How to stay in—without staying the course.

It is not time to pull out of Iraq, but it is time to get real. This means accepting, and acting on, several unpleasant facts.

First, the United States has run out of political and moral credibility. Every serious Iraqi faction, even those sympathetic to our aims and grateful for our intervention, is doing all it can to gain distance from us. Any program, policy, or proposal put forth by Americans—or anyone associated with Americans—is bound to be rejected on its face.

Second, further U.S. military offensives that involve killing Iraqis—however justifiably or unavoidably—are almost certain to trigger further uprisings.

Third, ignoring short-term disfavor and sending in another 20,000 troops for the sake of longer-term goals will do almost nothing to quash the insurgency, foster stability, or impose order. It would take several times that many forces to alter the balance of power in any significant measure, and such forces are simply unavailable.

Fourth, "staying the course" is pointless at best, damaging at worst. There seems to be no road that leads from our current position to the Bush administration's goal of a secular, secure, democratic Iraq whose thriving example transforms the Middle East. A year ago, Paul Wolfowitz, the deputy secretary of defense who hatched this dream, spoke loftily of a Mesopotamian de Tocqueville. Today, we bank our hopes on the moderating influence of a man whose title is "grand ayatollah" and who refuses to meet with Americans as a matter of principle.

So, what to do?

Retired Army Gen. William Odom—director of the National Security Agency under Ronald Reagan and now an analyst at the hawkish Hudson Institute—looks at the trends and concludes that we should leave, quickly. "We have failed," he told the Wall Street Journal recently. "The issue is how high a price we're going to pay. … Less by getting out soon, or more by getting out later?" Odom further argued that other countries would be more likely to send troops to Iraq if they knew ours were leaving.

His logic is tempting but terrible. What if he's wrong and other countries don't take our place? Iraq, whose security forces are meager at best, would fall apart. A blood bath would follow, as those who did risk cooperating with us would be condemned as quislings. Neighboring powers might send their own armies across the undefended borders, either to quell or exploit the ensuing chaos. Whatever the wisdom of starting this war, it is impolitic and immoral to wreck a country, light some extra bonfires, and go home.

In the latest issue of the New York Review of Books, Peter Galbraith proposes to end the internal conflicts—and furnish a decent exit strategy—by turning Iraq into a tripartite federation of Sunnis, Shiites, and Kurds. (He writes that he wouldn't go quite so far as Leslie Gelb's idea to split the place into three separate states, but his plan amounts to much the same thing in its details.) This notion, too, is attractive on the surface but deeply flawed just beneath. First, the Sunnis, who have no oil or other resources, get nothing out of such a deal; they wouldn't accept it, and, if it were forced upon them, they'd rebel against it. Second, Iraq's ethnic divisions don't precisely match geographic ones; ethnic cleansing would loom large. Third, Galbraith (a well-known advocate for the Kurds) never spells out the role or composition of a central government; his scheme makes the Articles of Confederation seem Hamiltonian by comparison. Finally, the resulting state or states would be weak, and borders itchily porous. A quasi-independent Kurdistan would be ripe for Turkish invasion. Shiites would face Iranian assimilation or worse. The dispossessed Sunnis would invite or prompt forceful assistance from the Saudis. If Odom's shrug leaves Iraq open to civil war, Galbraith's dotted lines threaten to inflame the entire region.

However, Galbraith's premise—the appeal of somehow isolating the three major ethnic groups, which the British cobbled into the artificial state of Iraq 85 years ago to begin with—does inspire a more limited, practical idea. Creating three separate (or very loosely federated) states may be an unworkable solution, but how about creating three separate (or differently composed) security forces?