Varying situations, different
decisions
Early warning of a potential genocide
Jorie Radstake 10005412
Research project:
Who belongs where: controlling movement and settlement Master thesis Political Science: International Relations
D. Vigneswaran
Second reader: A. Steele 24 June 2016
2 Table of contents
Abstract: p. 3
Terminology and actors: p. 4
Introduction: p. 5 Literature review: p. 8 Methodology: p. 24 Analysis: p. 38 Conclusion: p. 56 Discussion: p. 61 Bibliography: p. 63
Annex A: transcript interview Nigel Rodley: p. 74
Annex B: transcript interview Paul Helsloot: p. 80
Annex C: timeline Rwanda: p. 89
Annex D: timeline UN Rwanda: p. 93
Annex E: timeline Amnesty International Rwanda: p. 99
Annex F: timeline Darfur: p. 105
Annex G: timeline UN Darfur: p. 110
3 ‘’In the dark times
Will there also be singing? Yes, there will also be singing. About the dark times’’
(Brecht, 1969)
Abstract
Various actors are concerned with early warning of a potential genocide in specific countries. Early warning is based on a complex process in which various variables interact. In this research, the role of several factors will be examined with regard to the decision-making process of international organizations to spread an early warning message of potential ethnic and genocidal conflicts. These factors are divided into two types of factors; internal
organizational factors and external factors, which coincides with historical institutionalism on the one hand and constructivism, pluralism and multilevel politics on the other side. Two case studies are included; the situation before the genocide occurred in Rwanda in 1994 and the situation before the genocide started in Darfur in 2003. Besides that, two actors are involved; the United Nations as an international political body and Amnesty International as a
nongovernmental organization. Via these case studies, the decision-making process of the spread of early warning messages of international organizations is analyzed, which reflects the finding that decisions of various organizations are based on different factors in varying situations.
4 Terminology and actors
AI: Amnesty International
Al-Bashir: President of Sudan
Annan: Former Secretary-General of the UN
Baum: Former UN Special Rapporteur on human rights
Booh-Booh: Head of UNAMIR
CDR: Coalition pour la Défense de la République (Rwanda)
CIVPOL: Civilian Police Contingent of UNAMIR
Dallaire: Force Commander of UNAMIR
Egeland: Former United Nations Undersecretary-General for
Humanitarian Affairs and Emergency Relief Coordinator
EWS: Early Warning System
FEWER: Forum on Early Warning and Early Response
FPR: Front Patriotique Rwandais
Habyarimana: Former President of Rwanda
Helsloot: Manager media and political affairs Amnesty
International
HEWS: Humanitarian Early Warning System
HRW: Human Rights Watch
Hutu: Population in Rwanda, Burundi and Congo
Interhamwe: Hutu paramilitary organization
I.S. International Secretariat of Amnesty International
Janjaweed: Militia in Darfur
JEM: Justice and Equality Movement (Sudan)
Kapila: Former UN Resident Coordinator
MRND: Mouvement Révolutionaire National pour le
Développement (Rwanda)
Ndiaye: Former UN Special Rapporteur on extrajudicial,
summary or arbitrary executions
NMOGI: Neutral Military Observer Group one
NMOGII: Neutral Military Observer Group two
Rodley: Former UN Special Rapporteur on torture
RPF: Rwandan Patriotic Front
SHR: Science for Human Rights program
SLA: Sudan Liberation Army
SLM/A: Sudan Liberation Movement/Army
SPLA: Sudan People’s Liberation Army
Tutsi: Population in Rwanda, Burundi and Congo
UN: United Nations
UNAMIR: United Nations Assistance Mission for Rwanda
5 Introduction
On 12 January 1951, the Convention on the Prevention and Punishment of the Crime of Genocide went into force. From that day on, genocide or atrocities relating to genocide were officially punishable. The term of ‘genocide’ was first used by Lemkin, a Jewish lawyer who escaped from the Nazi regime in Poland during the Second World War (Lemkin, 1947). Lemkin developed the concept of genocide through a combination of the Greek word ‘genos’, which means race, and the Latin word ‘cide’, which means killing (ibid). He strongly focused on the intentional aspects of genocide, referring to genocide as a crime which is based on a systematic plan (ibid).
Gregory Stanton (the president of Genocide Watch) describes genocide as a process which consists of ten stages that are not linear, but which all happen during mass killings with the intent to destroy a (part of a) specific group (Stanton, 2013). It begins with classification (distinctions between people based on race, ethnicity, religion or nationality), followed by symbolization, discrimination, dehumanization, organization, polarization, preparation, persecution, extermination and denial (ibid). The international community has been criticized for paying solely attention at the ninth stage, when mass killings already occur. Various researchers and experts claim that attention and intervention is needed in earlier stages in order to prevent genocide (Adelman and Suhrke, 1996); (George and Holl, 1997);
(Kuperman, 2001); (Austin and Koppelman, 2004); (Dallaire, 2004); (Clough, 2005); (Grunfeld and Huijboom, 2007); (Stanton, 2013).
The emphasis in this thesis is on the preparation and the prevention of genocide, so the phase before an actual genocide arises. The preparation of genocide can be revealed via various signals, varying from military techniques (hunt and destroy operations rather than open armed combat in military trainings) to hate propaganda via mass media and trial massacres (a small number of civilians are killed in order to test whether the perpetrators will be punished) (Albright and Cohen, 2008). Action in the preparation stage of a potential genocide strongly relates with the prevention of genocide because signals of a preparation will be a sign for international actors to pay attention to the area in which a genocide might be planned. Stanton refers to the preparation stage of genocide as the planning of a final solution to target specific groups (Stanton, 2013). Terms such as ‘ethnic cleansing’ and ‘counter-terrorism’ are used and indoctrination which is based on fear of the victim group appears (ibid).
6 Risk assessment and early warning are important tools in order to investigate whether a potential genocide is planned and prepared (Albright and Cohen, 2008). Whereas early warning systems and models have been used in the past as tools in order to examine whether natural disasters such as storms or hurricanes would occur, early warning systems related to humanitarian disasters began to arise as well (Schmeidl and Jenkins, 1998). Several early warning models have been developed by various actors. States use early warning systems in order to examine the likelihood of humanitarian disasters in given areas, but various
organizations such as NGOs and the United Nations have developed their own systems as well.
Austin (2004) describes that early warning ‘’is a large field with many different
methodologies operating on different levels and with a wide range of issues’’ (Austin, 2004: 130) and that ‘’there are a broad variety of actors involved in these systems from grassroots projects to academics working on computer simulations’’ (ibid). But although early warning varies, the concept of early warning is the same in every single situation. Early warning consist of obtaining knowledge and using this knowledge in order to mitigate potential conflicts (Austin, 2004). This is also the case with early warning of a potential genocide. Initially, information about potential ethnic conflicts in specific countries or regions is gathered in specific ways via diverse systems. Issues such as ‘when will a conflict erupt’, ‘why will it erupt’ and ‘how will it erupt’ will be examined. This knowledge will then be used in order to mitigate these ethnic conflicts (ibid). The outcome is an early warning about a potential genocide in the form of a letter, a speech, a fax or other forms of communication. Moreover, an early warning message of a genocide can be defined as a report in which a warning of a potential genocide is given, or when an ethnic component of killings is emphasized. In this way, the words ‘genocide’ or ‘ethnic violations/killings’ have to be included in the report. Subsequently, an early warning of a potential genocide is developed through an analysis of information that is retrieved through specific systems, whereas reports about potential ethnic or genocidal violations are the outcomes.
Early warning systems for genocide have been developed on various levels, such as systems on a global level or local level (Matveeva, 2006). These various early warning systems select different types of data and they deliver different types of analysis. Quantitative data is for instance used, such as data sets. Quantitative data is useful for summaries and standardized analysis. But qualitative data is used as well, such as interviews with locals and testimonies
7 of witnesses. Qualitative data is beneficial because it provides in-depth information and it is actor- and context sensitive (ibid). Various actors incorporated different qualitative and quantitative methods in order to spread an early warning about a potential genocide. So, systems and models of actors vary, which influences the forms of early warning messages as an outcome as well. In this way, there is not a standard example of an early warning system of genocide or a common genocide early warning method, but various actors develop early warnings of genocide in their own way.
The main topic in this thesis is early warning of various actors, whereas early warning will be used in order to prevent a potential genocide. Several early warning models and systems of specific actors will be examined, as will be explained in the theoretical section. Moreover, the research question is; How do international actors decide that they should spread early
warning messages of a genocide? So, the decision-making process of early warning of a potential genocide of specific organizations is the central topic. Two case studies will be included as a reflection of the use of early warnings; the genocide in Rwanda and the genocide in Darfur. Obviously, early warnings of a genocide has to be spread before a genocide occurs, not during or after the genocide. So the situation before the genocides in Rwanda and Darfur occurred will be analyzed. The United Nations and Amnesty
International will be included as actors.
Studying early warning as a method of prevention of a potential genocide is relevant because whereas previous academic work focused primarily on the gap between early warning and early action (Adelman and Suhrke, 1996; George and Holl, 1997; Schmeidl and Jenkins, 1998; Harff, 2003; Grunfeld and Huijboom, 2007; Meyer and Otto, 2010), little is known about how early warnings are developed (Matveeva, 2006). So in other words, it is important to examine how international organizations decided that a warning about a potential genocide is needed because this decision-making process is not researched in a detailed way before (ibid). Therefore, it is useful to go back deeper behind the process of analyzing and deciding that deteriorating situations may or may not result in the onset of a genocide, in which competing definitions of genocide are not the most important rationales but where the decision to spread an early warning about potential genocidal violence matters the most. Moreover, the main focus in this thesis is the process of early warning. Thus, questions about what an early warning actually is and how various actors develop early warnings will be answered. This is relevant because early warnings are not clear and outlined concepts, but
8 rather messy concepts and interpretable in different ways. An early warning of a potential genocide of political institutions such as states or intergovernmental organizations differs considerably from an early warning of human rights groups or nongovernmental
organizations for example. Reports differ, conclusions of various actors diverge and other words are used to describe the same situation. It is important to look at the differences and similarities of early warning as a process because this will reflect the fact that early warnings cannot be viewed as a fixed concept but that the decision-making process of early warning and early warnings as an outcome vary.
Besides that, it is important to research how various actors such as nongovernmental and international organizations develop early warnings before a potential genocide occurs
because it might lead to academic recommendations about the creation of early warnings in a clearer and faster way. Furthermore, it is also important to study early warnings because transparency may lead to an increase in cooperation between various bodies. If the United Nations knows exactly how Amnesty International gathers its information, or how different signals are interpreted for example, the UN might want to work more closely with Amnesty International. Transparency may also lead to cooperation if information will be shared and if early warning systems are improved. Various actors may learn from each other and meetings about how to signalize a potential genocide might be set up. Lastly, it is important to study early warnings because more information about early warnings might result in an evaluation of early warnings from specific actors in the past. Lessons can be learned, which relates to the statement that improvements in early warning systems and the use of early warnings are strongly needed in the future.
Literature review
Although the Genocide Convention (1948, art. 1) obliges that genocide is a crime under international law and that perpetrators should be punished and genocide should be prevented, it does not describe exactly how genocide should be prevented. However, various preventive methods exist and have been used in the past. Diplomatic actions have for example been implemented in the way of peace conversations and mediation. Pressure has been extended in the form of economic and political sanctions (Ziegler, 2014). And in some cases, force has been used in order to prevent humanitarian crises (Albright and Cohen, 2008); (Bellamy and Williams, 2011); (Kuperman, 2012); (Pape, 2012).
9 Early warning contributes to the prevention of a potential genocide, regardless of the question of which actions will be used in order to prevent conflicts. In this chapter, the process of early warning and early warnings as a concept will be examined in a detailed way. Moreover, different academic early warning systems will be described. Besides that, several practical early warning models of various organizations will be discussed on the basis of policy documents and practitioner material. In other words, two early warning types will be distinguished; academic early warning and a practitioner version of early warning of
genocide. The different styles of early warning models and strategies reflect the central point that there is a lack of consensus about a standardized early warning method or a basic early warning model. Every organization has developed its own unique way of early warning, which coincides with the institutional idea that the behavior and strategies of organizations are imbedded in the character and policy of organizations themselves. But on the other hand, it could also be the case that shared ideas and norms are decisive factors in the decision-making process of organizations to spread an early warning. These competing rationales will be explained in detail.
Early warning
Early warning and early warning messages are no clear outlined concepts and opinions of academics vary. Early warning messages in the nineties are for instance described as expert opinions that are based on local information, or reports of ongoing atrocities (Harff, 2009). Harff (ibid) claims that early warning should be based upon schematic mechanisms and statistical information with specific signals of dangerous situations as an outcome instead of signals that are based on experiences and conversations with persons in risk areas. Others refer to early warning messages as informative messages that are based on systematic evidence and predictions about the likelihood of the occurrence of particular disasters in specific areas in the future (Schmeidl and Jenkins, 1998). In this sense, it is not entirely clear how the likelihood of events can be predicted, and systematic evidence does not necessarily solely have to include statistical information. But both approaches focus on the bridge between research and predictions.
Albright and Cohen (2008:17) state on the other hand that early warning messages should not be viewed as solely alarm bells that are ringing just before disasters occur. They also claim that although early warning messages that are communicated to policymakers and other relevant actors are the most visible outcome, these messages are based on an intensive
10 process (Albright and Cohen, 2008:18-19). Thus, they focus on early warning as a process. They do not exclude statistical information as a source, but they view early warning in all its aspects instead of focusing on just one single relation such as the relation between statistic information and a prediction about whether a specific situation is on the rise.
Adelman (1998) also describes early warning as a process. He distinguishes three main components in his definition of early warning; information gathering, strategic analysis and communication to policymakers. Information gathering refers to monitoring signals in risk areas, or data collection in other words. Strategic analysis is based upon interpretation of the data or monitored signals and predictions about whether a specific event will or will not occur in the near future. Communication to policymakers includes options in order to take action and further recommendations (ibid). But Krummenacher and Schmeidl (2001) on the other hand argue that the threefold description of the process of early warning is too narrow. They argue that early warning is developed in five stages; collection of information, analysis of information, description of case scenarios, report of response options and communication of the findings to policymakers (Krummenacher and Schmeidl, 2001). But conversely, these five stages can be summarized in the three stages of Adelman (1998) because a description of case scenarios can be viewed as a part of the analysis stage, whereas report of response options and communication of the findings both refer to communication to policy makers. So, the definition of early warning of Adelman (1998) seems accurate because detailed steps of early warning can be reduced to these clear and concise three stages of the process of early warning.
Totten (2008) follows this description of the process of early warning in his definition of early warning and early warning messages in relation to the likelihood of the occurrence of genocide. He states that genocide early warnings are ‘’those statements, actions and events that suggest in one way or another that a situation is possibly slouching towards genocide’’ (Totten, 2008: 85). The process of early warning of genocide consists of the collection of early warning signals, analysis of the data, and communication of the data and findings to actors such as the United Nations, states, human rights nongovernmental organizations, international journalists and citizens (ibid). This strongly relates with the definition of early warning as ‘’the collection, analysis and communication of relevant evidence and conclusions to policy-makers in order to enable them to make strategic choices’’ (Adelman and Schurke, 1996: 13).
11 Adelman and Suhrke (1996) describe that international actors did not know exactly what was going to happen in the period before the genocide in Rwanda started, but that they were aware of facts upon which an appraisal could be based that mass killings were likely about to occur in the near future. Thus, although nobody could predict the genocide in all details and although the extremely high number of victims could not be foreseen, information that referred to disastrous developments was available. This supports the statement that data collection, whether it is solely based on statistical information or not, is important in the process of early warning, but that the impact of interpretation and analysis of signals and indications is decisive. So, not only early warning systems of organizations differ, but analysis processes vary as well. In other words, early warning varies because every
organization interprets data in its own way. In this way, early warning is not a fixed concept and early warning processes differ between organizations. This is important because it could also have been the case that organizations differ in their decisions to spread an early warning because of the models they are using. But it is more difficult than that, because not only different data is used, but data is analyzed in different ways as well. So, the decision-making process of early warning of a potential genocide and ethnic violence is based on various models and ways of analysis, which refers to the complicated character of the decision-making process of early warning of organizations.
Adelman and Suhrke (1996) refer to five interference factors that influence interpretation and communication about facts that point to a potential genocide. First of all, incredulity takes place. In the late nineties and even nowadays, genocide is inevitably linked with the
Holocaust in the Second World War (ibid). Actors did not connect facts with the possibility of an upcoming genocide because another genocide seemed incredible. Next to that, mind-blindness or wishful thinking is linked to interpretation and communication of information, such as when actors focus on peace agreements and peacekeeping operations rather than on instability. Distraction influences interpretational and communicational processes as well, for instance if important events such as another war or crisis occurs elsewhere in the world. Attention lies somewhere else in this case, which results in a lack of analysis of the signals. And clearly, if analysis does not take place, publication about analyzed information is impossible.
Moreover, confusion about information hinders interpretation processes as well. If persons have a different understanding of genocide as a concept for example, confusion and
12 of killed persons. No standard amount or boundary of victims is formalized in juridical documents, so interpretations about when a genocide can be called a genocide are likely to vary. Lastly, desensitization also influences analysis and communication processes. Cruel conflicts have always occurred in Africa, so disasters can be expected in the future as well according to various persons. Signals that refer to killings in Africa could therefore be misinterpreted, because actors may think that killings are common in this part of the world (ibid). There is no shock or panic reaction anymore, because people are getting used to it. These five variables show that early warning is not only based on three stages (data collection, interpretation and communication to policymakers), but that these three stages themselves are based on complicated and intensive processes and that numerous factors influence these stages. Thus, early warning is based on an extensive process and not on a few clear and outlined variables.
Effectiveness of early warning messages
Effectiveness of early warnings is related to actual prevention of genocide in a retro perspective way. This is a problem, because the spread of early warnings is a speculative venture since nobody can forecast exactly what might happen in the future (Albright and Cohen, 2008). Some scholars referred to the success of early warnings as a dichotomy; early warnings were successful if genocide did not occur. Early warnings failed when genocide occurred or when risks areas continued to be dangerous (Adelman and Suhrke, 1996). But other researchers argue that it is idealistic to judge the effectiveness of early warnings solely on the fact that genocide did or did not occur (ibid). Early warnings and preventive measures are viewed as one of the most difficult of all policies to develop and use (ibid). But although the effectiveness or success of early warnings is difficult to examine, clear challenges to early warnings are identifiable.
As previously stated, early warning is based on three elements: collection of signals, analysis of these signals and communication to relevant policy-makers (Adelman and Schurke, 1996). One of the failures of early warnings is related to the link between signals and analysis of social conflicts (ibid). Albright and Cohen (2008) state that accurate warning analysis requires in the first stance reports of relevant actors and relevant events, which also can be framed as the information problem. Embassies and actors such as military officers and other international members should know exactly what happens in their host countries, but
13 accurate information about the situation in the capital of a country but not on rural areas, not in the last place because of resource constraints and because they live in the main cities (ibid).
On the other hand, news reporting in almost every single place in the world has increased considerably. But although this seems like a rather positive development, the volume of information has negative consequences for the analytical process of early warnings as well. The reporting overload can be overwhelming and it is complicated to examine the accuracy of the information (ibid). For example, it is hard for analysts to evaluate whether killings of civilians is based on intentions to kill a group or a part of a group, which refers to genocide, or whether the killings are an element of a civil war. When information of diplomats and other related actors is inadequate, ambiguous and contrasting, the information needs to be analyzed in an extensive and complicated way in order to carry out a clear and accurate early warning.
Moreover, successful analysis depends on the clarity of the signals as well (ibid). When there is a lack of proof of the intention to commit genocide, it is hard to claim that a potential genocide is on its way (George and Holl, 1997). But when signals are clear, such as when dead lists are spread and when political actors speak openly about the killings of members of a group for example, signals are easier to analyze. This correlates strongly with analysis of case-specific indicators (Albright and Cohen, 2008). It is therefore important to look at background information of a conflict in a specific area and the focus should not be limited to an evaluation of criteria on a specific check list. And specific indicators do not have the same effect in every situation. Hate radio for example contributed clearly to the preparation of a genocide in Rwanda and Nazi Germany, but it did not play a significant role in other situations (Adelman and Suhrke, 1996). Therefore, case-specific characteristics should be examined thoroughly as well.
Academic early warning models
Early warnings are based on various signals. Woolf and Hulfsizer (2005) emphasize factors such as group cultural history, situational elements, social psychological facts and context and interpersonal elements in order to investigate whether a potential genocide is likely to occur. Group cultural history refers to elements such as the use of aggression in policy, assumptions of antipathy with a perceived threat as conflict orientation and supreme ideologies based on dehumanization and institutionalization of resistance against cultural
14 diversity. Situational elements include destabilizing crises and authoritarian governments and actors. Social psychological facts and context are based on social cognition (a division of groups of people in ‘us’ and ‘them’) and social influence, such as indoctrination via
propaganda and mass media. Lastly, interpersonal elements refer to prejudice, aggression and altruism (ibid).
In contrast, other researchers did not focus on broad ranges such as cultural, situational, social psychological and interpersonal areas, but they examined whether concrete
components such as the regime type and economy of countries influenced the chance of the development of genocide. Various models which rest on differentiating risk factors exist. These models are based on long-term structural indicators that predict the likelihood of the onset of a potential genocide in current situations and in specific areas. Thus, most of these models are based on a list of factors which indicates whether a situation should be viewed as a dangerous and risky situation. One of the most well-known models is the model of Harff (2003), that consists of six risk factors. The first factor is political upheaval, which can be defined as past state failures. This definition is later replaced by the description of state-led discrimination against minorities as an indicator of political upheaval. The second risk factor is historical policies of genocide, followed by an exclusionary ideology of the ruling group, regime type, ethnic character of the ruling group and trade openness (ibid). More than sixty other factors were tested, such as economic, environmental and societal factors, but they had no significant effect on the accuracy of the model.
However, this risk model was improved in 2013 and 2015 due to updates and inclusion of missing data (Harff and Gurr, 2015). The list of risk factors was narrowed down to the following five variables: state-led discrimination against ethnic or religious minorities, existence of an exclusionary ideology used by a ruling elite, minority elite or rivalry between elite ethnicities, type of polity (autocracy versus democracy) and historical use of genocidal policies (ibid). Trade openness was thus no longer a significant variable. These two models have been and still are important tools in order to examine the likelihood of a potential genocide. The development of the models was a radical shift in the prediction of the signs of genocide. Risk assessments and early warnings were previously merely based on the opinions of experts. But with these models, a system based on quantitative and systematic analysis has been developed and it turned out to be significant and useful (Harff, 2009).
15 Rost (2013) agrees with the finding of Harff and Gurr (2015) that the type of polity
(autocracy or democracy) is a significant factor in order to examine the likelihood of the onset of a potential genocide. However, Rost (2013) does not focus on the dichotomy between an autocracy or a democracy, but he rather looks at anocracies, which has democratic and autocratic elements. Anocratic countries pay attention to political and democratic values (elections, voting standards) but they often fail to accommodate them, which results most of the time in conflicts and violence (ibid). Other factors are war, threats (assassinations and riots), the Cold War (one finding includes the notion that the likelihood of the onset of a genocide was higher during the Cold War), violations of human rights,
economy and national capability (based on consumption, production and population). Although this model was accurate and consistent in identifying the cases of genocide in the sample, a high number of false alarms occurred. So further improvements in order to use it empirically are needed (ibid). This clearly shows that although the model of Harff and Gurr (2015) is leading, other and new models are still developed in order to improve the system. But it takes a long time to develop such a system and new models have to be tested in an extensive way. Therefore, the model of Harff and Gurr (ibid) still takes the lead with regard to early warning models of a potential genocide.
But on the other hand, the proliferation of different models show that there is no consensus about a proper early warning model of genocide. Various risk factors are included in specific systems and models differ considerably. And even nowadays, early warning systems are improved and adapted to specific situations. In this way, different genocide early warning systems are used and a typical or common early warning system is lacking. This means that organizations can choose from a broad range of genocide early warning models, which relates to the practitioner use of early warning of (international) organizations.
Practitioner use of early warning
Early warning models are used by several actors in order to investigate whether signals of a potential genocide occur. The Central Intelligence Agency (CIA) of the United States was the first actor who used an early warning system when the State Failure Task Force was
inaugurated in 1994 (Totten, 2008). The mission of the Task Force included analysis of elements that might be signals of a potential genocide in specific states. Other actors, such as nongovernmental organizations (NGOs), the United Nations, other states and regional organizations began to develop and use various models of early warning as well.
16 The UN established the Humanitarian Early Warning System (HEWS) in 1992 after the call from the UN Secretary General for the reinforcement of a system for early warning for natural and political disasters (Harff, 2009). The system was developed by the UN’s Office for the Coordination of Humanitarian Affairs and it is based on a database of qualitative and quantitative data of country information (ibid). An extensive amount of sources is used in order to compose background reports (Verstegen, 1999). Specific countries are transferred to a watch list if statistical analysis and expert opinions pointed out that these countries are so-called risk countries, whereby a scale of conflict intensity is used in order to determine which countries on the list should be monitored more closely than other countries. These risk
countries are intensively monitored via reports of field offices or media reports. These reports, combined with country specific indicators, are used in a qualitative analysis and the outcome is a country report (ibid). Although HEWS got critique in the way that political actors used the system for their own goals (Harff, 2009), it is an innovate system because it combined quantitative and qualitative analysis (Verstegen, 1999).
Several other bodies of the UN use systems in order to analyze signals of potential
humanitarian disasters, but the Special Advisor on the Prevention of Genocide uses an early warning model that is explicitly developed in order to analyze whether a potential genocide is under its way in specific areas (Wisniew, 2012). The Special Advisor has the mandate to ‘’act as a catalyst to raise awareness of the causes and dynamics of genocide, to alert relevant actors where there is a risk of genocide, and to advocate and mobilize for appropriate action’’ (United Nations). The Office’s Analysis Framework consists of seven signals, which
cumulatively increase the risk of a potential genocide in specific areas; (1) tension between groups, including human rights violations and discrimination against a specific group; (2) lack of institutional capacity to prevent genocide, such as partial judges; (3) the presence of illegal arms; (4) a political, economic, social or military ideology that is based on the
assumption that a specific group has to be targeted; (5) circumstances that might influence the development of genocide, such as strengthening of the military force; (6) acts that are
referring to genocide, such as torture and rape of group members; (7) triggering factors, such as news events and elections (ibid).
On the contrary, the Forum on Early Warning and Early Response (FEWER) does not use a standardized model or a pre-specified list of signals in order to investigate whether specific countries should be considered as risk countries (Van de Goor en Verstegen, 1999). FEWER
17 is a coalition of various organizations that provides early warnings and promotion of early action strategies (Wisniew, 2012). FEWER was an innovative early warning model because various organizations worked together and because it did not only pay attention to early warning itself, but also to early political action (Van de Goor en Verstegen, 1999). The model consists of two parts; early warnings and early action. In the early warning part, specific situations in countries or areas are monitored through structural indicators, accelerators and triggers. After that, key indicators are analyzed and opportunities for peace are outlined. The early action part focuses on the identification of proper responses and instruments for conflict prevention (ibid). This case-specific model contrasts models that use lists of signals of a potential genocide and it could also be viewed as a framework for discussion (ibid). It is also important to emphasize the differences with early warning that is solely based on data gathering and analysis instead of a process of monitoring, analysis and communication of possible solutions and recommendations. Whereas some institutions focus on analysis of systematic data, other institutions focus on the entire process of early warning (monitoring of a situation, analysis and communication) in an broad way.
Another actor that uses an alternative model of early warning is the Sentinel Project (Wisniew, 2012). The Sentinel Project is a nongovernmental organization that is based in 2008 in Canada with worldwide members. The most important part of their mission is the prevention of genocide via effective early warning (The Sentinel Project). The Early Warning System (EWS) is based on four components, whereby the last two components are just in their concept phase. The first stage is risk assessment, where analysts examine various characteristics that may refer to a potential genocide. The characteristics are based on a risk factor list that rests upon information from the UN, NGOs, governments and academic sources. The second component is operational process monitoring, where events in risk areas are monitored via the media, NGO reports, correspondents and contacts on the ground. Social media will be inserted to raise attention about the situation in risk countries. The third
element is the vulnerability assessment, where the vulnerability of the attacked group will be examined. The fourth and last component is forecasting; the stage where analysts investigate when and how genocidal violence may break out (ibid). This is an example of an actor that uses a standardized list, but that also looks at other case specific elements in order to
investigate whether a genocidal event may occur. Although research on the stages still has to be finalized, a broad range of levels and variables are covered. This differs with HEWS because HEWS only analyzes information that is based on a database, whereas the Sentinel
18 Project focuses not merely on a risk factor list, but also on specific events in countries.
Research of the Sentinel Project is more case specific in this way and it could be linked with the intensive early warning process of FEWER because it tries to include as much as relevant strategies and variables as possible.
A NGO that has a long history and that has an outlined early warning system in comparison with the unfinished system of the Sentinel Project, is Human Rights Watch (Human Rights Watch). Human Rights Watch (HRW) actively researches and reports human rights violations in more than ninety countries. HRW is an independent body with more than eighty
researchers, and decides on their own which countries and issues they address. Early warnings are based on the work of researchers, who use a broad range of sources. A lot of researchers are field workers on mission, where they speak with victims, witnesses and other relevant actors. They constantly communicate with the local society, journalists,
governmental and national officials in order to gather information about human rights violations in specific countries. Interviewing is one of the methods that is used, combined with extensive background research that is based on data such as international or domestic law, nongovernmental organization reports and media stories. Other methods, such as forensic investigations, GPS coordination and satellite imagery are used as well. Thus, this NGO uses qualitative methods in order to examine the likelihood of (genocidal) violence in specific areas. Quantitative elements, such as a factor list, are not included in their research. The interpretational phase is thus based on stories of others, such as refugees, governments, other NGOs and others. This contradicts with early warning of HEWS, which analyzes and communicates information that is based on quantitative and qualitative research. And it also differs from the broad and varied early warning process of the Sentinel Project.
Another NGO that is concerned with human rights is Amnesty International. Amnesty International is a NGO that also uses merely qualitative methods in their analysis of human rights violations and early warnings, just as Human Rights Watch (Amnesty International). On a conference in 1982, 25 early warning signals were formulated; emergency situations, cooperation between courts and politicians, arbitrary arrests and detention, secret prisons, isolation of official prisons, judges who ignore juridical procedures, limitation of regular juridical procedures, increase in human rights violations, suppression of advocates, suppression of protests of family members, testimonies of survivors, refugees and
international correspondents, identification of violent target groups, human rights violators on high positions, lack of official research of violations, fake clashes with opposition groups,
19 fake cases of ‘murdered because of escape’, installation of military groups, closure of borders and information flows, increase in killings, unidentified graves, military vehicles on
cemeteries, suppression of human rights groups and the installment of military tribunals (ibid).
But rather than focusing on this list, Amnesty International uses other methods and sources as well such as testimonies of victims and their family members, reports of advocates, refugees and organizations (religious organizations and women organizations for example), media reports and field work in various countries. Amnesty International also uses the Science for Human Rights program (SHR), where geospatial technologies, such as satellite imagery, is used in order to get access to secured conflict zones, to obtain visual evidence and to report about the evidence in innovative ways. This program is thus used for monitoring human rights violations and prevention of conflicts (ibid).
Concluding, there are various actors that use specific systems, and processes of early warning vary among institutions. Models differ or have similarities with academic early warning models, just as the analysis stage. Besides that, some actors strive to be independent and collect their own information, whereas others share their information and use information from other actors. But although some actors share their information, and although
cooperation between various actors is pursued, various actors come to early warnings on their own way with their own specific models, analysts and reports.
The fact that actors have developed their own ways of early warning of a potential genocide relates to the idea that actors develop their own strategies. These strategies are based on internal decisions. In this way, strategies are related to the character of organizations and choices are in line with the policy of organizations. If this is true, organizations should be viewed as rational institutions that strive for their own ideals and principles. In this way, the role of other factors, such as third actors and constraining variables such as the context of situations, is negligible. Decisions are made because of internal logics, which means that the decision to spread an early warning could also be based on institutional factors.
The idea that organizations make decisions that are based on institutional factors, such as the policy of an organization or general behavior of organizations, could be an explanation for the fact that organizations use different early warning methods and that they analyze specific situations in different ways. Moreover, it also possibly explains why organization A spreads an early warning about a potential genocide, whereas organization B decides that an early
20 warning about ethnic conflicts is not needed in exactly the same situation. In this way,
decision-making processes of early warning vary between organizations. Organizations decide on their own whether it is relevant to spread an early warning about potential ethnic conflicts or not, which reflects the main thoughts of institutionalists.
Institutionalism
Institutionalism refers to the involvement of institutions in the creation of the international system (Steinmo, Thelen and Longstreth, 1992). Institutions can be defined as formal organizations, but also as informal procedures, habits and patterns that create order and behavior (ibid). Institutionalism is based on various types, but the most well-known types are rational choice, sociological institutionalism and historical institutionalism (Steinmo, 2008). Rational choice institutionalism refers to the principle that human beings are rational actors who strive for as much benefits as possible. Institutions, rules for example, are important because they frame the behavior of individuals (ibid). Sociological institutionalists on the other hand emphasize social patterns, social roles and social interaction. They do not think that actors are making choices because they want to gain as much as possible, but actors behave in a particular way because it is appropriate (ibid).
The third type of institutionalism, historical institutionalism, combines these two
perspectives. Historical institutionalists view actors as rational actors that pursue their own goals, but they also emphasize the influence of social patterns on the way an actor behaves. Historical institutionalists attach much importance to history as well. History matters in three ways; things happen within a historical context and that has consequences for decisions or occasions, actors learn from the past, and expectations and behavior are shaped by what has happened in the past (ibid). This is also called path-dependency, which refers to the fact that behavior and strategies of an organization are based on institutional decisions in the past, such as rules, law and habits (Aspinwall and Schneider, 2000).
Historical institutionalism and path-dependency coincide with self-reinforcement (Pierson and Skocpol, 2002). Self-reinforcing dynamics are based on specific patterns, rules and thoughts that are deeply imbedded in the way organizations and other actors behave. These ideas and patterns are developed in the past and are incorporated in such a way that they are a fixed part of the policy of organizations (ibid). Stinchcombe (1968) calls this historical causation. Specific events and processes trigger dynamics in a period in the past. These dynamics reproduce themselves over time, even when the original events and processes do
21 not occur (ibid). Thus, the same dynamics are repeated over and over again, while a repetition of the same conditions is lacking. In this way, choices, rules and behavior in the past are reproduced over time and policy and behavior of organizations is always influenced by this process.
It is relevant to emphasize whether institutions themselves determine that early warnings are necessary, or that other factors influence their decision-making process as well. Historical institutionalism points to the idea that decision-making processes of organizations and other actors are based on the decision-making processes of the same organizations in the past. In this way, early warning of organizations is based on internal organizational logics. Habits and rules in the past will for example influence the way research is conducted with regard to early warning. Organizations will ground their analysis on the way they analyzed information and facts in the past. And information is spread in a specific manner because organizations are used to spread information in this way. Needless to say, this does not mean that research, analysis or communication of organizations cannot change. But their methods, analyzing patterns and communication is based on the way organizations did research, analyze things and communicate about situations in the past. Thus, the decision-making process with regard to early warning is based on the policy of an organization, and on its rules and habits.
Organizational dynamics are reproduced, even when the context differs.
But on the other hand, it could also be possible that institutional logics do not explain entirely why organizations use different early warning models and why a specific organization
decides to spread an early warning, whereas other organizations decide that it is not useful to spread an early warning. Analyses of the likelihood of potential conflicts might also be based on other factors instead of exclusively on internal factors. The fact that some organizations decide to spread an early warning and that some organizations decide that an early warning of potential ethnic conflicts is not necessary could be based on institutional logics, but it is also possible that organizations decided collectively that they should spread an early warning of genocidal conflicts, whereas other organizations decided collectively that an early warning was not relevant.
Moreover, it is also possible that early warning models differ and that analysis processes vary but that parallels and connections can be made between the early warning processes of
international organizations. If this is true, groups of organizations base their decision-making process of early warning of ethnic conflicts on what other organizations in the same group do,
22 or what is collectively decided. If a human rights organization such as Amnesty International decides to spread an early warning about a potential genocide, it could also be possible that Human Rights Watch decides that an early warning is needed. In other words, organizations that are based on the same principals might cooperate or influence other organizations.
If this is true, shared ideas and norms will be important in the decision-making process of international organizations of early warning of ethnic conflicts. Collective ideas and group processes matter and organizations should not be viewed as unitary actors in the international system. In this way, organizations decide for instance to spread an early warning about genocidal violence because of shared norms which are incorporated in law and legal documents. Or organizations spread an early warning because they analyze that specific events should be monitored more intensively. It could for example be the case that an early warning is spread because organizations are concerned about instable elections in a specific country. In this way, ideas about events play an important role. In conclusion, it is not guaranteed that organizations decide to spread an early warning about a potential genocide because of internal logics. It is possible that external factors such as collective ideas and shared norms influence the decision-making process as well, which refers to principal ideas that are incorporated in constructivism.
Constructivism
While historical institutionalists focus on the influence of patterns in the past on the decision-making process of organizations and other actors, constructivists state that behavior and policy of organizations are constructed (Adler, 2013). Constructivism refers to the role of ideas, culture and norms that collectively construct interests, identities and behavior of actors (Finnemore and Sikkink, 2001). Ongoing processes of social interaction create behavior and identities of actors, so decision-making processes of organizations are constantly changing. Social interaction coincides with shared ideas of various persons. Something exists because an idea about it is shared among various actors.
This clashes with historical institutionalism because historical institutionalists argue that behavior is based on decisions and behavior of an actor in the past, while constructivists emphasize that behavior is constructed and based on interaction and shared ideas. Thus, historical institutionalists focus on internal dynamics of actors, while constructivists focus on a broad range of concepts and ideas. The latter emphasizes interaction with other actors and the involvement of other concepts, such as culture or religion (ibid). On the other hand,
23 constructionists mention that behavior could also be historically structured, so they do not deny the existence of path-dependency as a pattern. But much more processes and concepts are important, not only historical dynamics.
It is also relevant, besides the exploration of a historical institutional view, to examine whether the decision-making process of organizations is based on shared ideas, social interaction and norms. Law could be an interacting variable for example. Shared ideas are established in legislative rules, which possibly influence the decision-making process of organizations and other actors. In this way, internal organizational logics are not the only factor upon which decisions are based, but the role of contextual factors, such as law, ideas about events and interaction with other actors have to be considered as well.
Pluralism and multilevel politics
Moreover, pluralism refers to the involvement of other actors, ideas and opinions as well (Lijphart, 1975). Pluralists emphasize the influence of interaction between various actors on policy and decision-making processes. Cooperation is possible and decisions of actors are based on various persons or organizations. This coincides with constructivism, because in both paradigms is reflected that the decision-making process of organizations is not merely based on their own behavior and policy. Other actors are involved as well.
Besides that, multilevel politics is based on the entanglement and overlap between various actors (Bache and Flinders, 2004). Policy of actors on different levels, such as local actors, supranational institutions and international authorities, is strongly intertwined. The decision-making process of European member states is for example influenced by European decisions and international agreements (Smith, 2011). Decisions of an actor are not solely based on its own predictions, ideas and goals, but other actors on different levels influence its internal processes as well.
The principles of pluralism and multilevel politics are important to emphasize because decision-making processes could be entangled with the decision-making processes, opinions and policy of other actors. It is for example possible that an organization thinks that it is useful to spread an early warning message because other actors emphasize the urgency of a report as well. Or an organization cooperates for instance with other organizations with regard to early warning of a potential genocide if they collectively raise an alarm. In this case, the decision-making process of an actor is concerned with the role and ideas of other actors.
24 This results in dismissal of the statement that the decision-making process of an institution is based on its internal organizational logics, because the role of other actors is decisive as well. Concluding, the decision of organizations to spread an early warning about ethnic conflicts or a potential genocide could be based on two various factor types, internal factors or external factors. It is important to examine whether internal factors or external factors are decisive in the decision-making process of organizations because this examination provides an answer to the research question; How do international actors decide that they should spread early warning messages of a genocide? Internal factors matter when an organization decides to spread an early warning because it is in line with its policy. Besides that, internal logics are decisive in the decision-making process of organizations if an organization spreads an early warning because of their own research. But on the other hand, external factors matter the most if organizations decide to spread an early warning because of collective norms that are incorporated in law, or when organizations spread an early warning because of shared ideas about events. In this way, international organizations do not decide on their own that they should spread an early warning about a potential genocide or ethnic conflicts, but they cooperate with other organizations. Both rationales could be an explanation for the fact that early warning models and early warning analyses differ between international organizations, which reflects the lack of consensus about early warning of a potential genocide.
Methodology
This chapter provides an insight in the way how research is conducted in order to answer the research question; How do international actors decide that they should spread early warning messages of a genocide? First of all, the research question will be linked to two hypotheses, whereas the first hypothesis will refer to institutionalism and the second hypothesis reflects constructivism and pluralism as paradigms. Thereafter, a conceptual model will be outlined. Besides that, Rwanda and Darfur will be introduced as case studies and the choice for the inclusion of the United Nations (UN) and Amnesty International (AI) as actors will be explained. After that, the use of process tracing and interviews as methods will be described. Lastly, the research design will be outlined, which refers to the fact that four stories are made (UN-Rwanda, AI-Rwanda, UN-Darfur, AI-Darfur) in order to examine whether the decision-making process of the UN and AI of early warning of a potential genocide in Rwanda and Darfur was based on internal factors or on external factors.
25 Operationalization
A key concept in the research question (how do international actors decide that they should spread early warning messages of a genocide?) is decision-making. A decision-making process can be defined as an evolution with the selection of an option while alternatives are rejected as an outcome (Phillips, 1997). Options are in this research narrowed to two options; the possibility for an organization to spread an early warning message of potential ethnic or genocidal violence and the option of not spreading an early warning message. Moreover, a decision-process of an actor is based on various concepts. Historical institutionalists argue that decisions and patterns in the past influence decision-processes in the future (Aspinwall and Schneider, 2000). Policy of an institution is based on previous measures and decisions, which is also called path-dependency as is outlined before. In this way, decisions are based on policy, habits and rules of an organization. These logics are incorporated in the institutions themselves. In this way, organizations make decisions that are based on their own internal organizational logics. Hypothesis one refers to this expectation:
H1: The decision-making process of early warning of a potential genocide of international organizations is based on internal factors.
Internal factors refer to the mandate of an organization, its history, its goals and its structure. These concepts are all incorporated in the policy of an organization (Burstein, 1991). But in this research, the internal process of an organization is linked with early warning of a potential genocide. In this case, an internal process does not only refer to policy of an organization, but also to early warning itself. Early warning is based on three components; data collection, data analysis and communication (Adelman and Schurke, 1996); (Totten, 2008). In this way, the internal process of an organization is operationalized as internal policy and internal research, whereby the latter concept is subdivided in data gathering, analysis and communication.
On the other hand, the decision-making process of organizations with regard to early warning of a potential genocide could be influenced by other factors as well. According to
constructivists, decisions are constructed and they are based on shared norms and ideas (Finnemore and Sikkink, 2001). If this is true, the decision-making process of organizations is not (solely) based on the internal process, but (also on) contextual factors. Contextual factors refer to all kind of concepts, so specification is needed. Constructionists emphasize the role of shared norms and ideas, which are for instance incorporated in international
26 agreements (Finnemore and Sikkink, 2001). Contextual factors refer in this thesis to
legislation. Shared norms and ideas are incorporated in law, such as in human rights
agreements. But contextual factors also include events, such as a war, elections or rebellion. Shared ideas exist about such events. War should for instance be prevented, elections need to be fair and rebel groups should exclude the use of force according to a lot of (Western) countries. These ideas are also established in international legislation. Or they can be adopted in political resolutions or measures. If a war just destroyed a country for example, the United Nations could decide that they will send a peacekeeping force because all the member states share the idea that the country needs support. In this way, an event (the arrival of a
peacekeeping force) is based on shared ideas. Thus, contextual factors refer to legislation and specific events.
According to pluralism and multi-level politics, decisions of actors are based on other actors as well (Lijphart, 1975); (Bache and Flinders, 2004). Other actors could be anyone; other international organizations such as NGOs or political institutions, but also states, networks and individuals. It refers to all actors that are not a part of the organization itself. Actors can cooperate for example, information can be shared and reports can be written collectively. It is also possible that actors are influenced by other actors. Organizations can for example be inspired by other actors, or they could be forced to make specific decisions. The influence of other actors and contextual logics such as law and shared ideas about events compete with the idea that internal factors are decisive in the decision-making process of organizations.
Therefore, a second hypothesis is formulated:
H2: The decision-making process of early warning of a potential genocide of organizations is based on external factors.
In this way, two competing rationales are developed. Internal factors are decisive in the decision-making process of international organizations if organizations spread an early warning because they have analyzed that an early warning of ethnic violence is needed. In this way, organizations interpreted data in a specific internal way. Moreover, internal factors also refer to the policy of organizations. In this way, organizations spread an early warning because it is in line with their policy and because they are used to spread early warnings in specific situations because they have always done it in that way.
But if actors such as human rights organizations or political actors influence international organizations in their decision to spread an early warning of ethnic or genocidal violence, the
27 decision-making process of early warning of international actors will be based on external factors. This happens for instance when a specific organization spreads an early warning because other organizations claimed that it is necessary to spread an early warning. This reflects the idea that third actors influence organizations and that cooperation might occur. External factors also refer to the involvement of law and shared ideas about events. So, external factors are decisive if an organization decides to spread an early warning because a specific situation is not in line with legal international agreements. Or an organization spreads an early warning because it thinks that specific events, such as political elections, might influence ethnic violence on the short term. In this way, context matters, which competes with the idea that the decision-making process of early warning of genocidal violence is primarily based on internal factors.
Conceptual model
The two competing lines of explanation (the decision-making process of an organization is based on internal factors or on external factors) could be presented schematically as well, as can be viewed in figure one. The original explanation refers to the expectation that internal logics influence the decision-making process of organizations with regard to early warning of a genocide. Based on the policy and research of an international organization, an early
warning message is spread or not. If no evidence is found for this expectation, it could be possible that external factors affect the decision-making process of international actors. Legislation, other actors and (political) events result in this case in the decision to spread an early warning message of a potential genocide. So, internal organizational factors and external factors are independent variables and the decision-making process of early warning of international organizations is the dependent variable.
d
Figure 1.
Internal organizational factors: - Policy - Research: Data collection Analysis Communication Decision-making process: - Need for an early warning
message
- No need for an early warning message External factors: - Legislation - (Political) events - Influence other actors
28 Case studies
Case studies will be used in order to investigate whether the decision-making process of early warning of genocidal violence of international organizations is primarily based on internal factors or on external factors. Case studies are useful because specific phenomena such as behavior, situations, events and processes can be described, which is fundamental to social science (King, Keohane and Verba, 1994). Gerring (2007: 86) states that ‘’case study analysis focuses on a small number of cases that are expected to provide insight into a causal
relationship across a larger population of cases’’. In this way, singular cases represent a wide array of cases and a causal relation in one case study is likely to refer to the same causal relationship in other cases.
It is useful to adopt a case study approach because a lot of genocides have occurred in the past and early warning has been applied to various situations. Via the use of case studies, not every genocide and every early warning process needs to be examined. The decision-making process of international organizations of early warning in specific countries is examined in specific periods. These singular cases will represent a wide array of situations, genocides and early warning processes. In other words, case studies are beneficial because they provide not only an insight into the decision-making process of international organizations in these specific situations and periods, but they also provide an insight into the decision-making process of international organizations of early warning in other countries and in other periods. In this way, singular cases will reflect causal relations of a broader set of cases. This does not mean however that early warning in specific case studies coincides exactly with early
warning processes in other situations, but the case studies provide an insight in how international organizations decide that they should spread an early warning about potential ethnic violence in general.
As previously has been stated, various genocides have occurred in the past, such as the Holocaust and massive ethnic killings in Cambodia, Yugoslavia, Burundi, Armenia and Ukraine. Besides that, various early warnings of genocidal violence have been spread. From all these genocides, two case studies are selected; Rwanda during the period just before the genocide started and the situation in Darfur before the genocide occurred. This selection is based on a diverse-case method (Gerring, 2007). In this way, case studies are selected
because specific dimensions of these case studies vary. Rwanda differs from Darfur as a case study because the genocide in Rwanda took place in three months, while the genocide in
29 Darfur lasted several years. Moreover, there was a clear start and end of the genocide in Rwanda, which was not the case for the genocide in Darfur. The government was involved in the genocide in Darfur, but not in Rwanda. Besides that, the genocide in Rwanda took place throughout the entire country, whereas the genocide in Darfur occurred in one specific part of Sudan. Besides that, there was an peacekeeping force of the United Nations before the
genocide in Rwanda happened, while a peacekeeping force of the UN was only send to Darfur when the genocide already took place. In other words, Rwanda is a relevant case study because it is a typical example of a short genocide with a clear beginning and end, which occupied the entire country, where the official government was not involved in the killings (because almost all the members had been killed) and where a peacekeeping force of the UN was present. And Darfur is a useful case study because it is a typical example of a long and complicated genocide without a proper begin- and end date, which occupied only one part of a country, in which the government played an important role and where a peacekeeping force of the UN was lacking.
Moreover, another reason for the inclusion of Rwanda and Darfur as case studies refers to temporality as an issue. The genocide in Rwanda happened in 1994, while the genocide in Darfur began in 2003 (Williams and Bellamy, 2005). Thus, the genocide in Darfur happened almost a decade after the genocide in Rwanda. This is important, because these two case studies could show a potential development over time with regard to early warning of
genocidal violence. It is for instance possible that the early warning decision-making process of organizations changed after Rwanda because of technological improvements or because of a moral obligation for the international community to participate more actively in the
prevention of potential ethnic conflicts. Moreover, Rwanda and Darfur as case studies
represent general early warning over time. In other words, these case studies do not only refer to early warning in the nineties or to modern early warning, but early warning of potential genocidal violence is covered during different time periods.
First case study: Rwanda
In Rwanda, tensions between the ethnic tribes of the Hutu’s and the Tutsi’s had always exist (Reyntjens, 1996). In the Rwandese civil war around 1990, the Hutu-led government and the Rwandan Patriotic Front (RPF), which was based on Tutsis, fought against each other. Due to international pressure, a ceasefire was arranged and peace conversations started in 1993 (ibid). A transitional government would be installed, but resistance and agitation delayed the
