As the fifth anniversary of the United States’ second-longest (next to Vietnam) and second-costliest (next to World War II) war passes, the good news is that the counterinsurgency strategy of Gen. David Petraeus and Lt. Gen. Raymond Odierno seems to be working. The bad news is that it will probably not save Iraq.

Although the U.S. troop “surge” has had some effect, it is probably not the most important factor dampening violence back down to the levels of mid-2004. The United States had comparable force levels in Iraq (about 155,000 troops) in 2005, but the mayhem was worse than now and was increasing.

Furthermore, the carnage in Iraq started dropping even before the United States began the surge (and temporarily increased again as U.S. troops were being added). In part, prior ethnic cleansing that had more cleanly separated hostile Shiite and Sunni populations has likely caused the reduction. Even more important was probably Petraeus’ and Odierno’s exploitation of the fissure between mainline Sunni insurgents and al-Qaeda in Iraq.

Al-Qaeda in Iraq’s blindingly incompetent slaughter of fellow Muslim civilians, which brought rebuke even by al-Qaeda’s central leadership, caused Sunni insurgents to get fed up and turn against the group. Petraeus and Odierno cleverly exploited this fissure by driving a wedge between the two factions. Although guerrilla operations are the most successful form of warfare in human history and counterinsurgency forces seldom win over the long term, they do best when they can divide the rebel movement.

The United States was able to defeat the Greek communist insurgents during the 1947-49 period and Filipino rebels from 1900 to 1902 by splitting the insurgencies. In the latter case, the United States was able to persuade Emilio Aguinaldo, the most prominent rebel commander—perhaps by a cash payment—to surrender his forces. In Iraq, the United States is now essentially paying off former Sunni guerrillas in the “Awakening Councils” by funding, equipping and training them to fight al-Qaeda in Iraq and working with the formerly hostile Shiite Mahdi militia.

Although this strategy has merits by attenuating violence in the short term, it will likely exacerbate Iraq’s larger problems, thus eventually leading to a full-blown civil war. The Petraeus and Odierno strategy makes sense if the objective is to keep a lid on the violence until President Bush leaves office. When the tar baby is successfully passed onto the next president, Bush can then rerun the “Kissinger” argument from Vietnam. That argument goes something like this: “The United States would have won the Vietnam War if the Democratic Congress hadn’t cut off funding for it.” In Iraq, the similar Bush administration refrain will be: “The situation in Iraq was improving until we left office and handed over to power to President X.”

But Bush’s short-term strategy would likely aggravate Iraq’s central underlying problem—ethno-sectarian hostility. Had the Bush administration made a serious effort to consult experts on the Arab world before invading Iraq, it would have discovered that the country was one of the most fractured in the Arab world and would be one of the least likely to support and sustain a liberal democratic federation. Prior to supporting former Sunni guerrillas, the administration was only funding, equipping and training two sides—the Kurds and Shiites—in the ongoing civil war. Now the administration is supporting all three sides. The Shiite/Kurdish-controlled government is opposed to the U.S. program to support the Sunnis and has been reluctant to let them in the security forces.

Such deep underlying ethno-sectarian suspicions and fissures have been around for centuries in what is now Iraq and are unlikely to be rectified by passing a few benchmark laws. Given the history of Iraq—in which one group controlled the central government and oppressed the other groups—all groups, even including the formerly ruling Sunnis, are suspicious of central authority and will fight for control of it. Thus, societal cooperation, of which Iraq has little, must precede legislation or the laws will be disregarded. Even less credibility will accrue to laws passed under pressure from an outside occupying power.

The only way the United States can pull its finger out of the dike without the dam crashing down is to use the threat of withdrawal—pulling the backstop out from the corrupt Shiite/Kurdish government—to get the Shiites, Sunnis and Kurds to agree to formally decentralize the country. If the central government has only limited power, the groups would fear its potential oppression less and attenuate their fight for control of it. In a decentralized, loosely confederated Iraq, their militias could provide security over members of the their own groups in new autonomous regions (the country would probably have three or more of these regions based on ethno-sectarian or tribal affiliation). Also, judicial, resource (oil) management and most other government functions could reside at the regional level. The central government would be responsible only for diplomatic representation overseas and negotiating trade agreements with other countries and among regions.

Heretofore, the major sticking point in getting the three groups to support such a decentralization scheme was Sunni worries about meager oil resources in their region. The Kurds have had a de facto state in northern Iraq since the end of the Persian Gulf War in 1991. Many Shiite leaders also favor setting up an autonomous region, the possibility of which is guaranteed in Iraq’s constitution. Even the Sunnis, finally disabused of the fantasy that they are strong enough to once again rule all of Iraq, and having tasted oppression at the hands of the Shiite-dominated security forces, are becoming more favorable to decentralization.

To push the Shiite/Kurdish-dominated Iraqi government into gerrymandering regional borders—giving territory containing oil to the Sunnis to ensure their acceptance of decentralization—any new U.S. president must establish a timetable for the rapid withdrawal of U.S. forces, which prop up that dysfunctional government.

Because the Shiite have roughly 60 percent of the oil and about 60 percent of the population, the only border that might need to be gerrymandered is near the northern oil fields by Kirkuk between Kurdistan (about 20 percent of the population and approximately 40 percent of the oil) and Sunni-dominated areas (roughly 20 percent of the population and little oil).

The historical record on partitions illuminates dos and don’ts for any soft partition of Iraq into a loose confederation—the most important of which is that the Iraqis must do the dividing themselves for it to have crucial legitimacy in their eyes. In 1947, in partitioning India and Pakistan, Britain found out the hard way that the location of the partition line is vitally important and that an outside power drawing such a border arbitrarily can have disastrous and violent consequences.

Thus, the United States should avoid getting involved in the details of creating borders between regions, but some general lessons can be learned from past partitions. First, regional boundaries don’t have to exactly mirror ethno-sectarian areas, but they should come as close as possible. The case of Northern Ireland shows that a large minority (Catholics), which could be perceived as a threat by the majority (Protestants), should not be stranded on the other side of the borderline. A small minority on the other side of the line will probably experience little violence (Protestants in Ireland). Second, the case of Kosovo demonstrates that boundaries must consider ethno-sectarian or tribal shrines and sites. Third, although drawing borders along ethno-sectarian divides should minimize population movements, some migration will likely be necessary. Such movements must be voluntary, can be encouraged through incentives and must be protected (as the violence in Indian-Pakistan in 1947 showed).

Although a U.S. withdrawal and soft partition is not a perfect solution, Iraq is in some sense already partitioned, with forces primarily loyal to ethno-sectarian groups providing security. U.S. policy training of such armed organizations is merely reinforcing this de facto partition. Such an unratified partition is very dangerous and will likely lead to a full-blown civil war. Only a new American president signaling a rapid U.S. withdrawal could motivate the parties to formalize, adjust and make permanent the decentralized Iraq that already exists.