Most anyone who's worked in government has a story -- probably re-told often these days, given the Iraq debate -- about facing a big decision on the basis of information that then turned out to be wrong. My favorite is from August 1998 when, with Bill Clinton just three days away from a trip to Moscow, the Central Intelligence Agency reported that President Boris Yeltsin of Russia was dead.
In 1998 the news that Mr. Yeltsin had died was, of course, no more surprising than the news, in 2003, that Iraq had weapons of mass destruction. It matched what we knew of his health and habits, and the secretive handling of his earlier illnesses. Nor was anyone puzzled by the lack of an announcement. Russia's financial crash 10 days earlier had set off a political crisis, and we assumed a fierce Kremlin succession struggle was raging behind the scenes.
In the agonizing conference calls that ensued, all government agencies played their usual parts. The C.I.A. stood by its sources but was uncomfortable making any recommendation. National Security Council officials, knowing Mr. Clinton wasn't eager for the trip, wanted to pull the plug immediately. The State Department (in this case, me) insisted we'd look pretty ridiculous canceling the meeting because Mr. Yeltsin was dead -- only to discover that he wasn't.
Eventually we decided that the Russians had to let the deputy secretary of state, Strobe Talbott, who was in Moscow for pre-summit meetings, see Mr. Yeltsin within 24 hours or the trip was off. Nothing else would convince us: no phone call, no television appearance, no doctor's testimony. The next day Mr. Yeltsin, hale and hearty, greeted Mr. Talbott in his office, and two days later Bill Clinton got on the plane to Moscow.
When the trip was over, I phoned the C.I.A. analyst who had relayed the false report. He was apologetic -- sort of. ''You have to understand,'' he said. ''We missed the Indian and Pakistani nuclear tests last spring. We're under a lot of pressure not to miss anything else.''
Some of the lessons of this episode are the same as those emerging from the Iraq debate: sensitive intelligence is often too weak to guide important decisions; if the information fits what we already believe, or what we want to do, it gets too little scrutiny.
Yet Mr. Yeltsin's ''near-death experience'' of 1998 carries another lesson that unfortunately hasn't been part of the current controversy. When policymakers have imperfect information about a serious problem (which is almost always), what should they do? The answer, then as now, is to shift the burden of proof to the other guy. If we had been denied that meeting with Mr. Yeltsin, it would hardly have proved that he was dead. But we would have canceled the trip all the same. Russian uncooperativeness -- not our poor intelligence -- would have left us no choice.
Going to war and canceling a trip are vastly different matters, but what the Bush administration did with Saddam Hussein in the run-up to war followed the same rule: it challenged him to prove that American intelligence was wrong, so that the responsibility for war was his, not ours.
Clearly, President Bush and his advisers did not expect Saddam Hussein to cooperate in this test, and might still have wanted war if he had. But even if the administration had handled other aspects of the issue differently, it would still have been necessary to subject Iraq to a test. In our debate about the war, we need to acknowledge that the administration set the right test for Saddam Hussein -- and that he did not pass it.
When America demanded that Iraq follow the example of countries like Ukraine and South Africa, which sought international help in dismantling their weapons of mass destruction, it set the bar extremely high, but not unreasonably so. The right test had to reflect Saddam Hussein's long record of acquiring, using and concealing such weapons. Just as important, it had to yield a clear enough result to satisfy doubters on both sides, either breaking the momentum for war or showing that it was justified.
Some may object that this approach treated Saddam Hussein as guilty until proved innocent. They're right. But the Bush administration did not invent this logic. When Saddam Hussein forced out United Nations inspectors in 1998, President Clinton responded with days of bombings -- not because he knew what weapons Iraq had, but because Iraq's actions kept us from finding out.
A decision on war is almost never based simply on what we know, or think we know. Intelligence is always disputed. Instead, we respond to what the other guy does. This is how we went to war in Iraq. The next time we face such a choice, whether our intelligence has improved or not, we'll almost surely decide in the very same way.
Stephen Sestanovich is a senior fellow at the Council on Foreign Relations and a professor of international diplomacy at Columbia University. From 1997 to 2001 he was United States ambassador at large for the former Soviet Union.