Journal of Humanitarian Assistance
Analytical Framework
Subject matter
This report examines the effectiveness of international monitoring and management of the Rwanda conflict, which culminated in genocide of half a million to one million persons in the second quarter of 1994. The report has two distinct yet closely related foci: (i) the existence and effectiveness of early warning mechanisms anticipating escalations in violent conflict and mass murder; (ii) the nature and appropriateness of the responses of significant international actors towards the evolving conflict.[1] Since the 1990 invasion marked the opening of the conflict cycle that culminated in genocide, the study begins with the factors leading up to the invasion of October 1990 and ends in early 1995. Within this time frame, different phases of the conflict are identified and studied in varying degrees of depth. In the period after August 1994, consideration is given only to conflict management issues concerning internally displaced persons and refugees. Given the constraints of this study, early warning relating to providing humanitarian assistance is a peripheral rather than central focus.
Process
Data collection for the study commenced at the beginning of 1995 and has been carried out by Howard Adelman (York University, Toronto) and Astri Suhrke (Chr.Michelsen Institute, Bergen), with the assistance of research associates working on assigned topics. These included: Agnes Callamard, Shally B. Gachuruzi, Peik Johansson, Bruce Jones, Kate Halvorsen, Steven Livingston and Todd Eachus, Ogenga Otunnu, Gérard Prunier, Turid Lægreid and Amare Tekle. The research is based on secondary resources (books and articles by academics and journalists, media studies, reports, etc.) as well as considerable primary data collected through interviews and document searches in the UN system (New York and Geneva), the NGO community, and visits to relevant national capitals in Europe and North America (Paris, Rome, Brussels, London, Washington and Ottawa), and in Africa (Nairobi, Kigali, Kampala, and Dar-es-Salaam). Some of the most valuable documentation was made available for exclusive use of this study. Nevertheless, documents have limitations: a great deal of communication and key parts of the decision-making in this case were not recorded. As for interviews, personal recollections of participants always pose difficulties with respect even to basic facts; there is a need for caution, given the propensity to rewrite history in the reconstruction of memory. This difficulty was compounded by the traumatic nature of events in Rwanda and controversies about international responses and responsibilities. With these constraints, we have tried to provide coherence in reconciling the various perspectives on the crisis.
If the report does not adequately reflect the position of some parties, one reason is that documentation and personnel were not available. Most of the persons interviewed have been identified only by their institution (see Appendix), but some insisted on complete anonymity.
Although the research responsibility was initially divided according to issues - Adelman taking primary responsibility for early warning and Suhrke for conflict management - the final product is the responsibility of both. The dialectic of reaching a consensus on virtually all issues was a discipline in itself.
Premises
This study falls outside the bounds of evaluation in a conventional sense. To address the questions posed by the terms of reference required two tasks: sorting out the cause-and-effect relationships of complex historical developments and assessing their consequences, not only in relation to formal stated goals but equally in relation to higher standards of conflict management and international norms. In these respects, the study follows in the tradition of critical policy analysis. There is a recognition of competing goals, of unintended, unforeseen or indirect consequences, yet a presumption of some choice and consequent responsibility by the actors concerned.
The report is based on some general premises concerning the interpretation of a complex social conflict. The first premise is that historical actors, in comparison to the critical analysts of those acts, interpret events and their significance from very different perspectives. Though both base their assessments on inadequate information and communication, analysis of a conflict "from within", while it is evolving, differs significantly from ex post facto analysis. The analyst is in some respects better positioned to understand and interpret what occurred. In retrospect, the equivalent of early warning data may be interpreted differently and "more correctly" in light of what actually happened; signals that at the time were ambiguous may now seem compelling, and additional information may be available. The analyst may be constrained by lack of access to information sources but can view the issues from several perspectives. Herein also lies the danger of misrepresenting the ability of historical actors to interpret signals correctly within their own situational context. A related distinction lies in the set of filters that screen incoming information; the range of action that policy-makers consider realistic will influence the attention paid to information, how they interpret it, and to whom it is communicated. In a feed-back mechanism, this in turn influences the formulation of policy options.
Early warning is the collection, analysis and communication of the relevant evidence and conclusions to policy-makers to enable them to make strategic choices.[2] Unlike traditional intelligence, which also collects and analyzes information and communicates the results, the object of early warning is not primarily security for one's self or one's country, but the security of another; in early warning, the security is not self-directed. The other party or parties are not presumed to be adversaries, as is the case with intelligence analysis.
Human rights monitoring also involves the collection of information, analysis of that information and communicating the results. But the prime audience is the public, even though human rights reports are addressed to leaders. For human rights reports do not facilitate the decision-making of those leaders, but rather serve as a superego, a haunting conscience for people in power. Like early warning, it is other- rather than self-directed.
Early warning can be concerned with military conflicts, military coups, impending humanitarian disasters, such as starvation and flows of refugees, slaughters, and, at the extreme, genocide. In the Rwanda case, all of these were involved at different stages.
The spontaneous slaughter of one group by another, usually as a result of inter-ethnic strife, is not generally referred to as genocide. Genocide must be the result of deliberate policies of the political elite intent on wiping out the other group. According to the 1948 Convention, the "intent to destroy" is an essential requirement for identifying genocide.[3] Thus, understanding the political process that leads to the genocide is critical. The decision of a political elite to foster or permit genocide to occur may be the result of their implicit consent rather than explicit orders. In the Rwandese case, no such subtle distinctions between implicit and explicit policies are necessary. The genocide was actively promoted by political and military leaders, and there were apparently concrete plans to kill at least all the Tutsi in Kigali. Once the killing commenced, there is a debate on whether the leaders continued to manage the genocide until the end, or once the frenzy of killing started and - given the organization in place - it acquired an additional momentum of its own.
However, there need be no debate on whether the use of the term "genocide" is appropriate. According to the 1948 Convention, intent to destroy a group "in whole or in part" constitutes genocide, and the term was applied to describe even relatively small slaughters (300-2,000) in the 1990-93 period. In the public mind, by contrast, the term "genocide" is used only when substantial proportions of a group are slaughtered. What happened after 6 April 1994 meets both the legal and the popular meanings of the term.
According to what standards do we assess the anticipation of violent conflict? Were there relevant structures for collecting and analyzing signals, and, if so, did these function in an efficient, accurate and timely manner? What were the relevant structures? Who has - or should have - responsibility for collecting and analyzing information and translating that analysis into options for action? The assumption is that the responsibility for collecting and analyzing information flows from the responsibility for action. That is, actors who played a role in the Rwandese conflict had an obligation commensurate with their responsibilities to obtain and assess data about the evolving violence. In each case, the overriding consideration must be whether the system(s) caught and processed available information that at the time reasonably could be said to indicate that an event - an invasion, an escalation of the conflict, genocide - was a possibility, likely, or already under way.
In the ideal state, early warning serves prevention. In reality, preventive measures are among the most difficult of all policies to formulate and carry out. The next best, as it were, is to focus on conflict management. Here, the general criteria identified by the terms of reference are "containment and de-escalation". We see these as being derived from an ethical standard which holds that the international community should have attempted to reduce violence in Rwanda in general, and prevent genocide in particular. Given this premise, a specific response has to be assessed not only in relation to its formal goals (e.g. of UNAMIR I); the question must also be asked whether the formal mandate was appropriate in relation to a higher standard of conflict reduction extant in the international community.
A number of generally recognized international norms for assessing conduct appropriate to conflict reduction are relevant in the Rwanda case. The obligation exists for the parties to the conflict to attempt to settle their differences by peaceful means, utilizing when ever possible the appropriate international organization to which they belong. In refugee matters, countries of origin are obliged to allow their nationals the right to be repatriated.[4] States are not to permit invasions of other states from their territories.[5] There is an obligation not to intervene in the internal affairs of sovereign states except in cases of genocide. These norms are not always compatible, either in principle or in practice, nor were they in the Rwanda case. For instance, norms affirming the right of refugees to return, and norms against using military force from outside the country to enforce that right seemed to be at odds in practice. No similar incompatibility existed between the general strictures against intervention, on the one hand, and the explicit legal right and associated moral obligation to intervene to stop the mass slaughter of civilians according to the Genocide Convention.
If the first task of evaluation is to identify appropriate standards, a second is to assign causative responsibility for decisions and their consequences. Different degrees of responsibility may be assigned to different actors, and responsibility may vary over time. In the Rwandese case, that meant, in the first instance, the Rwandese parties to the conflict. It also meant, however, that the regional and European states most closely involved in the development of the conflict triggered by the October 1990 invasion (especially Uganda and France as the principal associates of the major Rwandese protagonists) bore significant responsibility. Third, regional actors (especially Tanzania and the OAU) were moved by self-interest and formal obligations, and, in fact, played an active and sustained mediating role in the pre-1994 period. Fourth, there were actors (other states, the Catholic Church, NGOs) that chose to become involved for various reasons; they also had commitments and responsibilities that varied with the extent of their involvement. Fifth, the UN and its agencies became involved in a major way as the crisis in Rwanda threatened regional peace and security. Finally, the media had a potentially important role, particularly in the development of public awareness and public policy.
As for responsibility in the normative sense of obligation to act, the premise of this report is that the principal though not exclusive responsibility for anticipating and managing conflict lies with the parties themselves and their principal associates. In particular, insofar as the principals are responsible for the development of the conflict, they have both a political/moral obligation and some actual capacity for regulating it. Beyond that, international organizations have formal obligations to assist in conflict management. In cases of genocide, those obligations are much greater and apply to the members as well as the organizations.
It must not be forgotten that those most responsible were the perpetrators of the crimes. They have been identified as extremists. In this context, extremists are presumed to have possessed two essential characteristics. First, their actions were determined by an ideological view of the world - in their case, the definition of Rwanda as a Hutu nation. Second, they regarded others who did not fit into this definition and who threatened its realization as legitimate targets for murder.
Information gathering, analysis and communication, as well as the actions based on this, must not be assessed only according to norms and attributed to agents based on their respective responsibilities. These assessments must take into consideration the limitations at work. For example, a system cannot be faulted for a failure at early warning if the information available was small, inconsequential, of doubtful value, or ambiguous. Further, not all conditions within Rwanda that led up to and accounted for the genocide were equally amenable to change. While there are various theories about how and why the genocide occurred, most distinguish between structural factors and situational variables. Structural factors, such as demographic density and the shortage of land, can create a context conducive to social violence, but policy can only alter that context in the long run and only with uncertain outcomes. Situational variables that had a triggering or accelerating effect, such as arming militias or scaling down UNAMIR, were subject to more immediate control through internal as well as external influences.
The same distinction between situational and structural variables is also applicable to those charged with interpreting and responding to the crisis. A realistic assessment of the international response to the Rwandese catastrophe must recognize the structural restraints of the existing international system based largely on competitive nation-states. Current capacity at state, regional (OAU) and international (UN) levels to identify and respond to conflicts of the kind that occurred in Rwanda constitute structural constraints on action. The broader foreign policy interests of the major powers provided the context within which specific policies towards Rwanda were formulated, both prior to and during the 1994 violence, and represented structural constraints of a different kind. Nevertheless, within these limitations, options were available; these constitute the focus of this report. Structural problems (e.g. on the UN level) will be dealt with only insofar as they affected the Rwanda case.[6]
The conclusions and associated recommendations are situation-specific. As such, they are not necessarily relevant for warning or management of the conflict in Rwanda as it is currently evolving, but rather to a set of circumstances and actions that have generic applicability.
Time frame for analysis
As recognized in the terms of reference, the period preceding the genocide is relevant for an understanding of the signals emitted about the evolving conflict, and efforts to contain it through diplomacy and military force. This report is organized around five phases of the conflict; these are similar to those suggested in the terms of reference, but adjusted after closer examination of the historical material.
Phase I covers the refugee problem in Uganda prior to October 1990. Chapter 1 examines how the refugee issues became a source of renewed violence, given the failure of the Rwandese government to permit return, the unwillingness of Uganda to accord full rights of settlement, and whether there were early warnings of the invasion itself.
Phase II includes the period from the October 1990 invasion until the signing of the Arusha Peace Agreement in August 1993. This period was characterized by an interlinked process of civil war, civil violence and preventive diplomacy. The organized violence against unarmed civilians in this phase constituted a kind of early warning to the effect that worse might be in store. Simultaneously, the intense diplomatic activity resulted in a peace agreement between the government of Rwanda (GoR) and the rebel Rwandese Patriotic Front (RPF). The political democratization within Rwanda during this period was related to the dynamic of conflict and its management.
Phase III focuses on the planning, authorization, and deployment of UNAMIR from mid-1993 until 6 April 1994, and its activities in early warning and conflict management against the backdrop of the progressive unravelling of the Arusha peace accords.
Phase IV covers the two weeks from 6 April until 21 April 1994. The immediate aftermath of the death of Rwanda's President in a suspicious plane crash - when it soon became evident that large-scale and targeted killings were under way - marks a distinct phase in terms of the international response. The report focuses on the information available that genocide was immanent and had, in fact, begun, and the corresponding responses, particularly as they relate to the question of what role UNAMIR should play. The dynamic interaction of the management of the genocide and the management of the conflict - or rather, the failure to distinguish between the two - is a central concern. The analysis focuses on the decision-making process in the UN over the future of UNAMIR, efforts by UNAMIR to carry out relevant parts of its mandate in the new situation, and the eventual decision by the Security Council to scale down the force, leaving only a symbolic military presence.
Phase V focuses on the subsequent decision to reverse withdrawal and re-engage the UN to protect civilians. The study assesses the efficacy of this process and its general impact on both the conflict on the ground and the violence against civilians. In this phase, another effort at retroactive conflict management was undertaken, i.e. the UN-authorized French Opération Turquoise. This section of the report relies on historical rather than sociological analysis, which would require micro-level assessments in the field.
Phase VI examines the issue of the security of the internally displaced within Rwanda and the establishment of militarized refugee communities in neighboring countries. This affected the security of the refugees and led to a massacre of several thousand persons at the Kibeho camp inside Rwanda. There were also wider implications in terms of conflict management. Social and political elements of the defeated regime who were responsible for the genocide regrouped to consolidate their power in exile. Thus, one element in a renewed cycle of conflict was formed.