The View From Olympus: On Intelligence

The Establishment’s latest hissy-fit over Donald Trump was sparked by his questioning of some assessments by the U.S. intelligence community (CIA, DIA, NSA, etc.). According to the Establishment, a president or other decision-maker must regard intelligence as hard fact. To do otherwise is to create a “crisis”. The January 6 New York Times hyperventilated on its front page,

Mr. Trump will have to say whether he accepts the agencies’ basic findings on the Russian role [in the U.S. election]–or hold to his previous contention that inept, politicized American spies have gotten the perpetrator of the hacking wrong. That would throw the intelligence agencies into a crisis of credibility and status with few, if any, precedents.

In fact, President-elect Trump’s doubts about the accuracy of our intelligence shows that he understands intel better than does the New York Times. Put simply, intel is never hard data. It is always a “best guess”, and history is littered with cases where it has been wrong.

The nature of intelligence is such that it is always incomplete. In any given situation, you do not know how incomplete it is. Further, some of it is always wrong, and you, the user, cannot know how much is wrong or what portions are wrong. Far from being hard data, intelligence is the world seen through a glass, darkly.

Obviously, the degree of incompleteness and the extent of intel’s error vary widely from case to case. If you have broken the other sides’ codes, your intel is probably more accurate than it otherwise would be–unless your opponent has realized you’ve broken his codes and is feeding you false information. But even when we were reading the traffic protected by the Germans’ Enigma machine in World War II, the Ardennes offensive of December, 1944 caught us completely by surprise. Suspecting we were reading their mail, the Germans kept their planning off the Enigma network. As is often the case, over-confidence in our own intelligence set us up to be surprised.

Intel is always incomplete and some of it is always wrong because of its nature. Gaining intelligence is a competitive action against a thinking opponent who tries to deceive you. Unless he is a complete moron, he sometimes succeeds. He either causes you to miss something entirely, or he fools you into believing something that is not so. His goal can be either making you uncertain, or making you certain but wrong; the latter is deception, which is hard to achieve but has big pay-offs.

As if all this were not enough to make intelligence a very squishy product, you next must consider the problem of bias. All intelligence agencies have biases, and those biases shape their findings. U.S. intelligence agencies are strongly biased toward telling a president what he wants to hear. Remember, the “findings” that Russia tried to elect Donald Trump were made under a president who sees Russia as an adversary. Then, the agencies are biased toward inflating the threat, because that supports their claim on more resources. Finally, their internal factionalism, such as the division between humint guys and photo interpreters, also creates biases and “filters” that distort findings.

If you put this all together, you realize that a president who is skeptical about intelligence products is probably going to be better anchored in reality than a president who accepts what the intel community hands him. If he is a clever president, he will develop his own sources of information beyond what the system gives him. Were I President Trump, I would read the Financial Times over breakfast every morning. When trying to figure out what is happening in places such as the Middle East, I might even pick up the phone to my good friend Vladimir and ask him, “So what are your intel folks telling you about this one?”

Admiral Rickover said, “You have to use the chain of command to pass your orders downward, but anyone who relies on the chain of command for his information is a fool.” As President-elect Trump has repeatedly demonstrated, he is no fool.