At the beginning of the Iraq war in 2003, there was one number that was crucial to American military officials as they planned airstrikes.
"The magic number was 30," said Marc Garlasco, who was the Pentagon's chief of high-value targeting at the start of the war. "That means that if you hit 30 as the anticipated number of civilians killed, the airstrike had to go to Rumsfeld or Bush personally to sign off." If the expected number of civilian deaths was less than 30, however, neither the president nor the secretary of defense needed to know.
Four years later, the U.S. military still has rules in place that permit the killing of civilians in airstrikes. In fact, the number of anticipated civilian deaths is carefully appraised beforehand in a calculation known as the collateral damage estimate, which is then reviewed by commanders and military attorneys who must decide if the benefits of the strike outweigh the cost in innocent civilian lives. The military's goal is to reconcile precision-bombing technology with the web of international treaties, including the Geneva Conventions, designed to protect civilians during wartime under a legal rubric called the Law of Armed Conflict.
But what these rules mean is that killing civilians is legal -- as long as the deaths are the result of a strike at a legitimate military target. And it also means that some unknown percentage of civilian deaths from airstrikes in Iraq and Afghanistan are not accidents.
"'Accident' is not the right word," said Sarah Holewinski, executive director of the Campaign for Innocent Victims in Conflict, which works to help remunerate civilians caught in the crossfire. "They call them accidental deaths, but they are not," said Holewinski. "They know what they are doing."
The decisions the bomb planners are making about who lives and who dies are increasingly important. While Washington's political elite wrestles with how to wind down the ground war in Iraq, the air wars over Iraq and Afghanistan are escalating sharply. In the first six and a half months of 2007, according to a Human Rights Watch database, the U.S. Air Force dropped 527,860 pounds of bombs in Afghanistan -- nearly equal to the 575,500 pounds during all of 2006, according to that database. In Iraq, the tonnage is lower, but the increase is more dramatic. In the first half of this year, 222,000 pounds of bombs fell in Iraq, compared to 61,500 during 2006.
Today, thanks to improved technology and intelligence, U.S. officials know with more certainty than ever that they may kill innocent civilians in an airstrike at a legitimate military target, like a key insurgent leader. But the same technology, they say, also permits them to minimize the inevitable civilian casualties. Col. Gary Crowder, the deputy director of the Combined Air Operations Center in Qatar, the command hub running the air wars in both Iraq and Afghanistan, calls it "precision as a method to decrease casualties rather than a method to hit the target." Within the past two years, the military has come to recognize that collateral damage is a crucial consideration in the larger battle for hearts and minds. Is the collateral damage estimate a particularly chilling example of military bureaucracy, or is it a modern and efficient tool for saving lives?
What is happening in Iraq and Afghanistan today is a far cry from the sweeping, indiscriminate "carpet bombing" of Vietnam 40 years ago, when nobody really knew when the strings of falling bombs from massive B-52s might be landing on civilians. It's also different from the situation in the first invasion of Iraq 16 years ago, when, despite all the publicity about "smart bombs," 90 percent of the aerial munitions used were blunt instruments -- old-fashioned, unguided "dumb bombs." Things have even changed a lot since the invasion of Iraq four years ago, when estimates of collateral damage were crude and based on outdated information.
The transformation began with the intersection of high technology and bad publicity in the first Gulf War. The pivotal moment was the airstrike that killed approximately 288 Iraqis sheltered in the al-Firdos bunker in Baghdad on Feb. 13, 1991. After that incident the man running the war, Gen. Norman Schwarzkopf, personally reviewed every proposed target and reportedly called off some strikes. The military realized that the advent of precision-guided munitions gave it the ability to minimize collateral damage as well as increase the percentage of targets destroyed. The first Gulf War gave birth to a modern military bureaucracy that could analyze and approve airstrikes based, in part, on anticipated civilian casualties.
A dozen years later, during the opening phase of the second Iraq war, there was a system in place for calculating those casualties. In 2003, collateral damage estimates of civilian deaths from planned airstrikes were done with computer models using outdated population density statistics. There was also that magic number, 30, that required a signature from someone in Washington.
According to Crowder, there was a higher tolerance for civilian casualties during the first phase of the war than there is at present. Garlasco, now a senior military analyst at Human Rights Watch, describes a callous approach back in 2003. "It is important to understand that in January 2003 when the target packages were 'finalized,' we had about 300 targets that were considered 'high CD,' or high collateral damage, meaning over 30," Garlasco wrote in an e-mail. "We had the Air Force play with the bomb angles, fusing, bomb tonnage, etc., and got that number down to about 25."
Garlasco said every high-CD target was ultimately approved except two structures holding foreign journalists, the Al-Rasheed hotel and Saddam's Ministry of Information.
The days of the "magic number" of 30 are over. The sign-off process now used to approve airstrikes in Iraq and Afghanistan is classified. But there is still methodology in place for collateral damage estimates. Planners must still ask and answer a set series of questions about each airstrike, including the number of expected civilian casualties.
The collateral damage must be proportional to the military advantage gained by the strike. "You can't deliberately target civilians," explained Georgetown law professor David Koplow. "But the proportionality judgment contemplates that sometimes civilians will be harmed in pursuit of a military strike."
The number of civilian casualties is still a key factor in determining which strikes are approved and who must sign off. "That will determine who the approval authority is," Crowder said. Strikes likely to involve higher collateral damage still require higher approval.
Four years after the invasion there is far better on-the-ground intelligence and far better technology. The skies over both Iraq and Afghanistan are filled with buzzing drones armed with powerful cameras. Military officials hundreds of miles away planning an airstrike can, in some cases, literally count the number of women and children near a target by staring at a screen. In other cases, troops on the ground calling in airstrikes can see nearby civilians with their own eyes. And individual "smart bombs" dropped in Iraq or Afghanistan are steered to pinpoint targets with stunning precision, sometimes guided by a laser shot by someone standing on the ground staring at the target.
We do know some of the details of the classified sign-off process. Many of the tough calls are made in Crowder's command center in Qatar, 700 miles south of Baghdad. It is the brain center in control of every detail of the air wars above Iraq and Afghanistan. In the air-conditioned comfort of that command hub, Crowder's team assemble explicit "air-tasking orders" for almost all of the airplanes flying over the two war zones. This includes targets that have been reviewed by commanders and vetted by attorneys.
There are two kinds of airstrikes. The first are preplanned strikes aimed at what are sometimes called "deliberate targets." An example would be a bridge, a Taliban leader, or an insurgent safe house quietly under surveillance from a circling drone. The military calls this new capability to eye a target for a long time "persistent look."
The cold, hard math of estimating collateral damage for preplanned strikes includes a calculation of the precise number of expected civilian deaths from each bomb, and it is made every day for every strike in the air wars over Iraq and Afghanistan. This grisly number is what must be weighed against the military value of the target in question. Garlasco calls that process "the macabre calculus of trying to determine how many dead civilians are worth a dead bad guy."
Planners say they go to great lengths to minimize collateral damage. The size of each bomb, ranging from 250 pounds to the extremely devastating 2,000-pound varieties, is minimized. The direction of the approach of the aircraft is altered to focus the explosive force away from civilians. And extreme prejudice is supposed to be employed when civilians are thought to be within 1,000 feet of a target, a situation referred to as "danger close." Military officials even consider what is on the ground underneath an approaching plane to minimize the damage in case a malfunction causes munitions to fall short of their targets. "We always look at our run-in heading and what is underneath the run-in heading," Crowder said. "This is a constant evolution of trying to get more and more precision."
When the assessment has been made, the planners submit the proposed airstrike to a military attorney whose job is to make sure the assessment has been done according to the law and that the anticipated casualties are within set limits.
These kinds of deliberate targets, however, represent a minority of the airstrikes in Iraq and Afghanistan. Much more common is the second category of airstrike: close air support. These are hastily arranged attacks from the air to relieve troops on the ground under enemy fire. They present a unique set of challenges with respect to protecting civilians.
Some planes are tasked to simply circle Iraq and Afghanistan, patrol air space, and wait for the troops on the ground, pinned down by enemy fire, to call for bombs. These situations are dubbed "troops in contact."
During "troops in contact" situations, the targeting for an airstrike is done by a person called a joint terminal attack controller -- the person on the ground authorized to aim the bombs. In some cases, the controller is out in the field right next to the troops in contact. And he is likely to aim those bombs using either a laser or a GPS system.
In the laser method, the controller on the ground aims the laser at the target. A pilot high above that target identifies the laser dot, and the coordinates of the dot are routed to the "smart bomb" under the nearby circling airplane. Once released, the bomb steers itself to the target.
A controller may instead rely on a GPS system and laser range finder to calculate the coordinates of the target. The controller reads the coordinates of the target to the pilot above via radio. The pilot punches in the numbers, reads the coordinates back to the controller -- and it is bombs away.
Controllers carry data on collateral damage ranges for each variety of bomb. But when troops are in a firefight and need immediate help from the air, decisions have to be made quickly, with greater emphasis on protecting U.S. forces. Steps are taken to minimize civilian deaths, but there simply is no time for a formal collateral damage assessment. Often there is only limited information on where any vulnerable civilians might be. "In those kinds of circumstances," Crowder said, "that is where you see most of the civilians being killed."
The result of improved technology and intelligence is that the military is more likely to hold its fire in 2007 than in 2003. In fact in the large majority of cases, Crowder says, commanders simply do not drop bombs at all if intelligence shows with some certainty that any innocent civilians are likely to be killed in a preplanned airstrike. "Our default is not to drop," he said in a telephone interview from Qatar. Rare situations do still arise when the excruciating decision goes the other way. "There are circumstances where we accept the fact that a target is of such value that there may be civilian casualties," Crowder said. "That is a hard calculation to make."
But all this effort to minimize civilian casualties does not stem from altruism alone. In the past two years, as American military officials have grasped the fact that they are dealing with a long-term insurgency, they've also become aware that the wars in Iraq and Afghanistan may hinge on the hearts and minds of the civilians who sometimes end up in the crosshairs. Crowder is studying "A Savage War of Peace," Alistair Horne's 1977 history of the French occupation of Algeria, illustrative of the dynamics of an insurgency. "It is an issue ultimately of growing the capacity of indigenous forces," Crowder said. "Every civilian casualty we create undermines the support for those forces and reduces the likelihood of obtaining our objective."
Other U.S. officials agree that civilian deaths from airstrikes erode support for U.S. forces and the governments they are trying to prop up. "You could win the battle and lose the war," said Air Force Maj. John Thomas, a spokesman for the International Security Assistance Force in Afghanistan, the NATO force trying to provide stability in that country and help the government led by President Hamid Karzai. "The fight is to not lose the support of the Afghan people or this fledgling Afghan government," Thomas said in an interview from Kabul.
Civilian deaths are a particularly sensitive issue for U.S. officials in Afghanistan, after being stung this summer by a string of harrowing headlines. News reports last month noted allegations of 90 civilian deaths from airstrikes over a 10-day period. Karzai held a news conference last month calling air operations over Afghanistan "careless." Then on June 19, the Agency Coordinating Body for Afghan Relief, representing nearly 100 humanitarian groups, released a statement of "strong concern" about the death toll.
Is the criticism fair? Does the military's system of collateral damage assessment work? The success of U.S. efforts to minimize casualties from airstrikes is extremely difficult to judge, since statistics for Iraqi and Afghan fatalities are scarce and hotly debated. In Afghanistan in particular, what is happening on the ground is difficult to assess. The terrain is rough and remote. There are few reliable witnesses or independent assessments of civilian deaths. But it is at least possible that civilian deaths from airstrikes are decreasing even as the tonnage of bombs dropped increases.
A study published in the British medical journal Lancet last October painted a famously bleak picture of spiraling violence in Iraq. The Lancet study came in at the high end of casualty estimates for the war, estimating that 655,000 more Iraqis have died since the invasion than would have died if the occupation had never occurred, and that fully 601,000 of those deaths were violent. The study said 31 percent of the violent deaths were attributable to coalition forces. The study, overseen by epidemiologists at Johns Hopkins and derided by the Bush administration as not credible, said the number of violent deaths attributable to coalition forces has increased every year since the invasion. But there was one exception: Compared with 2003 and 2004, the number of reported deaths from airstrikes decreased in 2006.
Anecdotally, U.S. officials also add some caveats to recent civilian casualty headlines. They acknowledge that mistaken intelligence has led to some truly accidental deaths, including the deaths of children. But in other cases, they say, Taliban or al-Qaida sympathizers appear to be delivering inflated reports of civilian deaths. The Taliban also fights among civilians, dramatically increasing the likelihood of civilian deaths when troops call for close air support. (U.S. officials say that during a recent firefight, innocent civilians were forced into a trench alongside Taliban fighters so that any airstrike would result in the significant loss of civilian lives).
The unreliability of civilian casualty numbers from Afghanistan is exemplified by an Associated Press report from Friday, July 27. The AP said that airstrikes in Helmand province, in support of NATO troops, had killed 50 Taliban and 28 civilians. The source was Gereshk district chief Abdul Manaf Khan, who based his claims on reports from villagers. The AP noted that Khan's claims could not be verified because of the remote location -- and that there was apparently no evidence. Though the clashes had started Thursday night, Khan said the bodies had already been buried, even as fighting continued Friday. The article also quoted Malim Mirwali, a member of Parliament for Gereshk, who claimed that more than 40 civilians had been killed in that airstrike.
In a telephone call from Kabul, Garlasco, the airstrike chief turned human rights advocate, said he is trying to get to the bottom of the overall civilian casualty numbers in Afghanistan. According to Garlasco, his research to date shows that the number of civilian deaths reported from airstrikes is inflated.
But one of the hurdles to determining whether the U.S. military is successfully limiting civilian casualties is that the military itself isn't doing a good job of counting. That's why advocates for civilians in Iraq and Afghanistan, while applauding the military's steps to minimize civilian deaths, remain skeptical about whether those steps are working.
The military is supposed to go back and review an airstrike using aerial footage and, when possible, conduct an inspection by troops on the ground. But thorough postmortems are rare. The military devotes resources to preventing civilian casualties before munitions are used, the advocates say, but it isn't devoting the same energy to see if its preventive measures are working. "You have to do this collateral damage estimate beforehand," said Holewinski, from the Campaign for Innocent Victims in Conflict. "And afterwards, in order to make sure you did it correctly, you are supposed to do a collateral damage assessment. And that is what they rarely do." Crowder said that the military reviews airstrikes from aerial footage, but admitted that on-the-ground investigations are far from guaranteed. "In some cases, you can't get there in a timely fashion."
And regardless of whether the American measures are effective, human rights activists are well aware that the NATO forces fighting in Afghanistan have tighter rules about collateral damage than the Americans. The statement issued by human rights groups last month, which cited "disproportionate and indiscriminate" air power -- language quoted from the Law of Armed Conflict -- directed most of its venom at "forces or agencies outside NATO command, often American forces in Operation Enduring Freedom."
In an unconventional and complicated arrangement in Afghanistan, roughly 32,000 NATO-led troops are conducting stabilization activities there. Meanwhile, approximately 21,000 U.S. service members are conducting counterterrorism operations in Afghanistan under Operation Enduring Freedom, currently spearheaded by the 82nd Airborne Division and special operations units. Commanders from NATO and Operation Enduring Freedom both call in airstrikes, carried out almost exclusively by U.S. aircraft getting their air-tasking orders from Crowder's shop in Qatar, the Combined Air Operations Center.
NATO forces conduct their own collateral damage estimates for preplanned targets like Taliban leaders. Thomas, the spokesman in Kabul, says NATO forces have adopted a "zero tolerance" policy on civilian casualties for preplanned strikes. This is because the NATO stabilization mission is particularly dependent on the good will of the Afghan people. (And NATO partners are extremely sensitive to civilian casualties.)
Thomas calls the airstrike protocol there "the most strict possible rules for civilian casualties that you could possibly have." Forces there will not carry out a strike "if there is any reasonable evidence or reason to think that there might be a single civilian in a compound or an area."
Even for close air support for troops fighting under the NATO flag, Thomas says, forces go to great lengths to avoid dropping bombs on innocent civilians. "Self-defense trumps everything," he said, noting that close air support is far more prevalent than preplanned strikes. "But even in a self-defense-type situation, the idea is to try to break contact some other way to avoid endangering civilians."
But Thomas also confirms that the standards for civilian deaths are "more strict" for NATO forces than for U.S. forces in Operation Enduring Freedom. "Our job is not a counterterrorism mission," Thomas explained. "That's not the same for other missions going on here or in Iraq."
"Yes, the Taliban are killing civilians," said Garlasco, summarizing the situation. "And yes, the U.S. is doing its darnedest to make sure they don't ... But when it really comes down to it, they are still dropping the bombs, so it is incumbent on them at all times to do their best and follow international law to make sure they are not killing innocent civilians." On whether they are succeeding in preventing civilian deaths, Garlasco said, "I really think they are trying very hard. It's tough."