A small number of politicians and analysts, including me, have been advocating for some time what has generally been loosely labeled the “partition” option for Iraq. Although at least one anonymous administration official has said that the Bush administration probably would end up there, and although administration policy is tending toward actions on the ground that would make such an outcome more likely, the administration has avoided publicly embracing the concept.

The problem lies, in part, with the label. Opponents of the idea link the term “partition” to the bloodbaths of Palestine and South Asia, and fear a repetition of those episodes. Yet there have also been successful partitions—for example, Czechoslovakia and the Soviet Union after the Cold War. Thus, the problem is not with the inherent concept of partition, but the way it is carried out.

In the case of Palestine, not all parties agreed to the partition, and that led to decades of violence. Palestinian Arabs, falsely promised independence if they would help Britain fight their Ottoman masters during World War I, perceived that their new British overlords gave much of their land to Zionist Jews returning to Palestine after an absence of almost two thousand years. After World War II, the world’s greatest power, the United States, recognized Israel 11 minutes after its declaration of statehood. Yet because all parties had not agreed to the partition, everyone expected—and got—war. For a partition to work in Iraq, all parties must agree to it.

The Kurds, who already have had a quasi-independent state in the north of Iraq since the first Gulf War in 1991, certainly favor a partition. Similarly, many Shi’ite leaders in southern Iraq have made noises about “autonomy,” and most of the Shi’a would probably accept such an arrangement. With both of these groups, it helps matters that they have oil on their territories. Even the Sunnis—with little oil currently on their territory (at least one geological survey indicates the presence of much untapped petroleum)—are coming around to wanting to run their own affairs. By arming Sunni militias in Anbar Province and elsewhere, who were fighting U.S. forces previously, the Bush administration is implicitly reinforcing an effective partition—with local militias providing security and services for the population in many parts of Iraq. Unfortunately, if this partition is not formally recognized and ratified by a nationwide conclave among Iraqi groups, the United States will have succeeded only in worsening the escalating civil war by arming all sides. The United States has already armed and trained the Iraqi security forces, which are made up of Kurdish and Shi’ite militia members.

The bloody South Asian partition of India and Pakistan offers a few things to avoid. The partitioning of the Indian province of Bengal, in the east, into East Pakistan (now Bangladesh) went far more smoothly than the partitioning of Punjab province, in the west, into West Pakistan (currently Pakistan). Most of the violence in South Asia was caused in the process of partitioning—that is, during the migration of 12 to 15 million people—not as a result of it. In the west, unprotected trains filled with migrants were repeatedly attacked. Furthermore, as the British had done when originally creating Iraq, the borderline between India and Pakistan was drawn on a map without regard to the reality on the ground.

In Iraq, the populations in mixed areas that have not already been ethnically cleansed are much smaller than in India. Violence in Iraq had already decreased prior to the beginning of the U.S. troop surge, primarily because much of the ethnic cleansing had already been done, according to some experts. This reality validates the common-sense notion that the separation of populations in any partition will reduce inter-group violence. To make the division work, the Iraqis themselves—not an outside power or international organization—would have to agree on very clear boundaries, based on local knowledge of ethno-religious enclaves and topographical and geographical features. Once the boundaries were established, monetary incentives could be used to encourage a region’s ethnic minorities to move to areas where their group was in the majority.

More positively, lessons can be learned from the Dayton settlement in Bosnia after its civil war in the 1990s. Instead of an outright partition, a loose confederation has been formed between a republic of Serbs and a federation of Croats and Muslims, with a weak national government allowing a veto by each group. This arrangement has established a fragile peace in Bosnia. In Iraq, with a history of the central government being used by one faction to oppress the other groups, the central government would have to be made even weaker—perhaps only policing an Iraqi free-trade area, negotiating free-trade agreements with other nations, and providing Iraqi diplomatic representation in other countries. In the case of Iraq, a confederation may have an advantage over an outright partition, because keeping around even a weak Iraqi central government might give the Turks a fig leaf to avoid intervening in Iraqi Kurdistan, to nix any desires by their own Kurdish population to secede and join the Iraqi Kurds.

In short, the division of Iraq has already happened—with militias providing local security and services in many areas—and must be formally recognized by the various Iraqi factions to avoid a full-blown civil war.