In response to:

The Reporter Resists His Government from the February 19, 2015 issue

To the Editors:

Thank you for Steve Coll’s thoughtful review of James Risen’s book Pay Any Price: Greed, Power, and Endless War [NYR, February 19]. Underremarked in these discussions is that the overwhelming bulk of classification is needless.

As someone inside the system, my personal experience may not be representative. I see high levels of classification on activities that if known would mostly be embarrassing for lateness, technical and scientific shallowness, and inadequacy. (I once enjoyed having a distinguished guest scientist with appropriate clearances confidently tell me that something we were already doing, using commercial components, was impossible.)

Whatever my personal experience, there are two fundamental problems with classification. First, it is rarely if ever contested. (In thirty-five years, I know of only a single case, and that was because the program authorities were trying to make a Gaussian distribution—a bell curve—top secret. It still took six months of dedicated work to rescind the classification.) Any system in which restriction is almost never challenged is biased toward restriction. Even fewer constraints exist on use of “OUO”—official use only—which bars release of the marked material under Freedom of Information Act requests. Last week I saw it on a proposal to store experimental data in a database. I suppose that was another defense from embarrassment, that taxpayers shouldn’t know that we’ve been doing expensive experiments for decades and haven’t systematically saved the data.

The second problem is culture. A closed culture, where only those inside can decide who else to let in, and which in general treats its mission as quasi-holy, shares characteristics with a cult. If prestige is strongly associated with access and what you know, then your bias is to classify to that level, regardless of merit. You will see merit if sufficiently motivated.

At a meeting with senior scientists at a national laboratory, I was told that an information system I built to help international analysts identify a particular nuclear security problem couldn’t mention the problem the system was meant to help with. If it did, it would have to be classified. OK to tell the analysts everything they needed to know to assess and understand the problem; but not what the problem actually was. Which they could find on Wikipedia.

James Kornell
Santa Barbara, California