Skip to content

The SNAFU Principle

Back a good long while ago, I was closely involved with The Intelligence Community, which contrary to popular belief isn’t really named by opposites. It is, however, a massive multiple-agency bureaucracy with roots in all the individual services, the Justice Department, and the Executive branch of FDR and Harry Truman. (And yes, the actual administration is important; government organizations acquire structure and organization based on who start them. The Department of the Navy is still clearly influenced by the organization at the time of John Paul Jones.)

It’s a marvelous example of my ongoing focus of interest: Big Organizations That Act Like Idiots. We will have plenty of examples drawn from the open literature on the intelligence business.

One of the things that people outside the business don’t really understand is that intelligence agencies never tell you the answer. Instead, you have to take a hint from Tolkien:

“Go not to the CIA for counsel, for they will tell you both no and yes.”

The question might be “why”?

Certainly what I observed was this: the actual intelligence collection goes on, and it’s relatively objective. You have a radio intercept that tells you that the 3rd Battalion of the 2nd Army is moving from A to B. All of these snippets are written down and transmitted back to the analysts, usually at CIA, where they are entered on 5×8 index cards. (Or the logical equivalent. When I got started with this business it really was literal 5×8 cards, but I understand they have more computerized methods now.)

But then, all these intercepts, along with reports from human spies, information culled from newspapers, dispatches from diplomatic posts, and so forth. These snippets are read and digested by analysts who are usually people with degrees in political science or history, along with some people with specific technical skills or other useful background, like military officers. The output of this process is that the analysts, every day or so, write a term paper on their area that goes to a senior analyst, who uses that information to summarize a higher level report. This goes up the hierarchy, being further summarized at area desks, in task forces and tiger teams, until it becomes input to various very senior intelligence people — like the Deputy Director for Intelligence, the Director of the Central Intelligence Agency, and now to the Director of National Intelligence. At each level, it’s summarized, reviewed, re-summarized, glossed, and rewritten until, without exception, it has two qualities:

  • it says nothing that can be proven wrong;
  • but it says exactly what is perceived to be what the boss wanted.

What’s more, when it’s presented to the Executive, it will then be shaded and summarized further by the presenter to more closely reflect what the presenter thinks is most advantageous. To that presenter.

I didn’t really understand this until, years later, I read a book by two guys who became my friends. This book is called Illuminatus!, originally published as the Illuminatus Trilogy in 1975. Shea and Wilson defined an illuminating notion called the SNAFU Principle, which they stated as “communication is only possible among equals.” The basic idea is simple: when two people do not perceive one another as equals, they will slant their communications — unconsciously and without any particular volition — to protect themselves in that disparate relationship.

So the analyst writes unconsciously to fit with the general assumptions, and the senior analyst favors things that confirm their own positions, and so on.

Wilson and Shea were both writers, and basically literary people; their summary is literate and verbal. But, being a computer scientist and at least somewhat literate in signal theory and information theory, it struck me that the SNAFU Principle wasn’t sufficiently precisely stated. It should be restated as followed:

“In any social hierarchy, the noise added to a communication between individuals in that hierarchy is directly proportional to the distance between them, and the factor of proportionality will be proportional to the perceived risk to them.”

This, finally, because to make things make sense. Let’s look at this restatement in some detail.

First, “In any social hierarchy….” The most important thing to observe is that to a first approximation there will be a social hierarchy established between any two people who are interacting. We very very rarely see another person as an exact equal, although we do sometimes find ourselves evaluating another as an effective equal. (So, for example, my friend Anil is a Professor of Biology; my area is Computer Science. Even though I know a fair bit of biology, and especially physiology, I defer to Anil on those topics, and vice versa.) Since we have been close friends for decades, we can maintain a relationship as equals and defer on to the other without friction on most topics. On the other hand, it’s very difficult — as any teenager will tell you — to speak fully and frankly to your parents or to a teacher.

Second, “… the noise added to a communication between individuals in that hierarchy is directly proportional to the distance between them….” “Noise” is the technical term, but it means more or less what you think it does: its what makes it more difficult to get the sense of some message. It’s literally the opposite of information; if you’re in a noisy bar trying to talk to a beautiful girl, the band that drowns you out is noise; on the other hand, if you later accompany the beautiful girl to a concert, and she insists on talking through the soloist’s performance, then what she is saying is noise.

So, this part really completes the original statement of the SNAFU Principle: complete communication is only possible among equals, because otherwise there is noise added that impedes communications. But there is a second observation to this that I added.

Consider if you are talking with someone to whom you feel significantly inferior in some hierarchy. In fact, consider if you work for GM as an assembly line worker, and in the presence of your boss, and her boss, and her boss’s boss, the President of GM asks you if everything is fine in your workplace. You might, depending on the union rules and such, say “No sir, we need more time off and higher pay.” You’re a lot less likely to say “No, sir, my boss is sleeping with her secretary and her boss is jealous, and it’s making life hard for us.” One answer is legitimate, and in a large corporation with a unionized workforce you’re unlikely to get a lot of repercussions. The other may be equally true, but it’s much riskier, and so it’s something you’re much less likely to say, even though in all likelihood it’s of more interest to the President of GM, and very probably presents a lot greater risk to the company.

What this means is simple:

  • the more hierarchy there is, the less likely it is for someone high in the hierarchy to be getting good information from the bottom levels;
  • the more risky it is to talk across levels, the worse the problem will be.

[Yes, it’s updated. I don’t know why I forgot that last line.]

{ 4 } Comments

  1. Wingborn | 2008-Jan-29 at 09:37 (@442) | Permalink

    Charlie, I think you’ve invented a new acronym:

    Big Organizations That Act Like Idiots.



  2. Charlie | 2008-Jan-29 at 09:50 (@451) | Permalink

    Thanks, I like it and I’ve added it as a category tag.

  3. Bernie | 2008-Feb-13 at 01:12 (@092) | Permalink

    Beautiful post. It has changed my thinking.

  4. Charlie | 2008-Feb-13 at 13:46 (@615) | Permalink

    Glad you liked it Bernie. Come back again.

{ 3 } Trackbacks

  1. […] it all in his head.” Communication becomes more difficult up and down the hierarchy (see the SNAFU Principle, below) and as more and more people join the endeavor who were not there from the start, the focus […]

  2. Carl’s Corollary | Explorations | 2008-Jun-15 at 19:50 (@868) | Permalink

    […] SNAFU Principle: “In any social hierarchy, the noise added to a communication between individuals in that […]

  3. […] Situation Normal, All Fucked UP […]

Post a Comment

You must be logged in to post a comment.