MEMO FROM ASD (INTELLIGENCE) TO NSC AND IC STAFF RE: STATUS OF STUDY OF REPORTING OF UNCERTAINTY IN INTELLIGENCE PRODUCTS

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP84B00506R000100150019-9
• Release Decision: RIPPUB
• Original Classification: S
• Document Page Count: 2
• Document Creation Date: December 9, 2016
• Document Release Date: June 8, 2001
• Sequence Number: 19
• Publication Date: October 3, 1975
• Content Type: MF

Approved For Rule se 2001/0909?EPF OO i o0,4110abs SFA
OFFICE OF THE ASSISTANT SECRETARY OF DEFENSE
WASHINGTON, D. C. 20301
lf~~ -4,j-
8 OCT 19('5
MEMORANDUM FOR Mr. Richard Ober, National Security Council
Intelligence Community Staff
SUBJECT: Status of IC Staff Study of Reporting of Uncertainty In Intelligence
Reference: Minutes of NSCIC/WG meeting dated Z July 1975
SD Declassification/Release Instructions on File
The NSCIC/WG originally sought Information on the reporting of
uncertainty data in current national intelligence products because serious
questions exist as to the reliability and meaning of many of the #iges
used in national intelligence estimates. The IC Staff has so far brfdfed IM
the.NSGIC/WG on the work it is doing to (a), study what the commnuii-r#y is
doing to improve part of its uncertainty data, or (b) conduct researeb -CD
into better methods of analyzing uncertainty. 7ttn
o
Mr. Ellsworth stressed in the meeting of the NSCIC/WG on JE3'uly Z
that the current IC Staff paper on expressing uncertainties in technical
judgements would answer only a small part of the requirement for P'AgresF,
in. this area.
This comment needs to be followed up. I am certain that you both.
agree that the lack of proper uncertainty data. represents a critical problem
in current national intelligence production. In far too many cases, consumers
have no way of knowing the reliability of the data they are given.
Further, the community seems to have failed to really address the
problem in rigorous analytic detail in many areas of key importance to SALT
and MBFR. Accordingly, when. it does provide uncertainty data, it t
difficult to understand what level of analytic effort went into the figures
or to place high confidence in the result. ers M
Soviet strategic yields, CEPS, and silo hardness are criticac
examples of such problem areas. So are Warsaw Pact military m ning,x(n
equipment strength, and readiness data. 0
-XI rPI
Classified b)r_... -__-.__M_
SUBJECT TO C NERAL DI;CLASSIFICA 'iOTI :>r;flr;i L1LE OF
ECtuTIVE OR,"ER 1162. AUT014ATICALL? D0WNGZADEb
`SP o YEAR 1hfil.iiV U S U~G1,Aa lr t3!,U usv.:.j__~~?_~~
Approved For Release 2001/09/03: CIA-R 15
 Approved For Relse 2001/09/040000T50g
Accordingly, I think it would be useful if, at the next meeting of
tit IhZSC'rG 'arGj. or. s~ei..);arnte staff. papor?.,_ the.l'c. Fir.-a..ff_c.OIJJd report
.. measuring the extent to which explicit quantitative state-
ments of uncertainty are provided for all figures shown in the NIT?, SNIEs,
and NIAMs.
L
Anthony H. Cordesman
S
w.a'~x vior t i.s
-wti , i s vCai%c t E:ra2
ODASD(R&?M) /rM
SECRET NO FOIIEIG?J UISSF.dq
- improving uncertainty data on force strengths and force
readiness as well as on technical data.
?~ prov .ding explicit summary statements of the major limitations
and uncertainties in the intelligence provided in NIEs. SNIrs, anci'N ,. .s.
correcting the grave problems which have existed in past
_ ff ) ' ]Y;+v A < ~' /t +.~' S{~ } n~~ J~~ y~r ~~~ 4 '~?^~'~r~ y v jii{
~~uL.?y ..~-w-fr f~?. it.iaY 'j ~+'* ~ ~.. 'Zy~~ mss. /} jt i ~~Y:? ~C~'~~?,t~ `cA~~.;rdA~~'..t..?'.f+St ~i ~.i /r~SGb L 1. ~,' - 1
A1JA 4~.. .i IvY.C' .w'~'-G akC ?v +~'' l.~i- ~ST?~~ ~
community
-personnel to replace loose adjoctivFti Cftr -rrat Ts Wirth i-r;,d-ern analytic
g~ c t,z~iiiu ~ s .
Approved For Release 2001/09/03 : CIA-RDP84B00506R000100150019-9
E