EXPERIMENTAL INTELLIGENCE PRODUCTS TO IMPROVE THE COMMUNICATION OF UNCERTAINTY

• Document Type: CREST
• Collection: Library of Congress
• Document Number (FOIA) /ESDN (CREST): LOC-HAK-558-14-4-3
• Release Decision: RIFLIM
• Original Classification: S
• Document Page Count: 2
• Document Creation Date: January 11, 2017
• Document Release Date: January 4, 2013
• Sequence Number: 4
• Publication Date: March 13, 1973
• Content Type: MEMO

No Objection to Declassification in Full 2013/01/04 : LOC-HAK-558-14-4-3
NATIONAL SECUR
SECRET/CODEWORD
MEMORANDUM FOR HENRY A. KISS
1355X
Information ~~-~ ah
3 c 9 3
SUB'J'ECT: Experimental Intelligence Promo Improve
the Communication of Uncertainty
Some while .ago I initiated a .program of experimental intelligence
products. The payoffs have been slim., but a few results are beginning
to appear. The purpose of this m.ernoranduxn is to show you some
samples of one particular line of ,products.
They have been designed to respond to:
-- An expressed view of yours that top level policy
making involves making complicated bets on th.e futuxe,
and that you wished .intelligence estimates were designed
to give you the betting odds.
-- Your saying that while you could not describe the
kinds of intelligence products you want, you would immediately
recognize them when you saw them.
Three products have been developed thus far: They are:
? -- The probability of Sino-Soviet hostilities in the
next twelve months. (See Tab A)
-- The outlook for hostilities to develop in Cyprus.
(See Tab B)
-.- A study .af the possibilities of a Canal treaty with
Panama in the next year.' (See Tab C)
SECRET/CODEWORD (XGDS) - 2, 3
DECLAS - Impdet
BYAUTH - H. A. Kissinger
No Objection to Declassification in Full 2013/01/04 : LOC-HAK-558-14-4-3
 No Objection to Declassification in Full 2013/01/04 : LOC-HAK-558-14-4-3
SECRET/CODEWORD
If you have time I recommend you look at these products as a first step
in responding to your needs. The. Sinn-Soviet study (Tab A) is the best
of the set.
The products:.
_- Attempt to produce estimates of betting odds.
-- Are put forward as the beginning of a survey of the
znaxket; to obtain answers to questions such as; Do you find
this new type of estimate useful? , etc.
The ,attached products 'are not fully developed. More tine thus far has
been spent in developing a mode of analysis that will produce estimates
of numerical probabilities attached to varfous possible fiatuxe outcomes,
than in developing alternative ways. in which such estimates might best
be communicated to top level policymakers. I now have the intelligence
people working an the latter problem.
I have tried the currently available products on other members of the
NSC staff. They have: '
-- Liked the idea of the numerical probability estimates.
-- Found the detailed probability trees intellectually
indigestible.
Do you find these products of any interest?
~~G,
Yes f
No
C oanment
Id'ave you any suggestions to make as to how to better display, or sum
marine the analysis results?
C omxnent
SECRET/CODEWORD
No Objection to Declassification in Full 2013/01/04 : LOC-HAK-558-14-4-3