How We Fight

WAR CRIMES AND MISDEMEANORS

On the morning of November 19, 2005, a U.S. Marine unit patrolling Haditha, a small farming town in the heart of Iraq's violence-plagued Anbar Province, was hit by a powerful roadside bomb. The explosion killed one marine and wounded two others. But that was just the start of the killing that day. Following the attack, four Iraqi college students and a driver who were approaching the scene in a taxi were also killed, as were 19 civilians in nearby houses, including an elderly man in a wheelchair and several children and the women who were trying to shield them.

The first official account of the incident claimed that only 15 civilians had died, all from the bomb blast. In March 2006, an extensive Time magazine investigation challenged that version of events. Subsequent inquiries found that the marines had probably killed all 24 Iraqi civilians themselves and that U.S. commanders should have conducted a more thorough investigation earlier. The marines involved in the incident appear likely to face criminal charges, including for murder.

If the marines' responsibility is confirmed, the Haditha killings will be the gravest violation by U.S. forces of the legal prohibition against the wanton targeting of civilians since the invasion of Iraq. A bedrock of the laws of war for more than a century, noncombatant immunity encompasses two key concepts: distinction and proportionality. (Although the United States has not ratified all the relevant international conventions codifying these rules, it treats most of them as binding.) Parties to hostilities are expected to always distinguish combatant forces and military objects, which are legitimate targets, from civilian populations and hospitals, schools, places of worship, and important cultural sites, which are not. Under no circumstances is force to be deliberately applied against strictly noncombatant targets. But according to the principle of proportionality, collateral damage to civilians and civilian objects is acceptable if it is the byproduct of attacks on legitimate military targets and the damage does not obviously exceed the anticipated military benefit.

If the allegations are proved true, the Haditha incident will represent an egregious violation of these principles. What is less clear is whether this incident was the result of the reprehensible actions of a few or just one instance of systematic misconduct by U.S. forces in Iraq. In other words, is Haditha the exception or the rule?

Judging by a June 2003 Pew Global Attitudes survey, many people are likely to conclude the latter. The poll found that over 90 percent of Jordanian, Moroccan, Palestinian, and Turkish respondents and over 80 percent of Indonesian and Pakistani respondents felt that the United States "didn't try very hard" to avoid civilian casualties in Iraq. That view was shared outside the Muslim world by over 70 percent of the Brazilians, French, Russians, and South Koreans polled.

And yet, despite some dark spots on its record, the U.S. military has done a better job of respecting noncombatant immunity in Iraq than is commonly believed. Over the past year, I have conducted dozens of interviews with commanders, judge advocates, and others who have served in Iraq; investigated operational "lessons learned" during a recent trip to Baghdad; observed the predeployment training of forces; and extensively reviewed unclassified Pentagon documents, official and unofficial histories, troops' memoirs and blogs, and human rights reports. I have found not only that U.S. compliance with noncombatant immunity in Iraq is relatively high by historical standards but also that it has been improving since the beginning of the war.

THE RULES OF THE GAME

In 1863, in response to the horrific treatment of civilians during the American Civil War, the U.S. armed forces adopted the so-called Lieber Code (named after its author, the political philosopher and jurist Francis Lieber), the first comprehensive set of regulations covering the conduct of land warfare. The Lieber Code recognized noncombatant immunity and provided the foundation for later treaties codifying the laws of war. Still, the U.S. military's commitment to fully institutionalize its norms and socialize U.S. forces remained slack for most of the twentieth century. U.S. forces engaged in extensive bombing (including two nuclear attacks) targeting civilians and civilian infrastructure during World War II. They clearly violated the principles of distinction and proportionality on numerous occasions during the Korean War and the Vietnam War. The 1968 My Lai massacre, in which U.S. soldiers killed hundreds of Vietnamese civilians, has become the most enduring symbol of these practices.

Since then, U.S. policy and practice have changed. An investigation into the underlying causes of the My Lai massacre, the Peers report (named after General William Peers, who headed the investigation), concluded that systematic shortcomings in the training of U.S. forces were partly to blame. In response to these findings, the Department of Defense adopted a directive in 1974 (reissued in 1979 and 1998) requiring that enlisted troops and officers alike receive instruction in the laws of war during basic training or in academies, at home stations and during field and mission-rehearsal exercises, and, once deployed, when being trained in the rules of engagement (ROE). The directive also dramatically elevated the role of judge advocates in advising on, monitoring, and enforcing compliance with the laws of war.

By the 1990s, it had became commonplace for U.S. military commanders and civilian leaders to emphasize their commitment to noncombatant immunity, and measures were put in place to minimize risks to civilians during the campaigns in the Gulf War and Kosovo. U.S. officials also referred to the norm during the wars in Afghanistan and Iraq. In January 2002, Secretary of Defense Donald Rumsfeld said of the U.S.-led campaign in Afghanistan, "I can't imagine there's been a conflict in history where there has been less collateral damage, less unintended consequences." More recently, the chairman of the Joint Chiefs of Staff, General Peter Pace, asserted that "no armed force in the world goes to greater effort than our armed force to protect civilians and to be very precise in the way we apply our power." Judging by three key markers -- the level of civilian casualties, the conduct of U.S. forces during operations, and the military's response to instances of noncompliance -- the actions of U.S. forces in Iraq have largely matched the rhetoric of their leaders.

CROSSFIRE

Drawing inferences from casualty data is a risky exercise. Especially in lengthy wars involving powerful armies and tenacious insurgents, large numbers of people, including civilians, are likely to be killed or wounded. This is, of course, the reason why war is hell, many wars seem morally suspect, and international laws were developed to protect civilians. Since accidents inevitably happen during conflicts and civilians often get caught in the crossfire, even full compliance with applicable norms cannot spare all noncombatants. That said, a very low (or very high) level of casualties relative to a war's scale is an indirect indicator of compliance (or noncompliance) with the laws of war.

There are numerous estimates of civilian casualties in Iraq, all of which are problematic. Some rely on questionable methodologies. Others are skewed by political agendas. All are plagued by incomplete data. The only group to systematically track Iraqi civilian deaths caused by violence since the beginning of the war and list the reported causes of death is the nonprofit organization Iraq Body Count (IBC). IBC's online database, which monitors media reports and reflects data from Iraqi morgues, suggests that 42,619 Iraqi civilians were killed by violence between March 19, 2003, and July 31, 2006. (This figure, which represents the midpoint between IBC's minimum estimate and its maximum one, is similar to that of the UN Assistance Mission for Iraq, which found, based on data from morgues in Baghdad and the Iraqi Health Ministry, that approximately 50,000 Iraqis -- including civilians, anticoalition forces, and Iraqi security forces -- had died through the end of June 2006, the vast majority from violent causes.) Some 6,434 Iraqi civilians were probably killed during the invasion; the rest have died in the "postwar" period that has followed. Of the civilian deaths recorded by IBC since President George W. Bush declared the end of major combat operations on May 1, 2003, about 2,558 appear to have been the result of U.S. attacks or crossfire between U.S. and anticoalition forces. That figure, however, is probably an underestimate, since it only includes 71 deaths from U.S. actions at checkpoints and during convoy operations (known as "escalation-of-force" incidents). The military's own accounting indicates that, on average, U.S. forces have killed about one Iraqi civilian a day in escalation-of-force events since the very beginning of the occupation. This information suggests that some of the entries in the IBC database that do not list a responsible party may actually represent deaths at the hands of U.S. troops. Taking into account the U.S. military's own estimate of deaths as a result of escalation-of-force incidents and adjusting for possible double counting, a more likely estimate of civilian deaths attributable to U.S. troops or crossfire is about 3,582. The remaining 32,603 deaths (90 percent of the total since the end of major combat) resulted from insurgent and terrorist attacks, sectarian fighting, and crime.

All sides have played a part in this unfolding tragedy. Insurgents and militias have placed civilians at risk by positioning their forces and arms caches in mosques and hospitals; using homes as shelters; firing mortars and rockets from yards, farms, and fields; and using ambulances, taxis, and other civilian vehicles to transport fighters and weapons. They have launched indiscriminate attacks that were certain to kill large numbers of noncombatants. And as the conflict has evolved, Sunni insurgents and Shiite militias have increasingly targeted civilians, triggering a spiral of sectarian violence. Civilian deaths in Iraq are also an indictment of Washington's failure to develop and execute an effective strategy for postwar contingencies and are evidence that U.S. forces have sometimes engaged in activities that put Iraqi civilians at high risk.

Still, the casualty figures for Iraq are much lower than those for many U.S. military campaigns of the last century. The number of civilian deaths during major combat in March and April 2003, including those caused by aerial bombing, was not significantly higher than it was in the 1991 Gulf War, even though U.S. objectives in the more recent conflict were far more ambitious and required extensive operations in densely populated Iraqi cities. During World War II, U.S. and British forces regularly engaged in strategic bombing against German and Japanese cities, killing more than one million noncombatants. In a single day of firebombing over Tokyo, on March 9, 1945, some 85,000 people, mostly civilians, were killed -- more than eight times the total number of civilians killed by U.S. forces or crossfire in the first three and a half years of the Iraq war. All told, the number of civilian deaths per ton of air-delivered munitions during major combat in Iraq was about 19 times lower than that in Dresden and 162 times lower than that in Tokyo. Some might argue that improvements in precision-guided munitions account for much of this difference. But many civilian fatalities caused by strategic bombing during World War II were the result of intentional efforts to destroy enemy morale rather than incidental byproducts of strikes on military targets. The real cause of the drop is a drastic change in U.S. policy.

Consider also the contrast between the Iraq war and other significant U.S. counterinsurgency efforts, such as those in the Philippines in the early twentieth century and in South Vietnam in the 1960s and 1970s. In both of those campaigns, civilians were directly targeted to some degree. Approximately 200,000 civilians (out of a total population of 7.4 million) died during the Filipino insurrection of 1899-1902, and over 520,000 civilians (out of a population of 16 million) were killed in South Vietnam. Available data for deaths in the Philippines and South Vietnam are imprecise about the cause of death, and so for the sake of consistency they must be compared with estimates for Iraq that include all violent civilian deaths, not just those arising from engagements involving U.S. forces. (Note, too, that the data for the Philippines include deaths from all war-related causes, including malnutrition and disease, not just violence.) Still, the difference is revealing. After adjusting for population size, the data suggest that 17 times as many civilians died per month in the Philippines and 9 times as many died per month in South Vietnam as have died each month during the counterinsurgency period in Iraq thus far. Even if the estimates for Iraq are off by a factor of two or three, the conflict's casualty count is far lower than that in previous U.S. counterinsurgency campaigns.

CONDUCT BECOMING

From the outset of the war, the U.S. military has put in place several mechanisms to ensure compliance with the principles of distinction and proportionality. In the lead-up to the invasion, the Pentagon developed the "joint target list," an inventory of all potential targets for coalition forces, which was vetted by judge advocates and other legal advisers. Certain operations directed against Saddam Hussein's regime were deemed off-limits because they targeted civilians or risked producing disproportionate damage to civilians and civilian infrastructure. Starting in late 2002, the Pentagon also enlisted UN agencies and nongovernmental organizations to help draw up a "no-strike" list including thousands of schools, mosques, sensitive cultural sites, hospitals, water-treatment facilities, power plants, and other elements of civilian infrastructure. The list placed significant constraints on air and land operations. During the initial ground invasion, for example, the artillery batteries used by U.S. forces were programmed with a list of sites that could not be fired on without a manual override.

The U.S. military has also tried to limit risks to civilians by reviewing its military targets with the collateral damage estimation methodology (CDEM), which uses computer software and human analysis to estimate possible civilian casualties for every target studied. The CDEM requires commanders and their legal advisers to ask themselves five questions to determine whether a given target is a legitimate one. Can they positively identify the person or the site according to the current ROE? Is there a protected civilian facility or significant environmental concern within the range of the weapon to be used? Can damage to that concern be avoided by attacking the target with a different weapon or a different method? If not, how many people are likely to be injured or killed in the attack? Must a higher commander be called for permission? When the targets considered represent a risk of "high collateral damage," the CDEM requires political approval by the secretary of defense and, during major combat, the president.

The ROE governing the use of force by U.S. troops in Iraq have also sought to balance each troop's legitimate right to self-defense with the importance of complying with the laws of war. They attempt to maintain this balance by providing troops with a clear sense of what constitutes a legitimate military target. During major combat in Iraq, the criterion was status-based. Individuals or groups, namely Iraqi military and paramilitary forces, that were "declared hostile forces" under the ROE, could be attacked on sight until they were wounded or they surrendered. As the war transitioned into a counterinsurgency mission and U.S. forces confronted adversaries who were largely indistinguishable from the civilian population, the criterion became conduct-based: U.S. troops must now positively identify a "hostile act" (such as the firing of an automatic weapon in their direction) or a "hostile intent" (such as the brandishing of a rocket-propelled grenade or the planting of an improvised explosive device) before they may fire their weapons.

Based on my interviews and the accounts of troops and embedded correspondents I have reviewed, U.S. forces in Iraq seem generally to have followed the requirement that they positively identify a hostile force, act, or intent before attacking. U.S. commanders have emphasized the need for visual confirmation of a legitimate target before firing. When the naked eye fails, U.S. forces rely on advanced optics. Forward observers are used to identify and "paint" targets, or provide coordinates for laser- or GPS-guided bomb attacks. Air force, navy, and marine fighter aircraft have also been retrofitted with new reconnaissance pods that allow real-time overhead surveillance and streaming video, thereby helping ground forces distinguish insurgents from civilians during raids and combat missions.

Furthermore, U.S. forces have been reluctant to use artillery against insurgents, even when they have been under mortar and rocket fire. Artillery systems (such as howitzers, mortars, and ground-launched rockets) have a large radius of destruction and so have a high potential for collateral damage, especially in densely populated areas. Instead, U.S. forces have either relied on close air support and ground forces that can keep their "eyes on the target" or refrained from attacking altogether. This summer, for example, as I was conducting interviews in the U.S. embassy in Baghdad, the Green Zone was hit by two rockets and two mortars. Although the attack's point of origin -- a residential neighborhood near Sadr City -- was determined quickly, the U.S. military chose not to fire back with artillery.

The U.S. ROE have also emphasized the importance of using proportionate force when engaging legitimate military targets in order to minimize collateral damage. The current rules explicitly require U.S. troops to respond to a hostile act or intent with "graduated" force. Under many circumstances, U.S. forces may engage in deadly violence only after warning their targets and trying nonlethal measures against them to no avail. The U.S. military also frequently engages in "weaponeering" -- selecting the most specifically tailored type and quantity of weapon necessary to produce a desired effect. When air strikes are required, it increasingly relies on precision-guided munitions, such as laser-, GPS-, and optically guided weapons. During the major combat phase of the war in Iraq, 68 percent of the munitions used were precision-guided, compared with 7-8 percent during the 1991 Gulf War, 30 percent in Kosovo, and 60 percent in Afghanistan. The U.S. military has also developed munitions with smaller payloads to ensure that it uses the smallest force necessary, and it relies heavily on penetrating munitions with delayed fuses to confine the damage caused by blasts and fragmentation to the area of impact. Additional mitigation techniques, including adjusting the timing, angles, and azimuths of attacks, have also been regularly employed to reduce risks to civilians.

The number of U.S. air strikes has also declined sharply since the end of major combat, indicating the U.S. military's heightened concern for proportionality. Official figures suggest that U.S. planes launched 18,695 strikes during the first 30 days of the war, compared with just 285 strikes in 2004 and 306 in 2005. Other estimates put the 2005 total at 654. Even with the higher number, however, there were almost as many air strikes per day during the major combat phase (an average of 623) as there were during all of 2005.

THE FOG OF WAR

Despite these steps, the United States' adherence to the norm of noncombatant immunity in Iraq has been incomplete. During major combat operations, the U.S. military went to great lengths to craft targeting procedures and choose weapons that reduced risks to civilians, but the execution was imperfect. When targeting high-value targets under time pressure, U.S. forces acted on intelligence that was insufficient to adequately protect civilians. For example, they relied heavily on intercepts from satellite phones to identify high-value targets of opportunity, according to a Human Rights Watch report, even though the technology used for tracking the coordinates associated with these phones was known to be inaccurate and they lacked the time and the human intelligence to either corroborate their targets' positions or estimate the likely collateral damage. In fact, not one of the 50 time-sensitive attacks U.S. forces waged against high-value targets in this time period killed the intended individual; together, however, they did kill and wound scores of Iraqi civilians. Human Rights Watch also suggests that ground-launched cluster munitions used against Iraqi artillery in residential neighborhoods in Baghdad, Hilla, Naiad, Nasiriya, and elsewhere killed or injured hundreds of civilians during the land invasion and that unexploded ordnance (or "duds") left over from these attacks killed or injured many more in the months following major combat.

After the end of the major fighting, some U.S. ground forces continued to engage in behavior that appears to have violated noncombatant immunity. Suicide attacks by pedestrians and civilian vehicles laden with explosives have continually threatened U.S. forces at checkpoints and temporary roadblocks or driving along supply routes. Sometimes, U.S. troops have fired on unarmed individuals or civilian vehicles that have gotten too close to their positions. Early on, many U.S. checkpoints and roadblocks were poorly marked, and frightened or confused Iraqi drivers often failed to stop their vehicles in time. Convoys of U.S. military vehicles typically display signs warning Iraqi drivers to stay back 50-200 meters or risk being fired on. Unfortunately, unwitting Iraqi drivers have occasionally attempted to overtake the convoys -- with deadly consequences.

Incidents have also occurred during raids and sweeps of Iraqi homes and businesses. Iraqis often complain that U.S. troops are arrogant and abusive during these operations. While searching for weapons and insurgents, U.S. soldiers have sometimes broken down doors, humiliated residents, and destroyed property. Too often, raids have been based on dubious intelligence gleaned from informants seeking to settle scores. And when soldiers and marines have encountered armed resistance from families seeking to defend their homes, they have sometimes responded with disproportionate force.

On other occasions, U.S. troops have interpreted "hostile act" and "hostile intent" in dubious ways. In some instances, merely spotting military-aged men engaged in suspicious activities or gathered in a questionable location has been enough for them to be treated as "bad guys" and "terrorists." In others, U.S. forces acting in legitimate self-defense during firefights, roadside attacks, and insurgent ambushes have responded with overwhelming force in the general direction of the attacks without taking sufficient care to positively identify their targets.

The U.S. military has also been roundly criticized, especially in Europe and the Middle East, for launching large-scale offensives against insurgents in densely populated urban areas with little regard for the inevitable civilian casualties. Nowhere has this criticism been more acute than in the case of Fallujah. In April 2004, approximately 2,500 marines assaulted the city in response to the death and mutilation of four U.S. contractors. The marines engaged in a series of ferocious close-quarters battles with scores of insurgents thoroughly mixed in with the civilian population. Reports from embedded news correspondents suggest that the marines did not intentionally target civilians during the offensive. And in my interviews, marine commanders who had overseen forces in the city claimed that their troops, unlike the insurgents, went to considerable lengths to protect noncombatants. They insisted that U.S. forces did not use artillery in the city, despite reports to the contrary, because they deemed its "area effects" to be too imprecise. Instead, the commanders relied on ground forces backed up by AC-130 gunships and 500-pound laser-guided bombs -- among the most precise air assets available to the U.S. military -- to strike hardened insurgent locations. And they called in air strikes only after lesser force had been employed without success.

Humanitarian concerns and widespread criticism forced the U.S. military to withdraw from Fallujah in late April. Almost immediately, the city became a stronghold of Sunni insurgents and a base for al Qaeda activities, eventually prompting another major U.S. attack. The second offensive, in November 2005, was massive. Between 10,000 and 15,000 troops assaulted the city, damaging or destroying 18,000 of the city's 39,000 buildings. Before attacking, however, marine and army forces surrounded Fallujah and launched an extensive information campaign urging residents to leave. Military and media estimates suggest that at least 250,000 of Fallujah's 280,000 inhabitants fled in advance of the onslaught. U.S. forces also collected much more specific intelligence about insurgent locations inside the city, relying extensively on unmanned aerial vehicles to identify potential targets. Even with improved intelligence, concern for civilian casualties was so great that senior U.S. commanders approved only a small percentage of the preplanned targets prior to the attack. Additional targets were struck only after positive sightings of insurgent fighters.

Despite the steps taken to spare civilians, over 1,000 Iraqi civilians may have been killed during the two Fallujah offensives (precise figures remain uncertain). The scale of destruction alone convinced many observers that the attacks were disproportionate, an illegitimate form of collective punishment that violated the norm of noncombatant immunity.

Many such instances of real or perceived noncompliance are at least partly attributable to the incredibly thick "fog of war" that enveloped U.S. forces as the swift invasion of Iraq turned into a frustrating counterinsurgency campaign. Even during the major combat phase of the war, the U.S. Army and the Marine Corps were confronting mostly irregular fedayeen fighters, not uniformed soldiers. With the thorough intermingling of insurgents and civilians since then, the increasing frequency of attacks on U.S. troops by individuals in civilian garb or civilian vehicles, and the insurgents' intentional use of civilians and civilian objects as shields, the landscape has become increasingly confused. To make matters worse, the U.S. military failed to adequately plan or train its troops for the stability operations and counterinsurgency missions that followed the invasion. Troops fresh from major combat and new ones deployed just as the major fighting was winding down had little practice or training in the types of nonviolent activities -- manning checkpoints, policing crowds, conducting searches -- that became so important after the invasion.

COURSE CORRECTIONS

Accidents and mistakes are bound to happen during war, and some individuals will always behave atrociously under the stress of combat. Occasional violations of the laws of war, even if egregious, do not in and of themselves constitute conclusive proof of general noncompliance. More telling than such instances of misbehavior is the U.S. military's response to them.

The U.S. military's record on punishing offenders and correcting practices is mostly promising, if somewhat checkered. It is not clear what steps, for example, the U.S. military has taken to avoid repeating the mistakes it made in hastily striking high-value targets at the beginning of the war. Although the United States seems not to have relied on satellite-phone intercepts to launch attacks during the counterinsurgency period, it has sometimes launched strikes based on dubious human intelligence, causing more civilian deaths.

The U.S. military has done a better job of trying to fix the problem of ground-launched cluster munitions. The Defense Department commissioned the Defense Science Board to study the problem of duds, and the resulting report identified a number of measures that could be taken to reduce the threat to civilians and friendly forces. The army has introduced new guidance systems to improve the accuracy of ground-launched cluster munitions and has developed a new unitary Guided Multiple Launch Rocket System that uses precision GPS-guided artillery that does not rely on dud-producing submunitions.

Progress on improving counterinsurgency practices is even clearer. First, the U.S. military has changed its official tactics, techniques, and procedures (TTPs). For example, in response to a number of incidents at checkpoints in Baghdad involving troops from the First Armored Division (which was among the units criticized in a 2003 Human Rights Watch report), the division's TTPs were promptly adjusted to reduce risks to civilian drivers. Although these practices were then adopted by many U.S. units throughout Iraq, they were applied inconsistently. Consequently, in late 2005 and early 2006, U.S. commanders began to place renewed emphasis on reducing escalation-of-force incidents. Signs at checkpoints, on U.S. vehicles, and on portable materiel have since been improved; new "dazzling" laser technologies to warn drivers are being tested; and steps have been taken to raise awareness within the Iraqi population about U.S. military practices. U.S. troops have also been instructed to pay more attention to cues that could help them distinguish actual bombers from innocent drivers. To monitor and enforce these changes, the U.S. military now requires commanders to report and justify the use of force up the chain of command every time a weapon is fired at a checkpoint or during a convoy operation. The resulting data are being tracked to further adjust TTPs.

Second, the ROE have grown much more restrictive since the beginning of the counterinsurgency period. A review of the evolution of the "shoot/don't shoot" vignettes used to train U.S. forces revealed that interpretations of what constitutes a hostile intent have been sharpened. Commenting on the current ROE, one captain from the Fourth Infantry Division told The Washington Post in February 2006, "What was allowed during the first tour in Iraq, isn't."

Third, troops' training and education have been substantially altered to emphasize counterinsurgency operations and the need to minimize harm to civilians. Cultural awareness is now emphasized, as is the importance of obtaining "actionable" intelligence. U.S. forces are encouraged to favor "cordon and knock" operations over indiscriminate sweeps. The evolution in training is especially evident at the three U.S. predeployment centers, the Joint Readiness Training Center (in Fort Polk, Louisiana), the National Training Center (in Fort Irwin, California), and the Marine Corps Air Ground Combat Center (in Twentynine Palms, California). These centers now house dozens of mock Iraqi villages populated with Arab Americans simulating civilians on the battlefield.

Together, these changes in TTPs, the ROE, and training appear to have had a positive effect. According to the military's statistics, by mid-2006, the number of Iraqi civilians killed at checkpoints, roadblocks, and alongside convoys had fallen sharply. Reports by embedded correspondents suggest that U.S. troops were behaving in a more restrained manner in 2006 than they had previously, even in insurgent strongholds such as Ramadi, where marines and soldiers have been involved in heavy combat. As one marine corporal conducting operations in Ramadi told the Los Angeles Times in July 2006, "I got shot at last night. We couldn't see where it was coming from, so we did not return fire. We can't spray 'n' pray."

Moreover, in response to criticism over the assaults on Fallujah, U.S. forces have refrained from waging additional urban offensives of such magnitude. The next-largest operation, the September 2005 assault on Tal Afar (an insurgent safe haven 40 miles from the Syrian border), was planned and executed with extraordinary care. Prior to the assault, the U.S. military used radio and television messages, loudspeaker broadcasts, posters and handbills, and airdropped leaflets to encourage residents in insurgent-heavy districts to evacuate. Iraqis who left the city were provided with prepositioned humanitarian supplies in outlying areas; those who stayed were directed to remain in their homes to avoid being mistaken for insurgents. U.S. forces then struck known insurgent safe houses and defensive positions with precision-guided munitions. Attacks were carried out with continual "eyes on the target" and timed to minimize the risk to noncombatants. The bulk of the operation relied on U.S. and Iraqi ground forces, who conducted house-by-house searches. U.S. and Iraqi forces killed or captured hundreds of insurgents, and, according to the U.S. military, only three civilians were caught in the crossfire.

Finally, beyond tactical and operational changes, the U.S. military has been taking steps to institutionalize the lessons learned from Iraq (and Afghanistan) so as to limit repeating its mistakes in the future. A November 2005 directive by the Defense Department states that stability operations "shall be given priority comparable to combat operations and be explicitly addressed and integrated across all [Department of Defense] activities including doctrine, organizations, training, education, exercises, materiel, leadership, personnel, facilities, and planning." This is truly revolutionary given the U.S. military's long-standing aversion to "military operations other than war," as the term of art goes. Other efforts are under way to adopt a new joint U.S. Army-Marine Corps counterinsurgency doctrine that is both more effective and more sensitive to the laws of war. The revised doctrine emphasizes the need for using, with great discrimination, the least amount of force possible when attacking insurgents.

FOR THE RECORD

Every war produces significant instances of misconduct, including war crimes. Such violations may be especially likely in a prolonged counterinsurgency campaign, due to the inherent stress of combat, the intense fog of war, the frustration of fighting an unseen enemy, the growing estrangement of the civilian population, and purposeful attempts by guerrillas to drive wedges between the counterinsurgent forces and the civilians these are supposed to protect. In the current conflict in Iraq, the massacre in Haditha stands out as the most serious case of misconduct allegedly committed by U.S. forces.

Still, a careful review of U.S. conduct during the Iraq war reveals no broad pattern of systematic civilian victimization by U.S. forces. U.S. compliance with noncombatant immunity in Iraq has been higher than critics often assert, and adherence has increased over time as the U.S. military has tried to correct its procedures in reaction to instances of noncompliance. Observed through the narrow lens of the laws of war, the U.S. military has gone to commendable lengths to comply with the principles of distinction and proportionality in Iraq.

But although the U.S. military has done more than is generally recognized, it should take additional steps. Under international law and Defense Department policy, it is currently obligated to thoroughly investigate and punish violations of the laws of war. As of mid-2006, the Pentagon claimed to have initiated over 600 investigations in Iraq and Afghanistan in response to allegations of misconduct, most of them relating to the abuse of detainees. But it has not opened investigations into numerous instances in which Iraqi civilians have been killed or injured at checkpoints, alongside convoys, or during combat. Of the investigations that have occurred, many appear to have been administrative inquiries meant to determine whether U.S. forces acted within the confines of their ROE -- and most have concluded that they had. Few criminal investigations have been opened, and few significant punishments have been doled out. Media reports suggest that in the first three years of the war, 16 marines and soldiers were charged with murdering Iraqis, 6 were convicted or pleaded guilty to murder, and a total of 14 were convicted on some charge in connection with the death of an Iraqi. (During the Vietnam War, in contrast, 95 soldiers and 27 marines were convicted of killing noncombatants.) Punishments for those convicted in Iraq have ranged from dismissal to life in prison. No convicted service member has received the death penalty even though murder is a capital offense under the Uniformed Code of Military Justice.

But the events at Haditha may have marked a turning point. The U.S. military initiated a criminal investigation into the alleged massacre in March 2006, and Pentagon officials familiar with the case have told the media that the marines involved will probably stand trial for murder and dereliction of duty. After the initial probe into the deaths was deemed to be highly suspect, moreover, the battalion and company commanders responsible for supervising the incriminated unit were relieved of duty, and additional disciplinary action against officers within the chain of command is possible.

Notably, reports about other significant abuses surfaced soon after Haditha. Several marines and a navy corpsman have been charged with dragging an Iraqi man from his home in Hamdaniya in April 2006, killing him in cold blood, and then attempting to cover up the crime by leaving an AK-47 and a shovel next to his body to suggest he had been planting an improvised explosive device. In another case, four soldiers from the 101st Airborne Division are accused of having released three male detainees in Salahuddin Province in May 2006 and then killing them. And in another still, five soldiers stand accused of raping an Iraqi girl and killing her and three members of her family in Mahmudiya in March 2006. The rash of reports may represent an upward trend in misconduct. More likely, however, it is evidence that the U.S. military has become more diligent about investigating and disclosing such incidents. In April 2006, Lieutenant General Peter Chiarelli, the operational commander of all U.S. forces in Iraq, ordered commanders across the country to begin investigating all incidents that result in the death or the serious wounding of an Iraqi civilian or that cause property damage in excess of $10,000.

In addition to conducting investigations, the military has taken steps to reduce the risk of war crimes in the first place. After Haditha, senior commanders in Iraq ordered refresher courses for troops to reinforce the ROE and the armed forces' "core values." In a letter to U.S. commanders, Chiarelli stated bluntly, "While I fully appreciate the stresses and pressures inherent in combat operations, I cannot, and will not accept behavior which is legally, morally, or ethically questionable." General Michael Hagee, the commandant of the Marine Corps, took the unusual step of traveling to Iraq in order to give a series of speeches to marines reemphasizing the importance of complying with the Geneva Conventions and U.S. ROE.

Significantly, however, these measures were initiated only after Time informed the U.S. military of the result of its own investigation into the Haditha deaths in February 2006 -- almost three months after the incident. The system worked reasonably well, in other words, but only once someone outside the U.S. military had pulled the alarm -- a sign that disclosure of potential violations may be hindered by camaraderie among troops.

The U.S. military should do more to overcome such obstacles to full compliance with the laws of war. For example, it should systematically conduct anonymous surveys of troops returning from Iraq and Afghanistan to ascertain why violations sometimes occur and whether a code of silence among U.S. forces masks them when they do. The military should also establish "civilian casualties response teams," comprised of judge advocates, intelligence officers, medical personnel, forensics and munitions specialists, engineers, and information officers, to study the effects of U.S. military operations on the Iraqi population. The findings should then be used to revise the rules and tactics governing the use of deadly force, inform planning and targeting procedures, and assist in criminal investigations.

More broadly, the U.S. military should further commit to tracking wartime civilian casualties and investigating all potential misconduct by U.S. forces. If nongovernmental organizations such as Human Rights Watch can conduct thorough investigations on the battlefield, so can the U.S. military. Recent efforts to keep tabs on escalation-of-force incidents are a step in the right direction, but they apply only to checkpoint and convoy operations, not to incidents that occur during patrols, raids, and combat. The U.S. military has gone to great lengths to comply with the norm of noncombatant immunity at the "front end" of operations. Now it needs an institution-wide commitment at the "back end" to better monitor and enforce adherence to the laws of war.

Tracking how U.S. operations affect Iraqi civilians is not simply a moral and legal imperative; it is vital to the United States' national interest. In Iraq's honor culture, every civilian death at the hands of U.S. forces feeds alienation and fuels the cycle of vengeance. The widespread perception that U.S. forces regularly misbehave in Iraq also helps violent jihadist groups everywhere recruit followers. Avoiding future Hadithas is thus crucial to tamping down the insurgency, winning hearts and minds, defusing sectarian strife, and stemming virulent anti-Americanism. In short, ensuring that U.S. forces comply with noncombatant immunity is not just the right thing to do for Iraqis; it is also the right thing to do for Americans.