HANDBOOK ON NATIONAL INTELLIGENCE ESTIMATES

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP84-00022R000200030014-5
• Release Decision: RIFPUB
• Original Classification: K
• Document Page Count: 6
• Document Creation Date: December 9, 2016
• Document Release Date: June 29, 2001
• Sequence Number: 14
• Content Type: MEMO

Approved For Release 2001 /09/0 ! fir- d ?-20~001 4-5
llheCoco r(gi;,Np'6I"tirll 4M
tinder the HISrO-;;CAi E-sr ~ t y~ ' t;4" f t'
R ) a M P 1 E -OF H O -At- - I # A 1 ~ A B , L r I M T t~PEP~ _ Ms d a been
C/ the t- - . ' , L r I M
IML of
estimates are written for: the Gwtral 'ttel:tiq Agency.
The President of the United States
The Secretary of State
f
aV-0 .,thase--people read any or all reports is
Aatp V~ 11~ ?r
no ef
;
Whether or not 1i_,
generally unknown to CIA and in any case is beside the point. Because They
constitute the distribution list, and since they make up the bulk, of those
who must have good intelligence in order to provide for the security of the
country, it should be assumed by CIA that they person y read its estimates.
To assume otherwise, would be to undervalue the estimates and thus-be tend to
vitiate them.
If top-level readers are to be the eaft target of estimates, it is
necessary to consider what sort of presentation best suits their needs. From
all that can be gathered, somethin g like the following can be assumed
regarding this problem:
1. The Readers are extremely busy people and either cannot or be-
lieve they cannot read etended studies, however important the matter
Approved For Release 2001/09/03 : CIA-RDP84-00022R00020003001~4-E 6
 Approved For Release 2001/09/03 : CIA-RDP84-00022R000200030014-5
covered. For this reason, they are. partial to summaries as has been s hown
costa--in their comments and requests. It can be inferred that a reader
who insists on a summary will be unlikely to read anything beyond the
summary unless tkztz what follows is itself very brief or concerns matters
of exceptional importance to the, in id reader. From the considerations
above have arisen not only the invariable inclusion of summaries.in CIA
reports, but the,tendency to treat subjects in a way that many find illogical,
with conclusions and forecasts given first instead of'last where"one would
normally expect them. This reverse order likewise relfects the desires of
people who can usually not afford to read beyond the minsmum they must know.
2. These poeple are in general interested in the conclusions reached
by CIA, not the processes by which those conclusions were reached. They
furthermore---unlike many readers of similar studies published in other
connections---do not demand all the evidence which supports the conclusions.
They are entitled to assume that, because all evidence available to the
Government is available to-CIA, and since CIA analysts and,.the best available
in the United States, conclusions reached in the papers will either be time
or at least as near the truth as it is possible for the Goverhuient to get,
Approved For Release 2001/09/03 : CIA-RDP84-00022R000200030014-5
 Approved For Release 2001/09/03 : CIA-RDP84-00022R000200030014-5
`t _T_
&t.th=time. Should CIA conclusions prove consistently wrong, it is to be
assumed that the Readers, rather than beginning to call for supporting
evidecni would=-speakng mildly---take steps to reorganize CIA. Readers
furthermore, must assume that behind the conclusions reached bSm?HA are all
the supporting facts, and that these facts could be made available at any time.*
zxft The participation of the IAC agencieat in the process of pre-
an
paring CIA reports, though estabilshed by NSC dtrective, is/incidental
rather than an essential feature of production. Though CIA may hope for
Agency concurrence and deplore substantial dissent, it 4 ld not become a
major effort to avdid the latter. Dissents can occur only when CIA believes
itself right. In such a case, a contrary opinion, while interesting, is hardly
devastating. The,dissent merely serves one of the purposes of coordination:
to allow a"&'intelligence"h'g*v*4 an opportunity to state divergent views
if bt so wishes. Whether or not the presence of dissents will incline Readers
primarily
to doubt CIA estimates will depen44 not on the frequency of of Agency
non-concurrence but on how, often CIA proves to be wrong.
It should be emphasized in this connection that NSC directives compie
Approved For Release 2001/09/03 : CIA-RDP84-00022R000200030014-5
 proved For Release 2001/09/03 : CIA-RDP84-00022R000200030014-5
,1
~.Z
Agency Agency
dmp at coordination, not dmpamtmomt concurrence. rYor this among
other reasons, analysts should never consider themselves as writing primarily
u
for the Agencies, sz in any way to service the A encies, or certainly to
gain the assent of the Agencies.
*Somewhat parallel case: the average reader of a newspaper looks at the
weather report in the upper left hand corner of page 1, which is a pure
conclusion; the exceptional reader, who happens to be interested in meteorology,
looks at the map and discussion on page 50; one in a million might avail
himsdaf of tke all supporting data which may be found in the Weather Bureau.)
4.0
tin )-"+' t~k
 Approved For Release 2001/09/03 : CIA-RDP84-00022R000200030014-5
Main elements of an Estimate:
1. conclusions. These will not be stated as such (i.e. in a section
trailing along the end marked "Conclusions")but should be draws first, before the
estimate is written. If an estimate is to be written, it is assumed that the Analyst
has before him a mass of evidence in various forms whieh which leads inescapably,
to conclusions which would be relevant to: (a) the formation of US policy; (b)
exxxexaa a warning of dangerous impending events; (c)' consequences arising from
actions undertakne or not undertaken by the US or a foreign power; (d) analyses of
situations which need to be clarified for a proper undertanding of world events;
(e) etc
The conclusions reached from the evidence, actually constitute the Estimate: a1Lx
other material appearing in the written paper may be justified only, because it
supports the conclusions or serves to clarify them.
Approved For Release 2001/09/03 : CIA-RDP84-00022R000200030014-5
 Approved For Release 2001/09/03 : CIA-RDP84-00022R000200030014-5
2. What constitutes_an_ORF,. First, any subject required of CIA by compbtent authority
(e.g. the NSC.) Such n subject automatically becomes an IM, an ORE, or both.
Otherwise, any subject which (a) comes under one of the headings in 1 above;
(b) should definitely be brought to the attention of the President; (c) represents
a problem capable of'solution, either through action or change of policy.
Point (b) above will be questioned. It is believed, however, that since
CIA was created primarily to furnish intelligence for the President of the United
xStates, and since he.are-n +tbep . ne..-o# all important distribution lists, it would
be a mistake to publish u;jless the paper, if he read it, would be of hse to him.
Whether or not he will read all or a part of any given pa study must be con-
jectural, but it -be--e:-ml taka'ta-'-assume ~enyth rig .else..than that t".,e' President
.w ..read every word e:fzan7zg1xexzstx'7 if for no other reason that that he might.
~~?~#,bat,_..havervrr;~zt~z~"z~a~xt~?s~ _".it. is-logics]: ta -suppose that to-, aim
lower _ixi .i ,tell. gene. production- d;_. end.;to,.,leaaen i~~aex it~c its quality.
Point (c) may be considered variable; yet in the last analysis, there is usually
little use in sending axnwpaxt out a publication which states a problem about which
nothing can be done. It is not, of course, the business of CIA to tell the G9v-
Approved For Release 2001/09/03 : CIA-RDP84-00022R000200030014-5
ernment what to do about its problems, but the Government should be able to derive