importantly, external political or military support. Armed groups
given sanctuary in a neighbouring country can organize completely differently from
those denied a safe haven.
23 Elisabeth J. Wood, “Rape as a practice of war: Towards a typology of political violence”,

Politics and Society (forthcoming); Amelia Hoover Green, “The commander’s dilemma:
Creating and controlling armed group violence”, Journal of Peace Research, Vol. 53, No. 5,
September 2016, pp. 619–632.
Despite these variations, a few basic requirements must be met should an armed group
wish to survive for more than a few months. These include: a) reliable fighters – requir-
ing recruitment; b) regular supplies of food and ammunition – requiring money and
logistics; and c) control over group members so that they follow instructions and do not
turn on their leaders – requiring military discipline and/or group loyalty. How a group
solves these challenges determines its structure, ranging from centralized hierarchies
with strict top-down discipline, to groups so decentralized they retain a role in the
community between bouts of violence. Group structure influences military capability,
the type of control exercised by leaders, and how combatants learn honourable from
dishonourable behaviour. While many factors shape group structure, the most deter-
minant concern how their three essential needs – recruitment, resources and control –
are met.
Recruiting new members and indoctrinating them into group norms are essential for
group survival. While the reasons for joining an armed group vary, Jeremy Weinstein’s
research suggests that access to resources strongly influences the type of recruit that
an armed group will attract, which in turn conditions relations between fighters and
the community.24 Resources such as diamonds or coltan attract opportunistic fighters
whose lack of a need for community support leads to harsher treatment of community
members. In resource-poor contexts, by contrast, fighters rely on the local population,
incentivizing better behaviour.
Sustaining an armed group is expensive, costing up to millions of dollars a year.25 To

capitalize on a given revenue source, a group needs to organize itself in a specific way.
This affects the type of control leaders exert and the group’s propensity to use violence.
To extract diamonds in Sierra Leone, for example, the Revolutionary United Front
(RUF) only needed to control a few mines and several hundred recruits. This required
a highly militarized and violent group (to protect the mines), with minimal levels of
hierarchy to monitor operations (to prevent diamond theft). By contrast, the Maoist
insurgents in Nepal – one of the world’s poorest and most inaccessible countries –
extracted revenue by taxing local communities and businesses. This necessitated a large
and disciplined organization. Unlike the RUF, the Maoists were known for their discip
line and adherence to a code of conduct. In both cases, the group’s pursuit of specific
resource strategies influenced their structure and propensity for violence or restraint.
Retaining control or fighter loyalty is also central to survival, as well as to military

effectiveness.26 Armed organizations are built around company-sized groups of (less
than 150) fighters. Shared combat experience unites group members, creating bonds
that can surpass family ties. For decentralized groups, this type of cohesion is the
main glue, holding each sub-unit together. Sub-units are then interlinked through
their leaders, allowing unit cohesion to survive changing alliances. Centralized mili-
tary structures build cohesion by fostering loyalty to the organization as well as to the
unit. This requires forging an identity based on group narratives and collective rituals.
Here, highly centralized State militaries leverage the history of regiment-size groups,
linked by nationalist ideals. In centralized non-State armed groups, ideology serves this
function. What matters most, though, is not the specific content of the ideas or values,
but their translation into practices that create a completely immersive experience and
alignment with the group.
24 Jeremy M. Weinstein, Inside Rebellion: The Politics of Insurgent Violence, Cambridge University
Press, Cambridge, 2007.
25 Achim Wennmann, “Grasping the financing and mobilization cost of armed groups: A new
perspective on conflict dynamics”, Contemporary Security Policy, Vol. 30, No. 2, 2009,
pp. 265–280.
26 Guy L. Siebold, “Key questions and challenges to the standard model of military group
cohesion”, Armed Forces & Society, Vol. 37, No. 3, June 2011, pp. 448–468.
Understanding organizational structure is important for humanitarian organizations

in two ways. First, it helps to identify key decision-makers within a group. In decen-
tralized groups, each sub-commander has significant authority over the unit’s oper
ations and conduct. Thus, local commanders are key contacts with whom to discuss the
conduct of their units’ members. In more centralized armed groups, sub-commanders
must follow orders, rendering senior leaders the central decision-makers and primary
contacts for addressing operational and humanitarian concerns. Weak monitoring sys-
tems can compromise oversight, however, giving sub-commanders more leeway in
interpreting central directives, so it may also be necessary to establish dialogue at the
sub-commander level.
Second, organizational structure can indicate the levers of influence that leaders have
at their disposal. Centralized armed groups rely on clearly established rules and values,
which are likely to be imparted to the rank and file through indoctrination and training.
Decentralized and community-embedded armed groups do not always have written
codes of conduct, drawing instead on shared values and traditions. Here, the source of
norm-influencing behaviour needs to be identified within the community, which might
not be obvious or accessible to outsiders.
For this report, the various types of armed forces and armed groups have been placed
on a spectrum ranging from highly centralized State militaries to armed groups that
are so decentralized they lack the organizational structure and responsible command
to be considered an armed group under IHL (see Figure 2). The latter type of group,
therefore, cannot be deemed a party to an armed conflict in the legal sense.
Figure 2: The spectrum of armed-group organization
INTEGRATED STATE CENTRALIZED NON-STATE DECENTRALIZED COMMUNITY-EMBEDDED

ARMED FORCES ARMED GROUPS NON-STATE ARMED GROUPS ARMED GROUPS
Four criteria were used to determine a group’s position on the spectrum: the locus and
type of authority; the nature of the hierarchy; the nature of discipline; and the degree
of social isolation. Observable indicators were used to determine comparable values for
each criterion (see Table 1).
Table 1: Internal structure of armed organizations

FEATURE INDICATOR CENTRALIZED DECENTRALIZED EMBEDDED
Locus Operational authority Top leaders Sub-commanders Negotiated among various
and type authorities both inside and
of authority outside the group
Decision-making Top-down Consensus among Joint decision-making with key
sub-commanders influencers
Nature of authority Bureaucratic (i.e. recognized Charismatic, with weak Charismatic
in every interaction, such as bureaucracy
saluting)
Nature of Levels of hierarchy in an Established hierarchy Limited hierarchy Flat hierarchy
hierarchy organization (9–17 levels) (5–8 levels) (1–4 levels)
Consistent use of rank across Highly regimented Limited Not present
an organization
Regulation of promotion from Highly regimented Informal Informal and fluid
one level to the next
Level of military coordination High Limited Low
Nature of Observable rules Clear signs of military discipline; Few signs of military discipline, No signs of military discipline
discipline regimented daily schedules; regimented schedule or internal
internal mechanisms for military mechanisms
justice
Codification of rules Rules explicitly documented and Some rules documented Rules unwritten and transmitted
referenced orally
Consistent application of rules Consistent across the Inconsistent No consistency across group
organization
Degree Interaction with individuals Tightly controlled; explicit Some interaction Embedded in social structures
of social outside of group permission required to leave (remain a community member)
isolation barracks or camp
Identifying where an armed organization falls on the spectrum provides a clearer view
of how they might operate, although this can change over time. The ELN in Colombia,
for example, went from a highly centralized armed group to a relatively decentralized
one following an internal purge and almost total defeat at the hands of government
forces, with a consequential change in behaviour (see Chapter 3).
Although seemingly united as one fighting force, many of today’s non-State armed
groups are, in fact, alliances of distinct groups. Despite common features or close vari
ations of the same name, these groups can be organized quite differently. Al-Qaeda
is a good example: founded in 1988, it now consists of more than 40 distinct groups,
each with its own structure and history. Collectively they operate as a movement with a
common identity and ideology, but demonstrate significant variation in their patterns
of violence and orientation to external entities. In movements of this type, there is a
centralized core with shifting authority over a number of decentralized smaller groups.
Like all typologies, this spectrum has its limitations. A linear model cannot capture every
variation. A decentralized group, for example, might have features of a centralized struc-
ture but still operate as an alliance. Conversely, a highly centralized armed group might
allow its members to be in regular contact with society, retaining decentralized aspects
to its organizational model. In Somalia, al-Shabab combines a centralized and highly
disciplined core with units whose members have strong clan-based loyalties; this tension
affects the group’s command and control. Thus, the spectrum merely offers a starting
point for analysing armed groups without being a substitute for detailed and context
ualized examination of their particularities.
3. SOCIALIZATION
In seeking to understand how norms of restraint develop and are propagated in armed
groups, the report explores socialization, the process by which people adopt the norms
and rules of a given community.27 The ICRC’s focus in the past has been on the formal
socialization mechanisms in armed forces and armed groups, ensuring at a minimum
that the rules of IHL are made known among all group members, are incorporated into
practical training and are backed by threat of punishment for non-compliance. The
ICRC has paid less attention to the informal norms, which can be as strong as for-
mal norms, even within highly professional State armed forces. Attesting to this is the
persistence of hazing rituals and sexual abuse in the Australian and US armed forces
described in Chapter 2, despite concerted efforts to stamp them out. It is well known
that in military and police forces, new recruits are introduced to the formal norms by
the institutional hierarchy and then shown how to interpret these norms in real-world
operations. The more that official group norms are supported or enforced at the peer
level, the more likely they are to be internalized.
There are three types of socialization identified that are of interest to us here.28 The first
(Type 0) involves no internalization of norms, just temporary norm adoption following
instrumental calculations of punishment or reward. The other two types involve dif-
fering degrees of internalization: learning and following a norm in order to conform to
group expectations and behaviour (Type 1); and fully internalizing the norm, so that it
becomes part of the individual’s identity – the “right thing to do” (Type 2).
Socialization is a process, and individuals can be socialized into committing violence

or showing restraint. But violence can itself be a socialization mechanism, particularly
among fighters forcibly recruited into non-State armed groups and obliged to commit
acts that will rupture family or community ties and form new bonds with the armed
group. Dara Kay Cohen has shown how gang rape can play this role, creating bonds of
loyalty and esteem among forced recruits.29
This study demonstrates the value of considering socialization mechanisms when look-
ing for ways to instil restraint – as it is interpreted under IHL – in soldiers and fighters.
The Roots of Behaviour in War study took the ICRC beyond simply raising awareness of
the law to promoting its integration at all levels of armed forces and vertically struc-
tured armed groups. The present study goes one step further, advocating creative
cooperation with integrated State armed forces and their partners and with non-State
armed groups (centralized, decentralized or community-embedded) to socialize their
soldiers and fighters to act with restraint towards civilians.
DISSEMINATION INTEGRATION SOCIALIZATION
27 See Jeffrey T. Checkel, “Socialization and violence: Introduction and framework”, Journal
of Peace Research, Vol. 54, No. 5, September 2017, pp. 592–605. This special issue on
socialization and violence contains nine excellent empirical studies on this theme.
28 Ibid.
29 Dara Kay Cohen, “The ties that bind: How armed groups use violence to socialize fighters”,
Journal of Peace Research, Vol. 54, No. 5, September 2017, pp. 701–714.
A. Dasiparu/EPA
CHAPTER 2
INTEGRATED
STATE ARMED
FORCES
Examples:
Australian
French
Russian
Philippine
UK ARMED GROUPS ARMED GROUPS
ARMED FORCES NON-STATE ARMED GROUPS
US armed forces
2.1 CHARACTERISTICS
The key characteristics of integrated State armed forces are:
•• Strictly hierarchical decision-making and authority
•• Codified, observable rules that are consistently applied
•• Observable signs of discipline (professionalism in uniforms, saluting, routines)
•• Separation from civilian life when on duty.
Members of integrated State militaries make up the bulk of fighting forces around the
world. Such forces have a strict vertical hierarchy through which authority flows from
the leadership to the rank and file. Rules are laid out in doctrine, socialized through
training and rituals, and enforced through threat of punishment. This does not mean,
however, that members of State armed forces do not engage in unordered or unauthor
ized violence. Whilst State armed forces share much in common, there is variation
between and within them in their socialization processes and sources of influence.
Not all State militaries are highly centralized. Some might be modelled on a centralized
structure, wear uniforms with insignia, and display a certain amount of discipline. But
a weaker central influence on the rank and file – owing, for example, to competing
clan or ethnic loyalties or irregular payment of wages – will place some State mili
taries towards the right of the spectrum, with their sources of influence and methods
of socialization closer to those of decentralized armed groups.
2.2 METHODS OF FORMAL SOCIALIZATION

Empirical studies have shown that training increases restraint on the battlefield.30
But not just any training. Andrew Bell has found that intensity matters: conflict data
from Afghanistan and Iraq suggest that US military units led by officers with more
30 Andrew Bell, “Measuring the effect of norm socialization on the treatment of civilians: An
analysis of U.S. army conduct in Iraq and Afghanistan”, paper presented at the annual meeting
of the American Political Science Association, San Francisco, 31 August–3 September 2017;
Christopher H. Warner et al., “Effectiveness of battlefield-ethics training during combat
deployment: A programme assessment”, The Lancet, Vol. 378, September 2011, pp. 915–924.
Integrated State armed forces 29
intensive training in norms of restraint engaged in less violence against civilians, even
when controlling for combat leadership capability. Research conducted for this study
similarly indicates that higher levels of IHL training result in greater adoption of norms
of restraint by combatants in the Australian and Philippine armies.
However, training intensity is only part of the story: evidence from the Australian
and Philippine militaries shows that mixed training methods, combining IHL brief-
ings, classroom discussions, case-study reviews and practical field exercises, are the
most effective in inculcating norms of restraint in combatants.31 The Australian Army’s
Royal Military College recently discovered the importance of testing ethical compliance
under duress: during a week-long training exercise in which cadets were sleep- and
food-deprived, instructors tried to enlist the cadets in simulated unethical and unlaw-
ful behaviour. Many acquiesced, demonstrating how fatigue and stress can lead to
ethical breakdown. The cadets themselves were shocked when anonymized recordings
of their actions were played back to them in the classroom and said that the experience
had taught them more than any other of the need to develop a strong moral compass
before facing the stress of the battlefield. Based on this experience, the Royal Military
College has since instituted an intensive, model ethics training programme that incorp
orates training while under duress, which has been found to significantly enhance the
adoption of norms of restraint by cadets.
Research also suggests that who it is who delivers the message makes a difference in
the socialization process. For Australian and Philippine soldiers, an effective instructor
in IHL requires credibility derived from operational experience: they need to be able
to draw on the dilemmas they have faced and explain the choices they made. Con-
versely, some combatants may give greater credence to people of certain backgrounds
with no combat experience but recognized IHL expertise: in the Philippines, junior
soldiers highly rated training by civilian lawyers from the ICRC. Perhaps troubling for
IHL training efforts, however, both Australian and Philippine soldiers generally rated
poorly the training conducted by military legal officers, finding such officers to be
generalists from higher echelons with no direct combat experience. Ultimately, such
research points to the need to understand organizational context in order to identify
the most effective training providers within an armed force.
Related research shows that the key moments to reinforce norms of restraint include
during immediate pre-deployment briefings and, most importantly, in the wake
of an incident in which a unit member has been injured or killed. Military expert
David Kilcullen suggests that restraint must be reiterated by the unit leader as soon as
it is feasible after the event: debriefings by army psychologists do not have the same
impact.32 Reinforcing norms of restraint must take place down to the lowest level.
A further area of formal socialization explored in the research was the role of pun-
ishment in encouraging compliance with the law. Survey data and interviews with
members of both the Australian and Philippine armies showed that the threat of pun-
ishment under domestic and military law exerts a much greater influence than that of
punishment under “IHL” per se. This finding confirms the importance of integrating
IHL norms into domestic law, standard operating procedures and rules of engagement.
However, although the threat of punishment under internal military law had a strong
influence on soldiers, particularly officers, this influence was surpassed by the social-
izing effect of informal norms and of “army values”.
31 This chapter is based on empirical research undertaken with the Australian and Philippine
armies by Andrew Bell.
32 Interview with David Kilcullen, Geneva, 29 November 2017.
2.3 INFORMAL SOCIALIZATION

The importance of the peer group’s informal norms in shaping the attitudes and behav-
iour of soldiers has received considerable attention in recent decades, demonstrat-
ing that social bonds of “brotherhood” among soldiers invariably trump patriotism
or ideology as a rationale to fight and kill.33 Unwritten norms such as “never leave a
man behind” are deeply ingrained in unit members across a wide range of State armed
forces. Data from the Australian and Philippine militaries show the vital role of such
informal norms and socialization processes.
The norm of brotherhood – or “mateship” in the Australian Army – is consistently

cited across ranks as having a fundamental influence on the views and actions of
soldiers. In interviews, soldiers noted that the decentralized nature of counter-
insurgency warfare, in which small units fight at a distance from central command,
further increases that influence. The bonds of brotherhood are especially pronounced
among members of special forces, who operate in small, tightly knit units that act
independently of conventional units.
The strength of informal norms in military forces is starkly illustrated in the persist
ence of sexual abuse and “hazing” rituals – the harsh and often humiliating initiation
processes to which new recruits are subjected, ostensibly to forge group cohesion – in
spite of military laws, reforms and disciplinary measures meant to stamp it out. Wood
and Toppelberg’s research on the US military points to informal mechanisms that trivi
alize sexual assault, establish it as an appropriate form of punishment, and condone
retaliation against those who report it.34 In Australia, 11 formal investigations, including
several parliamentary inquiries, were undertaken into hazing and other forms of abuse
in the Australian military between 1971 and 2009, and yet new cases of abuse continue
to arise,35 prompting a strong reiteration of values within the army.
Informal norms can be a double-edged sword, reinforcing or

undermining official organizational norms.
Informal norms can thus be a double-edged sword, reinforcing or undermining offi-

cial organizational norms. While there is no doubt that informal norms increase unit
cohesion, this cohesion becomes problematic when unit members begin “protecting
one another from the system, if or when you stuff up”.36 When loyalty to the group
supersedes loyalty to the organization as a whole, witnesses to unethical behaviour
are unlikely to come forward, compromising the ability of compliance mechanisms to
ensure adherence to the rules.
Lastly, survey experiments tested what proportion of a combatant’s peer group (25%,
50% or 75%) was needed to shift their opinion on conducting a hypothetical military
33 See, for example, Guy L. Siebold, “The essence of military group cohesion”, Armed Forces &
Society, Vol. 33, No. 2, January 2007, pp. 286–295; Charles Kirke, “Military cohesion, culture
and social psychology”, Defense & Security Analysis, Vol. 26, No. 2, June 2010, pp. 143–159.
34 Elisabeth J. Wood and Nathaniel Toppelberg, “The persistence of sexual assault within the
US military”, Journal of Peace Research, Vol. 54, No. 5, September 2017, pp. 620–633.
35 Richard Evans, “Hazing in the ADF: A culture of denial?”, Australian Army Journal, Vol. X,
No. 3, 2013, pp. 113–127, at 117.
36 Denny Neave and Craig Smith, Aussie Soldier: Up Close and Personal, Big Sky Publishing,
Wavell Heights, 2008, p. 52.
operation that would result in heavy civilian casualties. For both the Australian and
Philippine armies, the opinions of the peer group appear to play a significant role in
shifting combatant views towards restraint, with diminishing effects over the 50 per
cent mark. This suggests that those concerned with the promotion of IHL do not need to
ensure that all members of a unit internalize norms of restraint; even adoption by half
of the group or less can sensitize their comrades to the need to spare civilians.
Formal vs informal sources of influence

The graph below illustrates the comparative effects of formal mechanisms of IHL and
informal mechanisms of peer-group influence on the preferences of officers and unit
members in the Australian Army.
Australian Army
IHL, unit influence and acceptance of civilian casualties
Comparing officers and other ranks
Strongly agree
Average support for mission
Agree
Effect of IHL
and unit influence
Neither agree/disagree
Other ranks
Disagree Officers
Strongly disagree
o
eth L
l
ica
ica
nf
IH
un IH
oi
eth
es
+ tes
l/n
lat
Un
la
ro
Vio
Vio
nt
Co
In a survey experiment, participants were asked to advise their commander on

whether to target a high-value bomb-maker in a residential area. The hypothetical
operation was likely to incur civilian casualties. The control survey gave no further
information. Survey 1 added that the unit’s legal officer advised that the operation
would violate IHL. Survey 2 added that the majority of unit members believed that the
operation was unethical. Survey 3 provided both sets of information.
It is interesting to note that Australian officers were more influenced by both IHL
status and the opinion of their unit members than were enlisted soldiers. The graph
clearly shows that the combined effects of the formal and informal socialization
mechanisms had the greatest impact on shifting the views of the officers.
2.4 EMPHASIZING ETHICS

One way it seems that both the Australian and Philippine armies are trying to reconcile
the formal and informal processes of socialization is by emphasizing “army values”:
such values were frequently cited by all ranks as an important source of influence on
behaviour.37 These ethical values appear to play a strong complementary role to the law,
a hybrid of formal and informal norms that aims to discourage unwanted behaviour
not only because it is “against the rules” but also because it is “not who we are”. It is
a hybrid in the sense that the values are not formally enforced in the way that the law
is – unless behaviour that violates a value also violates the law – but career progression
and respect within the organization clearly depend on the extent to which these values
are embodied.
The use of values as a socialization tool in aligning behaviour to organizational

norms was demonstrated by the Australian Army in 2013 when it added “respect”
to its existing values of “courage”, “initiative” and “teamwork” in the wake of the
above-mentioned sexual abuse scandals within the Australian Defence Force (ADF).
In an appeal to soldiers’ ethical values, the then chief of army, Lieutenant General
David Morrison, said at the launch of “respect”:
No one has ever explained to me how a coward in barracks is a hero

on operations. And bullies who humiliate their comrades are cowards
– as are those who passively watch victimisation without the moral courage
to stand up for their mates.38
This appeal to personal honour resonated with soldiers from the Australian and
Philippine armies. In the words of one Australian soldier, “you need to be able to look
at yourself in a mirror” after the fight.
Thus, it is here that this study differs with the Roots of Behaviour in War conclusions,
which opposed invoking moral values, arguing them to be relativist and unreliable,
and instead advocated for a formalistic adherence to orders, discipline and hierarchy.39
It posited that the combatant is not morally autonomous, although this contradicts
rulings that do not allow the defence of “I was only following orders”. Military train-
ing does indeed seek to automate reflexes and limit the moral autonomy of individ-
uals; however, survey and interview data suggest that value-based motivation can in
fact be as powerful a motivator of combatant behaviour as the threat of punishment.40
37 Amelia Hoover Green, “Armed group institutions and combatant socialization: Evidence from
El Salvador”, Journal of Peace Research, Vol. 54, No. 5, September 2017, pp. 687–700; “The
commander’s dilemma: Creating and controlling armed group violence”, Journal of Peace
Research, Vol. 53, No. 5, September 2016, pp. 619–632.
38 Address by the chief of army, Lieutenant General David Morrison, AO, at the launch of the
fourth value of the Australian Army – “Respect”, Lavarack Barracks, Townsville, 4 July 2013.
39 For a very thoughtful, longer critique of this point, see Dale Stephens, “Behaviour in war:
The place of law, moral inquiry and self-identity”, International Review of the Red Cross,
Vol. 96, No. 895/896, December 2014, pp. 751–773.
40 In fact, the ICRC has not restricted its efforts to the law. In the 2016 commentary on
Article 47 of the First Geneva Convention, footnote 4 refers to the Roots of Behaviour
in War study and adds:
In order to be effective and to induce behaviour compliant with the law, international
humanitarian law must not be taught as an abstract and separate set of legal norms,
but must be integrated into all regular military activity, training and instruction. Such
integration should aim to inspire and influence the military culture and its underlying
values, in order to ensure that legal considerations and principles of international
humanitarian law are incorporated, as much as possible, into military doctrine and
decision-making.
This emphasis on organizational identity, “warrior’s honour” and ethical behaviour

allows soldiers to internalize such norms of restraint, encouraging IHL compliance
to a degree not possible through enforcement mechanisms alone. The internalization
of norms beyond IHL-based punishments is all the more necessary in decentralized
counter-insurgency warfare, where units operate far from commander oversight and
the legal enforcement mechanisms of higher command.
The research found that there is a need for both the law and the values underpinning it,
with the emphasis of each influence dependent on the target audience. The role of law
is vital in setting the standards, but ensuring that the values it represents are intern
alized seems to be a more durable way of promoting restraint. Despite the increased
legalization of military operations over the last decade – known as “lawfare” in some
quarters41 – the words of British historian John Keegan still ring true, especially in
counter-insurgency warfare:
There is no substitute for honor as a medium of enforcing decency on the

battlefield, never has been, never will be. There are no judges, more to the
point, no policemen at the place where death is done in combat.42
T. Aljibe/Getty Images
41 There is even a blog dedicated to the topic: https://www.lawfareblog.com

42 John Keegan, cited in Michael Ignatieff, The Warrior’s Honor: Ethnic War and the Modern
Conscience, Vintage, London, 1999, p. 118.
2.5 CHALLENGES TO SOCIALIZATION

The research also identified key issues that pose challenges to the socialization of com-
batants in norms of restraint. Perhaps the greatest challenge is the scepticism with
which junior soldiers serving on the front line view abstract principles of law and
ethics, particularly when confronted by the deadly risks they and their comrades face.
Moreover, maintaining military adherence to IHL in the face of consistent violations
by the opposing side remains a major obstacle to the observance of such principles on
the battlefield. Hence, emphasizing the identity-based nature of restraint could help
to encourage compliance in cases where soldiers question why they should respect IHL
when their enemies do not.
Identifying methods to overcome this scepticism is thus a major challenge for those
charged with promoting IHL and norms of restraint. Surveys and interviews show that
the example set by junior non-commissioned officers (NCOs) has the greatest influ-
ence on junior soldiers’ thoughts and behaviour. Such enlisted leaders must therefore
be central to efforts to promote and transmit norms of restraint among junior soldiers
within small operational units. In many ways, the junior NCO must become as much a
partner in IHL training as the senior battalion commander, for it is only when officers
at those levels adopt those norms that soldiers will experience formal and informal
socialization.
Additionally, evidence from this study shows that religious identification can be an
alternative focus for combatants’ loyalty, particularly for Muslim soldiers for whom
IHL resonates much less strongly than principles of Islamic law. To mitigate the poten-
tial for conflict between these two influences, trainers must emphasize the correl
ation between IHL and Islamic principles regarding restraint towards civilians and the
prohibition of the use of certain means and methods of warfare, using language and
references applicable to the particular context. The ICRC holds such seminars with State
and non-State entities throughout the Muslim world, emphasizing the shared prin
ciples between the two systems and pointing out that Islamic law precedes IHL by
over a millennium. In the Philippines, for example, a seminar on Islamic law related
to armed conflict and IHL and Muslim customs and traditions was held in early 2018
at the Philippine National Police Center for Law Enforcement Studies in Quezon City.
2.6 IMPLICATIONS
This research demonstrates that the “integration approach” has considerable ongoing
validity in seeking to shape the behaviour of combatants towards civilians, but it needs
to be fully tailored to the audience, taught with intensity and tested under duress.
Some of the findings are at odds, however, with those of the Roots of Behaviour in War
study, particularly in the emphasis on law over values, and suggests that a combination
of the two, through formal and informal socialization mechanisms, would provide a
broader basis on which to advocate for restraint.
B. Heger/ICRC
CHAPTER 3
CENTRALIZED
NON-STATE
ARMED GROUPS
Examples:
FARC-EP (Colombia)
LTTE (Sri Lanka)
INTEGRATED STATE CENTRALIZED NON-STATE DECEVNTRALIZED COMMUNITY-EMBEDDED
ARMED FORCES ARMED GROUPS MNLF (Philippines)
NON-STATE ARMED GROUPS ARMED GROUPS
Shan State Army (Myanmar)
3.1 CHARACTERISTICS
The key characteristics of centralized non-State armed groups are:
•• Leadership exercises tight command and control over subordinates through
a strict hierarchy
•• A prominent doctrine or ideology outlines goals, approaches and world view
•• Observable signs of discipline (professionalism in uniforms, saluting, routines)
•• Isolated from civilian population (housed in camps or barracks).
Centralized non-State armed groups share many of the structural characteristics found
in integrated State militaries, including a prominent hierarchy, elaborate doctrine and
strict discipline. But unlike State militaries, they do not benefit from State resources
and infrastructure, creating greater challenges for leaders to communicate with, and
monitor the behaviour of, field commanders and their units. To align their beliefs and
preferences with that of the leadership, group members are subjected to a system
of socialization and control penetrating almost every facet of daily life. This process
reshapes members’ identities and builds allegiance to the overall organization. Ideol
ogies espoused can be as diverse as communism or Salafi jihadism.
Many of the liberation movements in Africa and left-wing revolutionary armed forces
in Central America and Asia during the Cold War were structured in a highly centralized
manner, owing partly to their external funding as proxies in superpower rivalry. Today,
such groups are a small minority of non-State armed groups operating in the world,
but their military capabilities give them a prominent role in several armed conflicts.
Recent examples of centralized groups include the Communist Party of Nepal (Maoist),
which fought the Nepalese government from 1996 to 2006; the Liberation Tigers of
Tamil Eelam (LTTE), which fought the Sri Lankan government from the late 1970s to
2009; the Moro National Liberation Front (MNLF) in the Philippines; and the Islamic
State group in Iraq and Syria.
Centralized non-State armed groups 39
This chapter focuses on two non-State armed groups active in Colombia’s long civil
war, the Fuerzas Armadas Revolucionarias de Colombia-Ejército del Pueblo (FARC-EP),
which signed a peace agreement with the government in 2016 ending 52 years of con-
flict, and the Ejército de Liberación Nacional (ELN), which remains active.43 Both armed
groups established a highly centralized structure at inception, but the ELN decentralized
in the early 1970s following a brutal internal purge and almost complete annihilation by
the Colombian armed forces. The new ELN that arose opted for more collaborative
decision-making among commanders and sub-commanders and a decentralized struc-
ture so that no leader would again hold enough power to kill or evict comrades. Based
on these experiences, this chapter sheds light on the impact that an armed group’s
structure, ideology and institutions can have on patterns of violence and restraint.
3.2 D
IVERGENT PATTERNS OF VIOLENCE
AND RESTRAINT
According to available data, the FARC-EP and the ELN displayed similarities and dif-
ferences in their patterns of violence and restraint. Both groups employed a similar
repertoire of violence against civilians, including massacres, murder, kidnapping,
recruitment of minors, extortion, destruction of infrastructure and forced displace-
ment. Unlike the paramilitary groups, neither the ELN nor the FARC-EP engaged in
widespread rape or the seizure of goods or property. However, there were important
distinctions in sub-patterns of violence between the two groups. Even when accounting
for difference in group size – the FARC-EP at times reached two to four times the ELN’s
size – the FARC-EP committed seven times more civilian massacres, planted ten times
more landmines and recruited four times more minors than the ELN. The ELN, by con-
trast, committed almost as many kidnappings as the FARC-EP, despite being smaller,
and focused for a long period on destroying infrastructure, particularly oil pipelines.
Ideology and armed-group structure account for some of these variations. As shown
in Table 2, both armed groups espoused versions of Marxism and professed their
engagement in a people’s war. The ELN’s ideology, however, was heavily influenced by
Che Guevara’s “revolutionary humanism”, infused with Catholic values emphasizing
self-sacrifice, heroism and empathy with the poor, while the FARC-EP’s ideology was
pitched more in terms of class struggle. While both armed groups considered “dis-
respecting the masses” to be a serious offence,44 with stealing punishable by death,
ideological differences led to different group trajectories, and ultimately different
behaviour. The ELN, for example, initially prohibited on moral grounds coca production
in areas it controlled, thereby forgoing the financial windfall that the FARC-EP used
to finance its expansion into an army-like structure. For the FARC-EP, civilians were
viewed through the lens of class struggle rather than as illegitimate targets, and kill-
ings and kidnappings of “class enemies” were justified in these terms. The two groups
differed in their receptiveness to IHL: while the ELN supported “humanizing war”, the
FARC-EP only partially adopted IHL, preferring to end the war rather than humanize it.
The FARC-EP specifically opposed IHL’s prohibition on recruiting minors in its ranks,
considering 15 to be an acceptable age.
43 This chapter draws on field research led by Francisco Gutiérrez Sanín, based on a
methodology explained in the introduction to this publication, and on prior work, including:
Francisco Gutiérrez Sanín, “Telling the Difference: Guerrillas and Paramilitaries in the
Colombian War”, Politics & Society, Vol. 36, No. 1, March 2008, pp. 3–34; Francisco Gutiérrez
Sanín and Antonio Giustozzi, “Networks and armies: Structuring rebellion in Colombia
and Afghanistan”, Studies in Conflict & Terrorism, Vol. 33, No. 9, August 2010, pp. 836–853;
and Francisco Gutiérrez Sanín and Elisabeth J. Wood, “Ideology in civil war: Instrumental
adoption and beyond”, Journal of Peace Research, Vol. 51, No. 2, March 2014, pp. 213–226.
44 Chapter 1, Article 1, of the FARC-EP’s statutes. The ELN’s rules are even more elaborate, with
a whole chapter on mandatory rules regarding “behaviour vis-à-vis the masses”.
Table 2: Comparison between former FARC-EP and ELN

CRITERION FARC-EP ELN
Organizational blueprint Army Guerrilla
Ideology Communism Che Guevara and Christian liberation theology
Specific ideology with respect to civilians Debate with IHL; “war does not have to be “War has to be humanized”
humanized, it has to be ended”
Membership For life, difficult exit Relatively flexible
Structure of command High levels of centralization Relatively decentralized, but not network-like
In terms of organizational structure, the photos below highlight the differences in

the models adopted by the FARC-EP and the ELN. Both were disciplined and organ-
ized armed groups but to a different degree, with important repercussions on the way
behaviour was controlled. Having a strict hierarchy, strong disciplinary mechanisms
and an immersive socialization process, the FARC-EP was able to curb opportunistic
violence at the unit level. Therefore, most of its violence was ordered from the top.
The ELN had a harder time controlling its fighters, so not all violence accorded with
the group’s political objectives. In fact, patterns of violence varied across field units of
either group, indicating that intermediate commanders had leeway to interpret rules
and make decisions. Ideologies are full of ambiguities and potentially contradictory
cues, permanently teased out by force leaders, intermediary cadres and rank and file
members. To justify actions, certain aspects of the doctrine can be emphasized over
others, rules reinterpreted or subjects renamed (such as recasting “people” as “inform-
ers”). Thus, ideologies set the parameters of permissible violence and restraint, and the
institutions and mechanisms of socialization turn ideologies into observable practice.
FARC-EP soldiers display the military-style discipline of a highly centralized armed group, while the ELN’s more relaxed
approach is closer to that of the decentralized model.
C. Villalon/Getty Images
Reuters
3.3 SOCIALIZATION MECHANISMS

Both FARC-EP and the ELN engaged fighters in deep processes of preference and person-
ality transformation, emphasizing revolutionary morality. These were reinforced in the
FARC-EP by strict rules, army-like drilling, tight daily schedules, and other practices that
dominated members’ lives. First, membership in the FARC-EP was for life: deserters, if
caught, were executed. New recruits were given a three-month grace period, no more.
Second, in recruiting women to its ranks from the 1970s onwards, the FARC-EP created
a self-contained project, encouraging members to develop the whole of their personal
and sexual lives within the organization, although pregnancy was forbidden.45 Informal
contact with society was strongly discouraged. Third, daily drills, exercise and political
education sessions filled every minute of non-combat time, reinforcing the norms out-
lined in the FARC-EP’s doctrine and creating a shared sense of discipline and belonging.
This infusion of organizational culture (“cultura fariana”) aimed to shift the rationale for
restraint with respect to certain forms of violence such as rape from being “against the
rules” to being “not who we are”. Fourth, the FARC-EP created specific institutions so
that “trials” of both members and civilians were overseen by hierarchical superiors and
were not addressed by expedient, violent methods. These institutions played a significant
role in controlling behaviour in most – but not all – circumstances.
Each force’s level of control over its fighters can be illustrated by data on rape, a form
of violence commonly associated with armed conflict. Both the FARC-EP and the ELN
had strict rules against rape, without exception, even for the enemy or for commu
nities loyal to the enemy. Rape was punishable by death. Although incidents of rape are
notoriously underreported, existing data indicate low frequencies among the FARC-EP
and the ELN, especially compared with the paramilitaries. Narratives by victims of kid-
napping confirm this impression. Ingrid Betancourt, the political leader famously held
and harshly treated by the FARC-EP for six years, reported not a single case of rape.46
Other former hostages,47 none of whom had a favourable opinion of the FARC-EP, have
said much the same.48 Hence, both armed groups were able to practise restraint when
it suited them.
3.4 C
OMMUNITIES AS A SOURCE
OF RESTRAINT
As elaborated above, both armed groups’ institutions held decisive influence over
fighter behaviour. The FARC-EP and the ELN seemed less concerned by the political
costs of their violence, especially kidnapping, which the FARC-EP only ceased once
peace negotiations had begun. The ELN is still practising it. The final part of this
chapter examines a relatively recent area of study of interest to humanitarian organ
izations facing increasingly protracted conflicts: the role of communities in influencing
armed-group behaviour.
45 Francisco Gutiérrez Sanín and Francy Carranza Franco, “Organizing women for combat:
The experience of the FARC in the Colombian war”, Journal of Agrarian Change, Vol. 17, No. 4,
October 2017, pp. 770–778.
46 Ingrid Betancourt, No hay silencio que no termine, Aguilar, Bogotá, 2012; “Dura experiencia del
secuestro convirtió a Íngrid Betancourt en símbolo de la libertad en el mundo”, El Tiempo,
8 October 2008; “Así fue el secuestro de Íngrid Betancourt”, El Espectador, 3 April 2008.
47 There are far more narratives from people kidnapped by the FARC-EP than by the ELN.
48 See, for example, Centro Nacional de Memoria Histórica, Una verdad secuestrada: Cuarenta años
de estadísticas de secuestro 1970–2010, Imprenta Nacional, Bogotá, June 2013; and Clara Rojas,
Captive: 2,147 Days of Terror in the Colombian Jungle, Simon and Schuster, New York, 2009.
Oliver Kaplan’s work,49 focused mainly on Colombia, documents how civilian com-
munities can positively influence armed actors and limit violence – particularly if the
community is cohesive and well organized.
Communities employ several tactics to achieve this. First, by promoting a culture of

active neutrality and by demarcating safe zones, they can resist their members being
recruited or turned into informants for either side. Here, community cohesion is crucial
to avoiding certain members collaborating with, and receiving benefits from, armed
actors and others not. Over time, demonstrating non-partisanship can disincentivize
armed groups from entering demarcated zones.
Second, strong communities can implement local conflict-resolution processes. This

saves civilians from soliciting outside entities to address local disputes, particularly
where State institutions have a weak presence. It also avoids giving armed actors an
excuse to get involved in the community’s affairs and exploiting divisions to their
advantage.
Third, communities can establish local investigatory institutions to clarify accusations

made against suspected enemy collaborators. Armed groups may misinterpret civilians’
activities as aiding their enemies, or a rival may falsely denounce another community
member. By vouching for falsely accused suspects but not for confirmed collaborators,
a civilian transparency process can reduce violence against the wrongfully accused.
Such investigatory processes are established in dialogue with the armed groups con-
cerned to convince them to allow civilians to police their own communities.
The feasibility of these mechanisms largely depends on the strength of community

leadership and cohesion, and on the readiness of community members to show restraint
themselves. Just as communities’ positive agency is often overlooked by humanitar-
ian organizations, so too is their negative agency, evident in conflicts ranging from
Afghanistan and Colombia to Mali and Syria, where communities have demanded
violence against individuals considered to be a threat to the social order, particularly
criminals.
3.5 IMPLICATIONS
The research demonstrates the importance of armed-group organization and ideol-
ogy on the behaviour of members of centralized non-State armed groups, in many
ways mirroring the importance of doctrine and hierarchy in State armed forces. The
major differences are found in the weaknesses of monitoring mechanisms in non-State
groups, which can undermine adherence to orders from the central command, and in
the immersive socialization processes required to align members’ behaviour with that
of the leadership.
The strong ideological drive of centralized non-State armed groups provides the entry
point for discussions on whether their behaviour accords with the principles and object‑
ives espoused in the doctrine. This necessitates a good understanding of the fine print
of the doctrine, and possible contradictions, ambiguities and loopholes that can be used
to justify certain actions.
49 In particular, see Oliver Kaplan, “Protecting civilians in civil war: The institution of the ATCC
in Colombia”, Journal of Peace Research, Vol. 50, No. 3, May 2013, pp. 351–367, and Resisting War:
How Communities Protect Themselves, Cambridge University Press, Cambridge and New York,
2017. See also Juan Masullo, “A Theory of Civilian Noncooperation with Armed Groups: Civilian
Agency and Self-Protection in the Colombian Civil War”, doctoral dissertation, Department of
Political and Social Sciences, European University Institute, Florence, August 2017.
The role of communities in influencing armed-group behaviour warrants further

exploration, particularly given the correlation between the cohesiveness of a commu-
nity and its ability to protect its members from violence. Humanitarian organizations
are increasingly exploring ways to support communities’ self-protection initiatives,
particularly in protracted conflicts where navigating violence becomes the norm rather
than the exception. But aid organizations need to be mindful of the potential pitfalls
of their agency in this process, both for the communities involved and for perceptions
of their own neutrality, on which the safety of their personnel, access to communities
and ability to talk to armed forces and armed groups depend. This point is revisited in
Chapter 6.
I. Sanogo/AFP
CHAPTER 4
DECENTRALIZED
NON-STATE
ARMED GROUPS
Examples:
Movement for the Emancipation of the Niger Delta (Nigeria)
Anti-Balaka militia (Central African Republic)
Mouvement des Forces Démocratiques de la Casamance
ARMED FORCES ARMED GROUPS (Senegal) NON-STATE ARMED GROUPS ARMED GROUPS
4.1 CHARACTERISTICS
The key characteristics of decentralized non-State armed groups are:
•• An alliance of small armed groups, whose individual commanders retain
considerable decision-making power and responsibility over group members
•• Alliances are fluid as leaders and their groups may break away to form or join
new associations, without compromising group cohesion
•• Multiple decentralized groups can work together within a broader movement,
such as al-Qaeda or the Islamic State group, giving them local, regional and
global reach
•• Loose coordination within the alliance, including in military planning and
operations
•• Few observable signs of military discipline (e.g. absence of uniforms, saluting
and daily routines).
Decentralized non-State armed groups are generally composed of alliances of smaller

groups, whose individual sub-commanders retain significant authority. Although
appearing as a unitary group, in practice these alliances work as federations of small
autonomous groups. The shifting nature of the alliances and the small size of sub-
groups can lead to seemingly chaotic and disorganized violence. Yet such alliances have
seriously challenged more structured opponents in Libya, Syria and Yemen. Retain-
ing an image of disorganization can help commanders confuse the enemy, while dis-
tancing financial backers from responsibility for group actions. Today, decentralized
armed groups constitute the majority of non-State armed groups operating worldwide,
and are predominantly found in the Middle East and North Africa. Their decentralized
nature and unwritten codes of conduct make them especially challenging for humani-
tarian organizations to engage with, adding complexity to the operating environment.
Decentralized armed groups constitute the majority of non-State

armed groups operating worldwide.
Decentralized non-State armed groups 47
The proliferation of decentralized non-State armed groups and their resilience in the
face of stronger State militaries is partly owing to their horizontal structure, allowing
for a high degree of adaptability. Their structure makes them difficult to defeat, as
no one self-sufficient sub-group is critical to the survival of the broader alliance. While
the groups do not rely on a central command for directions and supplies, they may
strategically pool and disburse resources. A further advantage of this kind of federation
is that it permits the inclusion of groups emanating from local communities, linking
local grievances to broader political goals and providing a dynamic space in which a
herder can become a foreign fighter.
As this report will demonstrate, the more decentralized an armed group is, the stronger
the influence of sources external to the group. External actors draw upon religious,
social, political and economic authority to sway armed groups. Yet this authority
changes as power balances between armed groups and those entities shift.
To explore some of the dynamics operating within this extremely diverse category of
armed groups, this chapter compares two decentralized groups in Mali: Ansar Dine
and the Movement for Unicity and Jihad in West Africa (MUJAO).50 Although both were
inspired by al-Qaeda’s global call, each group used violence and demonstrated restraint
in different ways. This can shed light on how local actors can shape the behaviour of
decentralized groups even where both groups share a similar Salafi-jihadi method
ology. This chapter also explores how local clerics used references to Islamic jurispru-
dence indigenous to Western Sahara in their negotiations with Ansar Dine and MUJAO.
4.3 DIVERGENT PATTERNS OF VIOLENCE

AND RESTRAINT
In 2012, after a Tuareg separatist rebellion expelled local State governance from north-
ern Mali, a coalition of jihadi groups took control of the region. These jihadi groups
dislodged the Tuareg separatists through political tactics more than military victory,
siphoning off their combatants and striking alliances with influential civil leaders.
The research looked at how two of these jihadi groups – Ansar Dine, in the region of
Kidal, and MUJAO, in the region of Gao – took control of and governed areas of north-
ern Mali prior to the French military intervention in January 2013.51 Each group estab-
lished a system of governance drawing upon Islamic principles promoted by al-Qaeda,
and its two main figures, Osama Bin Laden and Ayman Al-Zawahiri. Yet each group also
portrayed itself as following a historically authentic Islam rooted in Western Sahara.
Governing northern Mali constituted a real-life laboratory for using, codifying and
controlling violence within the Salafi-jihadi project.52 Yet there is no unequivocal set
of rules for a sharia-regulated society, hence al-Qaeda’s and the Islamic State group’s
changing policies regarding takfir (declaring that a fellow Muslim is a non-believer)
and the targeting of Muslims.53
50 This chapter is based on research conducted by Yvan Guichaoua and Ferdaous Bouhlel.
51 The intervention followed the Malian interim government’s official request for French military
assistance and the adoption in 2012 of UN Security Council Resolution 2085. Operation Serval
began in January 2013 and was replaced by Operation Barkhane in August 2014.
52 The experience also arguably seeped into social practices, prompted new questioning on
governance, and left a persistent legacy, for example in redefining the role of local sharia
court judges (cadis).
53 As shown by Donald Holbrook, The Al-Qaeda Doctrine: The Framing and Evolution of the
Leadership’s Public Discourse, Bloomsbury Press, London, 2014, and Thomas Hegghammer,
“The ideological hybridization of jihadi groups”, Current Trends in Islamist Ideology, Vol. 9,
November 2009, pp. 26–45.
The two groups studied here used concepts such as takfir in different ways to justify
warfare against other Muslims or to strengthen alliances in Kidal and Gao. For instance,
Ansar Dine did not have MUJAO’s history of regularly perpetrating suicide attacks,
attacking other Malian armed groups or mistreating detainees. Ansar Dine also banned
forced marriage and the taking of foreigners hostage, while MUJAO’s governance policy
pursued both. What explains these dramatically different choices despite similar ideol-
ogies? And what role did local notables have in influencing these outcomes? The history
of each group and their socialization mechanisms provide some clues.
4.4 DIFFERING SOCIAL EMBEDDEDNESS

AND SOCIALIZATION MECHANISMS
While ideologically similar, Ansar Dine’s and MUJAO’s orientation to local social net-
works and socialization mechanisms differed. Ansar Dine was primarily a Tuareg move-
ment with historical links to local tribal networks in Kidal. The longstanding relationship
of the group’s leaders with local communities led to heavy recruitment from within
these communities, further strengthening mutual bonds. The territory Ansar Dine
would come to control was also the only region of northern Mali with a Tuareg
majority.54 This explains why Ansar Dine gained control of Kidal with little resistance
from Tuareg separatists. By contrast, the more diverse, regional commercial hub of Gao
had a bitter history of ethnic competition and violence between the majority Songhay
people and local Tuareg and Arab groups. MUJAO exploited these divisions and aligned
itself with the majority Songhay people to displace the Tuareg separatists controlling
the city.
MUJAO’s membership was more diverse than Ansar Dine’s, integrating fighters from
across the region. The group therefore had weaker ties to Gao, the city it would come
to control. MUJAO’s diversity was also reflected in its structure, a looser alliance than
Ansar Dine’s. It built cohesion by running training camps based on al-Qaeda doctrine
and military training manuals. Drawing on commanders’ experience in Afghanistan
and Algeria, the group used long-established al-Qaeda training regimes as a regional
recruiting tool and the group’s primary socialization mechanism. The capacity of these
camps to reshape recruits’ identities and forge allegiance to the organization was evi-
dent after MUJAO took control of Gao. The group expanded its recruitment pool, enlist-
ing and training local ethnic groups, which quickly formed a significant part of the
group’s offensive force as fighters and suicide bombers. By contrast, Ansar Dine’s goals
of Tuareg nationalism resonated specifically with the Tuareg community’s long history
of rebellion, enticing local fighters and echoing existing narratives of injustices. Local
socialization processes were quickly assimilated, and new recruits easily integrated,
courtesy of the significant, existing cohesion forged by previous experience of fighting
together. Ansar Dine’s political consolidation among Tuareg tribes also brought with it
the military benefit of uniting these small-scale groups.
54 International Crisis Group, Mali: Security, Dialogue and Meaningful Reform, Africa Report
No. 201, International Crisis Group, Brussels, 11 April 2013, p. 11.
P. Robert/Corbis Historical/Getty Images

4.5 S
OURCES OF INFLUENCE, AUTHORITY
AND LEGITIMACY
In both cases, local actors had significant influence over Ansar Dine’s and MUJAO’s
fortunes. Yet, as the power and authority of local actors fluctuated, so too did their
influence and that of the rules and social norms governing them. At one point, each
armed group relied heavily on community notables to achieve dominance – a move that
bore starkly different outcomes. Achieving dominance is an extremely complex pro-
cess, informed by local Islamic traditions, a history of local ethnic conflict, contested
control over trafficking routes, fragile military control, and community perceptions
of the Islamic ideology advanced by jihadist groups. Hence, it is useful to explore how
three types of actors leveraged their sources of influence and authority to shape events
differently, despite the strong influence of al-Qaeda on the two groups’ language and
policies.
LOCAL BUSINESS ELITE

To displace Tuareg separatists and rule, both armed groups relied, to different degrees,
on the support of local business elites. In Gao, the pre-war economy was booming,
bolstered by cross-border trade and investments by traffickers (one of Gao’s fast-
developing neighbourhoods was even named “Cocaïnebougou” – Cocaine Town). The
Tuareg separatist rebellion threatened these commercial interests, as well as Songhay
political dominance. In response to the separatists’ alleged behaviour of mass looting
and rape,55 the business elite, dominated by Arab traders, established a council to nego-
tiate with the armed groups occupying Gao.56 In this period, where neither armed group
controlled the city, each group alternatively cajoled, intimidated or victimized popu-
lations to try to achieve dominance. The groups’ fierce competition over support from
the community and business council demonstrates the importance of these influences.
55 Human Rights Watch, Mali: War Crimes by Northern Rebels, Human Rights Watch, Bamako,
30 April 2012.
56 The business group, known as the Conseil de Concertation des Sages, was set up to
communicate the population’s needs to armed groups occupying the city.
Eventually the Tuareg separatists’ undisciplined violence led the business elite to sup-
port MUJAO – financially and politically – in taking control of Gao, secure businesses
and protect the population. In Kidal, commercial interests were less powerful but still
played a supportive role in Ansar Dine’s consolidation of control.
LOCAL ISLAMIC SCHOLARS AND LEGAL INSTITUTIONS

Both Ansar Dine and MUJAO used Islamic norms and al-Qaeda methodology (manhaj)
to produce a governing system regulating the use of force against perceived enemies
and non-combatants. In both cases, norm formation was based on Islamic scriptural
sources and “inferred” sources that constitute the basis of Islamic jurisprudence (fiqh).
However, the implications of fiqh require interpretation through discussion and debate.
In Kidal and Gao, this was done through shura (consultation) councils. These councils’
decisions shaped public administration and the rules of daily life for those living under
the groups’ control. In Gao, debates also occurred in Islamic tribunals, enabling greater
popular participation. Significant differences emerged in how Ansar Dine (Kidal) and
MUJAO (Gao) drew upon or undermined existing legal systems.
In Kidal, Ansar Dine relied on the existing Islamic legal system to interpret rules around
local governance. This legal system included the local and centuries-old sharia judges
(qadis) originating from the Kunta tribe. Their influence had a tangible impact on
how rules evolved: for example, it was these judges who endorsed the decision not to
apply corporal punishment in Kidal. Based on interviews conducted for this study, the
group’s relative restraint, when compared with MUJAO’s, was strongly attributed to its
community links and the moderating voice of the local qadis. Yet, the strength of this
restraint deteriorated outside of Kidal and over time. There was also a dramatic shift in
the group’s behaviour following the French military’s arrest of alleged sympathizers of
Ansar Dine, camp searches and restricted access to some wells, among other actions.
In contrast, MUJAO undermined Gao’s existing judicial system. It appointed new judges,
causing confusion and competition over who had the legitimacy to make decisions.
Local religious leaders, drawing upon indigenous Islamic tradition had some success at
altering what they saw as a “wrong” application of sharia. Once MUJAO achieved com-
plete control, however, it established a set of rules (see box). Debate on how the rules
should be interpreted was stifled, with most of the authority in that regard delegated
to Islamic police. This led to sporadic resistance and, in turn, brutal repression by the
police chief. Local actors, including qadis, lost most of their influence. Yet the extent
to which they continued to be targets of intimidation and bribery suggests that MUJAO
was still concerned by their opposition.
GLOBAL SALAFI-JIHADI SCHOLARS

MUJAO and Ansar Dine were also influenced by the global al-Qaeda network, largely
mediated through the regional branch known as al-Qaeda in the Islamic Maghreb
(AQIM). AQIM did not have direct command and control over these groups but reached
them through strategic guidance and training. The arrangement was mutually bene-
ficial: while association with the groups increased AQIM’s regional influence, MUJAO
and Ansar Dine benefited from al-Qaeda’s regional and global legitimacy. The fluid
and changing nature of these “networks of networks” illustrates the extent to which
the relationships thus formed are political and strategic and highlights the front-line
role of sub-commanders in decision-making. Moreover, the three groups’ desire to
maintain distinct identities is suggested by their 2017 alliance as opposed to a merger.
Importantly, influence is often reciprocal – local groups such as MUJAO and Ansar Dine
can also have an impact on the character, policy and leadership of the umbrella group
and its regional branches.
MUJAO’s rules in Gao

In an interview, a Gao resident described what it was like living under MUJAO rule:
Anyone caught smoking one cigarette was punished with one lash; two cigarettes, two
lashes; ten cigarettes, ten lashes.
If you had a cardboard box with a photograph on it, the box would be torn apart. Rice
bags printed with a photo had to be covered.
The rules had to be obeyed without question: women had to wear the veil; shops had
to close on Fridays by the first call to prayer; the sale and consumption of cigarettes
were strictly prohibited; thieves had their hands cut off; unmarried men or women
who committed fornication received a hundred lashes; adulterers would be punished
by stoning; and highway bandits had their right hands and left feet cut off.
This last sanction was applied to young men from Fafa (a constituency of Gao on the
road to Niamey, Niger), who indulged in the practice of stopping vehicles heading to
Niger and robbing the passengers.
4.6 IMPLICATIONS
As these cases illustrate, decisions by commanders of decentralized groups can be
shaped by local, regional and global actors. In mapping sources of restraint, it is neces
sary to understand these different types of authority and levels of influence, as well as
the networks linking key commanders and their constituencies.
Our research also demonstrates the value of taking into account the principles and
world view of armed-group members. In the ICRC’s experience, familiarity with an
armed group’s history and ideological references is essential for effective dialogue with
that group. However, this is complicated by the fact that many groups operate within
broader movements, requiring coordinated engagement at the local, national, regional
and global levels.
In addition, the ICRC engages with the broader community of Islamic scholars to high-
light common ground between Islam and IHL and discuss issues of humanitarian con-
cern. This dialogue is essential to building mutual understanding and a prerequisite
for efforts to influence behaviour. This dialogue must be neutral if it is to be effective:
any attempts by States to co-opt and instrumentalize this engagement in the guise of
“countering violent extremism” are likely to undermine it.
G. Tomasevic/Reuters
CHAPTER 5
COMMUNITY-
EMBEDDED
ARMED GROUPS
Examples:
Arbakee militias (Afghanistan)
vigilante groups in Nigeria
INTEGRATED STATE CENTRALIZED NON-STATE DECEVNTRALIZED COMMUNITY-EMBEDDED
ARMED FORCES ARMED GROUPS communal fighters in Papua
NON-STATE ARMEDNew Guinea
GROUPS ARMED GROUPS
some Mai-Mai groups (Democratic Republic of the Congo)
5.1 CHARACTERISTICS
The key characteristics of community-embedded armed groups are:
•• Comprise 10–50 young men (and occasionally women) from a local community
•• Formed to defend community interests
•• Flat hierarchical structure; leaders are elected or nominated by the group and can
change frequently
•• Mobilization is not necessarily decided by the group, but by authority figures in
the community (traditional, religious or governmental)
•• Initiation rituals (coming of age ceremonies) and visible markers (scars, tattoos,
body ornaments) forge group cohesion
•• Mobilization is temporary, with group members returning to community life
between bouts of violence; may find themselves instrumentalized in support
of a party to a conflict
•• Codes of conduct are unwritten and reflect local values, customary laws and
traditions.
Members of community-embedded armed groups are usually young men initiated into
a (defensive or offensive) fighting role on behalf of their communities. They do not
remain mobilized as an armed group, but resume roles in the community between
engagements. Lacking the organizational structure and responsible command neces-
sary to be considered an armed group under IHL, community-embedded armed groups
have received less attention from the ICRC than those that do. However, their growing
number and the humanitarian consequences of their armed violence necessitates a bet-
ter understanding of their norms and sources of influence.
The wide variation in types of community-embedded armed groups makes it difficult

to generalize: there are few obvious commonalities between the Arbakee militias in
Afghanistan, vigilante groups in northern Nigeria, youth gangs in Central America and
communal fighters in Papua New Guinea’s highlands. This chapter will therefore focus
Community-embedded armed groups 55
on one type of embedded group – South Sudan’s titweng, gelweng and gojam armed
cattle-keeping groups – to identify characteristics and sources of influence that might
shed light on similarly structured groups elsewhere.57
5.2 NORMS OF RESTRAINT AMONG

THE CATTLE-KEEPING GROUPS
OF SOUTH SUDAN
South Sudanese living through the conflict unleashed in December 2013 claim that the
overall level of violence witnessed over the past five years represents a break from
previous ethical norms of restraint. Although the killing of non-combatants, including
women, children and aid workers, did occur previously, it is the scale and frequency of
the sexual violence, killing and mutilation of children and destruction of villages and
health facilities that are unprecedented. The titweng and gelweng militarized cattle-
keeping men from Dinka communities and the gojam or White Army cattle-herders
from Nuer communities have been clearly implicated in this violence.
Both the Dinka and the Nuer communities of South Sudan have codes of conduct that
regulate behaviour during violent conflict.58 Spiritual protection from ancestors and
deities was thought to depend on adherence to these codes. Fear of deadly pollution
such as “leprosy” among the Dinka and of nueer (spiritual contamination) among the
Nuer dissuaded breaches. Violations, when they occurred, were dealt with through
religious ceremonies often involving cattle sacrifices. Since their creation a hundred
years ago, customary courts have also actively addressed the consequences of violating
codes of conduct in war. These codes have shifted in recent decades, challenged by
youth using guns to access power or by political players manipulating and reinter-
preting traditional belief systems to their own advantage. One political leader in the
1990s, for example, argued that fighting against government forces did not carry the
same spiritual dangers as fighting against other communities, so there was no need for
restraint. This same argument has been used in the current conflict to justify attacks
against anyone allied with the government in Juba.
Although much of the violence witnessed over the last five years appears wanton and
chaotic, an impression reinforced by the lack of an obvious hierarchical structure
within the cattle-keeping community, there are different layers of authority seeking
to control it. Influence over the cattle-keeping groups is highly contested and fluid
and varies from group to group, making the identification of sources of influence at a
given moment challenging. But it is clear that politico-military elites use this image of
uncontrolled violence to shirk responsibility for arming and mobilizing these groups
and to resist calls to rein them in.
57 This chapter is based on research led by Naomi Pendle in South Sudan in 2016 and 2017. Not
all groups mentioned here are active in countries in which an armed conflict is occurring.
58 See Francis M. Deng, Tradition and Modernization: A Challenge for Law among the Dinka
of the Sudan, Yale University Press, New Haven, 1971; Francis M. Deng, Customary Law in
the Modern World: The Crossfire of Sudan’s War of Identities, Routledge, Abingdon, 2010;
Edward E. Evans-Pritchard, The Nuer: A Description of the Modes of Livelihood and Political
Institutions of a Nilotic People, Oxford University Press, Oxford, 1940; Sharon E. Hutchinson,
Nuer Dilemmas: Coping with Money, War, and the State, University of California Press, Berkeley,
1996; Jok Madut Jok and Sharon E. Hutchinson, “Sudan’s prolonged second civil war and
the militarization of Nuer and Dinka ethnic identities”, African Studies Review, Vol. 42, No. 2,
September 1999, pp. 125–145.
5.3 SOCIALIZATION MECHANISMS

The strong socialization processes of young men into these cattle-keeping groups
explains why influence over the group is so fluid and contested: although men can be
mobilized to fight in their hundreds or even thousands,59 their loyalty remains to the
small group whose childhoods they shared in the cattle camps (wuts) and the broader
community to which they belong. Cattle are essential to the livelihoods and mech
anisms of justice and order for both Dinka and Nuer, giving the cattle keepers a cen-
tral and esteemed role in the community. Boys are sent to the cattle camps at a very
young age, and are socialized through songs and stories into the norms and boundaries
of permissible behaviour and the spiritual and physical dangers of combat. Initiation
into age-sets or a broader category of adulthood often involve scarification, bonding
members of a cattle camp and marking their transition to community defenders. Over
the last few years, certain cattle-keeping groups have adopted “uniforms” or other
markers to distinguish themselves from others: some groups remove their t-shirts and
tie them around their arms,60 others have specific t-shirts made. Amongst the gelweng,
certain haircuts designate group membership.
The cattle camps appoint leaders democratically, with the most popular rising to the
top. This creates room for discussion and debate, and leaders can change at any time.
The leader has the authority to resolve small disputes, make decisions over the move-
ment of cattle, and represent the cattle camps among chiefs and governing author
ities. Most importantly, the leader will decide when and how the group fights, and will
participate in combat himself. But such decisions are subject to intense pressure from
different sources of authority.
Targeting status, not role

There has been a rise among some cattle-keeping groups in the number of targeted
revenge killings of non-combatants, particularly educated wage-earners, in their
homes or offices and in town. This is facilitated by the availability of guns: the use
of spears requires a larger group and might therefore spark a larger battle. Such
assassinations often follow a sociocentric logic of justice whereby revenge for an
individual’s act falls upon his or her group. Thus, if a gelweng’s family member
is killed, he will seek revenge against a member of the perpetrator’s family of an
equivalent value to the person killed, regardless of the intended target’s innocence.
Since education and a wage income are valued, educated and/or employed family
members are increasingly targeted. Aid organizations being a major employer in
South Sudan, many aid workers have been killed, not because of their affiliation with
the aid organization but because of family ties.
59 In 2014, some 4,000–5,000 gojam fighters allegedly marched on the town of Bor.
60 Interview with a titweng conducted in Dinka by a South Sudanese researcher, Greater Gogrial,
15 January 2016; Naomi Pendle, “‘They are now community police’: Negotiating the boundaries
and nature of the government in South Sudan through the identity of militarised cattle-keepers”,
International Journal on Minority and Group Rights, Vol. 22, No. 3, July 2015, pp. 410–434.
5.4 SOURCES OF INFLUENCE AND AUTHORITY

Authority is contested locally and nationally, inside and outside armed groups, between
national military and political leaders, local government administrators, chiefs and
religious leaders. All of these draw on different sources of influence, including commu-
nity history, ethnic identity, cultural norms and spiritual beliefs. Authority can change
over time, differ from one community to another, and might not be obvious to out
siders. Here we discuss three sources of influence.
POLITICO-MILITARY ELITES
The Sudan People’s Liberation Army (SPLA) has been arming and mobilizing the cattle-
keeping groups since the 1980s in response to attacks from Khartoum-backed northern
militias.61 It explicitly sought to break the strong ties binding men from the same com-
munities by banning age-sets among the titweng and formally incorporating the cattle
keepers into larger military formations. The titweng’s identity was also reframed, with
limited success, as “community police”, to bring them under government control.62
The 2013 outbreak of violence saw cattle keepers armed and mobilized to fight for and
against the government, often but not exclusively along ethnic identity lines. The SPLA
in Opposition (SPLA-IO) attempted to order the gojam into an army-like structure,
with ranks equivalent to that of the SPLA, but the armed men were given little formal
training, and authority over them was inconsistent.
Some politico-military elites have used patronage to gain authority over the cattle
keepers. To guard their own large herds, many have created their own wuts from a
gathering of trusted families. They provide ammunition, animal vaccination and the
promise of health care and food to the young men, thereby binding them to their
service. Accordingly, these powerful elites can steer behaviour towards violence or
restraint: for example, instructions to ensure that the herd grows larger incites cattle
raiding and combat, while inter-Dinka disputes between the Apuk and Aguok over rich
grazing land in 2009 were stopped by a powerful cattle owner.
CHIEFS
Dinka and Nuer chiefs have varying levels of control over their communities’ cattle
keepers. Nuer chiefs hold arguably less sway than their Dinka counterparts for two
reasons: the competing influence of Nuer prophets, discussed below; and the asso‑
ciation of western Nuer chiefs with the government owing to a rapid succession of chiefs
since the 1990s appointed by government authorities in a bid to exert local influence.
Dinka chiefs strongly influence the titweng’s and gelweng’s behaviour in two ways.
First, chiefs act as intermediaries between leaders of the cattle keepers and government
officials. As such, they can assert their opinions on requests made. Second, they preside
over the customary courts, deciding how cattle will be redistributed in compensation
claims and thereby pronouncing on the rights and wrongs of community behaviour.
To settle disputes between cattle keepers, the chiefs’ courts will even accompany larger
wuts during the dry season as they migrate over big distances in search of water. An
ad hoc chief’s court was established near Rumbek in 2012, for example, to end lethal
cattle-raiding between gelweng. In their rulings on compensation for deaths incurred
during one raid, the chiefs imposed a much harsher penalty for men who did not issue
a warning prior to the raid. It is an established norm that warnings be issued to allow
women, children and other non-combatants to flee.63
61 Luka B.D. Kuol, “Dinka youth in civil war: between cattle, community and government”,
in Victoria Brereton (ed.), Informal Armies: Community Defence Groups in South Sudan’s Civil War,
Saferworld, London, February 2017, pp. 19–26.
62 Naomi Pendle, “‘They Are Now Community Police’”.
63 Interviews with Dinka chiefs, Kuajok, 2012.
DIVINE AUTHORITY
Divine authority figures can be very powerful influences over cattle-keeping groups,
encouraging violence or restraint. Community members and high government officials
alike seek their guidance, and they can bestow protection on fighters in elaborate cere
monies before battle. One important Nuer prophetess was instrumentalized by politico-
military elites for this reason, receiving generous cattle donations in exchange for the
mobilization of gojam fighters.
Both the Dinka and Nuer believe strongly that nueer, or spiritual pollution, ensues from
the violation of certain rules. The Dinka bäny bith (masters of the fishing spear) and
the Nuer prophets and prophetesses are thought to hold the power, through sacrifice
and petition to the divinity, to cleanse fighters of nueer. Through this process, these
figures can reassert norms of restraint. Prophets have long played an important role
in promoting a moral order amongst western Nuer communities.64 Since the 1960s,
Gatdeang Dit – a Bul Nuer prophet of the divinity Deng – has used his influence to
prevent Nuer armed cattle keepers from carrying out violent raids against their Dinka
neighbours. Meanwhile, powerful bäny bith have used the threat of curse to influence
the behaviour of Dinka cattle-keeping groups.
The power of these spiritual authorities suffers, however, from two weaknesses. First,
a divinity’s presence in a prophet is not permanent. Ongoing empirical evidence of its
power is needed, and many prophets have lost their authority after failing to protect
cattle from a raid. Second, some gojam fighters feel so polluted by nueer that restraint
seems pointless. Three decades of war have divided the Nuer to such an extent that
a man might face his own brother in battle. Engulfed by the dangers of nueer that
this represents, many feel that they are beyond redemption and see no need to act in
accordance with traditional norms. In the words of one gojam: “Nuer are fighting for
the government that is contaminated by nueer … In this fight, Nuer are fighting along-
side the government and are killing their own people. They do inhuman things like
raping and torturing their own people. It is like they are cursed.”
5.5 CONTESTING AND REINTERPRETING

COMMUNITY NORMS
Over the last few years, the way in which communities have argued over the legitimacy
of violence against women and children offers some insight into the variety of beliefs
influencing the different groups, and their openness to debate.
Interviews with gelweng showed a strong reluctance to kill women and children. The
latter’s weakness would reflect badly on the killer, and the former’s “motherly” role
and “universality” would render their killing morally wrong (for women, not being
part of the patrilineal line, can move between families, clans and tribes). The gelweng
also wish to avoid the eruption of an unending feud should restraint be removed. It is
worth noting that reluctance to kill women does not preclude sexual violence: in the
words of a gelweng in January 2017, “We go at night. If we get a man we kill him. If it’s a
woman, we rape her. Then her husband will come to fight.” Previously, fear of spiritual
pollution (rot) for raping a married woman, which increases the chance of death during
battle, incentivized restraint. But some gelweng have reinterpreted this norm to justify
raping unmarried women – often still girls.
64 Douglas H. Johnson, Nuer Prophets: A History of Prophecy from the Upper Nile in the Nineteenth
and Twentieth Centuries, Clarendon Press, Oxford, 1994.
Among the titweng, by contrast, community debates over the legitimacy of killing
women and children arose after the practice appeared in the 2016 and 2017 conflict
(it had been absent during inter-Dinka disputes between Apuk and Aguok in 2005 and
2009). The killing of two children during a dawn cattle raid by Aguok titweng provoked
a tit-for-tat that saw several woman and children killed over rounds of fighting. Both
Nuer and Dinka believe that immortality is secured through future generations and so
that killing children (and women as child-bearers) brings lasting, total death. National
politicians joined debates in their home communities, with some arguing for restraint
and others for retaliation, particularly where they had an interest in disputed grazing
lands. Many chiefs advocated restraint for fear that the continued killing of children
would ruin all hope of peace. Unfortunately, a customary court case failed to resolve the
moral ambiguity over these killings and fighting continued.
Interviews with gojam echoed the gelweng’s moral repugnance at the idea of kill-
ing women and children, with women again considered as “universal”. In the past,
women were spared since attackers could marry them. But the brutality of raids in 2015,
which saw the killing and mutilation of children, prompted leaders to redefine women
and children of the opposing group as “government” and therefore not deserving of
restraint. This included anyone living in government-controlled garrison towns. Some
of the Nuer prophets have pushed back on these reinterpretations, insisting on the
continuity of customary beliefs. One even turned his home into a sanctuary for children
from all parts of western Nuer, irrespective of their political lineage. Still, the debate
among the gojam and their communities over the legitimacy and spiritual dangers of
such killings continues, with many gojam feeling, as mentioned earlier, that they have
broken too many taboos to be cleansed of nueer.
Creative ways of discussing norms of restraint

For several communities in South Sudan, wrestling competitions form a rite of
passage into manhood. Passionate discussions around wrestling provide an excellent
opportunity to introduce notions of IHL. During a first-aid training session with a
cattle-keeping group, an ICRC staff member asked a couple of young men to explain
the rules of a wrestling match. After they had done so, he sought clarification on
who was a legitimate opponent, pointing to an old woman and asking if she could
be challenged to a match. Predictably, derisive laughter followed as the young men
responded that women, old men and children were deemed too weak to partake in this
noble sport. Parallels were then drawn with protected categories of people under the
rules of armed conflict.
F. Zizola/Noor
5.6 IMPLICATIONS
Community-embedded armed groups remain part of their communities and can be a
formidable fighting force. As a consequence, local, regional and national authorities,
whether political, social, religious or economic, may vie for control over when, and for
what, such groups are mobilized. Understanding these competing authorities, and the
local beliefs and traditions invoked to influence them, is an important step in identify-
ing potential levers of restraint. These include local religious and social leaders whose
influence might not be apparent to outsiders. Innovative approaches have proven effect‑
ive in strengthening norms of restraint.
Humanitarian activities such as cattle vaccination, first-aid training, health-care

services and family reunifications bring humanitarian agencies into direct contact
with many of the community authorities who have influence over the behaviour of
community-embedded armed groups. As illustrated in the box, these activities can
serve as a springboard from which to explore local views on violence and restraint and
on legitimate and illegitimate targets. ICRC staff members with the least formal edu-
cation might be the best sources of information on and analysis of how to reinforce
norms of restraint and better respect for humanity among these groups.
ICRC
CHAPTER 6
IMPLICATIONS
AND
APPROACHES
This report has demonstrated that armed forces and armed groups vary significantly
in their organizational structure, capacities of command and control, socialization
mechanisms and openness to external influence. It has shown that patterns of violence
and restraint may differ both between and within armed forces and armed groups.
The research also found that the number of competing influences over armed groups
increases with the extent of decentralization and community-embeddedness of a
group. Sources of influence also change over time and in response to events. These
findings suggest that a detailed understanding of the inner workings of armed groups
is a prerequisite for identifying the sources of authority, the beliefs, the traditions and
the people steering their behaviour towards violence or restraint.
For the ICRC, gaining such understanding requires the broad, multidisciplinary par-
ticipation of its resident and mobile staff and the democratization of responsibility for
analysing armed groups. It is not only those with a military background or training
in negotiation who speak to armed forces and armed groups; frequently it is gener-
alists and technical specialists running programmes to repair water supplies, vaccin
ate livestock, distribute seed, restore livelihoods, establish health clinics or deliver
Red Cross messages who interact and build trust with communities and armed groups.
They need to understand the nature of the armed groups they encounter and contribute
their knowledge and experience to the analysis. Training manuals on engagement with
armed groups are predominantly based on rational actor models, identifying leverage
points based on their supposed economic or political interests.65 But findings from this
study demonstrate that behaviour is also shaped by values, traditions and ideology,
and communities debate acceptable and unacceptable behaviour. Understanding local
viewpoints and values starts by a deeper engagement with communities themselves.
The disparities between the different kinds of armed forces and armed groups revealed
in the research demonstrate the futility of a one-size-fits-all approach to improving
their compliance with humanitarian rules and principles. We have seen that organiza-
tional structure has a bearing on the sources of influence on armed forces and armed
groups. This chapter therefore proposes a “blueprint” for each category of armed group
based on its organizational structure. These blueprints aim to guide reflection on which
type of approach might be appropriate for which category. But first, we will review the
study’s overall findings and what these might mean for the ICRC and other humani-
tarian agencies.
6.1 MAIN FINDINGS AND IMPLICATIONS

On the ICRC’s current “integration approach”
The research found that the ICRC’s approach to instilling norms of restraint in
vertically structured State armed forces and non-State armed groups remains
highly pertinent. In State armed forces, the intensity of training and its adapta-
tion to the specificities of the audience increase its effectiveness. In centralized
non-State armed groups, ideology and its inculcation through all-encompassing
socialization practices shape behaviour to a large extent, suggesting that the
approach of the ICRC and Geneva Call,66 among others, to gain the commitment
of armed-group leaders to humanitarian norms continues to be relevant.
65 Gilles Carbonnier, Humanitarian Economics: War, Disaster and the Global Aid Market, Hurst,
London, 2015, p. 20.
66 See Geneva Call’s website for details on how the organization encourages non-State
armed groups to sign a “Deed of Commitment” to respect specific humanitarian norms:
https://genevacall.org/how-we-work/deed-of-commitment/
Implications and approaches 65
However, the research also found that the informal socialization processes of
the peer group can have as strong an influence on behaviour as formal mech-
anisms like training, and thus can strengthen or undermine adherence to IHL.
Hence, the ICRC could enhance its approach by gaining a better understanding
of those socialization processes and by considering ways to address informal
codes and practices that do not align with formal rules, such as marching songs
glorifying sexual violence.
On influencing behaviour in decentralized and

community-embedded armed groups
While it is difficult to extrapolate on findings from such a small sample of armed
groups, it seems clear that the behaviour of members of these groups is not as
chaotic or uncontrolled as often depicted; that there are clear sources of influ-
ence on their behaviour; and that the more decentralized the group, the more
these sources are external to the group. The Mali and South Sudan cases high-
light that there are competing sources of authority seeking to control the use
of violence at the local, regional and global levels and that these shift over time
and in accordance with events. This presents more entry points for dialogue on
behaviour but dilutes the impact of any one source on the armed group.
The complexity of alliances among many of the armed groups to have emerged
in the last decade – their manifestation as “networks of networks” – requires
a long-term vision of sustained dialogue with those identified as having influ-
ence on violence and restraint at a particular time. Sowing doubt in the minds of
armed-group members about their use of violence has been found in the ICRC’s
practice to temper behaviour over time. This long-term approach necessitates a
coherent strategy and strong institutional memory to compensate for shorter-
term staff turnover.
On the primacy of legal arguments

This study and the ICRC’s experience suggest that across all types of armed
groups, an exclusive focus on the law is not as effective at influencing behaviour
as a combination of the law and the values underpinning it. Linking the law to
local norms and values gives it greater traction. The role of law is vital in setting
standards, but encouraging individuals to internalize the values it represents
through socialization is a more durable way of promoting restraint. A downward
spiral of reciprocal IHL violations seems less likely to occur if norms of IHL
are intrinsic to a combatant’s honour. Thus, identifying historical and contem
porary references that resonate in local contexts enhances the persuasive power
of arguments for restraint.
On understanding armed groups

The research suggests that the organizational structure of an armed group pro-
vides important clues to the sources of influence on the behaviour of its mem-
bers. Analysing the patterns of violence of armed forces and armed groups – the
type of violence, and for each type, the target, frequency and method used (see
Chapter 1) – can in turn shed light on questions of command and control and
help identify where and when restraint is exercised. Monitoring instances of
restraint can in turn spark inquiry into why restraint was shown in one context
and not another, potentially broadening understanding of the dynamics and
personalities at play. Distinguishing between violence as a “policy” as opposed
to opportunistic “practice” can steer decisions over the appropriate level at
which to direct dialogue.
The following questions might help to provide further insights into an armed
force’s or armed group’s organizational structure:
1. Is there an obvious leadership that issues orders? Do those orders appear to
be followed?
2. Are there visible signs of hierarchy and discipline, such as uniforms,
saluting, consistent use of ranks across the group?
3. How does a group relate to a local community? Does it receive support
(political, social or economic)?
4. Who or what are the sources of influence over the group (political, social,
economic, spiritual or other)? How do they exert their authority?
5. Is the group’s ideology reflected in its practice?
6. How are the group’s rules socialized and reinforced in the group (e.g.
through training, rituals, speeches by influential figures, violent practices)?
7. Do units of a larger group exhibit rituals or practices that are at odds with
the group’s doctrine and stated purpose?
Not all armed groups will fit neatly on the spectrum of armed-group organ-
ization introduced in Chapter 1. Some are more akin to movements, with a
centralized core inspiring and guiding decentralized sub-groups with differ-
ing degrees of affiliation to the core. Analysing such alliances by examining
each member group individually advances understanding of their actual degree
of autonomy. Individual groups, for example, might have different sources of
influence, requiring engagement strategies at the local, national, regional and
global levels.
On understanding civilian agency

The role that communities play in influencing the behaviour of armed groups
was touched on in several of the cases studied. Civilians living in communities
are not passive entities but can influence armed-group behaviour in favour of
violence or restraint. In Colombia, cohesive communities with strong leaders
forged self-protection strategies that shielded them from armed violence and
abuse. But communities can also stimulate violence: in both Colombia and Mali,
communities called for violence against those seen to be jeopardizing business
interests or deemed socially undesirable.
A deeper engagement with communities can enhance understanding of mech-

anisms of restraint. Many opportunities present themselves in the course of
humanitarian activities to open dialogue with community members on their
norms of restraint (see the example of wrestling in Chapter 5). But in their
desire to support community self-protection initiatives, humanitarian organ
izations need to be mindful of the potential ramifications of such efforts. Inter-
ventions by outsiders can change and possibly undermine community dynamics;
aid projects can attract negative attention from armed groups; and supporting
the organization and advocacy of social groups is an inherently political process
that may harm perceptions of neutrality.
On trust as a prerequisite for engagement

The ability of humanitarian organizations to engage with, and try to influence
the behaviour of, armed forces and armed groups depends to a large extent on
the trust others have in the organization’s purposes and practices. Commanders
are unlikely to meet with, let alone listen to, representatives of organizations
they distrust. Rationalists might argue that the prospect of goods and services
in the regions under their control and potential legitimacy accrued from inter-
actions with aid organizations would outweigh considerations of trust.67 And
this might have been true of many armed groups in the two decades following
the end of the Cold War. But the reduced access by aid organizations to areas
controlled by non-State armed groups (and some governments) over the past
decade – through an absence of security “guarantees” and/or rejection of a
“Western” humanitarian presence – attests to a changing environment.
The importance of trust was strongly reflected in a survey undertaken for this
project among ICRC staff who engage with non-State armed groups. The survey
asked what respondents considered to be the most important factors deter-
mining the acceptance of the ICRC’s humanitarian action in areas controlled
by non-State armed groups. Fifty-one per cent of responses related to issues
of trust in the principles and working methods of the organization (such as
neutrality, independence, confidentiality and predictability), with only 21 per
cent stating reasons related to what the non-State armed group sought for itself
(such as goods and services for populations living in its territory, legitimacy,
medical supplies, and first-aid training for its fighters). Respondents empha-
sized the extent to which armed groups warn the ICRC that they are watching
that the organization’s actions are consistent with its claims, not just locally
but across different contexts. This suggests a need to further strengthen the
coherence and consistency of the ICRC’s humanitarian credentials in the prac-
tical and digital spheres.
6.2 BLUEPRINTS FOR ENGAGEMENT

The following “blueprints for engagement” summarize the main insights from the
research on each category of armed force or armed group, and the questions we should
ask ourselves when designing an engagement approach. They constitute a starting
point for analysing armed groups without being a substitute for detailed and context
ualized examination of their particularities.
67 For the various ways armed groups benefit from humanitarian action, see Fiona Terry,
Condemned to Repeat? The Paradox of Humanitarian Action, Cornell University Press, Ithaca and
London, 2002.
INTEGRATED STATE ARMED FORCES (Chapter 2)
CHARACTERISTICS INSIGHTS
• Strictly hierarchical decision-making and authority The intensity of training in IHL (frequency, methods)
• Codified, observable rules that are consistently applied makes a difference to battlefield conduct. The trainer
• Observable signs of discipline (professionalism in uniforms,
must be credible with the audience, whether through
saluting, routines)
• Separation from civilian life when on duty
experience or expertise.
SOURCES OF AUTHORITY AND RESTRAINT Training effectiveness is best tested under battlefield-like
Senior leadership conditions.
Junior officers and non-commissioned officers
Doctrine, standard operating procedures, rules of engagement Norms of restraint need to be reinforced at critical
and informal norms and values moments by the immediate superior.
Threat of punishment
SOCIALIZATION PROCESSES Formal socialization can be reinforced or undermined by

• Formal training, hierarchy and discipline informal socialization processes.
• Informal values and rituals (e.g. hazing, marching songs)
Norms of restraint are more likely to hold if they are

internalized as part of a soldier’s identity – beyond “it is
against the law” to “it is not who we are”.
CONSIDERATIONS APPROACHES
What events, legends, personalities and values form part Advise and assist in the integration of IHL into national
of the armed force’s identity? How do these shape formal laws and into military doctrine at all levels.
and informal socialization?
Assist in the development of IHL training tailored to the
How much influence do junior and non-commissioned audience. Find references that resonate with participants.
officers have on unit members’ behaviour and Recommend that training be tested under duress.
viewpoints?
Promote the socialization of values related to IHL by
What intersecting identities (e.g. religious, ethnic) supporting its integration into organizational culture.
do members of the armed force have? Do they create
other entry points for messages on restraint? Track patterns of violence and identify instances of
restraint. Investigate the sources of influence on restraint.
Do monitoring mechanisms weaken with distance from Distinguish between violence as a policy and as a
central command? How does this affect behaviour? practice.
What profile of trainer would be most credible with Encourage States allying with other State and non-State
particular audiences? forces to ensure that their partners socialize norms of
restraint among their soldiers or fighters.
CENTRALIZED NON-STATE ARMED GROUPS (Chapter 3)
• Leadership exercises tight command and control over Groups espouse an elaborate doctrine or ideology that
subordinates through a strict hierarchy, but monitoring specifies goals. They regularly publish or broadcast the
mechanisms may be weak group’s ideas and values to a wider public.
• A prominent doctrine or ideology outlines goals, approaches
and world view

• Observable signs of discipline (professionalism in uniforms, The rules stipulate the parameters and targets of
saluting, routines) permissible violence.
• Isolated from civilian population (housed in camps or barracks)
A weak capacity to monitor the behaviour of fighting

SOURCES OF AUTHORITY AND RESTRAINT
units leaves unit commanders with scope to interpret
Senior leaders and commanders of sub-units how norms are understood and applied.
Group ideologues and codes of conduct
Ideology, codes of conduct, discipline
Threat of punishment Group loyalty is forged through intense socialization
practices that aim to reshape members’ identities.
SOCIALIZATION PROCESSES
• Immersive regime (e.g. controlling all aspects of the daily routine)
• Initiation rituals and informal bonds
What is the group’s ideology? What does its code of Track patterns of violence and identify instances of
conduct say about violence and restraint? Where are the restraint. Investigate the sources of influence on restraint.
overlaps with IHL? Distinguish between violence as a policy and as a
practice.
Who articulates or interprets the group’s doctrine or
ideology? Discuss parallels between the group’s doctrine and IHL,
and seek further alignment.
How are group beliefs and rules socialized among
members? Discuss with the leadership any disparities between the
rules and observed behaviour. Advise on ambiguities that
Are there variations in patterns of violence between allow different interpretations of the rules.
different units of the same group? What does this convey
about command and control? Discuss with the leadership the informal norms that may
undermine formal rules, and the strength of monitoring
What is the relationship between the armed group and mechanisms.
local communities? Are communities able to resist being
drawn into the conflict? Discuss with communities ways in which they engage
with an armed group and how they shield community
What profile of trainer is most credible with particular members from violence and recruitment.
audiences?
DECENTRALIZED NON-STATE ARMED GROUPS (Chapter 4)
• Fluid alliances of small armed groups The more decentralized the armed group, the more its
• Individual commanders retain decision-making power over behaviour is influenced by sources external to the group.
group members
• Units may break away to join new associations, without
The conduct of individual units depends heavily on the
compromising group cohesion
• Multiple decentralized groups can work in a broader movement, commander’s preferences.
giving local, regional and global reach
• Loose coordination within the alliance, including in military Groups are integrated into local social networks
planning and operations (e.g. communities, local notables) and can retain links
• Few observable signs of military discipline
to regional or global armed groups.
SOURCES OF AUTHORITY AND RESTRAINT
The influence of local actors on the behaviour of the
Unit commanders
Local business, religious or cultural elites armed group fluctuates over time and in response to
Senior leadership events.
Ideological and religious texts
Threat of punishment Group values and rules can promote restraint, even in
the absence of monitoring systems.
• Extremely varied
• Can be based on local culture and customs
• Could include military and ideological training
• Strong informal socialization in the peer group
How does the alliance of armed groups fit together? Track patterns of violence and identify instances of
What is the nature of the relationships between restraint. Investigate the sources of influence on restraint.
small-group leaders and alliance leaders? Distinguish between violence as a policy and as a
practice.
What is the relationship between the armed group and
the local community? Do community/business/religious Prioritize dialogue with local commanders. These may
leaders exert influence on armed-group behaviour? change regularly.
Does the group draw on socialization processes based on Develop a nuanced understanding of the most important
local customs or traditions (e.g. coming-of-age rituals)? sources of influence over an armed group’s behaviour,
noting the type of authority they draw on.
How has the influence of key actors in an armed group
changed over time, and why? What is the source of their Engagement strategies need to mirror the structure of
influence (e.g. religious, financial, political or social). the alliance, interacting at the local, national, regional
and global levels.
What are the customary rules on warfare? What parallels
are found in IHL? The ICRC must be consistent, predictable and
transparent in all that it says and does.
COMMUNITY-EMBEDDED ARMED GROUPS (Chapter 5)
• Comprise 10–50 young men, and in some cases women, Group members do not remain mobilized, but return to
from a local community their roles in the community.
• Formed to defend community interests
• Flat hierarchical structure
• Mobilized to fight by community notables
Community-embedded groups may not choose when,
or politicians where or how they fight.
• Initiation rituals forge group cohesion
• Mobilization is temporary Local, regional and national actors may compete for
• Codes of conduct are unwritten and reflect local values,
influence and control over such groups.
customary law and traditions
SOURCES OF AUTHORITY AND RESTRAINT Traditional norms regulating violence and restraint
Traditional leaders may be subject to community debate.
Local politicians
Local religious leaders The image of chaotic, uncontrolled violence by these
Local business elite groups may mask who is really in control.
Leaders of local youth fighters
Community norms and values
Community debates over interpretation of norms
• Community coming-of-age rituals
• Local religious and customary practices
How do community-embedded armed groups fit into their Track patterns of violence and identify instances of
communities? restraint. Investigate the sources of influence on restraint.
Distinguish between violence as a policy and as a
How do group leaders emerge? On what does their practice.
authority lie? What is the extent of their direct influence
over the group? Acquire a deeper understanding of how
community-embedded groups relate to different types
Who influences when and how a group fights? of local and national authority figures.
What are the customary rules on warfare? What parallels Promote restraint through community norms, customary
are found in IHL? law or other legal frameworks (e.g. IHL and Islam).
How does the ICRC engage with group members when Pursue a cross-sectoral approach to understanding and
they are in their community role? Can we use this engaging with communities.
engagement to indirectly discuss behaviour during armed
conflict?
FURTHER READING
Abu Hanieh, Hassan and Abu Rumman, Mohammad, The “Islamic State” Organization: The
Sunni Crisis and the Struggle of Global Jihadism, Friedrich-Ebert-Stiftung Jordan & Iraq,
Amman, 2015.
Al-Dawoody, Ahmed, “IHL and Islam: An overview”, Humanitarian Law & Policy [blog],
14 March 2017: http://blogs.icrc.org/law-and-policy/2017/03/14/ihl-islam-overview/
Atran, Scott, Talking to the Enemy: Faith, Brotherhood and the (Un)making of Terrorists,
Ecco – An imprint of Harper Collins, New York, 2010.
Bangerter, Olivier, “The ICRC and Non-State Armed Groups”, Exploring Criteria & Con-
ditions for Engaging Armed Non-State Actors to Respect Humanitarian Law & Human Rights
Law, Conference Report, Geneva Call, Geneva, June 2007.
Bell, Andrew M., “Military Culture and Restraint Toward Civilians in War: Examining the
Ugandan Civil wars”, Security Studies, Vol. 25, No. 3, 2016, pp. 488–518.
Bongard, Pascal and Somer, Jonathan, “Monitoring non-State actor compliance with
humanitarian norms: A look at international mechanisms and the Geneva Call Deed of
Commitment”, International Review of the Red Cross, Vol. 93, No. 883, September 2011,
pp. 673–706.
Bourke, Joanna, An Intimate History of Killing: Face to Face Killing in Twentieth Century
Warfare, Granta Books, London, 1999.
Chirot, Daniel and McCauley, Clark, Why Not Kill them All? The Logic and Prevention of
Mass Political Murder, Princeton University Press, Princeton, New Jersey, 2006.
Dickinson, Laura A., “Military lawyers on the battlefield: An empirical account

of international law compliance”, American Journal of International Law, Vol. 104, No. 1,
2010, pp. 1–28.
Geneva Academy, Reaction to Norms: Armed Groups and the Protection of Civilians, Policy
Briefing No. 1, Geneva, January 2014.
Gutiérrez Sanín, Francisco and Giustozzi, Antonio, “Networks and armies: Structuring
rebellion in Colombia and Afghanistan”, Studies in Conflict & Terrorism, Vol. 33, 2010,
pp. 836–853.
Gutiérrez Sanín, Francisco and Wood, Elisabeth J., “Ideology in civil war: Instrumental
adoption and beyond”, Journal of Peace Research, Vol. 51, No. 2, 2014, pp. 213–226.
Gutiérrez Sanín, Francisco and Wood, Elisabeth J., “What should we mean by ‘pattern
of political violence’? Repertoire, targeting, frequency, and technique”, Perspectives on
Politics, Vol. 15, No. 1, March 2017, pp. 20–41.
Hoover Green, Amelia, “Learning Restraint: The Role of Political Education in Armed
Group Behavior Toward Civilians”, Simons Papers in Security and Development, No. 30,
School for International Studies, Simon Fraser University, Vancouver, December 2013.
FURTHER READING 73
Hoover Green, Amelia, “The commander’s dilemma: Creating and controlling armed
group violence”, Journal of Peace Research, Vol. 53, No. 5, September 2016, pp. 619–632.
Humphreys, Macarten and Weinstein, Jeremy M., “Handling and Manhan-

dling Civilians in Civil War”, American Political Science Review, Vol. 100, No. 3,
August 2006, pp. 429–447.
ICRC, “Spared from the Spear: Traditional Somali Behaviour in Warfare”, ICRC Somalia
Delegation, Nairobi, February 1998: http://blogs.icrc.org/somalia/2015/09/21/
spared-from-the-spear/
International Crisis Group, Exploiting Disorder: al-Qaeda and the Islamic State, Special
Report No. 1/Jihad in Modern Conflict, 14 March 2016: https://www.crisisgroup.org/
global/exploiting-disorder-al-qaeda-and-islamic-state
Jackson, Ashley, “Talking to the Other Side: Humanitarian Engagement with armed
non-State actors”, HPG Policy Brief, No. 47, ODI, London, June 2012.
Jo, Hyeran, Compliant Rebels: Rebel Groups and International Law in World Politics,
Cambridge University Press, Cambridge, 2015.
Jowell, Marco, “Cohesion through socialization: Liberation, tradition and modernity in

the forging of the Rwanda Defence Force (RDF)”, Journal of East African Studies, Vol. 8,
No. 2, 2014, pp. 278–293.
Kalyvas, Stathis N., The Logic of Violence in Civil War, Cambridge University Press,
Cambridge, 2006.
Kaplan, Oliver, “Nudging armed groups: How civilians transmit norms of protection”,
Stability: International Journal of Security & Development, Vol. 2, No. 3, 2013, pp. 1–18.
Maher, Shiraz, Salafi-Jihadism: The History of an Idea, Oxford University Press, Oxford,
2016.
Manekin, Devorah, “Violence against civilians in the Second Intifada: The moderating
effect of armed group structure on opportunistic violence”, Comparative Political Studies,
Vol. 46, No. 10, October 2013, pp. 1273–1300.
Marks, Zoe, “Sexual violence inside rebellion: Policies and perspectives of

the Revolutionary United Front of Sierra Leone”, Civil Wars, Vol. 15, No. 3, 2013,
pp. 359–379.
Meijer, Roel (ed.), Global Salafism: Islam’s New Religious Movement, Columbia University
Press, New York, 2009.
Mitton, Kieran, Rebels in a Rotten State: Understanding Atrocity in the Sierra Leone Civil War,
Oxford University Press, Oxford and New York, 2015.
Oppenheim, Ben and Weintraub, Michael, “Doctrine and violence: The impact of
combatant training on civilian killings”, Terrorism and Political Violence, Vol. 29,
No. 6, 2017, pp. 1126–1148.
Parkinson, Sarah Elizabeth, “Organizing rebellion: Rethinking high-risk mobiliza-

tion and social networks in war”, American Political Science Review, Vol. 107, No. 3,
August 2013, pp. 418–432.
Ron, James, “Savage restraint: Israel, Palestine and the dialectics of legal repression”,
Social Problems, Vol. 47, No. 4, November 2000, pp. 445–472.
Shapiro, Jacob N., The Terrorist’s Dilemma: Managing Violent Covert Organizations,
Princeton University Press, New Jersey, 2013.
Siebold, Guy L., “The Essence of Military Group Cohesion”, Armed Forces & Society,
Vol. 33, No. 2, 2007, pp. 286-295.
Sinno, Abdulkader H., Organizations at War in Afghanistan and Beyond, Cornell University
Press, Ithaca and London, 2008.
Slim, Hugo, Killing Civilians: Method, Madness and Morality in War, Columbia University
Press, New York, 2008.
Staniland, Paul, Networks of Rebellion: Explaining Insurgent Cohesion and Collapse, Cornell
University Press, Ithaca and London, 2014.
Stanton, Jessica A., Violence and Restraint in Civil War: Civilian Targeting in the Shadow of
International Law, Cambridge University Press, Cambridge, 2016.
Straus, Scott, “Retreating from the Brink: Theorizing Mass Violence and the Dynamics
of Restraint”, Perspectives on Politics, Vol. 10, No. 2, June 2012, pp. 343–362.
Straus, Scott, Making and Unmaking Nations: War, Leadership, and Genocide in Modern
Africa, Cornell University Press, Ithica and London, 2015.
Weinstein, Jeremy, Inside Rebellion: The Politics of Insurgent Violence, Cambridge Univer-
sity Press, Cambridge, 2006.
Wood, Elisabeth J., “Armed groups and sexual violence: When is wartime rape rare?”,
Politics & Society, Vol. 37, No. 1, March 2009, pp. 131–161.
Thematic journal issues of particular relevance
Engaging armed groups, International Review of the Red Cross, Vol. 93, No. 883,
September 2011.
Generating respect for the law, International Review of the Red Cross, Vol. 96,
No. 895/896, Autumn/Winter 2014.
Sexual violence in armed conflict, International Review of the Red Cross, Vol. 96,
No. 894, Summer 2014.
Socialization and violence, Journal of Peace Research, Vol. 54, No. 5, September 2017.
Understanding armed groups and the applicable law, International Review of

the Red Cross, Vol. 93, No. 882, June 2011.
War in cities, International Review of the Red Cross, Vol. 98, No. 901, April 2016.
The ICRC helps people around the world affected by armed conflict and other violence, doing everything it
can to protect their dignity and relieve their suffering, often with its Red Cross and Red Crescent partners.
The organization also seeks to prevent hardship by promoting and strengthening humanitarian law and
championing universal humanitarian principles. As the authority on international humanitarian law, it helps
develop this body of law and works for its implementation.
People know they can rely on the ICRC to carry out a range of life-saving activities in conflict zones,
including: supplying food, safe drinking water, sanitation and shelter; providing health care; and helping
to reduce the danger of landmines and unexploded ordnance. It also reunites family members separated
by conflict, and visits people who are detained to ensure they are treated properly. The organization works
closely with communities to understand and meet their needs, using its experience and expertise to respond
quickly, effectively and without taking sides.
4352/002 06.2018 1500 Cover photo: A. Hofford/EPA
facebook.com/icrc
twitter.com/icrc
instagram.com/icrc
International Committee of the Red Cross

19, avenue de la Paix
1202 Geneva, Switzerland
T +41 22 734 60 01
shop.icrc.org
© ICRC, June 2018

4352 002 The-Roots-Of-Restraint WEB

Uploaded by

Copyright:

Available Formats

4352 002 The-Roots-Of-Restraint WEB

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

4352 002 The-Roots-Of-Restraint WEB

Uploaded by

Copyright:

Available Formats

THE ROOTS

Research protocols were approved by an ethical review board composed of Gilles

In addition, we would like to thank the UK Department for International Development

Chapter 1: Norms of restraint, organizational structure and socialization.............. 17

Chapter 2: Integrated State armed forces................................................................ 27

Chapter 3: Centralized non-State armed groups..................................................... 37

Chapter 4: Decentralized non-State armed groups ................................................45

Chapter 5: Community-embedded armed groups................................................... 53

Chapter 6: Implications and approaches.................................................................63

Accordingly, the ICRC’s approach has traditionally relied on established systems of

Professor Gilles Carbonnier

This publication, based on two years of research by a group of distinguished scholars,

PURPOSE OF THE STUDY

MORE ARMED CONFLICTS,

The number of non-international armed conflicts has more than

Armed conflicts are increasingly fought in cities, leading to heavy

Chapter 2 summarizes quantitative and qualitative research conducted by Andrew Bell

Chapter 3 presents the findings of research in Colombia led by Francisco Gutiérrez

Chapter 4 explores sources of influence on two jihadi groups in northern Mali,

11 Alex Leveringhaus, “Autonomous weapons mini-series: Distance, weapons technology and

Sparing the Kodok hospital: An example of restraint

Restraint can be conceptualized in different ways.14 In his analysis of why violence

Identifying genuine restraint is challenging since it is essentially demonstrating a

Figure 1: Targets of JASDJ violence 2009–2016.22

Second, analysing patterns of violence allows us to draw comparisons between groups

Measuring influence via social media

23 Elisabeth J. Wood, “Rape as a practice of war: Towards a typology of political violence”,

Sustaining an armed group is expensive, costing up to millions of dollars a year.25 To

Retaining control or fighter loyalty is also central to survival, as well as to military

Understanding organizational structure is important for humanitarian organizations

Figure 2: The spectrum of armed-group organization

INTEGRATED STATE CENTRALIZED NON-STATE DECENTRALIZED COMMUNITY-EMBEDDED

Table 1: Internal structure of armed organizations

Socialization is a process, and individuals can be socialized into committing violence

DISSEMINATION INTEGRATION SOCIALIZATION

2.2 METHODS OF FORMAL SOCIALIZATION

2.3 INFORMAL SOCIALIZATION

The norm of brotherhood – or “mateship” in the Australian Army – is consistently

Informal norms can be a double-edged sword, reinforcing or

Informal norms can thus be a double-edged sword, reinforcing or undermining offi-

Formal vs informal sources of influence

In a survey experiment, participants were asked to advise their commander on

2.4 EMPHASIZING ETHICS

The use of values as a socialization tool in aligning behaviour to organizational

No one has ever explained to me how a coward in barracks is a hero

This emphasis on organizational identity, “warrior’s honour” and ethical behaviour

There is no substitute for honor as a medium of enforcing decency on the

41 There is even a blog dedicated to the topic: https://www.lawfareblog.com

2.5 CHALLENGES TO SOCIALIZATION

Table 2: Comparison between former FARC-EP and ELN

In terms of organizational structure, the photos below highlight the differences in

3.3 SOCIALIZATION MECHANISMS

Communities employ several tactics to achieve this. First, by promoting a culture of

Second, strong communities can implement local conflict-resolution processes. This

Third, communities can establish local investigatory institutions to clarify accusations

The feasibility of these mechanisms largely depends on the strength of community

The role of communities in influencing armed-group behaviour warrants further

Decentralized non-State armed groups are generally composed of alliances of smaller

Decentralized armed groups constitute the majority of non-State

4.3 DIVERGENT PATTERNS OF VIOLENCE

4.4 DIFFERING SOCIAL EMBEDDEDNESS

5.2 NORMS OF RESTRAINT AMONG

5.5 CONTESTING AND REINTERPRETING

6.1 MAIN FINDINGS AND IMPLICATIONS