How to Improve Intelligence

Robert Jervis writes in the Boston Globe that to improve intelligence, CIA investigators should stop thinking so intuitively, pay more attention to what they see in front of them, make assumptions that can be disproven, and realize that terrorists don’t see the world like they do. [%comments]

chris markl

wow this seems like a great plan not just for the CIA but for solving any problem that we encounter.


From that article:........"We face skilled adversaries who are trying to deceive us. "
This is called mythologizing your opponent - the author is guilty of what he accuses the CIA of.
"If people are often surprised to discover that their own spouses have been cheating on them, how could..." this is absurd on the face of it. "Surprised" is only the third act to fill out the drama of having known all along. The author exposes little evidence of having lived in reality.
And the crown jewel of authorial disconnect:...."many of their failures follow from intuitive ways of thinking"

Try reading the commission reports before stepping up to the podium.


I think that's an excellent article and applies, as chris says in #1, to all kinds of analysis. Explaining things to someone else makes the assumptions more clear and gives perspective. And just knowing that there might be dog that didn't bark is really helpful.


Use the Force, CIA, not.

Eric M. Jones

"make assumptions that can be disproven..." falsifiable hypotheses....

Neither science nor human thought works that way, only philosophers that think this is a clever notion even bother to mention it. Science is full of axiomatic non-falsifiable hypotheses. Scientists use the term "testable" and some things just aren't.

The reason the Japanese thought they could beat the US was based more on the actual armaments. In 1935 the US had the tiniest army...only larger than Germany's! Compared to the US, the Japanese had twice the active soldiers and five times the reserves, and was more than a match for the US in planes, submarines and battleships. The Japanese had two battleships (and three more being built) that were TWICE the size of the biggest US battleships.

Had the US not won the Battle of Midway....that would have been the end or the US in WWII.

Mark S.

Conclusion from the article: several US regulatory govt. agencies already do force their regulated industries to follow these approaches. One govt. institution does follow all or msot of the thinking described in the article.

1. FDA: demands evidence based decision making and verifiable data with a statistical rationale to back up any decision. Where data can't be obtained or not enough data is available the agency "wants" to see a risk management based approach. What the FDA "wants" is effectively a mandate even if it is not written into law. The downside for an intelligence/law enforcement environment: the data and conclusions per the FDA approach would not be timely.

2. FAA: recommends that pilot training have a team structure with a leader and a challenger, so the plane doesn't fly straight into a mountain because no one says anything.

3. CDC: Is aware and informed of outbreaks, medical emergencies and disease patterns far in advance of any media reports or most healthcare agencies. They are successful at this because they have close connections to many different professions that might be first responders in a medical situation. Their staff conduct training sessions for many different types of healthcare related professions (worldwide) and maintain contact. The reason the general public doesn't hear a lot about what they do is that they try to avoid issuing false positive alerts which would be equally damaging to missing an impending disaster.

Bottomline: the CIA (and the FBI and DHS and the other half dozen US intelligence agencies) could learn from other govt. agencies on how to improve their practices but there would be a fundamental problem in getting the police & military action oriented types to become data driven and analytical. Maybe they should just hire a bunch of behavioral economists.



On the continuum of pure logic to pure intuition, I'm sure the author is more or less correct.

On the other hand, the kind of person attracted to intelligence is probably predisposed toward "pattern recognition", and other heuristics. Telling them to watch what's going on against that grain is like telling people to pay more attention: It sounds good, but it isn't concrete enough to help.

I'd imagine one technique would be to start giving those quizes where the answers are more straightforward than one would believe, like "Does britain have a fourth of july?".

Otherwise, this is a lesson learned, and tossed into institutional experience.


I took "Conflict and Cooperation in World Politics" with Dr. Jervis when I was an undergrad. The 9/11 attacks occurred during my first week of classes freshman year (at Columbia), so a lot of the Poli Sci courses started having a terrorist angle, but he definitely knew his stuff. The CIA would be wise to listen to him!