Thursday, December 4, 2008

Top 5 Intelligence Analysis Methods (List)

(Note: I was recently asked to name and describe my top 5 intelligence analysis methods. As I began to think about it, what seemed like a fairly straightforward question morphed into what I could only think of a series of blog posts. So, here they are...)

Considerable emphasis has been put on improving the methods of intelligence analysis over the last six years. The 9/11 Report alluded to the need for it, the WMD Commission addressed it more directly and the DNI recently highlighted the continued requirement for advanced analytic techniques in its Vision: 2015 document.

Still, the intuitive method (also known as "read a bunch of stuff, think about it for a bit and then write something") remains the most popular method for producing intelligence analysis despite this method's well known tendency to permit a wide range of cognitive and systemic biases to corrupt the analytic product (see Heuer and Tetlock for excellent overviews of these problems).

Beyond the intuitive method (and the interesting defenses of it offered by books such as Blink and Gut Feelings), what, then, are the best methods for conducting intelligence analysis? Given the wide range of intelligence analysis problems (tactical, operational, strategic) and the large number of disciplines using intelligence analysis to support decisionmaking (national security, law enforcement and business) is there any chance that I can identify the five best methods?

My answer is, obviously, "Yes!" but before the fighting begins (and there will be fighting...), I intend to give myself a chance of convincing you by defining not only what I mean when I say "method", but also what makes for a good one.

What Is An Intelligence Analysis Method?

The word "method" is often used casually by analysts. When used this way, processes as different as brainstorming and Analysis Of Competing Hypotheses can both be seen as "methods" or ways to improve thinking. While such an informal definition might work at a cocktail party, it is not very helpful for professional purposes. "Method", in my opinion, should be reserved for processes that produce or substantially help the analyst produce estimative results.


It is simple, really. Estimative results are what decisionmakers want most from intelligence. It is nice to have a good description of an item of interest or a decent explanation of why something did or did not happen. Both provide useful context for the decisionmaker, but nothing beats a good, solid estimate of what the enemy or competitor or criminal is likely to do next. Defining method as something that produces estimative results means that I am connecting the most common term with the most desired result.

All the processes that help the analyst think but do not, by themselves, produce estimative results (such as brainstorming) I call "analytic multipliers". I get this from my military background, I suppose, where there are elements of combat power, such as armor or artillery, and combat multipliers, such as morale.

Analytic tools, then, are particular pieces of software, etc. that operationalize the method or the multiplier (or in some cases multiple methods and multipliers) in a particular way. For example, ACH is a method but the PARC ACH 2.0.3 software is a tool that allows the analyst to more easily do ACH.

I find these distinctions very useful in discussing the analytic process with students. If everything is a method -- if free association exercises are treated, linguistically, the same as multi-attribute utility analysis, for example -- then nothing, in the mind of the student, is a method. Clearly, not every process falls neatly into the method or multiplier camp (what is SWOT, for example, under these definitions?) but some generally agreed upon set of words to capture the large and easily recognizable differences between things such as ACH and brainstorming seems useful.

Tomorrow: What makes a good method?


Anonymous said...

I'm curious, what are the top 5 intelligence analysis methods or analytical multipliers?

Kristan J. Wheaton said...


This is the first in a series of posts on the subject. You can find all of the posts here:

Sorry for the confusion.