An Iraqi whom I will call Ahmed lives in Saidiya, an area in south Baghdad where, in the nineteen-eighties, the regime of Saddam Hussein built large houses for well-connected Army officers, most of them Sunnis. After the American invasion, in 2003, Saidiya became a base of Sunni resistance, but since last year vicious sectarian fighting has divided its streets between Sunni and Shia, with front lines crisscrossing the district; the highway separating Saidiya from the Shiite area of Bayya, to the northwest, now marks an impassable boundary. "It's just like the Great Wall of China," Ahmed said, during a recent phone conversation. A graduate of Baghdad University, with a degree in English literature, he worked before the war as a news translator for Iraqi state television.

Saidiya has one of the highest rates of sectarian killings in the city. Eighty-four unidentified corpses were found there between mid-June and mid-July, according to Zeyad Kasim, a researcher at IraqSlogger.com, a news-gathering Web site. Ahmed said that the number actually represents an improvement—earlier this year, he saw bodies lying in the streets even more frequently. The U.S. military "surge" launched this spring, in which thirty thousand additional American forces arrived in Baghdad, has helped to stabilize Saidiya's sectarian borders. The Americans don't often patrol Ahmed's neighborhood, but, when Iraqi Army forces call in air support, Apache attack helicopters can reach Saidiya within minutes.

Ahmed has little faith in the Iraqi Army itself. He said that the soldiers behave unprofessionally, don't respect the chain of command, and seem more concerned with their salaries than with their responsibilities. "Ninety per cent of my neighborhood think the Iraqi Army is hopeless," he said.

There is no functioning government in Saidiya. The power supply has dropped to less than two hours a day, and for a month Ahmed—a thirty-seven-year-old father of two who suffers from diabetes and a heart condition—could obtain water only from a hole that he dug in his back yard. His neighborhood is under the control of a Shiite militia claiming allegiance to Moqtada al-Sadr, the radical scion of a powerful clerical family, who has emerged as perhaps the most important political figure in Iraq. The militia employs the crippled and the poor to collect protection money, controls a black market for fuel, and forcibly recruits young men into its ranks as lookouts against the Americans. Its local "security" force consists of teen-agers brandishing AK-47s.

Ahmed, who has a Shiite father and a Sunni mother, considers himself a secular Shiite, and, in his view, the religious militias want to force people like him out of Baghdad. "Americans are the safe house for the whole situation in Iraq," he said. "Once they say they are going to withdraw, the whole country will become a hell." He went on, "I imagine that no Sunnis will be in Baghdad at all. Baghdad will be only for the Shiite man with the long beard and black imama—the turban. The Americans are representing the taboos, just like 'Lord of the Flies.' I imagine the Shiites will be just like that if the Americans have to withdraw. Who can fight will fight, who must leave will leave." He added, "Those who are weak, who are trying to avoid the savagery, those who are at the edge of being eaten by the Shiite specifically—that will be the end point, that will be their doomsday. The plan, as we hear it, is to make Baghdad empty of Sunnis."

This week, Ryan Crocker, the U.S. Ambassador in Baghdad, and General David Petraeus, the commander of the multinational forces in Iraq, will give their assessment of the surge to Congress—an event that, in Washington, has taken on the aura of a make-or-break moment for the Administration's policy. But their testimony is likely to be unremarkable. Administration officials, military officers, and members of Congress described their expectations of it in strikingly similar terms, and a few said that they could write it in advance: military progress, a political stalemate among Iraqis, more time needed.

The Petraeus-Crocker testimony is the kind of short-lived event on which the Administration has relied to shore up support for the war: the "Mission Accomplished" declaration, the deaths of Uday and Qusay Hussein, Saddam's capture, the transfer of sovereignty, the three rounds of voting, the Plan for Victory, the death of Abu Musab al-Zarqawi. Every new milestone, however illusory, allows the Administration to avoid thinking ahead, to the years when the mistakes of Iraq will continue to haunt the U.S.

The media have largely followed the Administration's myopic approach to the war, and there is likely to be intense coverage of the congressional testimony. But the inadequacy of the surge is already clear, if one honestly assesses the daily lives of Iraqis. Though the streets of Baghdad are marginally less lethal than they were during 2006, sixty thousand Iraqis a month continue to leave their homes, according to the International Organization for Migration, joining the two million who have become refugees and the two million others displaced inside Iraq. The militias, which have become less conspicuous as they wait out the surge, are nevertheless growing in strength, as they extend their control over neighborhoods like Ahmed's. In the backstreets, the local markets, the university classrooms, and other realms beyond the reach of American observers or American troops, there is no rule of law, only the rule of the gun. The lives of most Iraqis are dominated by a complex array of militias and criminal gangs that are ruthlessly competing with one another, and whose motives for killing are more often economic or personal than religious or ideological. A recent report by the International Crisis Group urged the American and British governments to acknowledge that their "so-called Iraqi partners, far from building a new state, are tirelessly working to tear it down."

After the string of bad decisions made by American leaders in the early years of the occupation, officials in Baghdad have made various technical corrections: training the new Iraqi Army in a more professional way, funding reconstruction projects that show faster results, and applying the methods of counter-insurgency to the war. But these improved approaches came much too late, and didn't quell the profound sectarian hatreds that emerged after Saddam's removal. A former Baghdad Embassy official told me, "If Iraqi leaders, in their own heart of hearts, don't share a vision, there's just not much you can do about it. I don't think accommodation is going to happen."

This political failure can't be attributed to the Iraqis alone. Iraq's leadership was originally installed, along sectarian lines, by the American-led Coalition Provisional Authority, and the chaos that followed the invasion drove Iraqis to seek safety in armed groups based on identity. The inability of Iraq's communities to reconcile doesn't absolve the United States of responsibility. Instead, it raises a new set of moral and strategic questions that are, in their way, more painful than at any other phase of the war. Facing these questions requires American leaders to do what they have not yet done—to look beyond the next three or six months, to the next two or three years. When America prepares, inevitably, to leave, what can we do to limit the damage that will follow our departure, not just for Iraq's sake but for our own?

White House officials are determined to present the surge as a dramatic turn in the war—as if the war could still be won. In interviews with me, they blamed the public's dissatisfaction on the Democrats, for "playing politics" with the war; on journalists, for being impervious to good news; and on the public, for having a short attention span. Peter Wehner, a former adviser to President George W. Bush who left the White House last month, acknowledged that the Administration had made many mistakes in Iraq. But he insisted that victory was still possible. Bush, he said, "has the stiffest spine in the Administration," and he described Petraeus as a man who could enter the military pantheon next to Grant, if only the American people would give him the chance. "What happens if, at the eleventh hour, we're witnessing one of the most remarkable feats in American history on the part of a general?" he said. "If that's the case, why do you want to give up now?"

Bush will likely use Petraeus's testimony, and his military prestige, to claim authority for sustaining the largest possible American presence in Iraq through the end of his Presidency. But how large could that presence realistically be? Currently, there are a hundred and sixty thousand troops in Iraq. The natural life of the surge will end in 2008, when the brigades sent earlier this year will finish their fifteen-month tours and return home. After that, it will become virtually impossible to maintain current troop levels—at least, for an Administration that has shown no willingness to disturb the lives of large numbers of Americans in order to wage the war. Young officers are leaving the Army at alarming rates, and, if the deployments of troops who have already served two or three tours are extended from fifteen to eighteen months, the Pentagon fears that the ensuing attrition might wreck the Army for a generation. Activating the National Guard or the reserves for longer periods could cause the bottom to fall out of public support for the war. Beyond these measures, there are simply no more troops available.

Several years ago, at the beginning of the insurgency, then Secretary of Defense Donald Rumsfeld decided that additional forces were not needed in Iraq, and he refused calls for a long-term expansion of the Army and the Marine Corps. With Rumsfeld gone and command in Iraq given to Petraeus—an innovative military thinker who oversaw the writing of the new counter-insurgency field manual—the Administration at last has a real strategy; but, having wasted four years, it now lacks the forces needed to sustain the gains made by the surge. According to a Pentagon consultant, a strategic-planning cell of colonels serving under the Army chief of staff, General George Casey, has estimated that the number of soldiers and marines who can be kept in Iraq into 2009 will be, at maximum, a hundred and thirty thousand. On Labor Day, at an airbase in western Iraq, Bush said, "If the kind of success we are now seeing here continues, it will be possible to maintain the same level of security with fewer American forces." What he didn't say was that some withdrawals will almost certainly have to come next year, regardless of the military's "success."

In the view of most Democrats, the inevitability of reduced troop numbers, the political stalemate in Baghdad, and the dwindling of public support in America require that a withdrawal begin soon. All the Democratic candidates for President have declared that they will end the war. And many Republicans in Congress have embraced the report issued last December by the Iraq Study Group, which called for regional diplomacy and for American combat units to be withdrawn; residual troops, likely numbering fifty to sixty thousand, would be limited to training and advising, counterterrorism, and force protection. The report did not impose a timeline, but others have tried to do so. In the Senate, two Democrats—Jack Reed, of Rhode Island, and Carl Levin, of Michigan—have introduced an amendment to a defense-appropriations bill which would require troop withdrawals to begin within four months of the bill's passage, leaving behind only a "limited presence." Among its co-sponsors are three Republican senators and three Democratic Presidential candidates: Hillary Clinton, Barack Obama, and Joseph Biden. In August, Senator John Warner, the Virginia Republican, called for some troop withdrawals to begin before Christmas.

The Democratic congressional leadership has refused to bring up for a vote bipartisan compromise proposals that lack a timeline for withdrawal. Senator Lamar Alexander, a Tennessee Republican, co-sponsored one such bill, which is based on the Iraq Study Group report; he said that, if the Democratic leadership allowed his proposal to come to the floor, it could get enough votes to compel the President's support. "You have the President being inflexible and the Democrats playing politics," Alexander said. If the legislative deadlock continues until the end of Bush's Presidency, the White House, by default, will get what it wants.

In Washington, the debate over the war is dominated by questions about troop numbers and timelines—that is, by immediate American political realities. The country seems trapped in an eternal present, paralyzed by its past mistakes. There is little or no discussion, on either side, of what America's Iraq policy should be during the next five or ten years, or of what will be possible as resources dwindle and priorities shift. If there is any contingency planning in the government, it's being done at such a secretive, or obscure, level that a repetition of the institutional disarray with which America entered Iraq seems bound to mark our departure.

Preparing a judicious withdrawal from Iraq will demand the integrated effort of the whole government, not just under this President but under the next one as well. "You just cannot pretend that the Iraq war never happened and everything can go back to how it was before," the former Embassy official told me. "The status quo before 2003 no longer exists. We have introduced fundamental new disequilibriums into one of the most sensitive parts of the globe. How do you contain it?" He added, "People have to start thinking about these things—small study groups with military, State, and intelligence people sketching out what are the core interests on a regional level, and working back from that to discuss some options. If that's been done, I don't know about it."

In June, a new center-left think tank, the Center for a New American Security, issued a report called "Phased Transition." It envisions a gradual shift from the current strategy of taking the lead in counter-insurgency operations to a "support" role in Iraq. The report goes into more detail about this transition than the Iraq Study Group did, and it proposes a timeline. Troop levels would be reduced to sixty thousand by January, 2009, with a third of those remaining involved in a greatly expanded advisory effort—U.S. soldiers would embed with Iraqi forces, help them plan and conduct operations, and act as intelligence sources, identifying capable, corrupt, and sectarian Iraqi leaders. This process would continue for several years before a complete withdrawal would begin, around 2012. At the moment, there are only some six thousand Americans training and advising Iraqis; after four years of war, the Pentagon has not committed the resources to broaden this effort, even though counter-insurgency warfare depends on the emergence of a capable indigenous Army. (A military spokesman disputed this, saying, "The Pentagon is going above and beyond the requirements it has been tasked to fill.")

Colin Kahl, a professor of security studies at Georgetown, contributed to the "Phased Transition" report. He warned that, in order for the plan to work, preparations would have to begin very soon. Advisory teams take considerable time to build. Officers and senior sergeants, culled from the limited number still available, would have to be assembled and trained months in advance of deployment; and skimming the cream of the Army and the Marine Corps for an advisory effort would reduce the efficacy of combat brigades. In other words, Kahl argued, President Bush needs to be forced to compromise now, or else the war will end in a precipitate, chaotic flight. Kahl told me, "If Bush keeps the pedal on the surge until the end of his Presidency, we will rocket off the cliff, and it guarantees that the next President will get elected on a pledge to get us out of Iraq now."

American officials, including Bush, talk a lot these days about decentralizing the military's efforts in Iraq—about building up local government and security forces, and pushing resources out of Baghdad. In the absence of a functioning national government, it's the only way to begin stabilizing parts of the country. America has achieved some successes in the past few months—alliances have been made with tribal and municipal leaders in Anbar and other Sunni areas—but these are only temporary fixes. And, without a functioning state in Iraq, U.S. support of these Sunni forces could easily lead to renewed violence and warlordism. It isn't clear that the tribal leaders behind the so-called "Sunni awakening," which emerged in opposition to the brutality of Al Qaeda in Iraq, see their present allegiances as anything other than tactical steps toward an eventual showdown with the ultimate enemy: the Shiite-led government, and its Iranian backers. Recently, there has been talk among American officers of applying a similar formula in the south of Iraq, by allying with Shiite tribes that have begun to resist the extremist religious militias. But supporting sectarian factions without fuelling an even larger conflagration would require a deftness that the American military has seldom displayed thus far. "This could go wrong in seven thousand ways," Kahl said of the new policy. "It's the 'sorcerer's apprentice' effect—we unleash forces we can't control."

Some leaders in Washington, such as Senator Biden, believe that the divisions between Iraq's ethnic groups are so intractable that the only solution is to divide Iraq into three autonomous regions: Shiite, Sunni, and Kurdish. But the idea of partition can't be imposed by outsiders and, so far, has no support from Iraqis. The past few years have made Iraqis deathly afraid of one another, but, apart from the Kurds, the war has not made most of them want to stop being Iraqis. A recent poll by academic researchers in Michigan found that the percentage of Baghdad residents identifying themselves as "Iraqis above all" more than doubled, across all groups, between 2004 and 2006. Civil war and sectarian rule have tarnished the prestige of religious parties and increased the appeal of a non-sectarian government. In August, the Shiite governors of two provinces in the south—a region that is almost entirely Shiite—were murdered, presumably by rival Shiite factions. This suggests that a partitioned Iraq would not be a peaceful or stable Iraq. Smaller civil wars would keep igniting, and the mixed cities would remain bloody, even if a way could be found to separate their populations. Nor is partition a formula for a quick American exit: to make it work, large numbers of troops would have to remain in Iraq, dividing and protecting the various factions and, in some cases, transporting people across sectarian lines. This, too, could go wrong in seven thousand ways. American troops could end up in a role similar to that of peacekeepers during the Bosnian civil war, supposedly guaranteeing safe passage to unarmed civilians who are targeted by their enemies for slaughter.

It's easy to fall under the illusion that a perfectly framed ten-point proposal could allow for a painless withdrawal. But what if there is no such thing as a "responsible exit" from Iraq? This is the view of Stephen Biddle, a fellow at the Council on Foreign Relations, who spent the spring in Iraq, as part of a strategic-assessment team of military and civilian experts. He said, "When you look at the spectrum of policy approaches in Iraq right now, the extremes"—maintaining the largest force possible or pulling out immediately—"make more sense than the middle." The "middle-ground policies," he argued, "tend to dramatically reduce our ability to control the environment militarily, because they all involve withdrawing about half of the troops. It's our combat activity that's currently capping violence around the country, and almost everybody would cut that out—which means the violence is only going to increase. And yet they leave tens of thousands of Americans in the country, to act as targets. Continued U.S. casualties, continued deterioration of the situation all around them: within two or three years, that's going to generate powerful pressure to go all the way to the zero option. Why not do it sooner, and save the seven to eight hundred lives you're going to lose to walk through this drill in the meantime?"

A military officer with extensive experience in Iraq was less polite. "I just think it's dishonest when people say we could go to advisory, get to fifty thousand troops, focus on training, still do the counterterrorism thing but not counter-insurgency," he said. The reality of Iraq is bound to defeat the fantasies of Washington, the officer suggested. "What about the enemy, man?" he said. "Are we going to ask them to conform to our plan?"

David Kilcullen, an Australian counter-insurgency adviser who served on Petraeus's staff in the first half of the year, said, "The real question is not withdrawal dates or troop numbers. The real question is: What do we want Iraq to look like once we don't have a hundred and sixty thousand troops there? And is what we want achievable?" This spring, Kilcullen also served on the strategic-assessment team, which was led by Colonel H. R. McMaster, of the U.S. Army; David Pearce, of the State Department; and Colonel James Richardson, of the British Army. The assessment implicitly contradicted the Administration's rhetoric by declaring the violence in Iraq to be driven by a communal power struggle in which the Iraqi government was one of the main actors. The goal that it set for the next two years was not a democratic Iraq but "sustainable security," with Iraqis in the lead. One participant told me that a majority of the team believed that it was too late to achieve this goal.

In Kilcullen's view, allowing the surge to run its course into next spring, while doing as much damage as possible to Al Qaeda in Iraq in the meantime, would make it likelier that a gradual withdrawal of troops would not leave behind the chaos of previous drawdowns—from Falluja and Mosul in 2004, from Tal Afar and Baquba in 2005, and from Baghdad in 2006. He said, "The longer you stay there doing police and counter-intelligence work, the more long-term stability there is once you leave." He compared the surge to a course of antibiotics: "You keep taking it as long as possible, even after the symptoms are gone, to kill the underlying infection."

While serving on the assessment team, Kilcullen drew up a list of core American interests in Iraq, which he later gave to senior officials at the White House and the State Department. In order of priority, the list contained the following items: maintain the flow of oil and gas in the region; prevent the establishment of an Al Qaeda safe haven in Iraq; contain Iranian influence; prevent a regional war; prevent a humanitarian catastrophe on the scale of Rwanda; and restore American credibility in the region and in the world (which Kilcullen called "the master interest," and which doing all the others would go a long way toward achieving). Some interests, he acknowledged to me, might be incompatible: for example, undermining both Sunni-led Al Qaeda and Shiite Iran.

Most proposals for withdrawal emphasize at least three of the interests on Kilcullen's list. One is counterterrorism. The Center for American Progress, a liberal think tank, released a plan in June that calls for ending support for Iraq's national-level security forces and removing all U.S. troops by the end of next year. It sets firm deadlines and abandons the effort to create a stronger central government and Army. This plan claims that it will actually be easier to fight Al Qaeda if American forces aren't pinned down in Iraq—a recruiting poster for jihadists around the world. "Redeploying U.S. troops would make Iraq a quagmire for our terrorist enemies and rivals in the region," the authors of the plan confidently assert. Pakistan, not Iraq, will likely remain the headquarters of international jihad, but the idea that a fragmented Iraq will be a hard place for terrorists to thrive willfully ignores the recent examples of Somalia, Afghanistan, Lebanon, and Pakistan. And, without large numbers of American troops in Iraq, counterterrorism efforts will be limited to U.S. electronic surveillance and intelligence from Iraqi sources, with operations carried out by the Iraqis, whose interests and agendas constantly shift.

A second strategic interest is preventing a regional war. Every plan for withdrawal puts a strong emphasis on diplomacy in the Middle East. Talking to other countries in the region, including America's enemies, is a lot more appealing than being targeted by them inside Iraq. Ambassador Crocker met his Iranian counterpart in Baghdad several times over the summer, but the talks thus far have been fruitless, with the Americans demanding an end to Iranian subversion in Iraq and the Iranians denying the charges. Unless the talks take place at the level of the Secretary of State, and broach the full range of antagonisms between the U.S. and Iran—nuclear weapons, regional security, Lebanon, Afghanistan—they stand little chance of succeeding.

The notion that Iraq and the Middle East will be more stable without an American occupation, as the Center for American Progress claims, misunderstands the role that America has come to play in Iraq: as a brake on the violent forces let loose by the war. America's diplomatic leverage will be weakened by a withdrawal, and Iraq's predatory neighbors will take advantage of the power vacuum to pursue their own interests. Even if regional interference doesn't take the form of Saudi troops crossing the border to defend their Sunni brothers, Iranian Revolutionary Guards infiltrating Iraq to secure Shiite power, and Turkish forces entering Kurdistan to prevent it from becoming independent, the combined effect of proxy fights, irregular incursions, and increased refugee flows will likely roil the Middle East for years. Stephen Biddle, of the Council on Foreign Relations, told me, "The likelihood is that it doesn't become a regional war, but there's a roughly thirty-to-forty-per-cent chance that it'll spread. During the Cold War, we spent trillions worrying about infinitesimally small risks. Thirty-to-forty-per-cent chance of a real, honest-to-goodness catastrophe is something that ought to factor into our policymaking now."

Zalmay Khalilzad, the Ambassador to the United Nations, who spent almost two years as the U.S. Ambassador in Baghdad, trying in vain to bring about a political compact, sketched the possible fallout in stark terms. Without Americans present, he asked, "could it intensify into a terrible situation in which you get massacres, and that not only leads to escalation in Iraq but affects others? The losing side may ask for help from its brethren next door. We cannot stand aside and let these terrible things go on."

Then there is the prospect of mass slaughter. Bush recently raised the spectre of genocide, in a speech suggesting that a withdrawal from Iraq could lead to killing on the level of Cambodia after the Vietnam War. Many Democrats were skeptical. On the campaign trail, Barack Obama said, "If that's the criteria by which we are making decisions on the deployment of U.S. forces, then by that argument you would have three hundred thousand troops in the Congo right now—where millions have been slaughtered as a consequence of ethnic strife—which we haven't done. We would be deploying unilaterally and occupying the Sudan, which we haven't done." The argument is shallow: by Obama's reasoning, America doesn't have an obligation to prevent large-scale massacres in a country it invaded and occupied, but it does have an obligation not to be hypocritical about it.

"Genocide," defined as an attempt to destroy a whole people on the basis of identity, has taken on the quality of a sacred category. It's more useful to talk about slaughter, which doesn't prompt the same amount of posturing. "Iraq's violence is already quite deadly," the Center for American Progress report says. "No single force will be able to truly gain an upper hand in the country." The first proposition is true, the second arguably true. Together, they amount to a neat self-exculpation in the wake of an American departure: violence is already widespread, and it can't get that much worse. Given the examples of Falluja and Baghdad—not to mention the unfortunate fates of Yazidis, Christians, Mandeans, and Gypsies in villages that America never occupied—the burden of proof lies on anyone who claims that Iraqis without Americans around won't be substantially worse off, and might even fare better. Even Iraqis who want American troops out immediately acknowledge that the result will be more bloodshed. If America decides to leave Iraqis to their fate, they should at least be spared the parting thought that it's for their own good.

Predictions that a departure will actually strengthen America's position are not so different from the wishful thinking of the Bush Administration. They see only the benefits of withdrawal; they refuse to face the brutal trade-offs that either staying or leaving would impose. A more honest argument says that it's simply not a core American interest to prevent Iraqis from being massacred: the result of a withdrawal may be a humanitarian tragedy but a strategic footnote. (Obama's statement implied as much.) This viewpoint has recently brought together hard-nosed realists, antiwar progressives, and isolationist conservatives. Even in narrow strategic terms, though, American interests would be harmed by large-scale slaughter in Iraq. The spectacle, televised around the world, would deepen the feeling that America is indifferent to human, especially Muslim, life. It would brand the U.S. as untrustworthy to potential allies and feckless to potential enemies. And it would destroy what's left of American prestige. Toby Dodge, an Iraq expert at Queen Mary College of the University of London, who also served on the strategic-assessment team, told me, "What has defeated America in Iraq, apart from the failure of the state and its own incompetence, are a bunch of radicals with nothing more sophisticated than reëngineered artillery shells and rocket-propelled grenades. That is a loss of cataclysmic proportions."

Dodge comes out of the British left and vehemently opposed the war. But this summer, when we met at his London office, he spoke of withdrawal as a prelude to catastrophe. "What are the U.S. troops going to leave?" he said. "They're going to leave behind a free-for-all where everyone will be fighting everyone else—a civil war that no one actor or organization will be strong enough to win. So that war will go on and on. What will result in the end is the solidification of pockets of geographical coherence. So if you and I were mad enough to jump in a car in Basra—pick a date, 2015—and we tried to drive to Mosul, what we'd be doing is hopping through islands of comparative stability dominated by warlords who, through their own organizational brilliance, or more likely through external support, have managed to set up fiefdoms. Those fiefdoms will be surrounded by ongoing violence and chaos. That looks a lot to me like Afghanistan before the rise of the Taliban. Or Somalia. That's where Iraq goes when Americans pull out."

Dodge's grim vision does not make an irrefutable case for staying in Iraq. But it's a reminder that the illusions and naïve hopes with which America started the war shouldn't accompany its end.

The dream of creating a democratic Iraq and transforming the Middle East lies in ruins. Any change in Iraq policy has to begin with the understanding that the original one failed, and that America's remaining power can only be used to limit the damage. But Iraq still matters to the United States, whoever is in the White House, and it will for years to come.

One way in which Iraq and Vietnam—two wars doomed to be endlessly compared—are not the same is in the implications of America's departure. Contrary to Bush's recent claim, the American exit from Vietnam didn't lead to the Cambodian genocide (U.S. actions during the war did), and, for all the bloodiness of the aftermath in Vietnam, it was not a strategic disaster. America's prestige was damaged, but the dominoes did not fall, and the civil wars in Southeast Asia did not affect the larger history of the Cold War. But Iraq, sitting in the geographical heart of the Middle East, on top of all that oil and radicalism, is unlikely to become marginal. In 1966, Senator George Aiken gave Lyndon Johnson some memorable advice about what to do in Vietnam: Declare victory and get out. In contemplating a change in American policy on Iraq, one former Bush Administration official turned the advice around: "Declare defeat and stay in."

This doesn't mean keeping large numbers of troops in Iraq indefinitely; that has become impossible. David Kilcullen argued that next summer, when the surge is scheduled to end, American forces could be reduced to a level—say, eighty thousand—that might allow most of the core interests to be protected. Such a move would involve difficult calculations: as American commanders pull back from more stable areas—starting in the northwest, the west, and the south, where there are fewer sectarian divisions—they will risk a return to higher levels of violence. On the phone from Baghdad, General Petraeus said, "There's an issue of what you might call 'battlefield geometry.' Where do you thin out and how do you do it? It's not as simple as 'Put in five brigades, one each month, take out five brigades, one each month.' You might want to thin out in one place and not another. As you do that, you do want to modify your mission." He added that "you may still be emphasizing protecting the population in one area," while in more secure areas American forces might take on a role of supporting and advising Iraqi Army units. The changes in mission will come sector by sector and incrementally, with commanders hoping that today's local ceasefire or the formation of a friendly Sunni militia in one town somehow holds and leads to long-term stability.

But, when the surge ends, there will have to be a strategic turn, away from Americans in the lead. An indefinite war in Iraq "costs us moral authority across the world," Kilcullen said. The occupation of Iraq remains hugely unpopular with America's democratic allies and throughout the Arab and Muslim world. "We need that moral authority as ammunition in the fight against Al Qaeda," he added. "If we're not down to fifty thousand troops in three to five years, we've lost the war on terror."

At the same time, America will need a political strategy that consists of more than simply reiterating support for the Iraqi government or maneuvering to replace it with another. The political deadlock in Baghdad, which has brought the government of Nuri al-Maliki close to collapse, is often seen by Americans as a perverse failure on the part of Iraqi leaders to do the right thing. Senator Levin, explaining why his amendment insisted on a timetable, said, "As long as the Iraqi leaders believe that their future is in our hands instead of theirs, they will continue to dawdle while their country is torn by bloodshed." This view fails to acknowledge the American role in empowering sectarian leaders, and it misunderstands the sources of Iraq's divisions, which run deeper than individual stubbornness, to a fundamental clash of visions and interests.

Toby Dodge admitted that anyone arguing against immediate withdrawal has to face the "killer question: Why should American troops continue to die when the chances for success are so low?" He offered his answer "with an honest recognition that it doesn't sound very plausible." Dodge's approach would bring the maximum pressure to bear on Iraqi politicians by persuading the region and the world—Iraq's neighbors, the European Union, the United Nations—to come into the Green Zone, not as tools of American policy but as equal partners in an effort to force a political deal, not unlike the U.N.'s role in creating a government in Afghanistan after the fall of the Taliban. This would imply an American confession of failure. Instead of pursuing more ambitious goals for democracy in the region, the U.S. would offer security guarantees to Iran and Syria in exchange for coöperation. "We then turn to the Iraqi government," Dodge went on, "and say, 'You've got to reform your government, make it more inclusive, less corrupt, more coherent, less sectarian.' So the Iraqi government is reconstituted within a multilateral framework where the E.U., the U.N., and the U.S. are all singing from the same hymnbook."

Ambassador Khalilzad recently set this proposal in motion when he introduced a resolution calling for the U.N. to return to Iraq. He told his colleagues that Iraq will be the world's problem for a long time. Khalilzad was part of the neoconservative group, led by Paul Wolfowitz, but after years of working on Iraq he sounds chastened. "We've made mistakes," he said at his New York office. "I have personally made mistakes. I grant you all that. Now, however, let's look at Iraq, at the future of the region." Until now, he said, the U.S. has "been like a five-hundred-pound gorilla, sucking the air out of getting any kind of coöperation, rather than giving others a chance." It is late in the day to try to "internationalize" Iraq, and other countries could be forgiven for doubting America's sincerity. But Khalilzad seemed to recognize that there's no longer any choice, and the new government of French President Nicolas Sarkozy has recently suggested that it is ready to play a significant role.

For Dodge, the only reason to give this long-shot strategy a chance is the awfulness of the alternative. "I wouldn't bet the house on it succeeding," he said. "But I would bet my hopes and fears for Iraq on it."

In the likely event that the Iraqi state continues to collapse and the withdrawal of American forces leads to further fragmentation and violence, the U.S. will have to adopt something like what the former Administration official called "a venture-capital strategy": using the available resources to invest in numerous small-scale efforts around Iraq, with the knowledge that many, if not most, of them will fail. "We have to shift from the Grand Plan to many initiatives," he said. "We've had enough of the grand strategy." Most of these initiatives will not be able to depend on backing by military force, as so much of American policy in Iraq has until now.

As the U.S. plans for the gradual withdrawal of forces, it should start preparing now for what it can realistically leave behind. The focus of most discussion is on the military side, but America's long-term influence in Iraq will more likely be political. In 2004, when the Coalition Provisional Authority transferred sovereignty to an Iraqi government, it shut down its offices in the provinces and opened four "regional embassy offices" around the country, effectively putting the U.S. out of the business of local development, even as Iraq became fragmented under the pressure of civil war and a collapsed central government. One of the less noticed aspects of the surge has been a belated effort to return American officials to the more obscure corners of the country in the form of "provincial reconstruction teams": joint civil-military efforts that funnel technical help and money from the American and Iraqi governments into the provinces, where political and economic development seems more feasible and responsive to the local population. The recent formation of local police forces and town councils in Anbar Province has had nothing to do with the central government and has been far more successful than previous attempts. These teams should be expanded during the life of the surge, so that they reach the self-sustaining, self-protecting size of a hundred and fifty people; this approach roughly follows the model of Afghanistan, where provincial reconstruction teams first developed several years ago have been an important, if insufficient, tool for extending development to the countryside.

Since Iraq is going to remain in pieces for years, different strategies must be pursued for different regions. Provincial reconstruction teams can provide support to local officials who share the vision of a less sectarian, more democratic Iraq, and they can serve as American eyes and ears. (Without American brigades nearby, these teams will have to depend on the extremely uncertain ability of Iraqi security forces to maintain enough order for them to operate.) This political and intelligence work wouldn't necessarily end if the teams have to be withdrawn, because they would have cultivated Iraqi allies in each province. Around the country, there are thousands of Iraqis who, in the past four years, have put themselves forward as candidates for local councils, city halls, government departments, and the Iraqi security forces. Many more Iraqis have formed organizations in support of issues like women's rights, created labor unions, set up educational programs, and established one of the Arab world's freest (and most endangered) presses.

Too many of these civic-minded citizens, though, have been killed by insurgents and militias who have led a brutal campaign to suppress the stirrings of civil society in Iraq. Lately, it has become virtually impossible for Americans to work with such Iraqis directly. The National Democratic Institute, a Washington-based democracy-promotion group affiliated with the Democratic Party, moved its operations and much of its Iraqi staff out of Baghdad after an American worker was killed in an ambush, last January. But several Iraqi staff members stayed behind to do similar work as a local, all-Iraqi organization, and the N.D.I. is providing technical help and money to get the new group up and running. The U.S. government and private groups should find ways to support Iraqi efforts like these, and, when possible, go back in to work directly with them. The years ahead will be exceedingly difficult for Iraqi democrats, but many of them will tell you that they have learned a great deal in the tumult of the past four years.

If, after a large-scale American withdrawal, Iraq becomes unlivable for such people, then it would be in America's interest to evacuate those who request it, and contingency plans for mass airlifts and expatriation should be made before a crisis comes. This evacuation should go well beyond the current, sluggish process of resettling Iraqis here who worked directly for the U.S. in Iraq. It should be a much more ambitious effort to save the core of what perhaps someday could become a more decent society in Iraq, including military officers, local politicians, and the many thousands of other Iraqis (and their families) who have worked closely and well with Americans. U.S. officials and officers should begin making lists for evacuation. Most Iraqis would be settled in neighboring countries, but as many as a hundred thousand should be brought here, especially if Iraq begins to resemble Cambodia in 1975 or Rwanda in 1994. Beyond the humanitarian imperative, the United States would be salvaging what's left of its enormous investment in Iraq and preserving the seeds of a better future.

When I asked Ahmed, the resident of Saidiya, how he saw Iraq's and his own future beyond the occupation, he wrote back from an Internet café, because his home connection wasn't working. "There might be something good in the future, but only after two decades," he wrote. "It was really a dream to get rid of Saddam, but nowadays the most difficult and unachievable dream, or to put it in other words, the dream that will not come true within the next twenty years, is to have a secular government that separates between religion and politics. I think this long process will outlive us. . . . And, yes, there will be a life for me in another country, in the U.S. specifically. I can be a link that will play a relatively small but effective role in the whole process of creating modern Iraq. I have more to say about that, but I am hearing just now far shooting . . . it is better for me to leave."

There are already at least two million Iraqi refugees in the Middle East. For a long time, the Administration's policy was to pretend that they didn't exist. When officials began to acknowledge the refugees, earlier this year, their statements gave the impression that these Iraqis were waiting out a bad spell and were ready to go back any day. Only recently has the Administration begun seriously to address what's become one of the worst humanitarian crises in the world. On August 28th, the State Department announced that thirty million dollars in aid will be given to help educate Iraqi refugee children in schools around the Middle East. This is a late and small start, but American policy on the refugees remains ad hoc and wishful: in the words of Ellen Sauerbrey, the Assistant Secretary of State for refugees, the answer to this immense and immediate problem is simple—"creating a safe and stable Iraq." A major American-led effort to help provide education, health care, and jobs to Iraqi refugees probably would require us to give aid to the governments that are hosting them—even, through intermediaries, to Syria. Unless the U.S. recognizes that Iraqis won't return home for years and are living precarious lives in Arab countries that are barely tolerating them, these refugees will become yet another radicalized population of homeless people in the Middle East, and one more strategic problem for the United States.

Radicalization is one of the dangers considered in "Things Fall Apart," a short book by Daniel Byman and Kenneth Pollack, published earlier this year by the Brookings Institution. (Pollack more recently co-wrote a controversial Times Op-Ed, called "A War We Just Might Win," which declared definitively that the surge was a military success.) "Things Fall Apart" examines the history of recent civil wars around the world and concludes that they usually last a long time, spill over into other countries, and end only through the military victory of one side, or through massive external intervention. Pollack and Byman outline a strategy for containing an "all-out civil war" within Iraq's borders. Their policy proposals include avoiding taking sides, keeping American troops out of urban warfare, and providing economic support and promoting political reform in the region to prevent it from being destabilized. Perhaps most questionably, they recommend setting up bases for American patrols along Iraq's borders with its Arab neighbors and Kurdistan (but not Iran), to prevent foreign fighters from getting in and refugees from pouring out. American brigades and regiments would protect refugee collection points called "catch basins" on the borders. Since these would likely become miserable camps infiltrated by sectarian militias, with Americans as vulnerable camp guards, this proposal is notably risky, and has garnered scant support. But at least the authors didn't simply wish the worst-case scenario away.

Last October, I was invited to sit in on a "war game" at the Brookings Institution, for which Pollack and Byman assembled a group of former Republican and Democratic officials to simulate the endgame in Iraq. As civil war drove American troops to the borders, the officials' policy debates zeroed in on the short-term winner, Iran. Some officials wanted to arm Sunni groups, with the backing of other Arab states, to wage an insurgency against Iran and its proxies in Iraq. Since that war game, the "Sunni awakening" of tribes in western Iraq and the American decision to sell arms to Saudi Arabia, a Sunni country, have nudged the U.S. closer to the Sunni side. It may very well become American policy to keep Iraq's Sunni Arabs strong enough to create a stalemate in the civil war and contain Iranian influence in the region. But, given that the war began in 2003 with the goal of bringing Iraq's Shiite majority to power as the leading edge of a democratic transformation in the Middle East, this return to balance-of-power Realpolitik, with the Saudis as America's most important allies, would represent the ultimate failure of the President's project.

The war was born in the original sins of deceptive salesmanship, divisive politics, and wishful thinking about the aftermath. The bitterness of that history continues to undermine American interests in Iraq and the Middle East today. President Bush will have his victory at any cost, with one eye on his next Churchillian speech and the other on his place in history, leaving the implementation of his war policy to an Administration that works at cross purposes with itself, promising freedom and delivering rubble. The opposition is plainly eager to hang a defeat around his neck and move on from what it always regarded as Bush's war. Before the U.S. can persuade the world to unite around a shared responsibility for Iraq, Americans will have to do it first. The problems created by the war will require solutions that don't belong to a single political party or President: the rise of Iranian power, the emergence of Al Qaeda in Iraq, the radicalization of populations, the huge refugee crisis, the damage to a new generation of Iraqis who are growing up amid the unimaginable. Whenever this country decides that the bloody experience in Iraq requires the departure of American troops, complete disengagement will be neither desirable nor possible. We might want to be rid of Iraq, but Iraq won't let it happen. ♦