IN late 1989, the Central Intelligence Agency predicted that Iraqi leader Saddam Hussein would not invade Kuwait in the foreseeable future. Nine months later, the Iraqi Army rolled into Kuwait City.
The CIA is supposed to have shadowy sources everywhere. Yet the mistake about Mr. Hussein is just one of the agency's famous flop forecasts - analysts thought the Soviet Union wouldn't invade Afghanistan either.
Pushed by such failures, plus the need to work more efficiently in an era of smaller budgets, the CIA is in the midst of sweeping reforms in the way it writes up and disseminates its intelligence, according to a top agency official.
These reforms represent sweeping changes in an agency already roiled by the implications of the Aldrich Ames spy case and attacked in Congress as bloated, unresponsive, and perhaps unnecessary.
Wrong predictions are only half of the problem, said Douglas J. MacEachin, CIA deputy director for intelligence, at a rare meeting with reporters. Past CIA products were often too long, too opinionated, and devoid of facts that policymakers could use to think for themselves, he said.
In short, CIA analysts had drifted away from thinking like government officials and had adopted the worst habits of university professors. The culture had become ``publish or perish'' - the lengthier and more esoteric the subject, the better.
``We listened to too many people who told us we were a university that happened to use classified material,'' bemoaned Mr. MacEachin, a crusty 29-year veteran of the CIA system.
Young analysts knew that the way to get ahead was to write a lengthy, hard-bound paper that was the definitive word on, say, Botswana's potash production.
Forecasts of future events were too often made flatly, as if they were the results of fortune telling. Fearing policymakers would begin to do analysis themselves, CIA analysts were reluctant to spell out the facts that led them to make a prediction.
Major forecast failures of recent years can be traced to wrong underlying assumptions, according to MacEachin.
CIA analysts thought Saddam Hussein wouldn't invade Kuwait because they judged his military had taken too much damage during its lengthy war with Iran, for instance. Analysts missed the Soviet invasion of Afghanistan because they thought the Kremlin wouldn't want to endanger the then-pending SALT II arms treaty.
As it turned out, the Soviet Foreign Ministry was vehemently against the action for just that reason, but it was overruled by ideologues in the ruling Communist Party, MacEachin says.
When it comes to crises, the White House and the State Department often don't really want predictions anyway, the CIA official said. What they want are facts and analyses that might show them a foreign-policy advantage.
Consider the CIA as equivalent to a football team's scouting squad, said MacEachin. The head coach doesn't want to hear a prediction about the next game's final score. He wants to know ``whether they've got a weak left safety,'' the official said.
After Gen. Colin Powell, chairman of the Joint Chiefs of Staff under President Bush, complained that the agency had misread Iraq's intentions, then CIA director Robert Gates began changing the system in 1992.
A task force produced a number of recommendations now being implemented, including sharp cutbacks in the number and length of CIA printed products. Facts and findings derived from them will be emphasized, rather than predictions.
An old saw goes that there's no bureaucracy like a secret bureaucracy, and MacEachin says he has run into tremendous ingrained resistance to change.
But as a longtime agency employee, MacEachin says the vehemence of the criticism sparked by the Ames spy case and lawsuits alleging CIA sexual discrimination trouble him also.
Incoming administrations look at CIA officials ``like you're a bunch of nuts,'' said MacEachin. ``I've had about enough of it, frankly. I'm a grandfather and I'm in a position to walk.''