RW-J-01: The Failure to Prevent — Three Accounts of Why the 1994 Genocide Was Not Stopped, and the Contested International Response (1993–2000)

1. Key Takeaways

• This document answers a different question from RW-B-01. The genocide anchor narrates what happened between 6 April and mid-July 1994 — the trigger, the decapitation, the machinery, the massacre sites, the RPF victory. This Block-J anchor asks why the genocide was not prevented, and why the international response failed once it began — and it answers by setting out three analytically distinct accounts, each grounded in a different scholarly tradition, each selecting different facts as load-bearing, and each implying a different locus of responsibility and a different counterfactual. The UN/international-failure account (Dallaire 2003; the Carlsson Inquiry 1999; the OAU Panel 2000; Power 2002; Melvern 2000) locates the failure in the Security Council and the P5 in the months and weeks around April 1994. The colonial/structural account (Mamdani 2001; Prunier 1995; Uvin 1998; Reyntjens 2013) locates it in the long structure — Belgian racialisation, the post-independence republics, the 1990–1994 political crisis — and argues that the "prevention window" was decades, not weeks. The RPF-and-contested-accountability account (Des Forges 1999 on RPF violence; the Bruguière and Mutsinzi instruments on the shoot-down; Reyntjens; Kuperman 2001) examines the conduct and subsequent narrative control of the force that won the war. The corpus presents all three with the same analytical care and adjudicates none of them.

• The early-warning record is the most fully documented of any modern mass atrocity, and the warning window was long. Between the 1990 RPF invasion and 6 April 1994 the warning record includes: the rehearsal massacres of 1990–1993 (Kibilira, the Bagogwe, Bugesera) documented contemporaneously by the international commission of inquiry that reported in March 1993; the report of UN Special Rapporteur Bacre Waly Ndiaye (August 1993), which explicitly raised the applicability of the Genocide Convention to the killings of Tutsi eight months before the genocide; the arming and training of militias and the RTLM incitement reported by UNAMIR and by the Belgian embassy through the winter of 1993–1994; and Force Commander Roméo Dallaire's cable of 11 January 1994 — the "genocide fax" — relaying an informant's account of arms caches, of Tutsi registration "for their extermination", and of a plan to kill Belgian peacekeepers to provoke withdrawal. The Department of Peacekeeping Operations, headed by Kofi Annan with Iqbal Riza as his deputy, instructed Dallaire not to raid the caches, judging the operation beyond UNAMIR's Chapter VI mandate, and to share the informant's intelligence with President Habyarimana — the head of the regime whose apparatus was the subject of the warning. No account of the prevention failure disputes this record; the accounts differ on what it proves.

• Account 1 — the UN/international-failure account — holds that the genocide was foreseen, foreseeable, and preventable at acceptable cost, and that the Security Council chose not to prevent it. Its load-bearing facts: UNAMIR was created weak (Resolution 872, 5 October 1993: ~2,548 personnel, a Chapter VI monitoring mandate, no enforcement authority) because the P5 — above all the United States after the 3–4 October 1993 Mogadishu battle — would authorise nothing stronger; Dallaire's January–March 1994 requests for authority and reinforcement were refused; and on 21 April 1994, two weeks into the genocide and after the killing of the ten Belgian peacekeepers had triggered Belgium's withdrawal, the Security Council voted unanimously in Resolution 912 to reduce UNAMIR to 270 personnel. The United States, operating under the restrictive peacekeeping doctrine codified as PDD-25 (signed 3 May 1994), instructed officials to avoid the unqualified word "genocide" — the "acts of genocide" formulation defended by State Department spokeswoman Christine Shelly on 10 June 1994 — and obstructed rapid reinforcement; the UNAMIR II force authorised by Resolution 918 (17 May 1994) deployed in strength only after the genocide had ended. The account's verdict was rendered by the UN's own Independent Inquiry (the Carlsson Report, 15 December 1999): the failure was systemic, the information available was sufficient, and responsibility ran through the Secretariat, the Security Council, and the member states.

• Account 2 — the colonial/structural account — holds that the genocide was the culmination of a racialised political structure built by colonial rule and reproduced by the post-independence republics, and that "prevention" framed as a 1994 peacekeeping question mistakes the symptom for the disease. Its load-bearing facts: the Belgian codification of Hutu and Tutsi as fixed, heritable, card-carried racial categories in the 1933–1934 identity-card census, ideologically scaffolded by the "Hamitic hypothesis" that cast the Tutsi as a foreign race; the 1959 revolution and the First and Second Republics' institutionalisation of anti-Tutsi quotas and the refugee diaspora whose exclusion produced the RPF; and the 1990 RPF invasion, which transformed a discriminatory but stable order into a wartime crisis in which the Hutu Power movement could cast every Tutsi as an internal enemy. On this account — argued in different registers by Mamdani (2001), Prunier (1995), Uvin (1998), and Reyntjens (2013) — the meaningful prevention failures were the failure to address the citizenship of the Tutsi diaspora before 1990, the failure of the Arusha process (RW-A-04) to bind the hardliners it excluded and threatened, and the failure of France in particular to condition its 1990–1993 military protection of the Habyarimana regime on dismantling the machinery of racial mobilisation. Belgium and France appear here not as bystanders who failed to act but as authors and underwriters of the structure that made genocide possible.

• Account 3 — the RPF and contested-accountability account — examines the one actor with forces on the ground that the other two accounts treat mainly as context: the Rwandan Patriotic Front. Its claims, each contested and each presented here with evidential grounding rather than endorsement: that the RPF's 1990 invasion initiated the war that radicalised the regime (a point Account 2 also registers); that the authorship of the 6 April shoot-down — the genocide's trigger — remains genuinely unresolved between the Hutu-extremist thesis (Mutsinzi Report 2009; the Trévidic–Poux ballistic findings 2012) and the RPF-responsibility thesis (the Bruguière ordonnance of 17 November 2006), as set out in RW-B-01 Section 3 and not re-argued here; that the RPF prioritised military victory over the protection of Tutsi concentrations and rebuffed proposals for a ceasefire-plus-intervention sequence that critics (Kuperman 2001; Reyntjens) argue might have widened the rescue window — a counterfactual the RPF and most genocide scholars reject as illusory, since the "ceasefire" on offer was with the génocidaire state itself; and that RPF soldiers killed Hutu civilians during and after 1994 — documented by Des Forges (1999), estimated in the suppressed UNHCR Gersony findings (1994) at [TBD-VERIFY: the commonly cited Gersony range is 25,000–45,000 killed April–August 1994; the report was never officially released and the figure cannot be confirmed], and acknowledged by the RPF only in limited, individualised form. These killings are an accountability question in their own right and a prevention-failure question insofar as fear of documenting them shaped the post-1994 international posture.

• The three accounts are not mutually exclusive, but they are not innocently complementary either — each is politically load-bearing in post-1994 Rwanda and beyond. Account 1 is the account the post-genocide Rwandan state and the RPF most fully endorse: it indicts the UN and the West, vindicates the RPF as the force that ended the genocide alone, and grounds the sovereignty-first doctrine ("never again will Rwanda wait for outsiders") that runs through RW-C-01 and RW-F-01. Account 2 is partially endorsed: the colonial construction of ethnicity is official Rwandan historiography (it underwrites the abolition of ethnic categories, RW-G-01), but its corollary — that the RPF invasion of 1990 was itself a cause of the radicalisation — is resisted. Account 3 is suppressed within Rwanda: the documentation of RPF killings sits at the boundary of the genocide-ideology and divisionism laws (2008, revised 2013), and scholars associated with it (Reyntjens most prominently) have been declared unwelcome. The analytical discipline this document models is to hold the distinction between Account 3 as documented contested record (Des Forges, Gersony, the ICTR Special Investigations) and the "double genocide" thesis — the claim of two equivalent genocides — which the corpus, following the overwhelming weight of scholarship and the ICTR's adjudicated record, treats as a denialist framing and does not present as a fourth account.

• The accountability record after 1994 split along the same three lines. The institutions of Account 1 produced inquiry and apology without sanction: the Carlsson Report (1999); the OAU Panel of Eminent Personalities' Rwanda: The Preventable Genocide (July 2000); President Clinton's Kigali airport statement of 25 March 1998; Belgian Prime Minister Guy Verhofstadt's apology in Kigali on 7 April 2000; Kofi Annan's acceptance of institutional and personal failure. No UN or member-state official was disciplined for the prevention failure. The structures of Account 2 were addressed rhetorically — the Duclert Commission's finding (26 March 2021) of France's "heavy and overwhelming responsibilities" and President Macron's Kigali speech of 27 May 2021 — but the colonial-inheritance argument has produced no reparative mechanism. The questions of Account 3 were institutionally foreclosed: the ICTR indicted no RPF member in its entire docket (1994–2015); Prosecutor Carla Del Ponte's Special Investigations into RPF crimes were terminated when the Security Council split the ICTY/ICTR prosecutor posts in 2003, a removal she attributed to her RPF inquiries; and the single transferred case — the 2008 Rwandan military trial for the June 1994 Kabgayi killings of clergy — produced convictions widely criticised as token.

• The counterfactual debate — could intervention have worked? — is the empirical core of Account 1 and the principal point of scholarly attack on it. Dallaire's contention, supported by the Scott Feil study for the Carnegie Commission (1998), is that a properly mandated force of roughly 5,000 deployed in April 1994 could have broken the genocide's momentum by securing Kigali and the main concentration sites. Alan Kuperman (Foreign Affairs 2000; The Limits of Humanitarian Intervention, 2001) counter-argues that given realistic intelligence lags and deployment timelines, even a determined Western intervention launched after the genocide was recognised could have saved at most a quarter of the eventual victims — and that the genocide's extraordinary speed, with the majority of killings concentrated in the first weeks, forecloses the rescue counterfactual. Critics of Kuperman (Melvern; Des Forges) reply that his clock starts artificially late: the warning record meant recognition should have predated 6 April, and the relevant failure was the pre-April refusal to resource and mandate the force already in place. The corpus records the debate as live; what no party to it disputes is that no serious attempt was made, so the counterfactual was never tested.

• The historiographical contest over the prevention failure has not closed; it has institutionalised — and it runs to 2026. France opened the Mitterrand-era archives under litigation pressure (the Graner case, Conseil d'État, June 2020) and through the Duclert process; Macron's April 2024 thirtieth-anniversary statement went further than 2021 in framing what France "could have" done [TBD-VERIFY: the pre-released Élysée formulation that France "with its Western and African allies, could have stopped the genocide" but "did not have the will" — the published video statement is reported to have been softer; the exact final wording requires confirmation]. The UN General Assembly fixed the official designation — the "1994 Genocide Against the Tutsi in Rwanda" — in its April 2020 resolution on the international day of reflection, itself a victory of the Rwandan state's framing over the older "Rwandan genocide" usage. Meanwhile the trials of the last fugitives (Félicien Kabuga, arrested in Paris in May 2020, found unfit to stand trial in 2023) and the universal-jurisdiction convictions in France, Belgium, and elsewhere kept the perpetrator-accountability track alive even as the prevention-accountability track — who answers for the failure to act — produced, across three decades, apologies but no sanction of any kind. That asymmetry is the enduring finding of this document.

---

2. The Warning Window: What Was Known and When (1990 – April 1994)

2.1 Why the Warning Record Matters to All Three Accounts

Every account of the prevention failure begins from the same evidentiary base: the record of what was known, by whom, and when, before 7 April 1994. The record matters differently to each account. For Account 1, it establishes foreknowledge at the Security Council and Secretariat level and therefore culpable inaction. For Account 2, it demonstrates that the warning signs were not anomalies but the visible surface of a structure that had been generating anti-Tutsi violence since 1959 — and that reading them required no special intelligence, only historical literacy. For Account 3, the same record is read for what it shows about the war: that the radicalisation accelerated in lockstep with the military confrontation that the RPF's 1990 invasion had opened, so that the warning window and the civil war are the same window. This section sets out the record itself; the three readings follow in Sections 3–5.

2.2 The Rehearsal Massacres and Their Contemporaneous Documentation (1990–1993)

The massacres of Tutsi that followed the 1 October 1990 RPF invasion — the Kigali round-ups of October 1990, Kibilira (October 1990), the Bagogwe massacres (January–March 1991), and Bugesera (March 1992), narrated in RW-B-01 Section 2 and RW-D-01 — were not hidden. They were documented while the Habyarimana regime was still in power by an International Commission of Investigation on Human Rights Violations in Rwanda, a four-organisation initiative (including the Fédération Internationale des Droits de l'Homme and Human Rights Watch/Africa) whose delegation, with Alison Des Forges among its members, worked in Rwanda in January 1993 and published its report in March 1993. The report documented state organisation of the massacres — local officials convening the killings, militia execution, central impunity — and concluded that responsibility reached the head of state's circle. Several commission members used the word "genocide" in characterising the pattern, with internal debate about whether the legal threshold had yet been met. The report's publication prompted a brief French statement of concern and no change in French military support; the regime issued denials; the killings' organisers suffered no consequence.

The United Nations' own machinery then produced what is, in retrospect, the single most pointed pre-genocide warning document. Bacre Waly Ndiaye, the UN Special Rapporteur on extrajudicial, summary or arbitrary executions, visited Rwanda in April 1993 and reported in August 1993. His report described the targeting of Tutsi as an ethnic group, considered the question "whether the term genocide should be applied", and explicitly invoked Article II of the 1948 Genocide Convention, noting that the victims were targeted "solely because of their membership of a certain ethnic group, and for no other objective reason". Ndiaye recommended concrete preventive measures, among them protection mechanisms for civilians and steps against the impunity of organisers. The report was filed in the UN human-rights machinery in Geneva and, as the Carlsson Inquiry later found, made no discernible impression on the political organs in New York. When the Security Council designed UNAMIR in the weeks immediately following Ndiaye's report, the report figured nowhere in its deliberations — a disconnection between the UN's human-rights and peace-and-security tracks that the Carlsson Report treated as emblematic of the system's failure.

2.3 UNAMIR as Designed: Resolution 872 and the Chapter VI Ceiling

The United Nations Assistance Mission for Rwanda was established by Security Council Resolution 872 of 5 October 1993 to monitor implementation of the Arusha Accords (RW-A-04): observe the ceasefire, assist with the demilitarised zone, monitor the security situation through the transitional period, and contribute to the security of Kigali — the last through the "Kigali Weapons-Secure Area" arrangement. The authorised strength was approximately 2,548 military personnel; the mandate was Chapter VI — consent-based monitoring, not enforcement. The design reflected the Council's post-Somalia temper precisely. The battle of Mogadishu (3–4 October 1993) had occurred two days before the resolution; the United States, whose Congress and administration were recoiling from peace-enforcement, pressed successfully for a minimal, cheap mission. Dallaire, who had led the August 1993 reconnaissance, had assessed a requirement closer to 4,500 with robust rules of engagement; the reconnaissance report itself, drafted under time pressure and with what Dallaire later acknowledged was inadequate intelligence preparation, underestimated the spoiler threat. Deployment was slow and under-equipped: the force that existed in early 1994 lacked serviceable armoured personnel carriers, ammunition reserves, medical capacity, and an intelligence cell worth the name. The Belgian contingent of roughly 440 — the force's professional core — was itself the residue of Belgium's failed effort to persuade other contributors to reinforce.

The mandate's interpretive ceiling was set in New York. The Department of Peacekeeping Operations — Under-Secretary-General Kofi Annan, his deputy Iqbal Riza, and military adviser Maurice Baril — read UNAMIR's security role narrowly and consistently instructed restraint, an interpretive posture shaped by the Secretariat's institutional memory of Somalia and by the knowledge that the Council would not reinforce a mission that got into trouble. The Special Representative of the Secretary-General, Jacques-Roger Booh-Booh, transmitted political assessments through early 1994 that were consistently more reassuring than his Force Commander's — a divergence that the Carlsson Inquiry documented and that gave New York a menu of readings from which the less alarming could be, and was, selected.

2.4 The Genocide Fax of 11 January 1994

On the night of 10–11 January 1994, Dallaire transmitted to DPKO the cable that has become the emblem of the prevention failure. Sourced to an informant code-named "Jean-Pierre", described as a trainer within the Interahamwe with access to its leadership, the cable reported: that the militia had been trained in accelerated three-week cycles in army camps; that the informant had been ordered to register all Tutsi in Kigali, and that he suspected the registration was "for their extermination", with his cells capable of killing "up to 1,000 Tutsis" within twenty minutes of receiving the order; that a plan existed to provoke and kill Belgian peacekeepers in order to "guarantee Belgian withdrawal from Rwanda"; and that a major weapons cache existed which the informant was prepared to reveal in exchange for protection. Dallaire announced his intention to raid the cache within thirty-six hours and closed with the now-famous flourish, "Peux ce que veux. Allons-y." ("Where there's a will, there's a way. Let's go.")

The reply, drafted in DPKO and signed out by Riza on Annan's behalf on 11 January, refused the operation: the contemplated raid exceeded UNAMIR's mandate; the information should instead be conveyed to President Habyarimana — with the demand that he act against the arms caches — and shared with the ambassadors of Belgium, France, and the United States in Kigali. The instruction was carried out; the régime's response was, predictably, nothing; the informant was not extracted [TBD-VERIFY: the fate of "Jean-Pierre" after January 1994 is not established in the public record]. Dallaire renewed his requests for authority to conduct deterrent weapons-seizure operations in late January and February 1994 and was refused each time. When the fax surfaced publicly in late 1995, it transformed the prevention debate: here was a contemporaneous, specific, operational warning — extermination registration, anti-Belgian provocation, caches — transmitted to the centre of the UN system twelve weeks before the genocide, every element of which was subsequently borne out. The fax is treated across the literature as the strongest single exhibit of Account 1; what it proves beyond Secretariat-level caution is precisely what the three accounts dispute.

2.5 The Final Quarter: January–April 1994

Between January and April 1994 the warning stream broadened. The Belgian embassy under Ambassador Johan Swinnen and Belgian military intelligence reported the radicalisation in detail, including RTLM's escalating incitement and lists of persons to be killed; Brussels lobbied — without success — for a stronger UNAMIR mandate. UNAMIR reported the blockage of the Broad-Based Transitional Government's installation, the January–February assassinations and militia violence in Kigali (including the killing of PSD minister-designate Félicien Gatabazi and the CDR leader Martin Bucyana in February), the distribution of weapons to civilians, and the regime-aligned media's open rehearsal of apocalyptic rhetoric. A CIA desk-level analysis of January 1994 is reported to have war-gamed the collapse of Arusha and projected catastrophic violence [TBD-VERIFY: Power (2002) reports a January 1994 CIA analysis projecting up to half a million deaths in a worst case; the underlying document has not been declassified in full]. On 5 April 1994 — the day before the shoot-down — the Security Council, in Resolution 909, extended UNAMIR's mandate while expressing "deep concern" at the transition's paralysis and threatening the mission's withdrawal within six weeks if the Arusha institutions were not installed: the Council's last pre-genocide act was thus a threat to leave, not to reinforce. The warning window closed at 20:25 the following evening. What was done within it — and by whom it should have been read — is the matter of the three accounts that follow.

3. Account 1: The UN/International Failure Account

3.1 The Account's Architecture and Sources

The UN/international-failure account is the dominant account in the Anglophone literature and in official post-mortems. Its principal sources are Dallaire's memoir Shake Hands with the Devil (2003), the UN's own Independent Inquiry into the Actions of the United Nations during the 1994 Genocide in Rwanda — the Carlsson Report, commissioned by Secretary-General Annan and delivered on 15 December 1999 under former Swedish Prime Minister Ingvar Carlsson, with Han Sung-Joo of South Korea and Lieutenant-General Rufus Kupolati of Nigeria — the OAU's International Panel of Eminent Personalities report Rwanda: The Preventable Genocide (July 2000, chaired by former Botswanan President Ketumile Masire), Linda Melvern's A People Betrayed (2000) and Conspiracy to Murder (2004), and Samantha Power's "A Problem from Hell": America and the Age of Genocide (2002), whose Rwanda chapter is built on the US documentary record. The account's structure is a chain of decision points, at each of which an identifiable institution chose the weaker option with sufficient information to know better.

3.2 The Chain of Decisions

The first link is the design decision of October 1993 (Section 2.3): a Chapter VI mission at half the assessed strength, adopted two days after Mogadishu because the United States would fund nothing more. The second is the January–March interpretive regime: DPKO's refusal of the cache raids, its instruction to share the informant's intelligence with the regime being warned against, and its consistent reading of the mandate at its narrowest — decisions taken by named officials (Annan, Riza) who later accepted responsibility for them, Riza telling the PBS Frontline documentary The Triumph of Evil (1999) that the Secretariat had judged the warnings within the pattern of previous ceasefire violations [TBD-VERIFY: the specific formulation "we get hyperbole in many reports" attributed to Riza in this documentary could not be independently corroborated in the fetch-blocked environment's search-only verification pass of 2026-08-29; the PBS Frontline interview transcript exists and Riza's on-record position in it is consistent with this characterisation, but the exact quoted words require confirmation directly against the transcript]. The third link is the first-week collapse: after the shoot-down and the killing of Prime Minister Uwilingiyimana and the ten Belgian paratroopers (RW-B-01 Section 4), Belgium — precisely as the January informant had predicted — withdrew its contingent, and then lobbied other capitals for UNAMIR's complete termination, on the reasoning that a Belgian withdrawal alone would be a national humiliation. Dallaire's requests in the first days for reinforcement and a mandate to protect civilians — his estimate, later elaborated, was that roughly 5,000 capable troops with enforcement authority could halt the killing — were not granted.

The fourth and most notorious link is Resolution 912 of 21 April 1994. Two weeks into the genocide, with tens of thousands already dead and the killing's character reported daily by UNAMIR, the Security Council voted unanimously to reduce UNAMIR to a token strength of 270 (the force in practice retained around 450 through Dallaire's foot-dragging). The Council's choice was framed by the Secretariat as among three options — reinforcement, token presence, or withdrawal — and the non-aligned members' appetite for reinforcement was defeated by US and UK resistance and Belgian lobbying for termination. The fifth link is the recognition refusal: through April and May the United States avoided the unqualified term "genocide" — the Pentagon's discussion papers warned that a genocide finding could commit the government "to actually 'do something'" [TBD-VERIFY: the quoted phrase appears in a 1 May 1994 Defense Department discussion paper released through the National Security Archive; verify exact wording], and the State Department's spokeswoman Christine Shelly produced the "acts of genocide" casuistry in the briefing-room exchange of 10 June 1994. Susan Rice, then on the NSC staff covering peacekeeping, is reported by Power (2002) to have asked at an interagency teleconference what effect the use of the word "genocide" would have on the November congressional elections [TBD-VERIFY: Power sources the remark to officials present; Rice subsequently stated she did not recall making it and that if she had, it was wrong]. The sixth link is the reinforcement that never arrived: Resolution 918 (17 May 1994) authorised UNAMIR II at 5,500, but the force generation collapsed into months of haggling — including the celebrated dispute over the lease terms for fifty US armoured personnel carriers, which reached Kampala only weeks later, unfit for immediate use — so that UNAMIR II deployed in strength only after the RPF victory. The chain closes with Opération Turquoise (Resolution 929, 22 June 1994), which Account 1 treats ambivalently: the only P5 deployment of the period, mounted by the government most compromised by its patronage of the perpetrator regime, and assessed in detail in RW-B-01 Section 8.

3.3 The Account's Verdicts: Carlsson, the OAU Panel, and the Apologies

The Carlsson Report's findings, delivered to the man who had headed DPKO during the genocide and who had commissioned the inquiry as Secretary-General, were unsparing: "The failure by the United Nations to prevent, and subsequently, to stop the genocide in Rwanda was a failure by the United Nations system as a whole." It found the information available before April 1994 sufficient to anticipate the catastrophe; criticised the handling of the 11 January cable; found Resolution 912 a decision that "will forever be remembered with regret" [TBD-VERIFY: exact formulation]; and distributed responsibility across the Secretariat, the Council, and the member states — without recommending consequences for any individual. Annan's response accepted the findings; he had already, on the report's release, expressed "deep remorse", and at the tenth anniversary in 2004 stated that he "could and should have done more to sound the alarm and rally support". The OAU Panel's Rwanda: The Preventable Genocide (2000) extended the indictment beyond the UN to the principal states — the United States, France, Belgium — and to the international financial institutions, and recommended reparations to Rwanda; the recommendation was never acted upon. The state apologies followed the same pattern of acknowledgement without sanction: President Clinton's Kigali airport statement of 25 March 1998 ("we did not act quickly enough after the killing began... we did not immediately call these crimes by their rightful name"), framed carefully as a failure of speed and nomenclature rather than of will, and accompanied by the claim that "all over the world there were people like me sitting in offices... who did not fully appreciate the depth and the speed" of the events — a claim the documentary record assembled by Power and the National Security Archive substantially contradicts; Belgian Prime Minister Verhofstadt's apology of 7 April 2000 in Kigali, the most unqualified of the cycle, following the Belgian Senate's 1997 commission of inquiry; and the French trajectory, which reached its reckoning only with Duclert in 2021 (Section 7).

3.4 The Counterfactual Contest

Account 1 stands or falls on the claim that intervention would have worked, and this is where it has been most rigorously attacked. The Feil study (Scott Feil, Preventing Genocide: How the Early Use of Force Might Have Succeeded in Rwanda, Carnegie Commission on Preventing Deadly Conflict, 1998), built on a panel of senior military officers, concluded that a force of roughly 5,000 deployed between 7 and 21 April with robust rules of engagement could have significantly altered the outcome by securing Kigali and deterring the killing's spread to the prefectures. Alan Kuperman's rejoinder (Rwanda in Retrospect, Foreign Affairs, 2000; The Limits of Humanitarian Intervention, 2001) reconstructed the killing's tempo — with the largest massacres concentrated in the first three weeks — and the realistic timelines of Western strategic lift, and concluded that even a maximal intervention ordered when the genocide's character became unambiguous could have saved perhaps 125,000 of the eventual victims, a quarter at most; prevention, he argued, would have required acting on pre-April warning, which he assessed as genuinely ambiguous before the fact. The responses to Kuperman (Des Forges in a published exchange; Melvern) attack precisely that assessment: the warning record of Section 2 was not ambiguous, the Ndiaye report and the January cable named the thing itself, and Kuperman's clock-start concedes the account's central claim — that the failure was one of attention and will before 6 April, not of logistics after it. The corpus's position is that the counterfactual is unresolvable in principle — no intervention was attempted — but that the debate's terms matter: Account 1's strongest form is not "the genocide could have been stopped in week two" but "the force already in place was designed, constrained, and then cut in ways that the available information made indefensible".

3.5 What Account 1 Does Not Explain

Account 1's limits define the space the other two accounts occupy. It treats the genocide's occurrence as given and asks only why outsiders failed to stop it — leaving unexamined the question of why Rwandan politics had produced an exterminationist apparatus at all (Account 2's question). It centres the Security Council and renders Rwandans — perpetrators, victims, and the RPF alike — as the background against which Western and UN choices are judged, a framing that both the Rwandan government's own usage and the postcolonial critique have noted, from opposite directions. And it has little to say about the one armed actor that did intervene — the RPF — whose conduct, victory, and subsequent custody of the historical record form Account 3. The Rwandan state's embrace of Account 1 is itself a historiographical fact of consequence: the account that indicts the UN and the West is also the account that most completely vindicates the RPF, and it underwrites the post-1994 sovereignty doctrine — never again will Rwanda outsource its survival — that recurs across RW-C-01, RW-D-08, and RW-F-01.

4. Account 2: The Colonial/Structural Account

4.1 The Account's Architecture and Sources

The colonial/structural account relocates the prevention question from the Security Council chamber to the longue durée of Rwandan state formation. Its foundational text is Mahmood Mamdani's When Victims Become Killers: Colonialism, Nativism, and the Genocide in Rwanda (2001), with the historical groundwork laid by Gérard Prunier's The Rwanda Crisis (1995), the political-economy variant developed in Peter Uvin's Aiding Violence: The Development Enterprise in Rwanda (1998), the post-genocide extension in Filip Reyntjens's Political Governance in Post-Genocide Rwanda (2013), and Mamdani's later restatement of the decolonisation-of-the-political argument in Neither Settler Nor Native (2020). Where Account 1 asks "why did the guardians fail to act in 1994?", Account 2 asks "what made a genocide of neighbours against neighbours possible — and what would prevention have actually required?" Its answer: prevention framed as peacekeeping addresses only the final hundred days of a catastrophe a century in the making; the meaningful prevention failures were structural and political, and several of the would-be rescuers of Account 1 were authors of the structure.

4.2 The Colonial Construction: Race, Cards, and the Hamitic Hypothesis

The account's foundation is the transformation of Hutu and Tutsi from fluid, locally variable social categories — bound up with cattle, clientage, region, and lineage, and crossable in both directions — into fixed, heritable, state-certified races. Pre-colonial Rwanda was a stratified kingdom in which the categories carried real hierarchy; the German and above all the Belgian administrations racialised that hierarchy. The intellectual scaffold was the "Hamitic hypothesis": the nineteenth-century European thesis (associated with John Hanning Speke and elaborated by missionary ethnography) that the Tutsi were a superior "Hamitic" race of northern origin — foreign conquerors of the "Bantu" Hutu — and therefore natural intermediaries of colonial rule. Belgium governed through this theory: it consolidated Tutsi chiefly monopoly in the administrative reforms of the late 1920s, and in the 1933–1934 census issued every Rwandan an identity card bearing a fixed ethnie — Hutu, Tutsi, or Twa — heritable patrilineally and unchangeable thereafter. Mamdani's formulation is that the colonial state made the Tutsi a race, not an ethnicity: a population defined as foreign to the nation, which is precisely the predicate genocide requires. The identity card survived the revolution of 1959, the republics, and every reform until 1994, when — as RW-B-01 Section 5 records — it functioned at the roadblocks as the genocide's sorting instrument. On Account 2, the card is the through-line: a Belgian colonial document, checked at a militia roadblock sixty years later, determining who died.

4.3 The Republics and the Institutionalisation of Exclusion

The second structural layer is the post-independence reproduction of the racial frame with the valence reversed. The 1959 revolution and the Belgian administration's late switch of patronage produced a "Hutu nation" state ideology (RW-A-03): the First Republic under Kayibanda cast the Tutsi as the inyenzi threat and conducted pogroms in 1959–1964 and 1973 that drove hundreds of thousands into exile; the Second Republic under Habyarimana formalised "ethnic and regional equilibrium" quotas in education and employment and froze the exiles out — refusing through the 1980s the return of a diaspora that had grown to some 600,000 [TBD-VERIFY: diaspora estimates by the late 1980s range across sources from ~400,000 to ~700,000], on the stated ground that the country was "full". Account 2's point is that the refugee question was the visible, decades-long, internationally legible prevention failure: a stateless population, denied return and citizenship, militarising in Uganda inside the NRA, while the international system treated Rwanda as a model development state. Uvin's Aiding Violence gives this its sharpest form: Rwanda in the 1980s was a donor darling — among the highest per-capita aid recipients in Africa — and the development enterprise built its projects through the discriminatory state, normalising the quotas, the cards, and the exclusion, so that the apparatus of international engagement reinforced the structure it would later profess astonishment at. Structural adjustment after 1990, the coffee-price collapse, and acute land scarcity (treated in RW-B-01 Section 2.5 as enabling conditions) complete the political-economy layer.

4.4 The 1990 Invasion and the Crisis of the Racialised State

The third layer is the transformation wrought by the 1 October 1990 RPF invasion (RW-B-02). On Account 2, the invasion — itself the product of the refugee exclusion — converted a discriminatory peace into a war in which the Hutu Power movement could fuse the racial frame with a security panic: every Tutsi an ibyitso, every concession at Arusha an existential surrender. Prunier's reconstruction of 1990–1994 traces how the regime's response to military pressure and democratisation pressure alike was the acceleration of racial mobilisation — the rehearsal massacres, the militias, Kangura and RTLM — and Mamdani presses the uncomfortable corollary: the genocide was carried out in the name of a majority that had been taught, by the colonial state and both republics, to understand itself as the native nation repelling a returning alien race. Prevention, on this reading, would have required addressing the structural crisis opened in 1990 — the citizenship of the diaspora, the survival guarantees of the regime's losing factions, the demobilisation of the racial frame itself — and the Arusha process (RW-A-04) attempted precisely this and failed: it produced a settlement that redistributed power on paper while leaving the Hutu Power coalition armed, terrified, and excluded, a configuration Account 2 reads as structurally explosive regardless of what UNAMIR was or was not authorised to do. The CDR's exclusion from the Arusha institutions, the 60/40 military integration formula that spelled mass demobilisation for the FAR, and the absence of any enforcement mechanism are, in this account, the prevention failures that mattered — committed not by DPKO in January 1994 but by the negotiators and guarantors of 1992–1993.

4.5 Complicity of the Patrons: France and Belgium in the Structural Frame

Account 2 reframes the two European patrons. Belgium appears twice: as the colonial author of the racial architecture, and as the post-independence patron that switched sides in 1959–1962 and sponsored the Hutu-nation state into being. France appears as the power that underwrote the structure's terminal phase: from October 1990 (Opération Noroît) through 1993, French troops, advisers, arms, and credit sustained the Habyarimana regime against the RPF — including, as the Duclert Commission later documented from the archives, sustained warnings within the French system about the regime's racial radicalisation that were subordinated to the Élysée's strategic commitment to the Franco-African sphere and its reading of the RPF as an Anglophone-Ugandan proxy. On Account 2, France's prevention failure was not Turquoise's lateness in June 1994; it was the unconditional patronage of 1990–1993, which removed every incentive the regime had to dismantle the machinery of mobilisation. Duclert's 2021 formula — "heavy and overwhelming responsibilities", but not complicity in the genocide itself — is, from this account's standpoint, an exact description of structural complicity: responsibility for the conditions, not the intent. The account thus dissolves Account 1's central dramatic frame — the bystander who failed to rescue — and replaces it with a different figure: the architect and underwriter who professed surprise at the collapse of the building.

4.6 What Account 2 Does Not Explain

The structural account's limits are the mirror of its strengths. It explains possibility, not decision: as RW-B-01 Section 2.5 records, the scholarship has largely rejected the reduction of the genocide to its structural conditions, because structure does not select its own outcome — the same colonial inheritance and the same land pressure did not produce genocide in 1973 or 1991, and the decisive variable remained the organised choice of an identifiable elite in 1994 (Des Forges; Guichaoua 2015). Pressed too far, the structural account risks a diffusion of responsibility that the perpetrators' own defence counsel at the ICTR attempted — blaming Belgium, the cards, and history for decisions taken by named men at the Ministry of Defence on the night of 6–7 April. It also generates no operational prevention doctrine: "decolonise the political identities" is a generational project, not an answer to the question of what should have been done in January 1994. And its account of 1990–1994 shares an unresolved boundary with Account 3: if the RPF invasion was a structural trigger of the radicalisation, the invasion's own justification — the indefensible exclusion of the diaspora — was equally structural, and the account provides no principled way of assigning weight between them. Mamdani himself draws from the analysis a conclusion that the post-1994 Rwandan state has partially adopted and partially refused: the abolition of the racial categories (adopted, as RW-G-01 documents) and a political settlement legitimate to both Hutu and Tutsi as political identities (refused, on the official ground that the categories themselves were the colonial poison — a disagreement that runs through the post-genocide governance debate in RW-D-07).

5. Account 3: The RPF and Contested Accountability Account

5.1 The Account's Status and the Discipline of Its Presentation

The third account is unlike the first two in its standing. Accounts 1 and 2 are contested in emphasis but rest on official inquiries and a settled scholarly base. Account 3 is politically explosive, partially suppressed, and entangled with genocide denial at its outer edge — which is precisely why the corpus's three-account discipline requires presenting it with the same evidential care, neither endorsing it nor allowing its abuse by denialist framings to justify its omission. Its sources span an evidential spectrum: at the most authoritative end, Alison Des Forges's Leave None to Tell the Story (1999) — the foundational documentary account of the genocide, which devotes a substantial section to RPF killings and was produced by the scholar least suspectable of minimising the genocide; the Gersony findings (1994); the ICTR record of the Del Ponte Special Investigations; the duelling shoot-down instruments (Bruguière 2006; Mutsinzi 2009; Trévidic–Poux 2012); Filip Reyntjens's body of work on the RPF's conduct and post-1994 narrative control; and Alan Kuperman's analysis of the RPF's April–July 1994 military choices. At the contested outer edge sit works such as Judi Rever's In Praise of Blood (2018), whose strongest claims the mainstream scholarship has not accepted and which this corpus cites only to map the controversy. The account has four distinct claims, which must not be run together, because they have very different evidentiary standing.

5.2 Claim One: The Invasion and the War as Preventable Origin

The first claim overlaps with Account 2's final layer: that the RPF's October 1990 invasion, whatever the justice of the refugee cause, was a chosen act of war that foreseeably endangered the Tutsi inside Rwanda — who had no vote in it — and that each subsequent RPF escalation (notably the February 1993 offensive that displaced up to a million people and hardened Hutu opinion) strengthened the Hutu Power coalition it claimed to oppose. Kuperman has pressed this furthest, arguing that the RPF leadership understood the retaliatory risk to Rwandan Tutsi and proceeded anyway, valuing victory over their protection [TBD-VERIFY: Kuperman's 2004 article "Provoking genocide: a revised history of the Rwandan Patriotic Front" advances this claim; its evidentiary basis — interviews with RPF figures — is contested]. The mainstream literature accepts the descriptive core (the war radicalised the regime; the February 1993 offensive was a political gift to the CDR) while rejecting the implied transfer of responsibility: the choice to answer invasion with the extermination of civilians belonged entirely to those who organised it, a point on which Des Forges, Prunier, and the ICTR jurisprudence are unanimous. The corpus records the claim in this disciplined form: the invasion is part of the causal history of the radicalisation; it is no part of the moral or legal responsibility for the genocide.

5.3 Claim Two: The Shoot-Down Question

The second claim concerns the trigger. If the RPF shot down Habyarimana's aircraft on 6 April 1994, then the act that activated the genocide's prepared apparatus was an RPF decision — with consequences for the prevention question, since the shoot-down destroyed the Arusha framework that every prevention strategy presupposed. The evidentiary state of this question is set out fully in RW-B-01 Section 3 and is not re-argued here; in summary: the Bruguière ordonnance (17 November 2006) issued international arrest warrants against members of Kagame's entourage on the RPF-responsibility thesis, producing a Rwanda–France diplomatic rupture; the Rwandan Mutsinzi Report (2009) and the independent ballistic-acoustic work of the French investigating judges Trévidic and Poux (January 2012) located the probable firing point in the FAR-controlled Kanombe zone, consistent with the Hutu-extremist thesis; the French investigation closed with a non-lieu in December 2018, upheld on appeal in 2020 [TBD-VERIFY: Paris Court of Appeal confirmation commonly dated 3 July 2020; a final cassation disposition is reported in 2022 — exact dates require confirmation]. The question remains formally unresolved; the weight of the technical findings since 2010 has run against the RPF-responsibility thesis without extinguishing it. For this document the essential point is methodological: Account 3's shoot-down claim is a genuinely open evidentiary question, and — as RW-B-01 insists — its answer, either way, does not alter the character or responsibility of the genocide that followed.

5.4 Claim Three: The Advance, the Ceasefire Question, and the Rescue Counterfactual

The third claim concerns the RPF's conduct of the war during the genocide. The RPF resumed operations within hours of the shoot-down and fought a continuous campaign to the fall of Kigali on 4 July. The critics' argument (Kuperman 2001; Reyntjens) runs: the RPF rejected ceasefire proposals during April — including those pressed by UNAMIR and regional mediators — on the stated ground that one does not negotiate a ceasefire with a government conducting extermination; it opposed, publicly and at the Security Council, proposals for a reinforced international intervention in late April and May, fearing a force that would freeze its advance short of victory [TBD-VERIFY: the RPF's 30 April 1994 statement opposing intervention on the ground that "the genocide is almost completed" and that an intervention force would be too late is widely cited; verbatim text and context require confirmation]; and its campaign plan prioritised enveloping movements and the destruction of the FAR over the immediate relief of the largest Tutsi concentration sites, with Kuperman calculating that a direct drive toward the south-west could have reached some massacre zones weeks earlier. Dallaire's memoir records his own anguished exchanges with Kagame over the advance's pace, including Kagame's reported observation that the war would entail sacrifices [TBD-VERIFY: Dallaire (2003) recounts Kagame remarking words to the effect that "there will be many sacrifices in this war" in response to concerns that the advance's pace was costing Tutsi lives; the verbatim quotation requires confirmation against the text]. The defence — which is also the mainstream position — replies that the counterfactual is incoherent: the proposed ceasefire partner was the génocidaire state, whose consistent use of every pause was further killing; the intervention the RPF opposed was, by June, the French Turquoise operation, whose effect in the south-west included sheltering the perpetrators' retreat; and the RPF's ninety-day campaign remains the only action by any actor that actually stopped the genocide. The corpus records both positions; the underlying fact — that the RPF's strategic choices were made by a belligerent fighting to win, not by a rescue force — is not in dispute and is the analytically useful core of the claim.

5.5 Claim Four: RPF Killings and the Foreclosure of Accountability

The fourth claim has the strongest documentation and the most consequential afterlife. That RPF/RPA soldiers killed Hutu civilians — in reprisals, in screening operations, and in massacres during and after the advance — is documented in Des Forges (1999), which estimated RPF killings in 1994 in the range of [TBD-VERIFY: Des Forges/HRW cite "perhaps 25,000 to 30,000" for April–August 1994 drawing in part on the Gersony findings; the figure is an estimate over contested data]; in the Gersony findings of September–October 1994, in which a UNHCR team under Robert Gersony reported systematic killings in RPF-controlled zones — findings the UN never published and whose existence it long declined to confirm, in what Des Forges documented as a deliberate suppression to protect the new government and the repatriation operation; and in the subsequent record (the Kibeho camp killings of April 1995 and the DRC dimension are treated in RW-A-01 and RW-J-02's planned mapping-report companion). The accountability record is asymmetric by design: the ICTR indicted no RPF member in its twenty-one years; Prosecutor Carla Del Ponte's Special Investigations into RPF crimes (2001–2003) were met by Rwandan governmental obstruction — including the disruption of witness travel — and ended when the Security Council, with US and UK support, split the ICTY/ICTR prosecutorships in August 2003 and did not reappoint her at the ICTR, an outcome she attributed directly to the RPF inquiries (Del Ponte 2008). The single emblematic case — the June 1994 killing of thirteen clergy including Archbishop Vincent Nsengiyumva at Kabgayi — was transferred to Rwandan military justice in 2008 and produced two convictions of junior officers with sentences widely assessed (Human Rights Watch 2008) as token, and acquittals of the senior accused. President Kagame has acknowledged that RPF soldiers committed killings and stated that perpetrators were punished, while rejecting any equivalence with the genocide [TBD-VERIFY: the specific 2014 acknowledgement referenced in the planning literature for this document — its exact text, venue, and date — has not been located; earlier acknowledgements in this form appear in Kagame interviews from the late 1990s and 2000s. Treat the 2014 attribution as unconfirmed]. The post-1994 state's control of the archive, the genocide-ideology and divisionism laws (2008, revised 2013), and the exclusion of researchers associated with this record (Reyntjens declared persona non grata; the BBC's 2014 documentary Rwanda's Untold Story banned and condemned by parliamentary resolution) mean that Claim Four's evidentiary base is effectively frozen at its 1994–2010 state — itself a historiographical fact this document records.

5.6 The Boundary: Account 3 Is Not the "Double Genocide" Thesis

The discipline of this account requires drawing one boundary without ambiguity. The documented record of RPF killings — Des Forges, Gersony, the Del Ponte investigations — establishes serious crimes that international law would characterise as war crimes or crimes against humanity, for which accountability has been foreclosed. It does not establish, and its principal documentarians have explicitly rejected, the "double genocide" thesis: the claim that the RPF conducted a genocide against the Hutu equivalent and parallel to the genocide against the Tutsi. Des Forges rejected the equivalence in terms; the scale, organisation, intent, and adjudicated record of the two bodies of violence are categorically different; and the equivalence claim functions, in the Rwandan and Francophone polemical context, as a minimisation of the genocide itself — which is why Rwandan law criminalises it and why this corpus, which treats the genocide as settled fact of the most adjudicated kind (RW-B-01), declines to present "double genocide" as a fourth account. The cost of that boundary, recorded honestly, is that the Rwandan state has at times deployed the denialism charge against the documented record itself — collapsing Des Forges's findings, Reyntjens's scholarship, and genuine denialism into a single prohibited category. Holding the boundary in both directions — RPF crimes are documented and unaccounted for; the double-genocide thesis is a denialist framing — is the precise analytical task this account exists to perform.

6. The Three Accounts in Dialogue: Convergence, Divergence, and Political Function

6.1 Points of Convergence

The three accounts converge on more than is usually noticed. First, on the fact of foreknowledge: none disputes the warning record of Section 2; the Carlsson Inquiry, Mamdani, and Reyntjens all treat the genocide as foreseen in the operative sense. Second, on the failure of the Arusha framework: Account 1 reads Arusha's collapse as the failure the peacekeepers were too weak to arrest; Account 2 reads the settlement itself as structurally unenforceable; Account 3 reads its terms (the 60/40 integration formula, the CDR's exclusion) as guaranteeing a hardliner backlash — but all three locate the catastrophe's proximate political origin in the August 1993 – April 1994 implementation void documented in RW-A-04. Third, on the marginality of the actual response: every account agrees that what the international system did between April and July 1994 — Resolution 912, the "acts of genocide" casuistry, the UNAMIR II force-generation farce, Turquoise's ambiguities — ranged from irrelevant to harmful, and that the genocide was ended by the RPF's military victory and by nothing else. Fourth, methodologically: all three accounts, in their serious forms, affirm the genocide's settled character and reject its relativisation; the divergence is about prevention and accountability, not about the event.

6.2 Points of Divergence

The divergences are structured, and they can be stated as a set of axes. The locus of agency: Account 1 indicts institutions that failed to act (the Council, the Secretariat, the P5); Account 2 indicts institutions that acted for decades — the colonial state, the republics, the patrons, the development enterprise — and built the structure; Account 3 examines the actor whose actions are usually treated as exogenous. The time horizon of prevention: weeks for Account 1 (January–April 1994), decades for Account 2 (1933–1994), and the war years for Account 3 (1990–1994). The operative counterfactual: a reinforced UNAMIR with enforcement authority (Account 1); a decolonised political settlement and a resolved refugee-citizenship question (Account 2); a paused advance, an accepted intervention, or an untriggered Arusha framework (Account 3) — counterfactuals that are not merely different but lodged in different decades and assigned to different deciders. The implied reform programme: Account 1 generated the actually existing institutional legacy — the Brahimi Report (2000), the responsibility-to-protect doctrine adopted at the 2005 World Summit, the UN Office on Genocide Prevention — while Account 2's programme (the decolonisation of political identity) was adopted only inside Rwanda and on the RPF's terms, and Account 3's programme (symmetrical accountability) was adopted nowhere. The beneficiary: each account, whatever its analytical merits, performs political work — and the work differs.

6.3 The Political Economy of the Accounts

The accounts' political functions deserve explicit statement because they shape what evidence each account's carriers have incentives to develop or suppress. Account 1 is the official account of post-genocide Rwanda and, in a penitential register, of the Western states themselves: for Kigali it underwrites the sovereignty-first doctrine, the standing rebuke available against any UN or Western criticism (deployed, as RW-D-08 and RW-F-01 document, against the UN Group of Experts reporting on the DRC), and the moral architecture of the Kwibuka commemorations; for Washington, Brussels, and the UN it offers the manageable currency of apology — acknowledgement of a failure to act being far cheaper than acknowledgement of complicity in structure (Account 2) or of selective accountability (Account 3). Account 2 is split by the post-1994 state: its colonial-construction core is official Rwandan historiography, taught in the curriculum RW-G-01 describes and inscribed at the Kigali Genocide Memorial, while its 1990–1994 corollary — the invasion as accelerant — is officially resisted; in France, conversely, the structural account became after 2021 the vehicle of the Duclert reckoning. Account 3 has no state carrier at all: its evidentiary development depended on individuals (Des Forges, Gersony, Del Ponte, Reyntjens) working against the institutional grain of both the Rwandan state and the Western governments invested in the post-1994 partnership — which is both a reason for epistemic caution about its gaps and a datum about why the gaps exist. A corpus reader should therefore expect Account 1 to be evidentially over-developed relative to its explanatory weight, Account 3 under-developed, and Account 2 selectively developed along the lines of its national sponsors' interests.

6.4 Can the Accounts Be Composed?

A composed reading — which several of the major scholars in practice adopt — runs as follows, offered here as synthesis rather than adjudication. The Belgian colonial state and the two republics built a racialised political structure that made an exterminationist project conceivable and administratively executable (Account 2). The 1990–1994 war, opened by the RPF's invasion of a state that had made the diaspora's return impossible, converted that structure into an active crisis, which the Habyarimana regime's inner circle resolved into a prepared genocide (Accounts 2 and 3, converging on the record in RW-D-01 and RW-B-01). The international system, holding warning sufficient to act and a force already in place, declined at every decision point between October 1993 and May 1994 to pay any material price to interrupt the preparation or the execution (Account 1). The genocide was ended by a belligerent fighting for victory, whose own crimes were then folded into the victors' settlement and placed beyond accountability — with the acquiescence of the same international actors whose failure the victors invoke (Accounts 1 and 3 in their uncomfortable mutual dependence). Each clause of that composition is contested at its edges; the composition itself is the nearest thing the historiography has to a stable consensus, and it assigns the responsibilities cumulatively rather than competitively — which is precisely what each account, read alone and politically, is built to resist.

7. Post-1994 Accountability: Tribunals, Apologies, Archives, and the Contested Record to 2026

7.1 The Perpetrator Track: ICTR, National Courts, Gacaca, Universal Jurisdiction

Accountability for the genocide itself — as distinct from the prevention failure — was institutionally vigorous, and its architecture is treated in RW-A-01 and RW-A-02; this section records only its bearing on the three accounts. The ICTR (Security Council Resolution 955, 8 November 1994, seated at Arusha) indicted 93 individuals and convicted 61 before closing in December 2015 [TBD-VERIFY: the commonly cited final tally is 93 indicted, 61 convicted/sentenced, 14 acquitted; reconcile against the ICTR legacy website], establishing the foundational jurisprudence (Akayesu, Kambanda, the Media Trial, Bagosora) catalogued in RW-B-01. Its residual mechanism (the IRMCT) carried the fugitive track to the period's end: Félicien Kabuga, the genocide's principal financier, arrested outside Paris on 16 May 2020 after twenty-six years; trial opened September 2022; proceedings indefinitely stayed in 2023 after the appeals chamber found him unfit to stand trial. Protais Mpiranya, the Presidential Guard commander, was confirmed in 2022 to have died in Zimbabwe in 2006. The domestic track ran through the Organic Law courts and the Gacaca system (2002–2012, ~1.2 million cases; RW-A-02), and the universal-jurisdiction track widened steadily: Belgium's "Butare Four" trial (2001) under its universal-jurisdiction law; the first French conviction (Pascal Simbikangwa, 2014) followed by a sustained series through the 2020s (Laurent Bucyibaruta 2022; Philippe Hategekimana 2023; Eugène Rwamucyo 2024) as French-Rwandan judicial cooperation recovered from the Bruguière rupture. For Account 1 this record is the system's partial redemption; for Account 3 its one-sidedness — every prosecution running in one direction — is the record's defining feature.

7.2 The Prevention Track: Inquiries and Apologies Without Sanction

Accountability for the prevention failure took the form of inquiry and apology, never sanction. The sequence: the Belgian Senate commission (1997), the most rigorous national inquiry, whose findings grounded Verhofstadt's 7 April 2000 apology in Kigali; the French parliamentary mission under Paul Quilès (1998), which acknowledged "errors of assessment" while rejecting complicity — a finding Rwanda answered with the Mucyo Commission (2008), which alleged active French complicity and named senior French figures, deepening the rupture the Bruguière warrants had opened; the Carlsson Report (December 1999) and the OAU Panel (July 2000), whose findings Section 3.3 records; and Clinton's 1998 Kigali statement. No official in any jurisdiction was disciplined, demoted, or prosecuted for any decision in the prevention chain — including the decisions documented in the UN's own inquiry. Three of the figures most criticised in the record were subsequently promoted: Annan to Secretary-General (1997) and the Nobel Peace Prize (2001), Riza to his chef de cabinet, and Susan Rice to Assistant Secretary of State for African Affairs (1997) and later National Security Advisor and UN Ambassador — career facts regularly cited in the Rwandan official press as the measure of Account 1's accountability deficit. The institutional legacy was doctrinal instead: the Brahimi Report (2000) on peacekeeping robustness, the responsibility to protect adopted at the 2005 World Summit (invoked thereafter in Libya 2011, with consequences that re-polarised the doctrine), and the UN Special Adviser on the Prevention of Genocide — each a direct descendant of the Rwanda post-mortems, and each, as the subsequent record from Darfur to Myanmar suggests, an answer to Rwanda's question that has yet to pass Rwanda's test.

7.3 The French Reckoning: Mucyo to Duclert to Macron (2006–2024)

The Franco-Rwandan strand ran longest and moved furthest. The Bruguière warrants (November 2006) caused Rwanda to sever diplomatic relations; the Mucyo Report (2008) and Rwanda's accession to the Commonwealth (2009) and shift to English-medium education marked the strategic rupture; the Trévidic–Poux findings (2012) and Kabuga-era judicial cooperation marked the thaw. The Duclert Commission — fifteen historians granted access to the Élysée, defence, and foreign-ministry archives — reported on 26 March 2021 that France bore "heavy and overwhelming responsibilities" (responsabilités lourdes et accablantes): it had sustained a regime it knew to be preparing massacres, had read the situation through an ethnicist and Anglophone-conspiracy frame, and had failed to draw consequences from its own internal warnings — while finding no evidence of complicity in the sense of intent to join the genocide. Rwanda's own commissioned study (the Muse Report, April 2021) reached compatible conclusions with a harder edge. President Macron, at the Kigali Genocide Memorial on 27 May 2021, recognised France's "responsibilities" and said France had "a duty to look history in the face", asking for the "gift of forgiveness" while stopping short of a formal apology — a calibration Kagame publicly accepted as "more valuable than an apology". At the thirtieth anniversary (April 2024) the Élysée pre-released a statement that Macron would acknowledge that France "together with its Western and African allies, could have stopped the genocide" but "did not have the will" [TBD-VERIFY: the published video message of 7 April 2024 is reported to have softened this formulation; the exact delivered wording requires confirmation]. The archival dimension advanced in parallel: the researcher François Graner's litigation won Conseil d'État access to the Mitterrand archives (12 June 2020), and the Duclert corpus moved substantial documentation into the public domain — making the French case the only branch of the prevention record where the archive has been substantially opened. For Account 2, Duclert is the account's vindication in official form; for Account 1, a late completion of the apology cycle; Account 3 notes that the same French judicial system that investigated the RPF shoot-down thesis closed that file in the same years the bilateral relationship was repaired.

7.4 The Custody of the Record: Narrative Control to 2026

The final accountability question is custody of the historical record itself. Inside Rwanda, the framework comprises the constitutional and statutory architecture against "genocide ideology", "divisionism", and genocide minimisation (the 2008 law, narrowed in its 2013 revision after international criticism of vagueness; the 2018 penal code provisions), the official designation "Genocide Against the Tutsi" — adopted domestically and carried internationally through the UN General Assembly's April 2020 resolution fixing the international day of reflection's nomenclature, and UNESCO's parallel usage — and the commemorative architecture (Kwibuka; the memorials; the curriculum) documented in RW-G-01. Defenders of this framework, including substantial survivor opinion and comparative scholars of post-Holocaust Europe, read it as the necessary defence of a settled record against an active, organised denialism operating in the FDLR diaspora and sections of Francophone polemics. Critics — Reyntjens (2013) centrally — read it as the juridification of a single permitted narrative in which the categories of "denial" and "divisionism" expand to cover the documented record of Section 5.5 and ordinary political opposition (the Victoire Ingabire prosecutions, treated in RW-D-05, turned in part on statements at the Kigali memorial about Hutu victims). The international scholarly infrastructure divides along the same line: the 2014 BBC documentary controversy produced duelling open letters of scholars; access to Rwandan archives and field sites conditions what research is possible; and the thirtieth-anniversary cycle (Kwibuka30, April 2024) displayed both the consolidation of the international consensus around the genocide's memory and the unresolved status of everything in Account 3. As of 2026 the position can be stated precisely: the genocide's record is settled and institutionally protected; the prevention failure's record is documented, apologised for, and unsanctioned; and the record of the victors' own violence remains documented but frozen — the corpus's three accounts preserved, in effect, in three different evidentiary states by the politics of the peace.

8. Conclusion

This anchor and RW-B-01 are a deliberate pair. The genocide anchor establishes the settled event; this document maps the contested explanation of why no one stopped it. The three-account discipline is applied here not because the facts are in doubt — the warning record of Section 2 is among the best-documented in the history of mass atrocity — but because the selection and weighting of those facts is irreducibly plural, and because each selection carries a politics. The UN/international-failure account, with its genocide fax, its Resolution 912, and its chain of named decisions, is the account with the strongest documentary spine and the most complete official confession — the Carlsson Inquiry remains the only instance of the United Nations formally convicting itself of systemic failure in a genocide. The colonial/structural account supplies what the first cannot: an explanation of why there was an exterminationist apparatus to fail to stop, and an indictment that reaches the authors and underwriters of the structure rather than only the bystanders of its collapse. The RPF-and-contested-accountability account supplies what both omit: an examination of the one actor that fought, won, ended the genocide, committed documented crimes of its own, and then assumed custody of the record by which all of this is judged.

The corpus's findings, stated within its discipline, are three. First, the accounts are cumulatively, not competitively, true in their well-evidenced cores: the composition offered in Section 6.4 — structure, war, prepared genocide, international abdication, belligerent victory, victors' settlement — is the nearest available approximation to the historiographical consensus, and the principal distortions in public memory arise from reading any single account alone. Second, the accountability asymmetry is the period's structural legacy: perpetrator accountability was pursued through four institutional generations (ICTR, national courts, Gacaca, universal jurisdiction), prevention accountability produced apology without sanction in every jurisdiction, and victors' accountability was institutionally foreclosed — an asymmetry that is not an accident of the record but a product of the post-1994 settlement, and one whose effects run forward through the genocide-ideology framework, the Rwanda–France reconciliation, and the standing rhetorical architecture of Rwandan foreign policy documented in RW-C-01 and RW-D-08. Third, the prevention failure's institutional descendants — R2P, the Office on Genocide Prevention, the peacekeeping doctrine reforms — constitute the international system's answer to Rwanda, and the record from Darfur to Srebrenica's anniversaries to Myanmar and Tigray indicates that the answer remains unproven; the Rwandan state's own conclusion, that survival is not delegable, is the one lesson of 1994 that has been acted upon without reservation, by the actor that learned it first.

For the corpus, the document functions as the Block-J methodological exemplar for Rwanda: the discipline of presenting a suppressed account (Account 3) with full evidential care while holding the boundary against its denialist abuse (Section 5.6) is the same discipline future J-block documents — the planned mapping-report companion (RW-J-02) and the elections three-accounts (RW-J-03) — must apply to their own contested records.

Sources

1. Dallaire, Roméo, with Brent Beardsley, Shake Hands with the Devil: The Failure of Humanity in Rwanda (Toronto: Random House Canada, 2003) — the UNAMIR Force Commander's memoir; the primary first-person source for the 11 January 1994 cable, the January–April requests and refusals, and the April–July collapse.
2. United Nations, Report of the Independent Inquiry into the Actions of the United Nations during the 1994 Genocide in Rwanda (the Carlsson Report), S/1999/1257, 15 December 1999 — the UN's own systemic-failure verdict; chaired by Ingvar Carlsson with Han Sung-Joo and Rufus Kupolati.
3. Organisation of African Unity, International Panel of Eminent Personalities, Rwanda: The Preventable Genocide (July 2000) — the OAU-commissioned inquiry chaired by Ketumile Masire; extends the indictment to the principal states and recommends reparations.
4. Power, Samantha, "A Problem from Hell": America and the Age of Genocide (New York: Basic Books, 2002) — the documentary reconstruction of US decision-making, the "acts of genocide" record, and the PDD-25 context.
5. Melvern, Linda, A People Betrayed: The Role of the West in Rwanda's Genocide (London: Zed Books, 2000; rev. eds. 2009, 2020) — the principal investigative account of the Security Council record, including the Council's closed deliberations of April–May 1994.
6. Melvern, Linda, Conspiracy to Murder: The Rwandan Genocide (London: Verso, 2004; rev. 2006) — the planning record; companion to the international-failure account.
7. Des Forges, Alison, Leave None to Tell the Story: Genocide in Rwanda (New York: Human Rights Watch / FIDH, 1999) — the foundational documentary record; its sections on the international response and on RPF killings ground Sections 3 and 5.5 respectively.
8. Mamdani, Mahmood, When Victims Become Killers: Colonialism, Nativism, and the Genocide in Rwanda (Princeton: Princeton University Press, 2001) — the foundational text of the colonial/structural account.
9. Mamdani, Mahmood, Neither Settler Nor Native: The Making and Unmaking of Permanent Minorities (Cambridge, MA: Harvard University Press, 2020) — the restated decolonisation-of-the-political argument, with Rwanda as a central case.
10. Prunier, Gérard, The Rwanda Crisis: History of a Genocide (London: Hurst, 1995; rev. 1997, 2008) — the principal early scholarly history; central to the 1990–1994 structural-crisis layer.
11. Uvin, Peter, Aiding Violence: The Development Enterprise in Rwanda (West Hartford: Kumarian Press, 1998) — the development-complicity argument within the structural account.
12. Reyntjens, Filip, Political Governance in Post-Genocide Rwanda (New York: Cambridge University Press, 2013) — the principal scholarly statement on post-1994 narrative control and the accountability asymmetry; with Reyntjens's article-length work on the RPF record (including "Rwanda, Ten Years On: From Genocide to Dictatorship", African Affairs, 2004).
13. Kuperman, Alan J., "Rwanda in Retrospect", Foreign Affairs, vol. 79, no. 1 (January/February 2000); and The Limits of Humanitarian Intervention: Genocide in Rwanda (Washington: Brookings Institution Press, 2001) — the principal counterfactual critique of Account 1 and the principal statement of Claim Three in Account 3; read with the Des Forges–Kuperman exchange in Foreign Affairs (2000).
14. Feil, Scott R., Preventing Genocide: How the Early Use of Force Might Have Succeeded in Rwanda (Washington: Carnegie Commission on Preventing Deadly Conflict, 1998) — the 5,000-troop counterfactual study.
15. Barnett, Michael, Eyewitness to a Genocide: The United Nations and Rwanda (Ithaca: Cornell University Press, 2002) — the inside-the-Secretariat account by a scholar serving at the US Mission to the UN in 1994; the institutional-culture reading of the DPKO decisions.
16. République Française, Commission de recherche sur les archives françaises relatives au Rwanda et au génocide des Tutsi (the Duclert Commission), La France, le Rwanda et le génocide des Tutsi (1990–1994) (26 March 2021) — the archive-based finding of "heavy and overwhelming responsibilities"; with the Republic of Rwanda's commissioned Muse Report, A Foreseeable Genocide: The Role of the French Government in Connection with the Genocide Against the Tutsi in Rwanda (April 2021).
17. Assemblée Nationale (France), Mission d'information sur le Rwanda (Quilès Mission), Rapport d'information (December 1998); and Sénat de Belgique, Commission d'enquête parlementaire concernant les événements du Rwanda (December 1997) — the two principal national parliamentary inquiries.
18. Bruguière, Jean-Louis, Ordonnance de soit-communiqué (Tribunal de Grande Instance de Paris, 17 November 2006) — the RPF-responsibility shoot-down instrument; read against the Republic of Rwanda, Report of the Independent Committee of Experts Charged with Investigation into the Crash on 06/04/1994 of Falcon 50 Aeroplane, Registration Number 9XR-NN (the Mutsinzi Report, 2009/2010), and the Trévidic–Poux ballistic-acoustic expert report (January 2012) with the December 2018 non-lieu.
19. Del Ponte, Carla, with Chuck Sudetic, Madame Prosecutor: Confrontations with Humanity's Worst Criminals and the Culture of Impunity (New York: Other Press, 2008) — the prosecutor's account of the ICTR Special Investigations into RPF crimes and their termination.
20. Human Rights Watch, Law and Reality: Progress in Judicial Reform in Rwanda (July 2008) and the HRW statements on the 2008 Kabgayi military trial — the documentation of the transferred-case outcome; with HRW/Des Forges materials on the Gersony findings.
21. United Nations Security Council, Resolutions 872 (5 October 1993), 909 (5 April 1994), 912 (21 April 1994), 918 (17 May 1994), 925 (8 June 1994), 929 (22 June 1994), and 955 (8 November 1994) — the primary Council record of the design, reduction, nominal reinforcement, Turquoise authorisation, and tribunal establishment.
22. United Nations, Special Rapporteur on Extrajudicial, Summary or Arbitrary Executions (Bacre Waly Ndiaye), Report on the Mission to Rwanda, 8–17 April 1993, E/CN.4/1994/7/Add.1 (August 1993) — the pre-genocide Genocide-Convention warning within the UN system.
23. International Commission of Investigation on Human Rights Violations in Rwanda since October 1, 1990 (FIDH, Africa Watch, UIDH, CIDPDD), Report (March 1993) — the contemporaneous documentation of the rehearsal massacres.
24. PBS Frontline, The Triumph of Evil (January 1999) — the documentary record including the Riza interview; with the National Security Archive's declassified US document collections on Rwanda 1994 (the "genocide fax" facsimile, the Defense Department discussion papers, and the State Department briefing transcripts).
25. Guichaoua, André, From War to Genocide: Criminal Politics in Rwanda, 1990–1994, trans. Don E. Webster (Madison: University of Wisconsin Press, 2015) — the granular reconstruction of the 1990–1994 political crisis and the April decision-making; the corrective against both over-structural and over-conspiratorial readings.

Related Documents

• RW-B-01: The 1994 Genocide Against the Tutsi — The 6 April Trigger, the 100 Days, and the End of the Genocide (existing) — the event anchor this document deliberately does not re-narrate; its Section 3 carries the full shoot-down treatment summarised here in Section 5.3, and its Section 8 the Turquoise treatment
• RW-B-02: The 1990 RPF Invasion and the Rwandan Civil War — Mutara to Arusha to the Hundred Days (existing) — the war whose dynamics constitute the warning window of Section 2 and the structural trigger of Account 2's final layer
• RW-A-04: The Arusha Accords — Power-Sharing Negotiations, the Broad-Based Transitional Government, and the Architecture of Failure (existing) — the settlement whose implementation void all three accounts identify as the proximate political origin of the catastrophe
• RW-D-01: The Habyarimana Era — The Second Republic, the MRND One-Party State, and the Trajectory to 6 April 1994 (existing) — the regime architecture whose radicalisation the warning record of Section 2 documents
• RW-H-PRES-01: Juvénal Habyarimana — A Biography (existing) — the president whose patrons, regime, and death organise Accounts 1–3 respectively; carries the parallel shoot-down treatment
• RW-A-01: The 1994 Genocide Against the Tutsi and the Aftermath — RPF State-Formation 1994–2003 (existing) — the successor anchor; the Gersony aftermath, Kibeho, and the refugee-war sequence that extend Account 3's record beyond this document's close
• RW-A-02: The Gacaca Courts — Community Participatory Justice for the 1994 Genocide (existing) — the domestic perpetrator-accountability architecture summarised in Section 7.1
• RW-G-01: Rwandan Education and Genocide Memory as Institutional Architecture (existing) — the commemoration-and-curriculum architecture through which the accounts' official standings (Section 6.3) are institutionalised
• RW-D-05: Rwandan National Electoral Architecture and the RPF Dominance (existing) — the Ingabire prosecutions and the genocide-ideology framework's political application referenced in Section 7.4
• RW-J-02: Mapping Report (2010) — DRC Allegations Against Rwanda (when written) — the contested-record companion extending Account 3's documentation into the DRC dimension
• RW-R-01: Rwanda Governance Books Canon (existing) — the bibliographic anchor for the three accounts' source traditions
• RW-N-01: Rwanda in International Perceptions — Donor Darling, Authoritarian Critique, and the Most Contested Reputation in Africa
• RW-I-01: The Rwanda Defence Force — The Army-State Relationship from Rebel Victory to Security Exporter
• RW-O-01: Rwanda Megatrends — The 2030s Questions
• RW-G-05: Mutuelle de Santé and the Rwandan Health System
• RW-M-01: Ndi Umunyarwanda and the Post-Ethnic Citizenship Project
• RW-F-02: back-reference added by symmetry sweep
• RW-H-PRES-03: back-reference added by symmetry sweep