Rwanda in Retrospect

A HARD LOOK AT INTERVENTION

Several years after mass killings in Bosnia, Somalia, and Rwanda, the United States is still searching for a comprehensive policy to address deadly communal conflicts. Among Washington policymakers and pundits, only two basic principles have achieved some consensus. First, U.S. ground troops generally should not be used in humanitarian interventions during ongoing civil wars. Second, an exception should be made for cases of genocide, especially where intervention can succeed at low cost. Support for intervention to stop genocide is voiced across most of the political spectrum.

Despite this amorphous consensus that the United States can and should do more when the next genocide occurs, there has been little hard thinking about just what that would entail or accomplish. A close examination of what a realistic U.S. military intervention could have achieved in the last clear case of genocide this decade, Rwanda, finds insupportable the oft-repeated claim that 5,000 troops deployed at the outset of the killing in April 1994 could have prevented the genocide. This claim was originally made by the U.N.'s commanding general in Rwanda during the genocide and has since been endorsed by members of Congress, human rights groups, and a distinguished panel of the Carnegie Commission on Preventing Deadly Conflict. Although some lives could have been saved by intervention of any size at any point during the genocide, the hard truth is that even a large force deployed immediately upon reports of attempted genocide would not have been able to save even half the ultimate victims.

PRELUDE TO GENOCIDE

Rwandan politics were traditionally dominated by the Tutsi, a group that once made up 17 percent of the population. Virtually all the rest of the population was Hutu, and less than one percent were aboriginal Twa. All three groups lived intermingled throughout the country. During the transition to independence starting in 1959, however, the Hutu seized control in a violent struggle that spurred the exodus of about half the Tutsi population to neighboring states.

The Hutu themselves were divided into two regional groups. The majority lived in the central and southern part of the country and supported the PARMEHUTU (Parti du mouvement et de l'émancipation des Bahutu), which assumed power upon independence, while a minority lived in the northwest, historically a separate region. During the first decade of independence, Tutsi refugees invaded Rwanda repeatedly, seeking a return to power. The ruling Hutu responded by massacring domestic Tutsi. In 1973, a northwestern Hutu officer, Juvenal Habyarimana, led a coup that shifted political power to his region. Northwestern Hutu came to dominate Rwanda's political, military, and economic life, engendering resentment from other Hutu as well as from the Tutsi. But large-scale violence against domestic Tutsi largely disappeared for 15 years in the absence of any further attempted invasions by refugees.

Stability began to unravel in October 1990, when an expatriate rebel force composed mainly of Uganda-based Tutsi refugees, the Rwandan Patriotic Army (RPA), invaded northern Rwanda. The RPA and its political arm, the Rwandan Patriotic Front (RPF), were led by battle-tested soldiers who had fought with the Ugandan guerrilla Yoweri Museveni to overthrow Uganda's government in 1986 before turning their efforts toward home. By early 1993, the rebels had made substantial inroads against the Hutu-dominated Rwandan Armed Forces (or FAR, in the French acronym). This military advance, combined with diplomatic pressure from the international community, compelled Habyarimana to agree to share power in the Arusha accords of August 1993.

The peacekeepers of the U.N. Assistance Mission for Rwanda (UNAMIR) then arrived, but for eight months the Rwandan leader obstructed and tried to modify the power-sharing provisions. The extremist wing of his northwestern Hutu clique viewed the accords as abject surrender to the Tutsi, who they feared would seize the spoils of rule and seek retribution. Habyarimana attempted to retain power by co-opting the opposition Hutu through bribery and appeals to solidarity against the Tutsi, and he succeeded in splitting off radical factions from the main opposition parties. But he and the extremists also developed a forceful option -- training militias, broadcasting anti-Tutsi hate radio, and plotting to kill moderate Hutu leaders and Tutsi civilians. On April 6, 1994, as Habyarimana appeared to be acquiescing to international pressure to implement the accords, his plane was mysteriously shot down. The genocide plan was put in motion.

In most areas of Rwanda, violence began on the following day. The government radio station and the extremists' counterpart -- Radio-Télévision Libre des Mille Collines -- urged the Hutu to take vengeance against the Tutsi for their alleged murder of the president. Led by militias, Hutu began to attack the homes of neighboring Tutsi, attempting to rob, rape, and murder them, and often setting fire to their homes. This initial step did not eliminate a high proportion of Tutsi, however, because their attackers were generally poorly armed. The vast majority of Tutsi fled their homes and sought refuge in central gathering places -- churches, schools, hospitals, athletic fields, stadiums, and other accessible spaces. Tutsi often passed through more than one such site to gather in larger concentrations, either voluntarily or at government direction. Within a few days, most of Rwanda's Tutsi had congregated at such centralized sites throughout the country, in groups ranging from a few hundred to tens of thousands.

At first, the assembled Tutsi gained a defensive advantage. The surrounding crowds of militia-led Hutu were generally armed only with swords, spears, and machetes -- or with the traditional masu, a large club studded with nails. By using walls and buildings for defense, Tutsi groups could often fend off attacks merely by throwing rocks. By contrast, individual Tutsi who attempted to flee were often killed immediately by the surrounding Hutu masses or caught and killed at roadblocks. For several days, this produced a standoff. Tutsi living conditions were deteriorating and supplies were dwindling, but most Hutu were unwilling to risk casualties by attacking.

This situation changed in most of Rwanda within a week, by about April 13, when better-armed Hutu reinforcements -- composed of members of the regular army, the reserves, the Presidential Guard (PG), or the national police -- began arriving at the Tutsi gathering sites. Although these forces were few in number at each site, they were armed with rifles, grenades, or machine guns, which tilted the balance of force. They would typically toss a few grenades on the Tutsi and follow with light-arms fire. Survivors who attempted to flee were usually mowed down by gunfire or caught and killed by the surrounding mob. Militia-led Hutu would then enter the site, hacking to death those still alive. Some Tutsi escaped in the initial mayhem or avoided death by hiding beneath their dead compatriots, but many were later caught at roadblocks and killed on the spot or taken to other central sites to face a similar ordeal. A few lucky Tutsi survived by hiding in places such as pit latrines or the homes of sympathetic Hutu, living to tell their harrowing tales.

Perhaps the most remarkable aspect of the genocide was its speed. According to survivor testimonies gathered by African Rights and Human Rights Watch, the majority of Tutsi gathering sites were attacked and destroyed before April 21, only 14 days into the genocide. Given that half or more of the ultimate Tutsi victims died at these sites, the unavoidable conclusion is that a large portion of Rwanda's Tutsi had been killed by April 21 -- perhaps 250,000 in just over two weeks. That would be the fastest genocide rate in recorded history.

Despite this generally rapid pace, two factors constrained the speed and extent of the killing in Rwanda. First, Hutu extremists generally avoided large-scale massacres when international observers were present -- as part of a comprehensive strategy to hide the genocide from both the outside world and Rwanda's remaining Tutsi until it could be completed. Wherever Tutsi were congregated under the watch of outside observers, the extremists favored an alternate strategy of slow, stealthy annihilation: Hutu leaders would arrive each day at such sites with a list of up to several dozen names, usually starting with the Tutsi political elite. These Tutsi would be removed under a false pretense such as interrogation before being taken to a remote location and executed. This occurred at several places across Rwanda: Kamarampaka Stadium and the Nyarushishi camp in Cyangugu prefecture, where Red Cross aid workers were present; the Kabgayi Archbishopric in Gitarama, under the watchful eyes of the pope's subordinates; Amahoro stadium in Kigali, where U.N. troops stood guard; and smaller sites in Kigali such as the St. Famille and St. Paul's churches. At such sites, the slower pace of killing meant that the vast majority of Tutsi there were still alive at the end of April, and a good number survived the entire ordeal.

Second, the killing varied among Rwanda's ten original prefectures. Byumba prefecture in the north was the base of the Tutsi-led rebels, who generally prevented large-scale massacres of Tutsi there. The two prefectures most dominated by Hutu extremists, Gisenyi and Ruhengeri in the northwest, also suffered relatively little killing because much of their Tutsi populations had fled prior to the genocide in response to earlier threats and harassment.

Two prefectures with high Tutsi populations and strong Hutu opposition movements also initially managed to avoid the genocide. Butare prefecture in the south was governed by a Tutsi prefect who managed to keep matters relatively calm until he was removed from office on April 18. Widespread killing then began with a vengeance, and tens of thousands of Tutsi perished in the next few days. Similarly, Gitarama prefecture, the heart of central Rwandan Hutu opposition to the northwestern Hutu regime, generally resisted implementing the genocide until government forces arrived to spur them on. Large-scale killing commenced there about April 21. Finally, the nature of killing in the capital, Kigali, also differed significantly from that in the rest of the country. During the first two days, a highly organized and thorough assassination campaign was carried out there against opposition politicians and prominent liberals such as human rights advocates. Unlike elsewhere, many of Kigali's initial victims of Hutu extremism were fellow Hutu.

Civil war also erupted in Kigali almost immediately. On April 7, an RPA battalion that had been stationed in the capital since December 1993 under the Arusha accords demanded a halt to atrocities against civilians -- and then clashed with government forces when its demand was ignored. With the president and the moderate opposition dead, war breaking out in Kigali, and radio broadcasts urging Hutu to kill their neighbors, the capital descended into chaos. Corpses began to pile up, totaling as many as 20,000 during the first week. Unlike in the countryside, however, Tutsi had a decent chance of gaining some refuge by reaching a central gathering site where foreigners stood guard. Although the extremists could not hide the chaos and violence in the capital, they generally avoided wholesale massacres before such witnesses in hopes of averting foreign military intervention.

By late April, only three weeks after the president's plane crash, almost all the large massacres were finished. The rebels themselves acknowledged on April 29 that "the genocide is almost completed." Human Rights Watch concurs that "in general, the worst massacres had finished by the end of April." By that time, it notes, "perhaps half of the Tutsi population of Rwanda" -- some two-thirds of the ultimate Tutsi victims -- already had been exterminated. Killing of the remaining Tutsi continued at a slower pace for another two and a half months until halted by the rebels' military victory and a belated French-led intervention.

Precise Tutsi death totals are difficult to determine because of several factors, including the inability to distinguish Tutsi from Hutu corpses. But estimates can be made by subtracting the number of Tutsi survivors from the number living in Rwanda immediately prior to the genocide. Estimated 1994 population figures, which are extrapolated from the 1991 census and account for annual population growth of three percent, indicate that Rwanda's pre-genocide population included approximately 650,000 Tutsi. There is no evidence for other, higher claims.1 After the genocide and civil war, some 150,000 Tutsi survivors were identified by aid organizations. Thus an estimated 500,000 Rwandan Tutsi were killed, more than three-quarters of their population. The number of Hutu killed during the genocide and civil war is even less certain, with estimates ranging from 10,000 to well over 100,000.

THE KNOWLEDGE GAP

Although U.S. intelligence reports from the period of the genocide remain classified, they probably mirrored those of the international news media, human rights organizations, and the U.N. -- because U.S. intelligence agencies committed virtually no in-country resources to what was considered a tiny state in a region of little strategic value. During the genocide's early phases, the U.S. government actually received most of its information from nongovernmental organizations. A comprehensive review of such international reporting -- by American, British, French, Belgian, and Rwandan media, leading human rights groups, and U.N. officials -- strongly suggests that President Clinton could not have known that a nationwide genocide was under way in Rwanda until about April 20.

This conclusion is based on five aspects of the reporting during the first two weeks. First, violence was initially depicted in the context of a two-sided civil war -- one that the Tutsi were winning -- rather than a one-sided genocide against the Tutsi. On April 13, the Western press accurately reported that Rwanda's Hutu interim government had fled the capital for refuge in Gitarama and that "the fall of Kigali seems imminent" (Paris Radio France International). When Western troops arrived to evacuate foreign nationals, the Tutsi rebels did not seek assistance but rather demanded that the troops depart immediately so as not to interfere with their imminent victory. The Canadian commander of the U.N. peacekeepers in Rwanda, General Romeo Dallaire, also identified the problem as mutual violence, stating on April 15 that "if we see another three weeks of being cooped up and seeing them pound each other" (The Guardian), the U.N. presence would be reassessed. In addition, until April 18 both the government and the rebels stated publicly that the FAR was not participating in massacres. (The government was engaged in a cover-up, and the rebels initially avoided implicating the FAR in the vain hope of winning its allegiance against the extremist Hutu.)

Second, the violence was reported to be waning when it actually was accelerating. Just four days in, on April 11, The New York Times reported that fighting had "diminished in intensity" and Le Monde wrote three days later that "a strange calm reigns in downtown" Kigali. The commander of Belgian peacekeepers stated that "the fighting has died down somewhat, one could say that it has all but stopped" (Paris Radio France International). On April 17, Dallaire told the BBC that except for an isolated pocket in the north, "the rest of the line is essentially quite quiet." Only on April 18 did a Belgian radio station question this consensus, explaining that the decline in reports of violence was because "most foreigners have left, including journalists."

Third, most early death counts were gross underestimates and never suggested genocidal proportions. Three days into the killing, on April 10, The New York Times quoted varying estimates of 8,000 or "tens of thousands" dead. But during the second week, media estimates did not rise at all. On April 18, the Times still reported only 20,000 deaths, underestimating the actual carnage at that point by about tenfold. The true scope of the killing emerged only on April 20, when Human Rights Watch estimated that "as many as 100,000 people may have died to date," followed the next day by a Red Cross estimate of perhaps "hundreds of thousands."

Fourth, the initial focus of international reporting was almost exclusively on Kigali, a relatively small city, and thus failed to indicate the broader scope of violence -- a consequence both of Hutu concealment efforts and of the Western evacuation of expatriates and reporters from the countryside. Although a few early reports of rural violence did trickle out to the West, these indicated military combat, mutual ethnic violence, or criminal looting rather than an extermination campaign. An RPF official told the BBC on April 12 only that "we want to stop the senseless killing that is going on in Kigali." The first international report of a large-scale massacre outside the capital did not emerge until April 16. As late as April 20, U.N. Secretary-General Boutros Boutros-Ghali still described the killings as "mainly in Kigali." This initial obsession with the capital, which contained only four percent of Rwanda's population, obscured the national scope of violence and thus its genocidal intent.

The rebels' own radio station did not report the nationwide scope of the violence until April 19. American newspapers failed to give any such indication until April 22, when they belatedly reported that fighting bands had reduced "much of the country to chaos" (The New York Times). Many foreign observers still could not conceive that a genocide was under way. On April 23, The Washington Post pondered why only 20,000 refugees had crossed the border -- even though half a million Tutsi had fled their homes -- and reported that aid workers had concluded that "most of the borders have been sealed by the Rwandan Army." Only on April 25 did The New York Times solve the riddle, reporting that violence had "widened into what appears to be a methodical killing of Tutsi across the countryside." The missing refugees "either have been killed or are trying to hide."

Fifth, no credible and knowledgeable observers, including human rights groups, raised the prospect that genocide was occurring until the end of the second week. In opinion articles published on April 14 and 17, Human Rights Watch gave no hint of an attempted nationwide genocide. The rebels did not use the term until April 17. Human Rights Watch finally raised the prospect in an April 19 letter to the U.N. Security Council. Other international observers remained considerably more cautious. The pope first used the word "genocide" on April 27. The U.S. Committee for Refugees waited until May 2 to urge the Clinton administration to make such a determination. Only on May 4 did Boutros-Ghali finally declare a "real genocide." Thus the earliest President Clinton credibly could have made a determination of attempted genocide was about April 20, 1994 -- two weeks into the violence.

THE MILITARY SCENE

At the time of Habyarimana's death, Rwanda hosted three military forces -- those of the government, the rebels, and the United Nations. Government forces totaled about 40,000, including the army, the national police, and 1,500 PG troops. But except for the PG and a few other elite battalions, this force was largely hollow, having expanded sixfold in three years responding to the rebel threat. Another 15,000 to 30,000 Hutu were scattered around the country in militias, but many apparently did not possess firearms or ammunition. Rebel arms were more primitive than the FAR's and included few motorized vehicles and no aircraft.

UNAMIR had about 2,500 peacekeepers, most either in Kigali or in the north near the demilitarized zone. Their presence was subject to the consent of the Rwandan government. Rules of engagement were somewhat ambiguous but were generally interpreted to bar the use of force except in self-defense or in joint operations with Rwandan national police.

On the first day of violence, the pg executed ten Belgian peacekeepers who were attempting to protect Rwanda's opposition prime minister. These deaths and the emerging chaos in Kigali prompted Western governments to evacuate their nationals. European troops began arriving on April 9 and evacuated several thousand Westerners before departing on April 13. On April 10, Dallaire also requested 5,000 more U.N. troops to halt what he perceived to be mutual killing confined to the capital. Instead, Belgium announced on April 14 that it would be withdrawing its UNAMIR battalion, which triggered unease among the other troop-contributors and led the U.N. Security Council a week later to cut authorized troop levels to a skeleton crew of 270.

Rebel forces, estimated at 20,000, had been constrained by the Arusha accords to a small area of northern Rwanda; the exception was one authorized Kigali battalion, which the RPF had reinforced clandestinely to about 1,000 troops. When the civil war was renewed on April 7, the northern-based rebels set out to help the stranded battalion in the capital and engage FAR troops elsewhere, making quick progress down Rwanda's entire eastern flank by late April. Thereafter, the war had two stationary fronts, in Kigali and Ruhengeri, until the end of June, as well as a broad mobile front moving westward through southern Rwanda. In just three months, the rebels captured most of the country -- Gitarama on June 9, Kigali on July 4, Butare on July 5, Ruhengeri on July 14, and Gisenyi on July 17 -- before finally declaring a cease-fire on July 18.

As reports of genocide reached the outside world starting in late April, public outcry spurred the United Nations to reauthorize a beefed-up "UNAMIR II" on May 17. During the following month, however, the U.N. was unable to obtain any substantial contributions of troops and equipment. As a result, on June 22 the Security Council authorized France to lead its own intervention, Operation Turquoise, by which time most Tutsi were already long dead.

POTENTIAL U.S. INTERVENTIONS

In retrospect, three levels of potential U.S. military intervention warrant analysis: maximum, moderate, and minimal. None would have entailed full-blown nationwide policing or long-term nation-building by American troops. Based on historical experience, full-blown policing would have required some 80,000 to 160,000 personnel -- that is, ten to twenty troops per thousand of population -- an amount far more than logistically or politically feasible. Nation-building would have been left to a follow-on multinational force, presumably under U.N. authorization.

Maximum intervention would have used all feasible force to halt large-scale killing and military conflict throughout Rwanda. Moderate intervention would have sought to halt some large-scale killing without deploying troops to areas of ongoing civil war, in order to reduce U.S. casualties. Minimal intervention would have relied on air power alone.

A maximum intervention would have required deployment of a force roughly the size of one U.S. division -- three brigades and supporting units, comprising about 15,000 troops and their equipment -- with rules of engagement permitting the use of deadly force to protect endangered Rwandans. After establishing a base of operations at Kigali airport, the force would have focused on three primary goals: halting armed combat and interposing itself between FAR and RPF forces on the two stationary fronts of the civil war; establishing order in the capital; and finally fanning out to halt large-scale genocidal killing in the countryside. None of these tasks would have been especially difficult or dangerous for properly configured and supported American troops once they were in Rwanda. But transporting such a force 10,000 miles to a landlocked country with limited airfields would have been considerably slower than some retrospective appraisals have suggested.

The first brigade to arrive would have been responsible for Kigali: coercing the FAR and RPF to halt hostilities, interposing itself between them, and policing the capital. The second brigade would have deployed one battalion in the north to halt the civil war in Ruhengeri and another as a rapid-reaction force in case American troops drew fire. The third brigade, supplemented by a battalion of the second brigade, would have been devoted to halting the killing in the countryside. Such an effort would have required roughly 2,000 troops to halt the war in Kigali, 3,000 to police Kigali, 1,000 to stop the fighting in the north, 1,500 for a rapid-reaction force, and 6,000 to stop the genocide outside Kigali -- a total of about 13,500 troops, in addition to support personnel.

The time required to deploy such a force would have depended mainly on its weight. A division-size task force built around one brigade each from the 101st Air Assault, 82nd Airborne, and a light army division can be approximated as the average of those divisions -- 26,550 tons, including 200 helicopters and 13,500 personnel. (The Marines could also have substituted for the one of the brigades.) Because Rwanda is a landlocked country in Central Africa, and because speed is critical in stopping a genocide, the entire force would have been airlifted. The rate of airlift would have been constrained by factors such as the delay in loading planes at U.S. bases, excessive demand for air refueling, fuel shortages in Central Africa, and the limited airfield capacity in Kigali and at the potential staging base at Entebbe in neighboring Uganda. At an optimistic rate of 800 tons daily, the task force would have required 33 days to airlift. Personnel, which are much quicker to transport than their cargo, could have been sent first -- but it would have been imprudent to deploy them into the field without sufficient equipment and logistics. Several additional days would also have been required for the delay between the deployment order and start of airlift, for the gradual increase in the capacity of theater airfields unaccustomed to such traffic, and for travel to and unloading at the theater. In addition, the rate of force deployment might have been slowed by the need to use limited airlift capacity for food, medicine, and spare parts to sustain the first troops to arrive. Thus the entire force could not have closed in the theater until about 40 days after the president's order.

Advance units, however, could have begun operations much sooner. Approximately four days after the order, a battalion or two of Army Rangers could have parachuted in and seized Kigali airport at night. Follow-on troops could have expanded outward from the airfield to establish a secure operating base. Within about two weeks, sufficient troops and equipment could have arrived to halt the fighting, form a buffer between the FAR and the RPF in Kigali and northwest Rwanda, and fully police the capital. Only later, however, could the intervention force have turned in earnest to stopping the genocide in the countryside as helicopters, vehicles, and troops arrived.

Some observers have suggested that the genocide would have ceased spontaneously throughout Rwanda upon the arrival of Western enforcement troops in Kigali -- or possibly even earlier, upon the mere announcement of a deployment. They claim that the extremists would have halted killing in hopes of avoiding punishment. But these Hutu were already guilty of genocide and could not have imagined that stopping midway would gain them absolution. More likely, the announcement of Western intervention would have accelerated the killing as extremists tried to finish the job and eliminate witnesses while they had a chance. Such was the trend ahead of the RPA advance, as Hutu militias attempted to wipe out remaining Tutsi before the rebels arrived. During the genocide, the ringleaders even trumpeted false reports of an impending Western intervention to help motivate Hutu to complete the killings. Although the Hutu generally held back from mass killing at sites guarded by foreigners to avoid provoking Western intervention, they would have lost this incentive for restraint had such an intervention been announced.

The 6,000 U.S. troops deployed to the countryside would have been insufficient to establish a full police presence, but they could have found and protected significant concentrations of threatened Rwandans. Ideally, helicopter reconnaissance could have identified vulnerable or hostile groups from the air and then directed rapid response forces to disperse hostile factions and secure the sites. Alternately, ground troops could have radiated out from Kigali in a methodical occupation of the countryside. Displaced Rwandans could have been gathered gradually into perhaps 20 large camps for their protection.

Depending on the search method, large-scale genocide could have been stopped during the fourth or fifth week after the deployment order, by May 15 to May 25. Interestingly enough, this would have been before the task force's airlift had been completed. Based on the genocide's progression, such an intervention would have saved about 275,000 Tutsi, instead of the 150,000 who actually survived. Maximum credible intervention thus could not have prevented the genocide, as is sometimes claimed, but it could have spared about 125,000 Tutsi from death, some 25 percent of the ultimate toll.

A more modest intervention would have refrained from deploying U.S. troops to any area in Rwanda in which FAR and RPA troops were actively fighting. In late April, this would have limited U.S. troops to a zone consisting of six prefectures in the south and west of Rwanda. A single reinforced brigade would have sufficed given the reduced territory, population, and threat of potential adversaries. Ideally, the ready brigade of the 101st Air Assault Division would have been designated and supplemented by two additional light-infantry battalions, supporting units for peace operations, and additional helicopters and motorized vehicles: a force of 6,000 personnel, weighing about 10,000 tons.

For such an action, three main objectives would have been set: first, to deter and prevent entry of organized military forces into the above-mentioned zone; second, to halt large-scale genocide there; and third, to prepare for a handoff to a U.N. force. Strategic airlift would not have relied on Kigali airport, which was still a battleground in the civil war, but rather on neighboring Bujumbura in Burundi and Entebbe in Uganda -- which would have further constrained the deployment rates. Still, facing little military threat in the zone, these troops probably could have stopped large-scale genocide there within three weeks after the deployment order, by May 11, 1994. About 200,000 Tutsi from the zone could have survived, as opposed to about 100,000 from this part of Rwanda who actually did. Elsewhere in Rwanda, genocide would have continued until stopped by the RPA, as occurred, leaving only 50,000 survivors outside the zone of intervention. Moderate intervention thus could have spared about 100,000 Tutsi from death, or 20 percent of the ultimate toll. Surprisingly, moderate intervention in this case would have saved almost as many lives as the maximum alternative, because by avoiding combat areas the interveners could have turned sooner to counter genocide in the zone where most Tutsi lived.

The third alternative, a minimal intervention, would have attempted to mitigate the genocide without introducing U.S. ground troops into Rwanda, relying on airpower alone from bases in neighboring countries. For example, the United States could have threatened to bomb the extremist ringleaders and the FAR's military assets unless the killing was halted -- and then followed through if necessary. But if the threat alone failed to coerce, U.S. pilots would have had difficulty locating the ringleaders or hitting FAR positions without killing rebels as well. Even if air coercion had succeeded in Rwanda, a follow-on ground force would have been needed to keep the peace. Alternatively, the United States could have pursued airborne policing, which would have attempted to interdict physically and intimidate psychologically the perpetrators of the genocide throughout Rwanda. Significant numbers of U.S. attack helicopters and fixed-wing aircraft could have patrolled Rwanda daily from bases in neighboring countries. If armed factions threatening large groups of civilians were spotted, air-to-ground fire could have dispersed the assailants, at least temporarily. Such air patrols would have continued until deployment of non-American ground troops or until the RPF won the civil war. But airborne policing could not have prevented smaller acts of violence in the meantime.

Another minimal approach would have been to help the Rwandan Tutsi escape to refugee camps in bordering states -- Burundi, Tanzania, Uganda, or Zaire -- by using helicopter patrols to ensure safe passage. Rwanda has only about 600 miles of paved roads. Assuming a team of 20 helicopters with standard maintenance needs, five helicopters could have been kept aloft at a time, with each responsible for 120 miles of roadway. If these helicopters flew at a ground speed of 120 miles per hour, each section of roadway could have been patrolled approximately every hour. Airborne broadcasts and leaflets would have directed the Tutsi to the exit routes. Air-to-ground fire would have broken up roadblocks and dispersed armed gangs to ensure the free flow of refugees. But this strategy could not have saved those Tutsi unable to reach major roads and would have caused a major refugee crisis.

Each of the airpower options would have had drawbacks, including the risk of losing airborne personnel to anti-aircraft fire, but each also had the potential to save tens of thousands of Tutsi. Coercion might have stopped the genocide quickly, potentially facilitating a cease-fire in the civil war. Airborne policing could have allowed more Tutsi to be saved by France's Operation Turquoise or a similar follow-up deployment. Free passage also would have kept more Tutsi alive, albeit as refugees, and they might have returned home quickly after the RPF's victory. About 300,000 Tutsi still were alive in late April 1994, of whom about 150,000 subsequently perished. If minimum intervention had been able to avert half these later killings, it could have spared about 75,000 Tutsi from death, or 15 percent of the genocide's ultimate toll.

A WESTERN FAILURE OF WILL?

Many observers have claimed that timely intervention would have prevented the genocide. Some even asserted at first that UNAMIR itself could have done so, although most now acknowledge that the peacekeepers lacked sufficient arms, equipment, and supplies. Conventional wisdom still holds that 5,000 well-armed reinforcements could have prevented the genocide had they been deployed promptly when the killing began -- and that the West's failure to stop the slaughter resulted exclusively from a lack of will. Rigorous scrutiny of six prominent variations of this assertion, however, finds all but one dubious.

Human Rights Watch makes the boldest claim: Diplomatic intervention could have averted the genocide without additional military deployment. These advocates contend that a threat from the international community to halt aid to any Rwandan government that committed genocide would have emboldened Hutu moderates to face down the extremists and extinguish violence. As proof, they note that moderate FAR officers appealed for support from Western embassies during the first days of violence, and that the intensity of massacres waned after the West intensified its condemnations in late April.

However, this argument ignores the fact that virtually all of Rwanda's elite military units were controlled by extremist Hutu, led by Colonel Theoneste Bagosora. These forces demonstrated their power and ruthlessness by killing Rwanda's top political moderates during the first two days of violence. By contrast, moderate Hutu officers had virtually no troops at their disposal. The moderates avoided challenging the extremists not because of a lack of Western rhetorical support but because of mortal fear for themselves and their families. This fear was justified given that the extremists stamped out any nascent opposition throughout the genocide -- coercing and bribing moderate politicians, removing them from office or killing them if they did not yield, shipping moderate soldiers to the battlefront, and executing civilian opponents of genocide as "accomplices" of the rebels. The decline in massacres in late April is explained simply by the dwindling number of Tutsi still alive. International condemnation did little except compel extremists to try harder to hide the killing and disguise their rhetoric. Even these superficial gestures were directed mainly at persuading France to renew its military support for the anti-Tutsi war -- hardly an indication of moderation.

The only way that the army's Hutu moderates could have reduced the killing of Tutsi civilians would have been to join forces with the Tutsi rebels to defeat the Hutu extremists. This was militarily feasible, given that the Tutsi rebels alone defeated both the FAR and the Hutu militias in just three months -- but it was politically implausible. By April, Rwanda had already been severely polarized along ethnic lines by four years of civil war, the calculated efforts of propagandists, and the October 1993 massacre of Hutu by Tutsi in neighboring Burundi. Even moderate Hutu politicians once allied with the rebels had come to fear Tutsi hegemony. Although the moderate Hutu officers sincerely favored a cease-fire and a halt to the genocide, they could not realistically have defected to the Tutsi rebels -- at least until the FAR's defeat became imminent.

The second claim is that 5,000 U.N. troops deployed immediately upon the outbreak of violence could have prevented the genocide. But this assertion is problematic on three grounds. It assumes such troops could have been deployed virtually overnight. In reality, even a U.S. light-infantry ready brigade would have required about a week after receiving orders to begin significant operations in the theater and several more days for all its equipment to arrive. Further delays would have resulted from reinforcing the brigade with heavy armor or helicopters, or from assembling a multinational force. Even if ordered on April 10, as requested at the time by Dallaire, reinforcements probably could not have begun major operations to stop genocide much before April 20. Moreover, it is unrealistic to argue that urgent intervention should have been launched on April 10 -- given that the international community did not realize genocide was under way until at least ten days later.

Intervention advocates, such as the Carnegie Commission, also erroneously characterize the progression of the genocide. The commission claims that there was a "window of opportunity É from about April 7 to April 21" when intervention "could have stemmed the violence in and around the capital [and] prevented its spread to the countryside." In reality, killing started almost immediately in most of Rwanda, and by April 21, the last day of this purported "window," half the ultimate Tutsi victims already were dead. Even if reinforcements had arrived overnight in Kigali, Dallaire was unaware of genocide outside the capital and thus would not have deployed troops to the countryside in time to prevent the massacres.

Furthermore, 5,000 troops would have been insufficient to stop genocide without running risks of failure or high casualties. Only 1,000 troops would have been available for policing Kigali -- some three troops per thousand residents, which is grossly inadequate for a city in the throes of genocide. In the countryside, U.S. commanders would have faced a stark choice: either concentrate forces for effective action, leaving most of the country engulfed in killing; or spread forces thin, leaving troops vulnerable to attack. To avoid such painful choices in the past, U.S. military planners have insisted on deploying more than 20,000 troops for interventions in the Dominican Republic, Panama, and Haiti -- all countries with populations smaller than Rwanda's.

A third claim is that U.N. headquarters had three months' advance notice of genocide and could have averted the killing simply by authorizing raids on weapons caches. Critics cite the so-called genocide fax -- a January 11, 1994, cable from Dallaire to U.N. headquarters in New York that conveyed a Hutu informant's warning that extremists were planning to provoke civil war, kill Belgian peacekeepers to spur their withdrawal, and slaughter the Tutsi with an Interahamwe militia of 1,700 troops that the informant was training. The cable also reported an arms cache containing at least 135 weapons, which Dallaire wanted to seize within 36 hours.

Dallaire, however, raised doubts about the informant's credibility in this cable, stating that he had "certain reservations on the suddenness of the change of heart of the informant.É Possibility of a trap not fully excluded, as this may be a set-up." Raising further doubt, the cable was the first and last from Dallaire containing such accusations, according to U.N. officials. Erroneous warnings of coups and assassinations are not uncommon during civil wars. U.N. officials were prudent to direct Dallaire to confirm the allegations with Habyarimana himself, based on the informant's belief that "the president does not have full control over all elements of his old party/faction." Dallaire never reported any confirmation of the plot.

Even if the U.N. had acquiesced to Dallaire in January 1994, it is unlikely the weapons cache could have been seized or that doing so would have prevented the genocide. The U.N. actually did reverse itself barely three weeks later, on February 4, 1994, granting Dallaire authorization to raid weapons depots. But his forces failed in every attempt, even after an informant identified three new caches on February 7. By mid-March, six weeks after receiving authorization, the peacekeepers had captured only a paltry total of 16 weapons and 100 grenades; their rules required cooperating with Rwandan police, who tipped off the extremists. If the U.N. had permitted Dallaire to act without consulting local authorities, Kigali could have responded under Chapter VI of the U.N. Charter (which governs consensual peacekeeping operations) by simply expelling the force. The peacekeepers also were vulnerable to violent retaliation, as they were dispersed and still lacked armored personnel carriers at the time. In addition, Dallaire's cable identified a cache of only 135 weapons -- a tiny fraction of the 20,000 rifles and 500,000 machetes imported by the government over the preceding two years. Even had Dallaire managed to seize this cache without prompting expulsion or retaliation, he could not have derailed the wider genocide plot without significant reinforcements.

A fourth claim holds that quickly jamming or destroying Hutu radio transmitters when the violence broke out could have prevented the genocide. A Belgian peacekeeper who monitored broadcasts testified, "I am convinced that, if we had managed to liquidate [Radio Mille Collines], we could perhaps have avoided, or in any case limited, the genocide." A human rights advocate characterized the jamming as "the one action that, in retrospect, might have done the most to save Rwandan lives." But radio broadcasts were not essential to perpetuating or directing the killing. By April, Rwandans had been sharply polarized along ethnic lines by civil war, propaganda, and recent massacres in Burundi. Habyarimana's assassination was a sufficient trigger for many extremist Hutu to begin killing. Moderate Hutu were usually swayed not by radio broadcasts but by threats and physical intimidation from extremist authorities. Furthermore, orchestration of the genocide relied not merely on radio broadcasts but on the government's separate military communications network. Silencing the radio might have had most impact prior to the genocide, when broadcasts were fostering polarization, but such action would have been rejected at the time as a violation of sovereignty. Even if hate radio had been preventively extinguished, the extremists possessed and used other means to foster hatred.

The fifth variant of the intervention argument is that the Western forces sent to evacuate foreign nationals during the first week could have restored order in Kigali -- and thereby prevented the genocide had they merely been given the orders to do so. Just four days after Habyarimana's assassination, some 1,000 lightly armed Western evacuation troops, mainly French and Belgian soldiers, had arrived in Kigali, where Belgium's 400-troop UNAMIR contingent was already stationed. Another 1,100 reserves were less than two hours away by air. But it is doubtful that this small force, lacking the right equipment or logistical support, could have quickly quashed violence in the capital -- or that doing so would have stopped the genocide elsewhere. The Western evacuators had to commit half their force to guarding the airport at the town's outskirts and a few key assembly points, leaving few available for combat. In addition, coordinated action would have been inhibited by the widespread perception that France and Belgium sympathized with opposite sides in the civil war. Moreover, Kigali was defended by 2,000 elite Rwandan army troops and several thousand regulars equipped with heavy weapons, another 2,000 armed fighters of the Hutu militia, and 1,000 national police. Also located there were more than 1,000 Tutsi rebels who had access to surface-to-air missiles and had explicitly threatened to attack the evacuators if they extended their mission. Even if the small Western force had somehow halted the violence in Kigali, it lacked the equipment and logistics to deploy troops quickly to the countryside. Rural killing probably would have continued unless the ringleaders were captured and coerced to call off the slaughter. Such a search would not have been a quick or simple matter for any force, as demonstrated by the failed search for the Somali warlord Mohamed Farah Aidid by U.S. troops in 1993. Ill-equipped evacuation troops could have wasted weeks looking for the ringleaders while genocide continued at a torrid pace in the countryside, where 95 percent of Rwandans lived.

The sixth claim is most realistic: Had UNAMIR been reinforced several months prior to the outbreak of violence, as Belgium urged at the time, genocide might have been averted. More troops with the proper equipment, a broad mandate, and robust rules of engagement could have deterred the outbreak of killing or at least snuffed it out early. Such reinforcement would have required about 3,500 additional high-quality troops in Kigali, armored personnel carriers, helicopters, adequate logistics, and the authorization to use force to seize weapons and ensure security without consulting Rwandan police. This would have been the 5,000-troop force that Dallaire envisioned -- but one deployed prior to the genocide.

Under the U.N.'s peacekeeping rules, Rwanda's government would have had to consent to such a change -- and probably would have. Prior to the genocide, its cabinet still was dominated by the Hutu opposition moderates who had negotiated the Arusha accords, which called for a neutral international force to "guarantee [the] overall security of the country." The U.N. Security Council had watered down implementation of this provision, authorizing UNAMIR only to "contribute to the security of the city of Kigali." As tensions mounted in early 1994, the Rwandan government again asked the U.N. to dismantle armed groups, but the peacekeepers were too weak. Belgium pleaded for reinforcements and a new mandate from the Security Council in January and February 1994 on the grounds that UNAMIR could not maintain order. But the United States and Britain blocked this initiative before it could even reach a vote, citing the costs of more troops and the danger that expanding the mission could endanger peacekeepers -- as had occurred in Somalia the previous October.

The Rwandan government, however, almost certainly would have welcomed a reinforcement of UNAMIR prior to the genocide. Five thousand troops in the capital would have meant 16 troops for every thousand Rwandans, a ratio historically sufficient to quell severe civil disorders. Such a force might well have deterred the genocide plot. Failing that, well-equipped peacekeepers could have protected moderate Hutu leaders and Tutsi in the capital and captured some of the extremists during the first days of violence, thereby diminishing the chance of large-scale massacres in the countryside. Indeed, such early reinforcement of UNAMIR is the only proposed action that would have had a good chance of averting the genocide.

LESSONS

The most obvious lesson of Rwanda's tragedy is that intervention is no substitute for prevention. Although the 1994 genocide represents a particularly tough case for intervention in some respects -- such as its rapid killing and inaccessible location -- it would have been a relatively easy mission in other respects, including the limited strength of potential opponents. Yet even an ideal intervention in Rwanda would have left hundreds of thousands of Tutsi dead. To avert such violence over the long term, there is no alternative to the time-consuming business of diplomacy and negotiation. Tragically, international diplomatic efforts in Rwanda prior to the genocide were ill conceived and counterproductive.

Whether pursuing prevention or intervention, policymakers must use their imagination to better anticipate the behavior of foreign actors. In Rwanda, Western officials failed to foresee the genocide, despite numerous warning signs, in part because the act was so immoral that it was difficult to picture. Increased awareness of such risks demands that any peacekeeping force deployed preventively to a fragile area be adequately sized and equipped to stop incipient violence -- rather than be sent as a lightly armed tripwire that serves mainly to foster a false sense of security. If the West is unwilling to deploy such robust forces in advance, it must refrain from coercive diplomacy aimed at compelling rulers to surrender power overnight. Otherwise, such rulers may feel so threatened by the prospect of losing power that they opt for genocide or ethnic cleansing instead. Western diplomacy that relies mainly on the threat of economic sanctions or bombing has provoked a tragic backlash not just in Rwanda, but also in Kosovo and East Timor over the last few years as local rulers opted to inflict massive violence rather than hand over power or territory to lifelong enemies. In each case, Western military intervention arrived too late to prevent the widespread atrocities.

Obviously, time is of the essence once large-scale attacks against civilians begin. Most such violence can be perpetrated in a matter of weeks, as was demonstrated in Rwanda, Kosovo, and East Timor. Despite this reality, domestic politics often prevents an American president from quickly launching a major intervention. Thus U.S. defense planners should be more creative in developing limited alternatives. The case of Rwanda underscores that lighter intervention options that avoid combat areas and focus mainly on stopping violence against civilians could save almost as many lives if pursued seriously and expeditiously. Rapid responses would be facilitated by the development of pre-prepared plans for known trouble spots and by better coordinating intelligence from available sources, including nongovernmental organizations.

That said, tradeoffs are inevitable if the United States hopes to increase its effectiveness in humanitarian military intervention. To deploy troops faster, additional "ultra-light" units (like the Tenth Mountain Division) would have to be created, either by converting existing heavier units intended for major contingencies or by increasing defense spending. The Pentagon's recent proposal to trim some heavy mechanized forces down to medium-weight units would not solve the problem, because they would still be too heavy for a quick airlift. Lighter units probably could save more lives abroad but would also be subject to more casualties and potential failure. Such tradeoffs should be made only after rigorous debate, which to date has been virtually absent in the United States.

Finally, no policy of humanitarian military intervention should be implemented without a sober consideration of its unintended consequences. Recent interventions, whether in Bosnia, Kosovo, or East Timor, have been motivated by the impulse to provide humanitarian aid to a party visibly suffering in an internal conflict. But intervention in those cases also resulted in the weaker sides being bolstered militarily. This pattern creates perverse incentives for weaker parties in such conflicts to reject compromise and escalate fighting because they expect foreign intervention or hope to attract it. The result is often tragedy, as intervention arrives too little or too late to protect civilians. Thus a policy of intervening to relieve humanitarian emergencies that stem from internal conflicts may actually increase the number and extent of such emergencies -- a classic instance of moral hazard.

Inevitably, decisions on whether and how to intervene in specific cases will be caught up in politics. But this challenge should not deter hard thinking on when and how such intervention can be most beneficial -- or detrimental. If Rwanda demonstrates nothing else, it is that thousands of lives are at stake in such decisions.

1 Some accounts claim that one million Tutsi lived in Rwanda before the genocide, making up 12 percent of the population, which would correspond with the estimate of 850,000 killed. But historical demographic data suggest otherwise. In 1956, a Belgian census counted almost 17 percent of the population as Tutsi, but half of those fled or died in the violence that accompanied independence. The remaining 9 percent subsequently had a lower fertility rate than the Hutu, reducing the Tutsi population to the 8 percent reported in the 1991 census.