The $22 million radioactive paper trail

Researchers at the National Archives have been busy reviewing millions of declassified documents to find “inadvertent disclosures of classified nuclear weapons-related information.” The Federation of American Scientists Secrecy News blog has a good summary of a recent DOE report (PDF) to the U.S. Congress on the costs and progress of classifying historical and current data related to the U.S. nuke stockpile. Apparently, if enemies of the U.S. know how many nuclear weapons we had in 1967, we’re done for. 

The GWU National Security Archive breaks down the program on a cost-per-page basis:

Number of pages in the national archive reviewed for sensitive information since 1999:	204 million
Cost to review 204 million pages:	$22 million
Cost per page:	$.09
Number of pages found that reveal classifed data:	2,766
Cost to find one page containing classifed data:	$8,000

After the JUMP: What nuclear secrets lurk in the National Archives? 

According to the original report, the following types of restricted data have been found in the National Archives:

• Naval nuclear propulsion information
• Mass or dimensions of fissile materials
• Efficiency of nuclear materials
• Thermonuclear weapon design or function
• Tamper design and performance
• Other weapon designs or Restricted Data (RD) information
• Stockpile quantities
• Storage locations
• Nuclear weapon yield
• Weapon height and depth of burst
• Emergency disablement
• Other weapons or Formerly Restricted Data information 

The DOE report concludes that:

Information regarding older nuclear weapons is of significant value since it is often technically less sophisticated. These designs would be most readily used by a would-be nuclear proliferant to obtain its first nuclear weapon.

But William Burr of the National Security Archive counters that:

The effort to retrieve [classified] ‘RD’ nuclear weapons design information is understandable (although whether adversaries would actually have seized opportunities to find the needle in the archival haystack is a problem worth considering).”

Check out FAS for more analysis.

[Edit: corrected the blog post title]