Snowden Archive
——
The SIDtoday
Files
Browse the Archive

Letter to the Editor: Using Scientific Methods in Analysis

SUMMARY

How can analysts share arguments justifying conclusions in order to avoid incomplete inferences like the Iraq WMD example? A technical director suggests using the scientific method by sharing "how we know what we know" articles to the SIDtoday.

DOCUMENT’S DATE

Nov 17, 2006

PUBLICLY AVAILABLE

May 29, 2019

1/1
Download
Page 1 from Letter to the Editor: Using Scientific Methods in Analysis
DYNAMIC PAGE -- HIGHEST POSSIBLE CLASSIFICATION IS TOP SECRET // SI / TK // REL TO USA AUS CAN GBR NZL (U//FOUO) Letter to the Editor : Using Scientific Methods in Analysis FROM: the SIDtoday editor Unknown Run Date: 11/17/2006 (U//FOUO) Here's a letter in response to the article A&P Unveils Strategy for FY07-FY13 : Comment: (U//FOUO) Carl Johnson/SID's Deputy Director for Analysis & Production (S2) stressed in a SID today article that increased efforts must be made to "put analysis first" and ensure that tools, techniques and knowledge become readily available to the analytic workforce. (S//SI//REL) Well, since its inception in the mid-20th century, NSA has largely reported intelligence based on transcription of foreign-language voice and facsimile, telling our many and varied customers what we know from classified sources. To be sure, the arduous process leading to product reporting indirectly helped ensure consistency on what ultimately bore the stamp of NSA. Now, in this post-9/11 era we are continually asked to provide deeper arguments justifying our conclusions, mindful of incomplete inferences like the "slam dunk" Iraq WMD example. The scientific tradition may have something to offer here. (U//FOUO) Those of us who work in mathematics, signals analysis, engineering, algorithm development, computer science, or physics occasionally contribute a piece of the puzzle worth reporting in its own right, not only to strengthen intelligence conclusions but to also explicitly outline what drives our claims. Scientific methods have stood the test of time for half a millennium, and their entrance into formal reporting can only improve our product. (U//FOUO) But what should be the vehicle? Last month an e-mail from Thomas Fingar/DDNI for Analysis , encouraged members of the IC to " share analytic insights as early as possible in a virtual workspace where analyst contribution is evaluated based on ideas presented. " (S//SI//REL) Presenting technical articles to SID today -- part of SID's "virtual workspace" -may well be a good way to start. This medium, only available via our secured intranet, offers opportunities to share "how we know what we know." Admittedly, Oversight and Compliance issues may need to be addressed, but hopefully not at the price of stifling originality, independence of thought, and, yes, even technical controversy. Let the marketplace of ideas judge relative merit. (U) Science has made much progress by working seemingly paradoxical evidence, even though Nature neither conceals nor reveals. The same cannot be said of our targets, making our Indications & Warning mission incalculably more difficult; all the more reason to present technical arguments supporting inference in an open, transparent, accountable -- and protected -- medium. From: Adolf Cusmariu ) "(U//FOUO) SIDtoday articles may not be republished or reposted outside NSANet without the consent of S0121 (DL sid_comms)." DYNAMIC PAGE -- HIGHEST POSSIBLE CLASSIFICATION IS TOP SECRET // SI / TK // REL TO USA AUS CAN GBR NZL DERIVED FROM: NSA/CSSM 1-52, DATED 08 JAN 2007 DECLASSIFY ON: 20320108