"We judge with high confidence that in fall 2003, Tehran halted its nuclear weapons program; we also assess with moderate-to-high confidence that Tehran at a minimum is keeping open the option to develop nuclear weapons."
This sentence is a key part of the latest National Intelligence Estimate on whether Iran is seeking to develop nuclear weapons. Its highly stylized language reflects an effort to be more precise about what American spy agencies know – and don't know – about what Iranian policy makers are doing and thinking.
The unclassified portion of the report that was just released to the public contains a page of explanation, enclosed in a box, about what its language is intended to mean.
The language of the report is called "estimative," meaning that it conveys judgments, not "facts, proof or knowledge." That's emphasized throughout the report whenever the writers say we judge, we assess, or we estimate.
The wording of the report is designed to do two other things:
— Show how likely the intelligence community thinks an event may be.
— Show how confident the analysts are in their judgment.
How likely is it?
The writers of the NIE use a scale of terms designed to give a closer idea of how probable a development or event may be:
—If there's a better-than-even chance that something will happen, the writers use terms such as probably/likely, very likely, or almost certainly.
—If there's a less-than-even chance, the writers use terms such as unlikely, very unlikely or remote.
—Even if there's a less-than-even chance that something will happen, the writers might want to mention it because it could have significant consequences if it did happen. In that case, they use phrases such as we cannot dismiss, we cannot rule out, or we cannot discount.
—If the writers simply don't have enough good information to determine the likelihood that something will happen, they use terms such as might or may.
How confident are they in their judgments?
It all depends on the quality of the information and where it comes from. Intelligence officials try to reflect that by assigning one of three levels of confidence:
—High confidence means the intelligence community generally thinks a judgment is based on high-quality information or that the issue is one that allows for a solid judgment.
Moderate confidence means that the information is plausible and comes from credible sources, but that there's not as much support for its accuracy.
Low confidence means that intelligence analysts question the credibility or plausibility of the information, or that they're concerned about the sources.
Even where intelligence officials have high confidence in their information, they stress that judgments are not certainties, and that there's still a risk that the conclusion could be wrong.
A National Intelligence Estimate concluding that Iran stopped work on its nuclear bomb project in 2003, contradicting an earlier report on the subject, was made public this week.
Here's a breakdown of the NIE:
What is a National Intelligence Estimate?
It's a report that represents the collective opinion of the U.S. intelligence community on a specific issue. NIEs are typically requested by the administration, military commanders, or members of Congress on perceived threats to U.S. national security.
Examples are the 2002 NIE "Iraq's Weapons of Mass Destruction Programs," the 2005 NIE on whether Iran was pursuing a nuclear weapons program, and the 2006 NIE "Trends in Global Terrorism."
Who produces the report?
It's produced by the National Intelligence Council, a think tank within the Office of the Director of National Intelligence. The council brings together analysts and advisers from the 16 separate U.S. government agencies that collect information related to national security and foreign relations. The best known is the CIA — the Central Intelligence Agency. Others include the intelligence-gathering units of the various military branches, the FBI, the State Department and the Treasury.
Where does information for an NIE come from?
The Office of the Director of National Intelligence lists six kinds of sources for information. Among them are:
Open-Source Intelligence: gained by studying information that's openly published in other countries, such as scientific journals and news reports.
Signals Intelligence: gained by intercepting communications, such as telephone calls, e-mail or military radio traffic.
Human-Source Intelligence: provided covertly — by spies — or overtly, by U.S. diplomats and military attaches overseas.
Imagery Intelligence: gained from satellite photos, radar, infrared sensors and other devices.
How are the judgments reached?
Council members meet to decide on the terms of reference including what the review will achieve, who will take part in it, and what questions they will consider. The agencies contribute their analyses of the issue, compare notes, and write a report that tells what they commonly believe — and how much confidence they have in each judgment. Agencies that disagree with an analysis can include their dissenting opinions in footnotes.
What happens when policymakers don't like or don't accept the conclusions of an NIE?
The estimates aren't necessarily the final word on a subject. The 2002 NIE on Iraq said that Iraqi leader Saddam Hussein had chemical and biological weapons and that he was reconstituting his nuclear program. After Iraq was invaded in 2003, the NIE conclusions proved to be wrong. Those alleged "weapons of mass destruction" were never found in Iraq. A 2004 Senate investigation concluded that the NIE was flawed because analysts began with the assumption that Iraq had the weapons — and were led to "ignore or minimize evidence that Iraq did not have an active and expanding program."
The 2005 NIE that concluded Iran was trying to develop nuclear weapons has been directly contradicted by the 2007 NIE, which says that Iran stopped work on its nuclear bomb project in 2003.