A review on "Psychology of Intelligence Analysis by Richards J. Heuer Jr."


Richards J.Heuer Jr.is one of the people who revolutionised the way intelligence content is produced in the Central Intelligence Agency of the U.S.A. Speaking crudely, his main contribution was the introduction of the "Scientific Method" into the everyday routines of the CIA analysts. This book is partly his self-reflection on this transformation, and partly a list of heuristics that any intellectual worker could employ to improve his own efficiency (and self-satisfaction). I found it very good. It clarified quite a bit of concepts I had been only vaguely aware of, and helped me hone a few of my own ideas.

I actually recommend reading it to everyone, and perhaps would even suggest studying it at school, because it is hard to find a skill of more generality than a skill of thinking. And the intelligence aroma just makes the book more exciting for kids.

If you are interested in more detail, welcome under the cut.

What is intelligence?

In English, "intelligence" can mean several things, the best known being the skill of thinking, and the profession of preparing reports about the world to the Head of a State. The coincidence of these two words is, although probably accidental, very illustrative.

It is astonishing, how living beings, when acting in groups, tend to replicate themselves on a larger scale. A company or a country can be seen as having a digestive system (economics), a fighting system (the military), an immune system (police), the nervous system (the government), as well as, obviously, a thinking system. To my own disgrace, I used to consider the academia to be this thinking system, however, after reading Heuer's book, I am much more sceptical about this attribution The Intelligence agencies, as well as political parties, seem to be much more akin to the "thinking subsystem" than the ivory tower people disconnected from the empirical world.

The Central Intelligence Agency is, perhaps, the most well-known thinking body in the world. It is not that being famous is a benefit for a secret service, but I guess they could not have avoided that. In the seventies they became so big that they had to develop and internal self-reflection mechanism, which eventually became the "Institute for Study of Intelligence".

The institute had a huge role in the CIA's internal reform, which in turn led to a ( self-proclaimed) huge boost in efficiency.

This book, which is written by a prominent player in this reform, summarises several observations, which are supported by quite a solid (for social sciences) scientific base, aimed at precisely identifying what exactly is the process of (human) thinking, where it fails and how it can be improved.

What does intelligence work look like?

Intelligence is, first an foremost, bureaucracy. I know, it seems like quite an obvious thing to say, but I'm always impressed by how much exactly I underestimate seemingly obvious things. I had to interpolate quite a huge cognitive gap between myself and the author by imagining hordes of CIA clerks writing thousands of pages of… can I call them reports? Maybe the word "encyclopedias" would have been a better fit?

Imagine a government official needing to make a decision. Nobody, even the smartest one can decide independently on every issue that arises. Moreover, it is probably impossible to even select a subset of issues and to research those carefully, since the urgency of the issues is not directly related to their importance, some (most?) are pushed in by the political agenda, which cannot be just ignored.

Therefore a super ignorant politician is in a desperate need of something that he can ask for "what shall we do with this super urgent thing that I have no idea about". God Bless Google and Wikipedia for being able to deliver context to the executives. But even having the context is not enough, because after the context is established, the politician is very likely to ask "what happens if we do X in this context Y?". And this is a question Wikipedia cannot answer. Moreover, it is probably just as unlikely to be able to produce an high quality answer to such a question on-the-spot, because answering such questions is usually a slow job. Therefore, there needs to be someone who "pre-caches" answers to such questions.

My imagination already paints immense cupboards filled with identically bound books, only differing by the titles on their spines: "What shall we do if North Korea invades South Korea in 2010 (2011, 2012, 2013, 2014… )", all very similar to each other, but ready to be placed at the U.S. president's (or, indeed a Chinese general secretary) desk in case the event occurs.

Essentially you have writers whose imagination is too poor to entice the readers, or journalists whose writing style is too dry, writing almost identical texts on the almost identical subject, day after day, year after year. Such extremely boring people are called "intelligence analysts".

There are also people who's job is to invent questions to ask them, as well as people who's job is to assess where the reports they are making bear as much resemblance to reality as possible.

If all of this sounds an awful lot like a very sad version of cosplaying Google – that's because it is.

Thinking as an algorithm.

So, the intelligence people have invented Google a long time before Google. And Google's question-answering system is, perhaps, the closest approximation to the "artificial intelligence" that we may think of nowadays. This leads us to the question of what exactly is artificial intelligence?

In fact this thought hasn't left me all the time that I have been reading this book. This book is almost like writing an interpreter for a peculiar (military-styled) query language, in an peculiar (psychologically-styled) language, producing reports in a peculiar (politically-styled) language.

And a language interpreter essentially consists of a database (and the books speaks about the knowledge representation in human brains), and a set of subroutines that operate on these data. Just that simple.

Naturally, as every programmer knows, machine primitives available to the programmer are far from being ideal. Most programmers are used to integer overflows when adding numbers. Far less people are prepared for cognitive biases in humans when writing instructions, manuals and guidelines for people.

Indeed the book deals quite a lot with these two topics: how to ensure proper data management in an analyst's brain, and how to program analysts in order for their reports to be written taking into account the limitations of the human brain's thinking subroutines.

A brief overview of the guidelines.

I will not write a lot about the guidelines themselves. After all, the book is all about that, and I have prepared a list of key points for myself, in a separate document. (https://gitlab.com/Lockywolf/linuxbugs-lwf/-/blob/master/notes/2020-09-03_Richards-J-Heuer-Jr_Psychology-of-Intelligence-Analysis/2020-09-03_richards-j-heuer-jr_psychology-of-intelligence-analysis_notes.org)

But I need to give points to the author for writing a book in a way that is pleasant for reading. Chapters are well structures, every section has a summary prepended to the main text. (I revised the book content after reading by doing a second pass and reading summaries only.) There are several problems given, which illustrate some of the book concepts on the reader himself. Nice!

In general, however, the three most general advices the book gives would be:

  • Trust nothing, nobody, and foremost, don't trust yourself.
  • Do care about your own mental data structure.
  • Write, write, and write. Bureaucracy rules the world.


One of the best books on the skill of thinking. I would definitely recommend it as a high school textbook for kids, as a gentle introduction into how messy the thinking process actually is. I will probably also read it later again, for a recap and to spot the unnoticed jewels.

The only drawback being that this book is already 20 years old, perhaps there is something more advanced already? Suggestions welcome.