Missing: Four Million Pages of Secrets

by Alex Wellerstein, published May 7th, 2012

The indomitable Steven Aftergood at Federation of American Scientists reported last week that the National Archives and Records Administration has, well, lost nearly 2,000 boxes of classified documents:

More than a thousand boxes of classified government records are believed to be missing from the Washington National Records Center (WNRC) of the National Archives and Records Administration (NARA), a three-year Inspector General investigation found.

But there are no indications of theft or espionage, an official said.

An inventory of the holdings at the Records Center determined that 81 boxes containing Top Secret information or Restricted Data (nuclear weapons information) were missing.  As of March 2011, an additional 1,540 boxes of material classified at the Secret or Confidential level also could not be located or accounted for, the Inspector General report on the matter said.  Each box can hold approximately 1.1 cubic feet or 2000 to 2500 sheets of paper.

So that's probably between three or four million classified pages that have just gone missing. Whoops.

(An Anecdotal Aside: The Washington National Records Center is in Suitland, Maryland, just over the South-East border of DC. It's not a great part of town. I once called them to see about doing some research down there, and they basically told me that it would be in my best interest to find some way — any way — to avoid actually going out to their facility. I didn't get the feeling they were trying to keep secrets; it was more like they were trying to avoid any bad headlines, e.g. "SCIENCE HISTORIAN STABBED ON THE WAY TO THE ARCHIVES – WAS DEVOTED TO WORK, ECCENTRIC, SAY FRIENDS." But like all places with bad reputations, there is probably a little exaggeration for effect here. And strictly speaking, K Street and its environs probably transacts a much higher volume of illegal activity.)

So, what to make of this? The first thing is to just state the likely and boring explanation: these boxes are probably just sitting on a shelf in a government archive, somewhere. They were probably moved — or not moved when they were supposed to — and someone lost the tracking information, or entered it in incorrectly. Or checked the "feel free to burn these records or send them to the dump" checkbox by accident.

The tracking of this kind of historical data is still pretty low-tech, and I can speak from a little experience on this point. As part of a graduate fellowship I had with the Department of Energy awhile back (I was the "Edward Teller Graduate Fellow," a title which I relished), I helped them improve their internal holdings database, which is just a big Microsoft Access database. It's not a bad database, to be sure, but it isn't some super-advanced NSA creation, and it can be pretty cryptic when you are trying to figure out what they have and where they have it.

You remember that scene from the end of Raiders of the Lost Ark, where they "hide" the Ark of the Covenent in that giant government warehouse? It's not entirely far-fetched, though the NARA stacks that I've seen are a bit cleaner looking, and have lower ceilings.

As the Library of Congress' web system likes to say, "Item not on shelf."

Let's banish from our minds the idea that terrorists or criminals are somehow trucking around archival boxes. Doesn't happen. These things don't usually contain stuff that is that interesting to ne'er-do-wells. It takes a lot of work to find the occasional interesting document, take it from me. It's just far-fetched. Activists and journalists and maybe even an historian or two? It's been known to happen. But people who want to actually do bad things to the fellow man? I wouldn't bet on it.

That doesn't mean, of course, that these missing records are actually being stored correctly, or safely, or securely, or in the manner which is mandated by the law. But I seriously doubt there's any great matters of security at stake. Especially since, as Steven points out, most of that stuff is pretty old, and probably should have been reviewed and declassified a long time ago (which requires resources — money, time, will, etc.).

But this isn't likely a security problem. It's not necessarily even an organizational problem. It may just be a complexity problem. If you have enough records, you're going to lose track of a few of them.

How many is "a few"? It's actually a quantifiable number. The problem is known as inventory control: how much physical stuff (in this case, boxes full of paper) can you actually control at any given time? In various industries and contexts, studies have shown that losses and misplacements have real minimal limits. I don't know what they are for records of this sort — in controlled facilities — but every place has an unavoidable loss rate. This is the rate of things going missing that happens even if you have extremely advanced inventory control systems — sophisticated databases, tagged items, really elaborate systems of checking and re-checking inventory. Because there are fundamental limits on how much stuff any organization can actually keep track of.

The NARA College Park facility has at least 3,340 boxes of materials in Record Group 326 (Atomic Energy Commission) alone. This is just one agency, and a lot of that agency's files are still held at the Department of Energy archives in Germantown, or at other NARA facilities. (This count is taken from RG 326's Master Location Register, which, I might add, as I look it over again, has quite a few "item not on shelf during last inventory" notes in it, many for classified entries.)

So let's imagine that the unavoidable loss rate is something that feels quite low, like 1-2%. That's 33-66 boxes out of this one Record Group, many of which are classified to one degree or another. (Is the loss rate per year or a constant? I don't know, but either way, it adds up.) Let's imagine that the loss rate is higher than the unavoidable rate, because people are using these records, the organization is not as perfectly well-oiled a machine as possible, or the record-keeping is sloppier than it ideally could be. What if the loss rate is as high as 5%? That's 167 boxes.

Now multiply that across the entire organization, all of its records, all of its secrets. Quite a lot of boxes. So 1,540 boxes doesn't quite surprise me, nor does it look all that surprising that you'd run into that sort of problem.

I'm not trying to exculpate anybody. It isn't like this fact makes losing classified records right. But it does make it expected. 

FBI filing room, 1944. Imagine how much paper 1% of these records adds up to.

There's a nice quote about excessive secrecy from Michael Hayden, former chief of the CIA and NSA, which is reproduced in Dana Priest and William Arkin's Top Secret America:

"This is just a reflection of complexity, not any vice."

This jibes with my own feeling on the matter pretty well, and likely applies well to this specific case. This is what happens when you have billions of (physical) pages of classified information accumulated over the course of the last seventy years or so. You inevitably lose a few of them. And "a few" of a billion is a lot

The only way to avoid losing track of thousands upon thousands of boxes of secrets on a regular basis is to not have as many secrets to lose track of.

If they had already declassified these boxes, it'd be a minor thing — the only people who'd be really upset are historians, and we all know how important a lobby they are. But they didn't, and now their loss is now a big, huge, ugly deal. Expect FBI involvement, expect Congressional hearings, expect a lot of man-hours spent tracking them down, expect a lot of finger-pointing and maybe even some finger-wagging.

7 Responses to “Missing: Four Million Pages of Secrets”

  1. Ologe says:

    When the whole “wikileaks US diplomatic cables” thing happened, I remember Hillary Clinton saying government business is impossible without the guarantee of secrecy. I agree, but why so many? Billions in 70 years is millions per month!

    Does this speak of a problem with the project of governance? Or is it just one of those “it’s not as bad as it seems”? Or is it a case of “you can’t understand it unless you’ve been there”? I certainly hope its one of the latter two.

  2. Tim says:


    If you took this post and substituted the words ‘document,’ ‘record’ and related terms with ‘nuclear weapons’ you’d be making an argument almost identical to Scott Sagan’s in ‘The Limits of Safety.’ I don’t know if this was intentional or not, but either way it is a neat exercise in ‘comparative bureaucracy.’



    • Tim: It’s a wonderfully adaptable concept when you have things that you’d rather not lose. I first became acquainted with the literature on inventory control through concerns about “MUF” (Material Unaccounted For) in reprocessing plants. If 5% or so of your plutonium is inevitably going to be “unaccountable,” once you start multiplying that by large numbers of kilograms, you quickly end up with many Significant Figures of plutonium per year. Cheery!

  3. Mike Lehman says:

    Hmmm, excessive secrecy not a vice? That’s an interesting argument to make, given the incredible costs, the often questionable basis, and the government’s general tendency to ignore their own rules when it comes to secrecy and declassification.

    And that doesn’t even touch use of secrecy to cover up bad policy or just stupid stuff.

    I will contribute my own anecdote about a visit to NARA about a decade ago Given my own proclivity for certain aspects of Air Force intelligence, I happened to meet with as much of a foot in the door as a historian without a clearance can get, which was to politely inquire if I at least had the RGs of interest about right. I knew what the answer to my inquiry most likely was, so wasn’t taken aback at being told that “someday” was the best idea they could give me about potential declassification timelines.

    “And that doesn’t include all the stuff the Air Force came in and took back from us right after 9/11…” is basically a paraphrase of what I was told about the situation, perhaps just to reinforce the hopeless nature of my quest. Maybe in the Air Force’s haste, they grabbed stuff back and didn’t document their hasty withdrawals well? That’s probably more likely than some spy getting at it, as already noted.

    Your recall of conditions in the stacks is interesting: “(This count is taken from RG 326′s Master Location Register, which, I might add, as I look it over again, has quite a few “item not on shelf during last inventory” notes in it, many for classified entries.)”

    Obviously no way to know if there’s a connection, but does make one wonder how much of that went on and if it was all properly documented. After all, as you point out and someone else said perhaps only a little more eloquently, we have met the enemy and he is us when it comes to secrecy.

    Interesting tidbit: Before the Bomb, secrecy of this scale in the government was most commonly associated with signals intelligence. Once the AEC set the standard with nuclear information, then every other potential user of secrecy in government was soon comparing their own situation in terms of their ‘secret powers’ to those of their nuclear cousins. Whose is stronger? Is ours ‘almost’ as secret as nuclear data? Massive secrecy envy. The Secrecy Race operated in parallel with the Arms Race it supported. Good thing we’re mostly better at keeping track of the hardware than the paperwork of the Cold War.

  4. […] to figure out what each site was doing that led to these different assessments. The problem of Material Unaccounted For never really goes away, but it’s interesting that it shows up this early in the game. […]

  5. Canageek says:

    I wonder what would happen if you brought over the inventory control from a company that handles a lot of inventory; Fedex, Costco, something like that. RFID tags, even GPS locators, scan in/scan out systems….