I've been thinking about what happens when a society goes to war for a limited objective but then comes to face what seems to be an unlimited cost.

The more blood and treasure a society spends on such a war, the harder it becomes to acknowledge that it can't be won. After spending so much, a retreat would be painful: The society would have to acknowledge that the entire enterprise — the lives lost, the money spent — was a waste (worse than a waste, really, because of opportunity costs and unintended consequences).

Any society would want to avoid the pain inherent in such acknowledgment. It would prefer to believe there is still some chance of winning. If victory is possible, even if securing it turns out to be costlier than first believed, at least the society would have something to show for what it paid.

If you were a member of the administration that launched such a war, and you understood these dynamics, what would you do? Even if you knew, up front or deep down, that the war couldn't be won, would you bring the troops home?

Not likely. You would have to take the entire blame for the failure, with no room for face-saving or rationalization. Most people wouldn't be able to face such an unarguable personal failure. Instead, consciously or unconsciously, such an administration would seek to defer the withdrawal to a successor. Doing so would obscure the administration's personal and historical culpability for the war: Members would always be able to say, "We could have won if our successors hadn't lost their nerve." And who could "prove" them wrong?

I expect such an administration would continue the war, trying to keep U.S. casualties close to levels the public had already proven willing to accept. Periodically, the administration would announce "turning points," the achievement of which would imply that the nation is indeed on the road to victory. As each previously declared turning point is reached and revealed to have no effect on the course of the war, the administration would articulate a new one, thereby maintaining the public's hope that there is still some purpose to the enterprise — that the war can still be won. Simultaneously, the sunk costs of the war would be increasing, deepening the society's need to win, somehow, if only to justify the increasing costs.

This is a potent political combination: undiminishing casualty levels, constant infusion of new hope, increasing sunk costs. Because this combination is relatively stable while the pain of a "we can't win" acknowledgment gets worse the longer the war drags on, the status quo would prevail for a long time. Eventually, the war could be passed on to the next administration. Blame for losing it could be passed on as well, or at least shared and obscured.

At some point, during the tenure of the administration that launched the war or of one of its successors, the war will have dragged on long enough to force the conclusion that victory isn't possible. It's not so much that the pain of what has been spent becomes overwhelming; it's the sense of nothing but further pain ahead, for no possible gain, that would bring about a new consensus on the war. Vietnam illustrates the point. I don't think what happened was, "We've lost 58,000 Americans and that's enough." It was more like, "We've lost 58,000, and even with another 58,000 I still don't see how we can win this." In other words, the pain of acknowledging failure was finally outweighed by the prospect of more pain for no gain. When a society reaches this point, it abandons the war.

In trying to articulate these dynamics, I've deliberately avoided mention of current events. Sometimes you can see more clearly by taking a step back from the matter at hand. But obviously I do think what I've described above applies to the Bush administration and Iraq. Maybe the question isn't just, "Is the war winnable?" but rather, "Even if it's not winnable, what will the administration and our society do then?" It's that second question that's important to answer.

Barry Eisler is the award-winning author of Killing Rain (Putnam) who spent three years in the CIA's Directorate of Operations. This essay first appeared on the blog Confessions of An Idiosyncratic Mind.