If we want to succeed in Iraq, we should quit looking for military solutions, discard the naïve notion that time will transform the warring factions into democrats, accept the likely division of the country, and introduce its leaders to the dark side of American politics: gerrymandering.

What should be obvious by now is the simple fact that Iraq’s violence is driven mainly by internal factors and long-held animosities. That’s why all the “solutions” the United States has tried to date have failed—why last month’s new strategy, this month’s revised strategy, next month’s updated strategy, and the new strategy that will follow that, also will fail. We can press the Iraqis to do all the right things and Iraqi leaders can agree in principle to all such conditions, but at the end of the day the factional nature of Iraqi society will make it impossible to implement such compromises.

So what do we do? If we really hope to see a stable, peaceful Iraq, we need to eliminate the two main causes of the violence: the U.S. occupation and the fact that all three groups in Iraq—the Shi’a, Sunni and Kurds— fear control of the central government by anybody else.

The U.S. military occupation is easy to fix: say we’re going, and go. The second is far more difficult, given Iraq’s recent history, where one faction, the Sunnis, used the power of the central government to oppress the others.

While they certainly don’t want a repeat of that, what the Kurds and Shi’a really would prefer is autonomy. Only the Sunnis—now out of power—want to keep Iraq unified, primarily because most known oil reserves are in the Kurdish north and Shi’ite south. If the country fractures along ethnic lines, the Sunnis become the big losers.

Surge or no surge, if the United States continues current policies, all it does is delay the eventual chaos. If we hope to prevent this, we need to think creatively—to consider the trick former Massachusetts Gov. Elbridge Gerry taught us.

The Web site fairvote.org defines gerrymandering—a manipulative political art dating back to 1812, when Gerry oversaw the creation of a political district “that looked like a salamander” (thus the term gerrymander)—as “the deliberate rearrangement of the boundaries” of political districts to affect the outcome of elections. In Iraq, that same creative arrangement of political boundaries might be the key to cobbling together a country that’s not at war with itself. The United States should show the way, then step out of the way.

First, the United States should start preparing for the inevitable troop withdrawal. The leaders of Iraq’s various factions need to know that judgment day is coming: They can either prepare for an all-out bloody civil war, or they can start putting the architecture together for a post-U.S. occupation government everybody can live with. The only type of government that would meet such a test would be a loose confederation of autonomous regions.

Iraq already is divided into such autonomous areas, with Sunni insurgents and Kurdish and Shi’ite militias governing them. What is now a de facto division of territory and power needs to be formalized.

The major obstacle to this is getting the Sunnis to agree. That’s where the fine art of gerrymandering comes in.

The Sunni position is understandable. In the past they controlled virtually everything. Today, because most of the oil wealth is in Kurdish and Shi’ite territory, they could end up with nothing. With creative map drawing, however, the Sunnis could be given oil fields in the northern and southern parts of the country, even if the Sunni-controlled region ends up looking like a salamander, or like Arizona’s 2nd congressional District—a contorted and twisted example of gerrymandering at its creative best.

Merely agreeing to “share” oil revenues among the regions probably won’t work because the Sunnis would be suspicious that the Kurdish and Shi’ite governments would someday cut them off from the proceeds. So don’t promise them a share of oil revenues; give them oil fields.

A loose confederation of Iraqi mini-states—with a central government that has clearly delineated, but limited power—could work.

In such a confederation, the weak central government might only have the power to conduct diplomatic affairs and trade negotiations with other nations, prohibit internal barriers to commerce within the confederation, and provide a judicial venue where disputes could be aired and resolved. The regional governments would provide security and most other government functions.

The Bush administration and its allies have no other viable choice than to try to help the Iraqis cobble together a new form of government that recognizes the deep divisions—and longstanding distrust—among the Iraqi factions. At this late date, even this solution may not work because the factions are splintering, may not be able to control their followers and may not be able to enforce agreements with the other groups.

Nevertheless, a decentralized solution may be Iraq’s last best hope. And if it works, it will not be because the Iraqis adopted the best traditions of American politics, but because they utilized one of its worst.