Operation Enduring Freedom:
Why a Higher Rate of Civilian Bombing Casualties
Project on Defense Alternatives Briefing Report #13
Despite the adulation of Operation Enduring Freedom (OEF) as a "finely-tuned" or "bulls-eye" war,1 the campaign failed to set a new standard for precision in one important respect: the rate of civilians killed per bomb dropped. In fact, this rate was far higher in the Afghanistan conflict -- perhaps four times higher -- than in the 1999 Balkans war. In absolute terms, too, the civilian death toll in Afghanistan surpassed that incurred by the 1999 NATO bombing campaign over Kosovo and Serbia; indeed, it may have been twice as high. Key among the factors shaping this outcome were (i) the mission objectives of Operation Enduring Freedom, (ii) some of the operational and tactical features of the bombing campaign, and (iii) the mix and technical characteristics of the weapons employed.
Through 10 December the total numbers of attack sorties and weapons expended in Operation Enduring Freedom (OEF) were far less than those in the 1999 Balkans campaign, Operation Allied Force (OAF): approximately 4700 attack sorties in OEF versus 13,000 in OAF; and, 12,000-plus weapons in OEF versus 23,000 in OAF.2 Nonetheless, credible reports of bombing mishaps and accidental civilian casualties suggest a level of civilian fatalities in Afghanistan greater than that experienced in the 1999 Kosovo war. At least 500 civilians were killed in the Kosovo war by the NATO bombardment.3 In Afghanistan, it is very likely that the bombing campaign claimed 1000-1300 civilian lives. (See Appendix 1. Estimation of Civilian Bombing Casualties: Method and Sources).
Given that fewer weapons were expended, a higher level of civilian fatalities in Operation Enduring Freedom implies that the bombing campaign in Afghanistan was less accurate than the one associated with the 1999 Balkans war.
Of course, there is a propagandistic aspect to the accounting and reporting of civilian casualties that complicates any effort to discern the facts.4 At the end of October, the Taliban asserted that more than 1600 civilians had been killed in bombing raids during the first three weeks of the war -- an average of 533 per week. US defense officials discounted these claims as lies, preferring to talk in terms of dozens of casualties (when they were willing to offer estimates at all).5 However, in early November, a report by British intelligence, directly countered the Taliban claims, estimating that only 300 civilians had been killed during October -- an average of approximately 90 per week.6 (Notably, bombing was relatively light during two of the weeks covered by this report; it became both more intense and more free-ranging after 20 October.)
Another insight on bombing casualties has been provided by Dr. Marc Herold, a University of New Hampshire economics professors, who has compiled a database of hundreds of articles on the war from the world press. This database includes accounts of more than 4,000 civilian deaths from bombing during the period 7 October - 1 January 2002.7
As noted above, the present study uses a lower estimate: between 1000 and 1300 civilians killed in the bombing campaign through 1 January 2002. This estimate relies on a press review that is less extensive than the Herold review, but that applies a more stringent accounting criteria in order to correct for likely reporting bias.8 Regarding an upper-end estimate of casualties: the present study finds it difficult to reconcile a civilian death toll from bombing that is much higher than1300 with the conditions being reported currently by journalists on the ground in Afghanistan -- although this may change when (and if) more comprehensive and systematic surveys are conducted. (See Appendix 2. Resolving Discrepancies in Casualty Accounts.)
The estimate used in the present study is broadly consistent with two other recent reviews: one by Human Rights Watch, which calculated at least 1000 civilian deaths, and one by Reuters news agency, which concluded that perhaps 982 people were killed in 14 incidents.9 It is also broadly consistent with an extrapolation of the estimate made by British intelligence at the end of October.
The high likelihood that 1000-1300 civilians were killed in the OEF bombing campaign directly contradicts the notion that the campaign was "cleaner" than other, recent ones. Instead, in terms of the rate of civilian deaths per bomb or missile expended, there seems to have been a distinct deterioration from the standard set in Operation Allied Force (1999), in which fewer civilians were killed and more munitions used.
The remaining sections of this report explore why Operation Enduring Freedom might have incurred a higher casualty rate. Stated more formally: the next sections examine the prior plausibility of the hypothesis that the Afghanistan bombing campaign imposed a higher rate of civilian casualties than did the bombing of Serbia and Kosovo in 1999.
A reliable and comprehensive empirical accounting of the civilian death toll may be a long time coming; indeed, it may not come at all or it may be left to non-governmental organizations with limited resources, as was the case following the Kosovo war. Nonetheless, there are good reasons, primae facie, to expect that the Afghanistan bombing campaign was less accurate than the one executed as part of Operation Allied Force.
Many features of the bombing campaign in Afghanistan distinguish it from the 1999 campaign in the Balkans, although not all pertain to the issues of accuracy and collateral damage:10
Several of these distinctive characteristics of Operation Enduring Freedom may have contributed to an increased rate of civilian casualties relative to the experience of the 1999 Balkan war. Three are examined below: weapon mix, mission objectives, and targeting methods. By looking at these we can assess the prior plausibility of the civilian casualty rate in the Afghan war being greater than that in the 1999 Kosovo war.
One development that distinguishes the Afghanistan bombing campaign that should have led to fewer civilian casualties than in the 1999 Kosovo war was an increase in the percentage of "smart" weapons used. However, within the category of 'smart" weapons, there also was a switch in emphasis from laser-guided bombs to bombs directed by means of the Global Positioning System (GPS).
Most current GPS directed weapons, such as the Joint Direct Attack Munition (JDAM), are simply less accurate than laser-guided bombs.15 Indeed, GPS-directed weapons are not routinely called "precision" weapons at all, but "accurate" or "near precision" ones. Under test conditions, JDAMs have been able to reliably achieve a Circular Error Probable (CEP) of approximately 10-13 meters -- meaning that fifty percent of the JDAMs dropped will hit within 32-42 feet of their programmed coordinates. By comparison, laser-guided bombs routinely achieve CEPs of 3-8 meters. Even a difference as small as an 8-meter versus a 10-meter CEP equates to being able to put 50 percent of expended weapons within a 2100 square foot circle versus being able to put them in a circle of 3300 square feet. Should an intended target sit among a cluster of buildings, the difference between these two circular areas is significant. And, of course, in either case 50 percent of the weapons fall outside the circles.
There are several reasons for the increased US emphasis on GPS-directed weapons: they are cheaper than laser-guided bombs, they can be used in all weather conditions, they can be launched from much greater distances and, because their targets need not be designated while the weapons are in flight, they can be dispensed in large batches. As substitutes for "dumb bombs" they promise greater precision in attack. But if the present generation of GPS-directed munitions are used to substitute for more precise laser-guided weapons, the aggregate effect will be a reduction in the average accuracy of the smart weapon mix.
What we have seen in the Afghanistan campaign relative to Operation Allied Force is an increase in the percentage of "smart" weapons but a sharp decrease in the percentage of these that are laser-guided. This correlates with an increased dependence on heavy bombers, which mostly delivered GPS-guided and unguided munitions.16
Through 10 December bombers flew approximately 11 percent of all strike sorties in Operation Enduring Freedom -- that is, more than 500 sorties out of total of approximately 4700. By comparison, in Operation Allied Force heavy bombers flew only about 2 percent of the 13,000 strike sorties. They nonetheless dropped more than a third of the munitions in the Balkan campaign.
In the first week of Operation Enduring Freedom bombers dropped more than 80 percent of all munitions: 500 GPS-guided bombs, 1000 Mk82 dumb bombs, and 50 cluster bombs -- out of a total of about 2000.17 But during this first week they accounted for approximately 40 percent of all strike sorties; as noted above, their proportion of sorties has declined since then. Still, they will probably account for close to half of all air-dropped munitions in this war.
In addition to the ordinance delivered by bombers, Navy fighters dropped fewer than 300 weapons in the first week -- a mix of GPS-directed and laser-guided ones. Thus, the mix of weapons during the first week of the war was probably about 55 percent unguided, 35 percent GPS-directed, and 10 percent laser-guided. Although Navy fighter-bombers played a bigger role after the first week, they also increased their use of unguided weapons.18 Also, carpet-bombing by heavy bombers did not begin in earnest until late October.
By 10 December approximately 12,000 munitions had been dropped or launched over Afghanistan. Of these, 6,732 -- or 56 percent -- were precision or near-precision weapons (including GPS-directed).19 At present, a reasonable estimate for the final mix of air-delivered munitions used in the Afghanistan war is 40 percent unguided, 20 percent laser-guided, and 40 percent GPS- directed. This compares to the experience of Operation Allied Force in which approximately one-third of the weapons were "smart weapons" (mostly laser-guided) and two-thirds unguided. This should lead one to expect, primae facie, a substantial improvement in the average accuracy of attack. But the switch in emphasis from laser-guided to GPS-guided weapons in the "smart category" would lessen the expected margin of improvement.
Also affecting the precision of attack in Afghanistan would be the increased percentage of cluster bomb usage. Information provided by the United States to the UN Mine Action Program indicates that there were 103 submunition strikes in Afghanistan by the end of the year. Complete reports are available on only 78 of these strikes in which a total of 1,210 cluster bomb units were used. This suggests that the total number of these weapons employed in Afghanistan might surpass 1,600 -- which is the approximate number used in the 1999 Balkans war. Even the lower number would imply a higher percentage of cluster bomb usage in Afghanistan. And while some cluster bombs can be directed to a target by precision means, they are designed to disperse their sub-munitions over a broad area on arrival -- typically 100 X 50 meters. (See Endnote 11.)
Demographic differences are important in considering the relative impact of bombing campaigns in different countries. Obviously, a greater population density favors increased casualties, other things being equal (such as target sets and bombing methods and accuracy). Because the population density of Afghanistan is only one-third that of Yugoslavia, it would seem, prima facie, that Afghanistan could absorb more bombing errors than could Yugoslavia without suffering a higher rate of collateral casualties. But Afghanistan's population is more unevenly distributed than Yugoslavia's -- and the US bombing campaign largely followed the pattern of population distribution.
Much of Afghanistan's population resides in only a third of the country's territory. Another third of the country is virtually unpopulated. Population density in the most populated third of Afghanistan is comparable to that of Spain. Especially densely populated are the areas around a few major cities: Herat, Jalalabad, Kabul, Kandahar, Kunduz, and Mazar-i-Sharif. Kabul has a population about two-thirds as large as inner city Belgrade, but covers an area only one-half as large; thus, Kabul's population density is approximately one-third greater than Belgrade's.
The fronts of the ethnic war in Afghanistan have tended to form in proximity to heavily populated areas. And these areas were also the sites of most of the US bombing activity through mid-December. Almost all of the US bombing activity occurred in only nine of Afghanistan's 32 provinces. These nine provinces comprise less than 25 percent of the country's territory but host more than 50 percent of its population. Moreover, in one of the largest of these provinces -- Herat -- bombing activity focused on only a small portion of the province's territory.
On balance, differences in population density still weigh against the likelihood of a higher casualty rate in Afghanistan -- but not nearly so much as national-level statistics might imply.
Unlike the campaign in the Balkans, which principally used strategic attack as a lever of coercive diplomacy, Operation Enduring Freedom had "regime removal" as one of its key objectives. One way of accomplishing this was to broadly target Taliban leaders and cadre -- not just military facilities, equipment, and front line troops. Given the relatively informal nature of leadership in Afghanistan and the "militia" character of the Taliban armed forces, this entailed targeting residences among other sites and, thus, some residential areas.20 Of course, Al Qaeda residences -- as best they could be identified -- were also considered fair targets. Indeed, the elimination of suspected Al Qaeda members -- wherever they were and whatever they might be doing -- was a prime campaign objective. Unfortunately, targeting Taliban and Al Qaeda leaders and cadre "at home" increased the likelihood that their families would be caught in the attack as well. And targeting residential areas meant a much reduced margin for error in attack, generally.
Thus, mission objectives, operational plans, and the character of the enemy might interact to exert considerable upward pressure on the civilian casualty count. And this would have been true even if the techniques and tools of the bombing campaign were as precise on average as those employed in the 1999 Balkans war -- which they were not.
Targeting also depended to an unusual degree on "intelligence" from local sources that were at times less than entirely trustworthy. Given the campaign time line, some intelligence relationships had to be developed very quickly; the time available for testing those relationships was limited. This made the US bombing campaign unusually susceptible to being coopted by local rivalries that might be either irrelevant or even detrimental to US campaign objectives. Two possible instances of such co-optation are the 20 December bombing of a convoy and village in Paktia province and the 14 December bombing of Pul-e-Khumri, a town in northern Afghanistan.21 In both cases, US attacks may have depended on local intelligence sources that were involved in active, local rivalries. The attacks impacted significantly on these rivalries, although the victims of the attacks insist that there were no Al Qaeda or Taliban among their number.
Another reason that error and casualty rates may have been higher in Afghanistan than in the Balkan war is that the Afghan air campaign focused much more on mobile, emerging, and opportunistic targets.22 When aircraft engage such targets, ground-based air controllers or spotters may be unavailable or unable to provide good GPS-coordinates or to designate the targets with lasers. Rapid engagement may also preclude air crews taking time to derive and input GPS coordinates or laser-designate a target -- thus compelling a trade of accuracy for time.23 Increased emphasis on emerging targets led Navy pilots to increase their use of dumb bombs. Because these do not require laser designation or the input of GPS coordinates, they allow a faster reaction time.
B-1 and B-52 bombers also employed flexible targeting during the Afghan campaign -- a capability they gained and began practicing only 18 months before the war began. Among US bombers, the B-2 has an unsurpassed capacity for achieving both flexibility and accuracy in bombing. It has been capable of flexible targeting for years. B-2's delivered almost all the GPS-directed munitions used in the 1999 Balkans war. In Afghanistan, by contrast, B-2's essentially retired from the war after the first week, leaving 18 B-1's and B-52's to conduct the vast majority of heavy bomber sorties. One technique for flexible bombing employed by B-52's in OEF had Navy F-14's acquire emerging targets, determine their coordinates, and pass the information along to B-52's by way of AWACs aircraft.24
Initial field reports suggest that the Afghanistan bombing campaign directly claimed more than 1000 civilian lives -- or more than one for every 12 bombs or missiles expended. By comparison, in the 1999 Balkans campaign the rate may have been as low as one civilian death for every 46 bombs dropped.
A variety of factors may have contributed to there being a higher rate of civilian casualties in the Afghanistan war than in the 1999 Balkans conflict. The most likely include campaign objectives, the nature of the enemy, the reliability of some intelligence sources, and the emphasis on and methods for the rapid engagement of mobile and emerging targets.
The weapon mix used in Afghanistan should have allowed greater precision in attack, although there were important countervailing factors. While the percentage of "smart" weapons was twice as high as in the 1999 Balkans war, there was a change in the mix of smart weapons used: GPS-guided weapons were much more prevalent in Afghanistan. Also, among all weapons, the proportion of cluster bombs used in Afghanistan was probably in the 10-15 percent range, compared with 7 percent in the 1999 Balkans campaign.
The estimate of civilian bombing casualties used in this report -- 1000-1300 -- draws on media sources much as the Herold study does, but it applies a stricter criteria to screen these sources and correct for likely reporting errors and distortions. In deriving the 1000-1300 estimate only Western press sources were used for hard numbers -- principally wire services (Reuters, Associated Press, and Agence France-Presse) and the British press (BBC News, the Independent, The Times, and the Guardian). These sources seemed more attuned to the issue of civilian casualties than were US newspapers, while also being disinclined to accept on face value official Taliban reports or accounts from the Pakistani press.
Within the large body of press accounts reviewed for this study, the estimates were anchored to a subset: (i) journalist eyewitness accounts of damage, injury, and burial and (ii) journalist interviews with medical personnel, aid workers on the scene, and individuals who lost family members. These well-investigated cases cover less than 25 percent of the reported incidents and they provide strong evidence of more than 300 deaths. These cases provided a yardstick for evaluating other reports.
Also important in deriving the 1000-1300 estimate were journalist interviews with refugees, although numerical estimates given by refugees of civilian deaths outside their own families were not taken at face value. In three cases where refugee reports of non-related casualties could be checked against journalist or other investigations on the scene of a reported bombing, the refugee recollections disagreed with the scene reports by a factor of more than three, on average. Similarly, official Taliban tallies of casualties (which were often broadcast via the Pakistan press) seem to disagree with journalist scene reports by a factor of more than four, on average. (The discount factor for Taliban reports is based on cases in Kabul that were investigated by reporters for the Agence France Presse.)
One need not assume duplicity as the explanation for exaggeration. In fact, mis-perception and exaggeration should be expected in the recollection of traumatic events by frightened or injured people. As for the quality of official reports: even assuming honest intent, this depends on the investigative capabilities, carefulness, and good-functioning of bureaucracies. In the best of times, the Taliban might be expected to under-perform. And, of course, they had little incentive to "err on the conservative side" in estimating civilian casualties for the press during the war.
For the purposes of the present study, when Afghan refugee or government reports were expressed in vague terms, the following reduction factors were used to derive an estimate: "some or a few" deaths was interpreted as 1, "a dozen or more" was interpreted as 3-4, "dozens" was interpreted as 8-10, "scores" was interpreted as 10-15, "hundreds" was interpreted as "40-60". When accounts explicitly mixed "dead" and "wounded" or gave a combined total for "casualties", only 25 percent of the estimate was treated as "dead".
If all Taliban government and Afghan refugee accounts of the numbers of civilians killed or wounded in the bombing campaign are taken at face value they would suggest a total of more than 5000 killed and 10,000 wounded. As noted above, it is likely that the actual toll is less than one-quarter as many. This discrepancy, although large, is not particularly surprising. In the United States, official estimates of the number of people killed on 11 September were initially twice as high as where they sit today. It took more than a month for the figures to be adjusted downward and more than two months before they came close to the present official estimate. As of 22 December, the official toll of those killed on 11 September still disagrees with other authoritative accounts (by the media and charities) and probably will be revised downward by another 15 percent before stabilizing. The final accounting of the deaths suffered on 11 September will probably be 50 percent below the estimates that prevailed during the first month after the attack.
Sources on civilian casualties in Operation Enduring Freedom (reverse chronological order):
Rory Carroll, "Deaths blamed on US blunder; Pressure mounts on Karzai to call for end to US air strikes," The Guardian (London), 28 December 2001;
Paul Salopek, "U.S. bombs leave wasteland; Fierce attacks anger villagers, raise questions," Chicago Tribune, 28 December 2001;
Norimitsu Onishi, "Even Precision Bombing Kills Some Civilians, Tour of a City Shows," New York Times, 26 December 2001, p. B2;
Stephen Farrell, "Graves testament to off-target bombings," The Statesman (India), 12 December 2001;
Tom Bowman, "'Friendly fire' kills 3 U.S. troops; Incident illustrates perils of air strikes on nearby positions," Baltimore Sun, 6 December 2001, p. 1;
Toby Helm and Stephen Robinson, "US bomb hits Afghan PM as deal is signed," Daily Telegraph (London), 6 December 2001;
Richard Lloyd Parry I, "Civilians Abandon Homes after Hundreds Are Casualties of US Air Strikes on Villages," The Independent (London), 5 December 2001, p. 14;
Alex Spillius, "Thousands flee shell shocked Kandahar," The Age (Australia), 5 December 2001;
John Donnelly, "Unintended Victims Fill Afghan Hospital," Boston Globe, 5 December 2001;
Richard Lloyd Parry, "A village is destroyed. And America says nothing happened," The Independent (London), 4 December 2001;
Philip Smucker "Villages pay price as US bombs go awry," Daily Telegraph (UK), 4 December 2001;
Chris Tomlinson, "Afghan village riddled with bomb craters; 155 villagers said killed,"
Associated Press, 3 December 2001;
Tim Weiner, "Villagers Dying Under U.S. Bombs, Anti-Taliban Forces Say," New York Times, 3 December 2001;
Richard Lloyd Parry, "US bombs hit wrong target for second time in two days in Jalalabad," The Independent (UK), 3 December 2001;
Richard Lloyd Parry, "Scores killed by B-52s," The Independent (UK), 2 December 2001;
"15 killed as US mistakes private jeep for military vehicle," AFP, 2 December 2001;
David Filipov, "US bombers gone, but bombs remain; Villagers return to ruined homes and deadly risks," Boston Globe, 2 December 2001;
Megan K. Stack, "Villagers Near Targeted Caves Voice Plight; Afghanistan: Residents plead for an end to U.S. bombings. The air raids on suspected Bin Laden hide-out killed scores of civilians, officials and witnesses say," Los Angeles Times, 2 December 2001, p. 3;
Rory McCarthy, "US planes rain death on the innocent; 'Precision' raids kill residents in capital city," The Guardian (London), 1 December 2001;
Kathy Gannon, "Residents flee as U.S. bombers pound Kandahar," Associated Press, 1 December 2001;
Chris Tomlinson, "Residents say village leveled by US bombs, civilians dead; U.S. says it didn't happen," Associated Press, 1 December 2001;
Alan Freeman, "Cluster-bomb casualties mount," Globe and Mail, November 28, 2001 Page A5;
Justin Huggler, "Legacy of civilian casualties in ruins of shattered town," The Independent (London), 27 November 2001, p.1;
James Rupert, "Bomb Victims' Pain Endures; Civilians die in U.S. raids," Newsday (New York, NY), 24 November 2001, p. 19;
"Afghanistan: Accountability for civilian deaths," Amnesty International News Service No. 189, 22 November 2001;
Andrew Maykuth, "Civilian casualties far fewer than reports suggested," Knight Ridder News, 21 November 2001;
Will Englund, "Amid Afghan suffering, hard questions of war; A child soldiers on as villagers ask why America bombs them," Baltimore Sun, 20 November 2001, p. 1A;
Justin Huggler, "Carpet Bombing 'Kills 150 Civilians' in Frontline Town," The Independent (London), 19 November 2001, p. 1;
"50 US cluster bombs found in Pakistan," The Times of India, 15 November 2001;
Tim Reid, "Bomb destroys al-Jazeera's office in Kabul," The Times (London), 14 November 2001;
William Arkin, "Bombing the Red Cross," WashPost.com, 4 November 2001;
Chokar Karaiz, "Merciless US bombing obliterates village: 60 killed," Dawn (Pakistan), 2 November 2001;
Rajiv Chandrasekaran, "Villagers Describe Deadly Air strike; U.S. Calls the Site A Rightful Target," Washington Post, 2 November 2001;
"Cluster Bombs in Afghanistan," Human Rights Watch Backgrounder, October 2001;
Bassam Hatoum, "U.S. Jets Damage Kandahar Hospital," Associated Press, 31 October 2001;
"Afghanistan: New Civilian Deaths Due to U.S. Bombing," Human Rights Watch Backgrounder, 30 October 2001;
Kathy Gannon, "U.S. air strikes target Taliban; 13 civilians reported killed," Associated Press, 29 October 2001;
Anthony Loyd, "Rogue bomb destroys village's joy," The Times (London), 29 October 2001;
Andrew Buncombe, Richard Lloyd Parry, and Phil Reeves, "Bombing 'errors'," The Independent (London), 29 October 2001;
John Nichol, "The myth of precision," Guardian Unlimited (UK), 29 October 2001;
Steve Gutkin, "Scene of devastation at site of apparent stray U.S. strike: 'The world shook'," AP, 28 October 2001;
Kathy Gannon, "Scenes of despair as dead of latest air strikes mourned in Kabul," AP Worldstream, 28 October 2001;
Mohammad Azam Said, "Bombing blunders kill more civilians as US-led campaign enters fourth week," Agence France Presse, 28 October 2001;
Steve Gutkin, "Afghans: Three Villages Hit by US," AP Online, 28 October 2001;
"Girls killed as US bomb strikes village, Red Cross stores razed," AFP, 26 October 2001;
John Zubrzycki and Roland Watson, "Stray US cluster bombs kill nine," The Australian, 25 October 2001, p. 9;
Dudley Althaus, "Afghans pouring across border; Refugees detail civilian deaths, say Taliban hiding among villagers," Houston Chronicle, 25 October 2001, p. 1;
Patrick Moser, "US military strikes criticized, Afghan civilians killed by cluster bombs," Agence France Presse, 25 October 2001;
Rupert Cornwell, "Pentagon admits US jets bombed old people's home in Herat," The Independent (London), 24 October 2001;
Said Mohammad Azam, "UN says US bombs struck mosque, village as civilian casualties mount," Agence France Presse, 24 October 2001;
Tom Bowman, "Pentagon concedes errors over weekend; Bombs strayed, hit near homes," Baltimore Sun, 24 October 2001, p. 1A;
"List of incidents of US bombs striking non-military targets," Agence France Presse, 24 October 2001;
"US bombs hit civilian districts in chase for Taliban troops: UN," Agence France Presse, 23 October 2001;
Kathy Gannon and Amir Shah, "Kabul residents say 8 civilians killed in midday U.S. strike," Associated Press, 21 October 2001;
"US bombs leave 10 dead in Kabul neighborhood: witnesses" Agence France Presse, 21 October 2001;
Stephen Farrell and Zahid Hussain, "US missed most Kabul targets, aid workers say," The Times (London), 20 October 2001;
Kathy Gannon and Amir Shah, "U.S. jets strike Kabul, civilians reportedly killed," Associated Press, 19 October 2001;
"At least six dead as US bombs hit Afghan homes: witnesses," Agence France Presse, 18 October 2001;
Kathy Gannon and Amir Shah, "U.S. strikes in Afghan capital hit homes, killing at least 5 civilians," AP, 18 October 2001;
"US bomb hits boys school, does not explode," Agence France Presse, 17 October 2001;
Said Mohammad Azam, "Red Cross warehouse bombed, US switches to low-level raids," Agence France Presse, 16 October 2001;
Kathy Gannon, "Afghanistan's Female Bombing Victims," Associated Press, 15 October 2001;
Kathy Gannon, "Fresh graves, fresh ruin in village Taliban claims was hit by U.S. strike," AP, 14 October 2001;
Stephen Farrell and Zahid Hussain, "Bombing victim tells how US raid hit village," The Times (London), 13 October 2001.
Paul Gallagher And Simon Pia, "Bombs Were Dropping and There Was Fire Everywhere,"
The Scotsman, 13 October 2001, p. 3;
Richard Lloyd Parry, "Air Strikes on Afghanistan: Casualties - Witnesses Confirm That Dozens Were Killed in Bombing," The Independent (London), 13 October 2001, p. 3; and,
Kathy Gannon and Amir Shah, "Afghan civilians mourn dead in Kabul," AP Online, 9 October 2001;
The difficulty of estimating the civilian toll of the bombing campaign is illustrated by the case of Kandahar, where estimates of the civilian bombing toll ranged from only 50 or 60 up to many hundreds. The discrepancies can be partly attributed to differences of interest among those reporting the tallies: Taliban versus anti-Taliban authorities, for instance. Different news organizations also to seem to have employed different initial hypotheses in conducting their investigations. The American press -- notably the Washington Post and Los Angeles Times -- was more sensitive to reports of low casualties and light damage; the British press -- notably the Guardian and Independent -- seemed to have an ear for reports of more substantial collateral effects. But other factors are also at work and these make it possible to reconcile some of the discrepancies.
Reports in the Pakistani press suggest that the total number of civilian bombing fatalities in the Kandahar area exceeds 500 -- but these reports substantially reflect Taliban claims. Conversely, Gul Agha Shirzai, the anti-Taliban military commander and new head of Kandahar, has reported the official tally of local civilians killed in the bombing as 50 to 60. Although it is difficult to find strong evidence of the Taliban claims, the tally put forward by the anti-Taliban commander also is suspect. Very well-documented incidents of collateral damage suggest a higher toll than reported by the new government. These incidents include an attack on the village of Daman outside Kandahar that killed 26 civilians, according to surviving family members. Civilian deaths also occurred when warplanes hit the Kila Jedid ammunition dump, Taliban positions near the airport, and a military vehicle park. Abdul Ali, a Taliban opposition leader who directed US air strikes from inside the city (using a cell phone), reports 14 civilians killed when one of the attacks he called down (on a police station) went astray. He also confirms three civilians dead from the ammunition dump attack.
According to other eyewitnesses numerous civilians also were killed in the attacks on the Taliban and Al Qaeda compounds in Kandahar, which included family living quarters and abutted residential areas. Security guards posted by the new government estimated that 40 families had been living in and around the main compound and that 100 individuals had been killed in the bombing. Most of the men killed in this complex were Taliban and, thus, they may have been excluded from the official civilian death toll. But their wives and children were also among the dead and may have constituted the majority. One of the guards explained to a Los Angeles Times reporter:
There were about a hundred killed here. You have to understand, this headquarters is surrounded by civilian houses. About 70% of the people who lived in this village were killed, and more than 100 injured.
Another guard reflected: "These were the families of our enemy. But you know, the children were not our enemy." This comment calls attention to a substantive issue: who counts as a civilian? The practice of strategic bombardment tends to allow many civilian targets under the categories of command and control sites and leadership sites. And, if the residences of political leaders and military personnel are to be targeted, it seems certain that some of their family members will die. This raises legal and moral questions not addressed in the present analysis. However, what is pertinent to a simple accounting of civilian deaths is the status of those family members. The dichotomy between military personnel and civilians has meaning only if it is held distinct from the dichotomy between friends and enemies. One might argue that the death of Taliban and Al Qaeda family members was unavoidable given the exigencies of war. They should nonetheless count as civilians in any estimate of the cost of this war.
Differences regarding the number and accounting of civilian casualties are apparent in the reports of medical personnel, too. Faizal Rabi, a doctor at Mirwais, Kandahar's largest hospital, disputes reports of numerous casualties. He told one reporter: "We heard there were some civilians killed. But I don't really know how many." However, the head nurse of Mirwais, Abdullah Mohammad, told other reporters that the hospital had received 40 to 50 wounded every day during the bombing, which was most intense after the fall of Kabul. According to Mohammad the hospital recorded 5-10 people killed daily with no more than ten percent of them being Taliban or Arabs. With heavy bombardment having occurred for three or four weeks, the head nurse's report implies a civilian death toll of between 100 and 250 people.
Two factors (other than political orientation) that can help explain the discrepancies are differences in the scope of the reports and differences in their evidential base. Some reports seem to focus on just Kandahar proper; others include its outlying districts. Some reports seem to exclude the wives and children of the Taliban and their foreign allies from the "civilians" category; others include them. Finally, some reports include injury and fatality estimates from hospitals in Quetta, Pakistan, such as the Sandeman Provincial Hospital, which have received many wounded from Kandahar and surrounding areas. (Ambulance drivers for one aid organizations reported bringing an average of six dead daily across the border, half of them civilian).
Increasing the scope of reporting interacts in an important way with the bombing strategy used by US air forces in their attacks around Kandahar. Attacks on the city proper seemed carefully targeted on obvious centers of Taliban power or sites designated by agents within the city. These sites included Taliban government buildings and suspected Al Qaeda meeting places, military installations and barracks, suspected Taliban and Al Qaeda housing compounds, electrical and telecommunications facilities, headquarters of the religious police, and the madrassah (Islamic school). Many of these sites were attacked repeatedly.
Attacks on areas outside the city and along roads leading out of the city were more free-ranging and opportunistic, with the aim of preventing the escape of Taliban and Al Qaeda fighters. And this may have interacted with refugee flows in an unfortunate way: everyone, soldiers and civilians alike, traveled by similar means. Yousuf al-Shouly, a correspondent for TV station al-Jazeera, told British journalist Jonathan Steele that F14 Tomcats played the lead role in these attacks. In al-Shouly's words, they hovered "like vultures looking for targets." Stripped of its bias, this is a not bad description of how tactical aircraft engage emerging targets. It also accords with refugee reports, such as those of a young woman, Rukia, who was attacked while fleeing Kandahar by car. Speaking to a reporter for the Atlanta Journal and Constitution, she conveyed the impression that the aircraft were "bombing anything that moves." Similarly, refugees fleeing Tirin Kot, told the Agence France Presse of losing 20 people from their group when the trailer they were riding was attacked from the air.
The preceding analysis suggests that some of the discrepancies in reports of civilian bombing victims can be reconciled by taking differences of scope into account and by assessing how the US bombing strategy might produce different "ground-level" experiences in different places. In light of the US bombing strategy and the large number of people fleeing Kandahar (100,000 by some estimates), it is not difficult to reconcile reports of fewer than 100 civilian bombing deaths in the city proper with a higher toll for the province as a whole. An estimate of 200-250 bombing deaths in the wider area is consistent with a variety of the narrower reports -- on the assumption that the wives and children of Taliban and Al Qaeda fighters count as civilians.
The death toll in Kandahar province cannot simply be extrapolated to the rest of Afghanistan, however. Kandahar province is home to slightly less than six percent of Afghanistan's population. But the province undoubtedly suffered more than average from the bombing campaign. Indeed, almost all the bombing activity occurred in only nine of Afghanistan's 32 provinces. The total population of the heavily targeted provinces (including Kandahar) is almost eight times that of Kandahar province alone.
Sources on Kandahar:
"US pounds Taliban frontline as UN warns of civilian casualties," Agence France Presse, 24 October 2001;
Scott Baldauf, Afghans flee Kandahar bombing, Christian Science Monitor, 6 December 2001, p. 6;
Justin Huggler, "War in Afghanistan: US Bombers Guided by Spy with a Phone; Our Man Behind the Lines," The Independent (UK), 16 December 2001, p. 15;
Tasgola Karla Bruner, "Wounded civilians bitter toward US," Atlanta Journal and Constitution, 8 December 2001, p. 11;
Kim Murphy, "US Bombing Spares Much of Kandahar," Los Angeles Times, 13 December 2001, p. 22;
John Pomfret, "Kandahar Bombs Hit Their Marks; Few Civilian Deaths Evident," Washington Post, 12 December 2001, p. 1;
Jonathan Steele, "How bombing and diplomacy eased the Taliban's grip on Kandahar, The Guardian (UK), 7 December 2001; and,
Juan O. Tamayo, "Kandahar in limbo: Old and new, Taliban not gone but rebels not in control," Knight Ridder/Tribune News, Service 13 December 2001.
1. Fred Kaplan, "New Warfare; High-tech US Arsenal Proves its Worth," Boston Globe, 9 December 2001, p. 34; Thomas E. Ricks, "Bull's-Eye War: Pinpoint Bombing Shifts Role of GI Joe," Washington Post, 2 December 2001, p. 1; Eric Schmitt and James Dao, "Use of Pinpoint Air Power Comes of Age in New War," New York Times, 24 December 2001, p. 1; and, Ann Scott Tyson, "US is prevailing with its most finely tuned war," Christian Science Monitor, 21 November 2001, p. 1.
2. "Operation Enduring Freedom: Operations," Global Security.org, 20 December 2001, internet; and, William M. Arkin, "How smart was this war really," MSNBC, 12 June 1999, internet.
3. For an accounting of the civilian bombing toll in Allied Force, see: "Civilian Deaths in the Nato Air Campaign," Human Rights Watch Report, Volume 12, No. 1, February 2000, <
4. How Many Dead? Major networks aren't counting, Fairness & Accuracy In Reporting, 12 December 2001 < Scott Baldauf, "In the war of spin, US opens a new front," Christian Science Monitor, 8 November 2001, p. 7; Kevin Canfield, "No Winners in Battle to Gauge Cost of War; Civilian Death Count Becomes Another Weapon," Hartford Courant, 26 December 2001, p. A1; Andrew Gumbel, "Who Is Winning the War of Lies?", Independent (London), 4 November 2001, p. 5; Holger Jensen, "Propaganda War Escalates along with Military Action," Rocky Mountain News, 8 November 2001, p. 33A; Howard Kurtz, "CNN Chief Orders 'Balance' in War News; Reporters Are Told To Remind Viewers Why U.S. Is Bombing," Washington Post, 31 October 2001, p. C1; Steven Livingston, "The Battle for Information: Can We Know What's True?", Newsday, commentary, 14 October 2001, p. B4; and, Jay Rayner, "Propaganda and Media: How much can we believe in the news campaign?", The Observer (London), 14 October 2001.
5. Anne Barnard, "Rumsfeld says air strikes must go on," Boston Globe, 5 November 2001, p. A8; Edward Cody, "Taliban Claims Large Civilian Casualties; Afghan Rulers Increase Efforts to Win Support From Islamic World," Washington Post, 12 October 12 2001, p. A23; Edward Epstein, "U.S. battles to justify bombings to Muslims; It rejects Taliban tab of civilian deaths," San Francisco Chronicle, 16 October 2001, p. A1; Damon Johnston, "Casualty claim 'ridiculous'," Daily Telegraph (Sydney), 16 October 2001, p. 5; and, Jonathan Weisman, "Civilian death count disputed," USA Today, 16 October 2001, p. 3A.
6. Macer Hall and David Wastell, "Truth and lies of Taliban's death claims; A report by military intelligence reveals how the Taliban are deliberately distorting the numbers of civilian casualties," Sunday Telegraph (London), 4 November 2001, p.14.
7. Seumas Milne, "The innocent dead in a coward's war: Estimates suggest US bombs have killed at least 3,767 civilians," The Guardian (London), 20 December 2001, p.16. Professor Marc W. Herold, A Dossier on Civilian Victims of United States' Aerial Bombing of Afghanistan: A Comprehensive Accounting, (10 December 2001). Database available at: http://www.cursor.org/stories/civilian_deaths.htm
8. See Appendix 1. Estimation of Civilian Bombing Casualties: Method and Sources.
9. Murray Campbell, "Thousands of Afghans likely killed in bombings," Globe and Mail (Toronto), 3 January 2002.
10. The terms precision and accuracy are used interchangeably here, although formally they refer to different measures. According to the USAF Intelligence Targeting Guide, "'Precision' is the closeness with which repeated measurements made under similar conditions are grouped together, and 'accuracy' is the closeness of the best estimated value obtained by the measurements to the 'true' value of the quantity measured." In terms of weapon delivery: precision measures the scatter of shots aimed at an identical point; accuracy measures the deviation of the center of a scatter pattern from the true location of the supposed target. "Section 13.3.2. Precision and Accuracy," Air Force Pamphlet 14- 210: USAF Intelligence Targeting Guide (Washington DC: HQ USAF, February 1998).
11. The United States has provided information to the UN Mine Action Program on a total of 103 sub-munition strikes in Afghanistan. "Of these, 78 were struck by a total of 1,210 CBUs, equaling a total of 244,420 sub-munitions." Information on the other strikes was still forthcoming. The estimate for the total number of cluster bombs used in the 1999 Balkans war comes from a report by Human Rights Watch. "UN to Clear Coalition Cluster Bombs," press release, UN Integrated Regional Information Network, 2 January 2002, available at: < and, Human Rights Watch, World Report 2001, < chapter: Arms/ Weapons and the Conduct of War/ Cluster Bombs.
12. William Arkin, "Dropping 15,000 Pounds of Frustration; US weapons superiority meets its match in Afghanistan's difficult terrain and local allies with their own agendas," Los Angeles Times, 15 December 2001, p. 10.
13. "DoD news Briefing; Secretary of Defense Donald H. Rumsfeld," M2 Presswire, 24 December 2001; and, "Hunt for Bin Laden Will Enter a New Phase with Use of 'Thermobaric' Bombs in Caves, St. Louis Post-Dispatch, 22 December 2001, p. 5.
14. James Dao, The New Air War: Fewer Targets, More Hits and Scarcer Targets, New York Times, 29 November 2001, p. 1; David A. Fulghum and Robert Wall, "U.S. Stalks Taliban With New Air Scheme," AWST, 15 October 2001; and, Elaine Grossman, "US Challenge in Targeting Afghanistan: Is there enough to bomb?," Inside the Pentagon, 27 September 2001.
15. Sources on GPS munitions:
Air Force Pamphlet 14- 210: USAF Intelligence Targeting Guide (Washington DC: HQ USAF, February 1998), Chapter 13, The Target Location;
"Joint Direct Attack Munition (JDAM)," in Director, Operational Test & Evaluation, FY00 Annual Report (Washington DC: DOTE, 2000), http://www.dote.osd.mil/reports/FY00/airforce/00jdam.html;
"Smart Bombs Made Dumb? Did Faulty Batteries Cause Failure Of Precision Guided Weapons?", cbsnews.com, 6 December 2001;
"Why bombing can go wrong," BBC News, 16 October 2001, internet;
Carlo Kopp, "GPS Part IV: US Direct and Indirect Attack Munition Programs," Australian Aviation (November 1996);
Kopp, "GPS Part III: US Direct Attack Munition Programs," Australian Aviation (October 1996);
Major Keith J. Kosan, Precision Engagement Against Mobile Targets: Is Man in or Out?, thesis, School of Advanced Airpower Studies, Air University, Maxwell Air Force Base, June 2000;
Paul Richter and Peter Pae, "High-Tech U.S. Bombs Are Precise but Not Perfect; munitions can go astray because of electronic, mechanical and human errors," Los Angeles Times, 24 October 2001, p. 16; and,
CDR Ronald J. Unterreiner (USN), et. al., "Close Air Support (CAS) in 2025: Computer, Lead's in Hot," in Air Force 2025 Final Report, Volume 3, Power and Influence (Maxwell Air Force Base: Air University, December 1996), Chapter 3, System Description.
16. David A. Fulghum and Robert Wall, "Heavy Bomber Attacks Dominate Afghan War," AWST, 3 December 2001; and, William M. Arkin, "In praise of heavy bombers," Bulletin of the Atomic Scientists ( July/August 1999), p. 80,
17. William M. Arkin, "A Week of Air War," WashingtonPost.com, 14 October 2001.
18. Robert Wall, "Navy Adapts Operations For Afghan War Hurdles," AWST, 19 November 2001.
19. Richard Newman, "Inside the War Room: How the war in Afghanistan is being run from Florida," US News & World Report, 17 December 2001, p. 18.
20. Two such incidents are reviewed in Appendix 2; For more detail, see: Justin Huggler, "War in Afghanistan: US Bombers Guided by Spy with a Phone; Our Man Behind the Lines," The Independent (UK), 16 December 2001, p. 15; Kim Murphy, "US Bombing Spares Much of Kandahar," Los Angeles Times 13 December 2001, p. 22; and, Jonathan Steele, "How bombing and diplomacy eased the Taliban's grip on Kandahar, The Guardian (UK), 7 December 2001.
21. David Filipov, "Another Deadly, Errant US Attack is Alleged," Boston Globe, 24 December 2001, p. 1; Sadaqat Jan, "Afghan elder warns Karzai over convoy bombing," Reuters, 23 December 2001; John Otis, "Airstrikes pose a deadly dilemma; Some say Afghan tribal rivalries can taint intelligence, cause errors," Houston Chronicle, 7 January 2002, p. 1; and, Kim Sengupta, "Americans 'duped' into attack on convoy," The Independent (London), 24 December 2001.
22. Elaine Grossman, "Air Force Chief Launches Major Effort to Improve Targeting Speed," Inside the Pentagon, 8 November 2001; and, Vernon Loeb, "Technology Changes Air War Tactics, Washington Post, 28 November 2001, p. 16.
23. Robert Wall, "Navy Adapts Operations For Afghan War Hurdles," AWST, 19 November 2001.
24. David A. Fulghum and Robert Wall, "Heavy Bomber Attacks Dominate Afghan War," AWST, 3 December 2001; Robert Wall, "F-14s Add Missions In Anti-Taliban Effort," AWST, 21 November 2001; Wall, "Navy Adapts Operations For Afghan War Hurdles," AWST, 19 November 2001; and, Frank Wolfe, "Bombers Able To Use Flexible Targeting For Afghan Campaign," Defense Daily, 26 October 2001.
Citation: Carl Conetta, Operation Enduring Freedom: Why a Higher Rate of Civilian Bombing Casualties, Project on Defense Alternatives Briefing Report #13. Cambridge, MA: Commonwealth Institute, 18 January 2002.
The Project on Defense Alternatives, The Commonwealth Institute
Copyright © The Commonwealth Institute. All Rights Reserved.