Nor is it clear that a move to partition would result in the neat division of Iraq into three smaller states, as many of its advocates seem to assume. As noted above, the Sunnis and the Shi‘a are highly divided and are likely to fight amongst themselves, leading to regular war within the communities and a probable fracturing of power in areas where they predominate. Many militia leaders, particularly the Sadrists, have made clear that they intend to fight for all of the land they believe is "theirs", which seems to include considerable land that the Sunnis consider "theirs." Baghdad is one area of contention between Sunnis and Shi‘a, but many other major cities are also home to multiple communities. Some of Iraq's oil also lies in areas that are not peopled exclusively by one group.

The partition model most observers seem to have in mind is the former Yugoslavia. There, however, years of fighting preceded the partition, clarifying the relative balance of power of the parties involved and reducing the number of ethnic enclaves through brutal "ethnic cleansing." These bloody battles helped convince many Yugoslavs that making compromises might actually be preferable to additional rounds of slaughter. These perhaps most important of all, the communities had a degree of unity and clear leaders-Slobodan Milosevic and Franjo Tudjman, for example-who could command their followers to stop the fighting. Nuri al-Maliki and other Iraqi leaders cannot issue similar orders even if they wanted to. Iraq's civil war is just not yet "ripe" for a solution like partition, and therefore to impose it upon Iraq would require a far greater military commitment by the United States than the present one-closer to the troop-to-population ratio required to police the Bosnia partition, where the conflict actually was ripe for solution when Richard Holbrooke sat down at the negotiating table in Dayton.

In the end, after years of bloodshed and ethnic cleansing, a massive civil war in Iraq may eventually create conditions for a stable partition. And the United States should be prepared for this possibility. However, a major U.S. effort to enact partition today would be likely to trigger the massacres and ethnic cleansing the United States seeks to avert.

But beyond the "don'ts", the United States must also consider other steps to minimize spillover. All of these options are difficult and carry their own sets of costs as well as benefits. For each choice, policymakers must recognize and accept the trade-offs if we are to go forward.

Some costs are relatively straightforward-but will require the United States to spend much more in aid and technical assistance to shore up allies in the region who are absorbing the brunt of spillover. This could make a considerable difference to Bahrain and Jordan. Although it is often lumped in with the other Arab Gulf states, Bahrain's standard of living cannot compare to Saudi Arabia, Kuwait or the UAE's because its hydrocarbon production is a fraction of theirs. While it does receive considerable subsidies from its fellow Gulf Cooperation Council members, Bahrain is still the poor relation of the Gulf, and the country is already feeling the heat from radicalization of its majority Shi‘a population from Iraq. Bahrain is also particularly vulnerable to anti-Americanism because it has been a reliable U.S. ally and hosts the headquarters of the U.S. Fifth Fleet. Jordan is a small, poor country already overburdened by its long-standing Palestinian refugee population, and trying to absorb hundreds of thousands of Iraqi refugees as well could be the straw that breaks the back of the Hashemite monarchy. Large numbers of additional refugees at the very least will strain Jordan's already weak economy. They may also bring the Iraq War with them, increasing violence in Jordan itself and perhaps polarizing the population. Economic assistance to both countries could help dampen internal problems there derived from, or exacerbated by, all-out civil war in Iraq. In addition, both need help in policing refugee camps and ensuring a robust counter-terrorism capacity.

The Refugee Challenge

INDEED, ONE of the most pressing issues is dealing with the refugee question-not only because of its negative impact on stability within Iraq but also the dangers posed to neighboring states. Because of our moral responsibility for the suffering in Iraq, many will want the United States to try to "do something." Strategic necessity should reinforce our moral obligations.

One approach that will inevitably be suggested will be to create safe havens in Iraq's cities. If the surge fails, however, this would be a mistake. The various United Nations forays into Bosnia in the 1990s should remind us of how difficult such a strategy would be and how easily it could turn into a disaster. As the tragic experience of Bosnia demonstrates, Iraqi cities would require huge numbers of troops to keep them safe. In fact, this was the principle behind the first Baghdad security plan, which kicked off in the summer of 2006. That plan sought to increase security in the capital as the first step toward a gradual strategy of stabilizing the country and enabling reconstruction. It eventually failed because Washington did not provide adequate numbers of American and properly trained Iraqi troops (as well as the political and economic support to lock in the security gains) to make the capital safe. Violence in Iraq's population centers cannot be controlled on the cheap and would require substantial commitments of both men and material, as the latest (and much larger) Baghdad security plan is already demonstrating.

At the very least, the United States should provide technical assistance to Jordan, Saudi Arabia and Kuwait to help them ensure that refugee camps do not become insurgent operating bases. Whenever possible, camps should be set up far from the borders. Policing is essential. In her study of refugee-linked conflicts around the world, Sarah Lischer contends that host governments must aggressively ensure that warlords do not run the camps and refugees are disarmed.³ For many regional states, however, management skills and military capacity are weak. U.S. aid can help bolster this.

Another option might be to create a system of buffer zones coupled with large refugee-collection points or safe havens ("catch basins") on Iraq's periphery. The safe havens themselves would be located on major roads, preferably near airstrips along Iraq's borders-the principal routes that refugees would take to flee the country-with a good logistical infrastructure in place to house, feed and protect tens or even hundreds of thousands of refugees. The idea would be to create positive incentives to persuade Iraqis not to leave their country by giving them a better alternative. This would help reduce the problem of refugees radicalizing neighboring states. The catch basins could also mitigate other important spillover effects by hindering or preventing attacks back into Iraq by militia groups, inhibiting insurgents from crossing both to and from Iraq, as well as simultaneously deterring (or preventing) the armed forces of most of Iraq's neighbors from intervening in an Iraqi civil war. The catch basins would also serve as military launch points to collect intelligence against various terrorist groups inside Iraq (particularly Al-Qaeda) and mount strikes against them whenever possible or desirable.

Nevertheless, such potential benefits would likely come at a high price and would also entail significant risks. Disarming refugees and pacifying refugee camps is a politically and logistically onerous task that would require the continued presence of large numbers of U.S. troops in the country for many years. Otherwise, these catch basins could easily become militia bases or be subjected to reprisal attacks. In addition, catch basins could not be set up along the Iranian border for political and logistical reasons. Thus, alone they could not solve this part of the problem and could even be construed as turning over Iraq to Iranian influence by preventing neighboring states from intervening to counter Tehran. Finally, the catch basins could be a public relations nightmare for the United States, as U.S. forces would be perceived as standing by impassively while the slaughter in Iraq continues only kilometers away.

A final option would be to resettle refugees from Iraq outside the region, including in the United States. This could greatly reduce the strain on Jordan, Syria, Lebanon and other regional states. But neither Europeans nor Americans are eager to embrace Iraqi refugees, whose fate so far has stirred little compassion in either area. But if the situation deteriorates further, many Iraqis, like many South Vietnamese, compromised by their close association with the U.S. administration in Iraq, will need to be extracted.

No matter what course of action the United States chooses, there will be costs.

Managing Additional Options to Spillover

NO MATTER what happens in Iraq, an overriding U.S. national interest will be to limit the ability of terrorists to use Iraq as a haven for attacks outside the country, especially directed against the United States. The best way to do that will be to retain assets (airpower, special operations forces and a major intelligence and reconnaissance effort) in the vicinity to identify and strike major terrorist facilities like training camps, bomb factories and arms caches before they can pose a danger to other countries. Washington would need to continue to make intelligence collection in Iraq a high priority, and whenever such a facility was identified, whether Shi‘a or Sunni, American forces would move in quickly to destroy it. When possible, the United States would work with various factions in Iraq that share our goals regarding the local terrorist presence. These same factions, however, would want U.S. money and support for their own political agendas, and many of them would be involved in brutal actions of their own.