That’s what happens when you try to run the world while excluding 99.8% of the population

Daniel Immerwahr writes:

Of the C.I.A.’s thirty-eight Soviet analysts in 1948, only twelve knew any Russian.

Whaaaaa?

There were a lot of Russian speakers in the U.S. in 1948. From Immerwahr’s article, my impression is that the problem was that the CIA was restricting itself to upper-class Ivy League types—and not many of those people knew Russian.

Assuming the story of the 26 out of 38 non-Russian-speaking analysts is correct (it’s so hard to believe, maybe the reporter got this one wrong?), wow. An amazing example of the narrowness of America’s ruling class.

Can we put a number on it? I did a quick google to see how many students were graduating from the Ivy League at that time. I couldn’t find any convenient table or graph, but I did come across this news article from 1950 saying that Harvard was giving out 1144 undergraduate degrees an that spring. I can’t readily find the numbers from Yale and the others, but let’s just say that there were roughly 5000 Ivy League graduates that year. To compare to the general population . . . 2.2 million Americans were born in 1930. So, roughly (only a rough approx because I’m excluding foreign-born in the denominator), if you restrict your recruiting to the Ivy League, you’re only targeting 0.2%, or 1/440th of the population. And that other 99.8% is where most the Russian-speakers are hanging out.

OK, those 99.8% would’ve included lots of incompetent people. The CIA couldn’t just hire at random; they’d need to do some interviewing. But the 0.2% seems to have contained a fair number of incompetents too! Maybe screening based on social class and grades in school wasn’t such a great idea.

I’m reminded of some procedures in statistics, where researchers screen their results using strict statistical significance thresholding based on noisy data. A big selling point of this approach, beyond its apparent guarantee of rigor, is that it has enough researcher degrees of freedom that you can get whatever result you want out of it. The other selling point is that you can take what you find as having great scientific merit, as it has survived this very difficult selection process.

Kind of like hiring an Ivy League graduate in 1948. He may not speak Russian, but he’s one of nature’s noblemen.