STATUS REPORT - DEVELOPMENT OF REGULATIONS RELATING TO CIA RELATIONSHIPS WITH THE ACADEMIC COMMUNITY

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP80M00165A000800080001-8
• Release Decision: RIPPUB
• Original Classification: K
• Document Page Count: 60
• Document Creation Date: December 15, 2016
• Document Release Date: March 8, 2004
• Sequence Number: 1
• Publication Date: July 1, 1977
• Content Type: MF

Approved For elease 2004/03/23 : CIA-RDP80M001 00080
DDA 77-3802 Kf ~- /'
JUL 1977
MEMORANDUM FOR: Deputy Director of Central Intelligence
FROM John F. Blake
Deputy Director for Administration
SUBJECT Status Report - Development of Regulations
Relating to CIA Relationships with the
Academic Community
1. In your memorandum of 29 January 1977 dealing with
the IG Survey of the DDS&T, you stated that the DDA should
establish a task force chaired by a DDA representative and
composed of representatives of the concerned Agency compo-
nents to deal with the subject of developing a consolidated
CIA regulatory issuance governing the Agency's relationships
with the academic community.
2. Such a task force was established under the chair-
manship of of my staff. This task force
began meeting in February 1977 and dealt with the defini-
tions and scope of activities which it would be proper for
the proposed Agency regulation to cover. It was determined
that even these areas proposed significant problems and that
our progress toward developing a consolidated regulation
would be slow.
3. Subsequent meetings were held and :Limited progress
was made up until the point when we were advised that Harvard
University was proposing a set of formal. guidelines to deal
with the relationships between Harvard and CIA. At this
point, many of the members of the original task force were
asked to participate in dealing with the proposed Harvard
guidelines. It was also generally agreed that the original
task force should pause and await the outcome of the negotia-
tions between Harvard and CIA regarding the guidelines so
that we could incorporate into our basic Headquarters regu-
lations any policies that were developed.
L ?%T
IT'sIlRy R1
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
4. As you are aware, the Harvard guidelines affair
took more time than was anticipated. As a result, the task
force was delayed. Now that the Harvard matter has been
dealt with, we are picking up the pace. The task force met
during the week of 20 June and is scheduled to meet again
on 6 July to discuss a proposed regulation which has been
drafted. We have established 29 July as a goal to'present
to you a coordinated Agency regulation dealing with this mat-
ter. We are aware of the external pressures which relate to
this matter and the need for timely response. Nonetheless,
we feel,that to shorten the deadline would increase the risk
of an inadequate regulation being provided for your considera-
tion.
j 4p,ho F, Blake
John F. Blake
AI/DDA do (1 .July 1977)
Distr' u on:
Original? Addressee
vl' ER
- 2 -
Approved For Release 2004/03/23 : CIA-RDP80MOO165A000800080001-8
 Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
Next 3 Page(s) In Document Exempt
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 25 August 1977
CIA ]RELATIONS WITH ACADEMIC COMMUNITY
In May of this year, Harvard University published guidelines for
relationships between the University and intelligence agencies. In brief, the
guidelines state that:
? the existence of any CIA-university contacts must be public
? private consultation arrangements must be reported in writing to the
dean and president
? CIA recruiters must be identified to dean, president, and placement
office in writing
? Harvard community members may not volunteer names of other
members without their permission
? Harvard community members should not undertake intelligence
operations for the CIA.
In correspondence with President Bok of Harvard on this subject, I have
made the following points:
"... American scholars who have been willing to share information
and interpretations of developments in the international arena often have
contributed valuably to intelligence support of the U. S. foreign policy-
making process. Without the continuing assistance of the academic
community, our ability to provide the President and other senior officials
with objective and enlightened analysis and estimates would be hampered.
I believe strongly that in this increasingly complex and competitive world
it remains in the best interests of both the academic and intelligence
communities to expand and refine their contacts in a spirit of mutual
"
respect and understanding.
" .. Current CIA policy covering our relatio t~A,~,e
and i adb~redcFohRel~sjerS004/.Q3/2l::-DIA:FZDP80rg0w
airea
fi M 'T-8
y, to a. large
 Approved For Release 2004/03/23: CIA-RDP80M0d'f65A000800080001-8
degree, consistent with the Harvard guidelines. Present Agency policies
may be summarized as follows:
All of our contracts with academic institutions are
entered into with the knowledge of appropriate senior
management officials of the institution concerned.
All recruiting for C [A staff employment on campus is
overt.
It is against our policy to obtain the unwitting services
of American staff and faculty members of U.S.
academic institutions."
" .. I take exception to the provision in your guideline which requires
your faculty members to report such arrangements in writing to the dean
of their faculty.... I believe that attempts to regulate the private lives of
our citizens in a manner discriminatory to any particular group, profession
or segment of society poses serious risks. I believe that we would be far
safer not to single out any group, despite what may be transient
enthusiasm for so doing. In point of fact, it is our policy in these cases to
suggest to individual scholars that they inform appropriate officials at
their universities of their relationship with CIA. Frequently, however,
scholars object to advising any third parties on the understandable
grounds that to do so would violate their constitutional rights to privacy
and free association and possibly expose them to harassment and damage
to their professional careers.... Thus, the decision on whether to advise
their institution of a relationship with CIA is left to the discretion of the
individual. We intend to continue respecting the wishes of individuals in
this regard."
This issue, of course, transcends the relationship with academics. All
American citizens must continue to have the freedom to choose whether or not
they want to cooperate with any government agency, and, if they choose to
assist the CIA in its work, we must be able to ensure the confidentiality of that
relationship.
PROPOSED INTERNAL REVENUE ACTION ON
ALLOWANCES
I share the concern of our operating components and the employees
overseas with respect to the possible repeal of Section 912 of the Internal
Revenue Code, which would mean that allowances paid to employees at
foreign posts would become taxable income to them. The matter of employees'
entitlements has been discussed at the President's Cabinet meetings, and I
assure you that the agencies and departments concerned are acutely aware of
the implications and the impact it can have on morale. The Secretary of State
and I have sent letters to the Secretary of the Treasury to urge that the
pr i e~fFoS~ asl~ l~Dot to3, in6ti4-I M?P1t Q0~$4 4At-$been
 Approved For Release 2004/03/23 : CIA-RDP80M00T65A000800080001-8
disseminated to the overseas posts and to the Headquarters components which
support them, and I assure you that the Agency will do whatever it can to
preserve the entitlements of our employees overseas.
STANSFIELD TURNER
Director
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
COMMENTS ON
THE GEOMETRY OF THE ARMS RACE
BY BRAMS, DAVIS AND STRAFFJCN
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
The paper, "The Geometry of the Arms Race," by
Brams, Davis and Straffin, is interesting and is indicative
of an important subject. Unfortunately, it is of limited
practical value as it stands. However, it does seem to be
a useful vehicle to focus and stimulate interest that could
ultimately lead to results of practical significance.
The following discussion indicates the deficiencies in
the results as presented in the paper.
Criticism of the paper falls naturally into three
categories: mathematics, assumptions, and conclusions.
The categories are discussed in that order. The mathematics
in. the paper is correct. With the single exception of the
game theoretic payoff matrix presented in Figure 1, there
are no errors in the mathematical concepts or the derivations.
In Figure 1 there is an interchange of subscripts, and the
Disarm
A
Arm
[A2IB2
A4Bl
[_AlB4
A3,B3
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
In the body of the paper, the subscripts always appear
correctly. Hence, the derivations are correct.
Certain assumptions are presented which. link the
physical reality of arms races to the mathematical formal-
ism of Prisoners Dilemma game theory. Some of these
assumptions are made for convenience, and can be easily
modified without changing the fundamental results. Other
assumptions are a necessary part of the main theme of the
paper. The assumption that both parties have the same
detection probability, p, is a matter of convenience. Un-
equal probabilities, PA and pB' could have been retained
throughout with only an increase in complexity. The assump-
tion that each is greater than 50%,-however, is more funda-
mental to the paper and deserves some attention. While it
is certainly the case that a flip of a coin will result in
p = .5, it does not follow that this lower bounds the de-
tection probability. A rational player will only flip a
coin if he knows that his detection probability is less
than 50%. In a covert armament program, however, player
B might force player A's probability well below .5 while
player A still believes it to be well above .5. Thus,
the game becomes one not only of imperfect information but
one of misinformation. This changes player A's views on
the security level achieved by a policy decision.
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
Another assumption which is central to the paper is
that each side is willing to accept an expected value
criterion of the worth of a policy. While this assumption-
can be changed to, say, a minimax criterion, such a change
makes the mathematical conclusions of the paper inapplicable.
Perhaps of most importance, however, is the main as-
sumption of conditional cooperation of the two players,
the "I will disarm if I do not detect a violation by my
opponent" policy. In view of the way many analysts consider
the arms race, and international conflict in general, this
appears to be an unlikely assumption. Unfortunately,
it is central to the paper and is thus intimately related
to the results. While possibly a valid assumption,-the
results of the paper need to be considered within the con-
text of this assumption. Discarding this assumption pro-
duces radically different conclusions from this methodology.
The conclusions of the paper, which tie the mathematical
formalism back to the reality of arms races are less strong
than the authors assert, as they rely heavily on the previous
assumptions. Thus, while a pooling of verification intelli-
gence or technology might be a good idea in its own right,
that conclusion from this paper is conditioned on the assump-
tion of conditional cooperation. Otherwise, it, itself,
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
becomes an uncertainty which must also be monitored. This
is directly related to the issue of a pla'yer's knowledge
of his own misclassification probability.
In summary, this paper should be taken as an inter-
esting attempt to bring some game theoretic ideas to the
arms control verification problem. The interest of the
paper lies in its methodology rather than its results.
Game theoretic analogies to arms control are not new,
rather it is the introduction of probabilisitc knowledge
that renders the paper novel and is an important step.
The fact that previous game theoretic models of arms control
have not had a dramatic inpact on reality should indicate
the difficulty of the problem, not the futility of,-the
approach. Perhaps the ideas generated in this paper coupled
with both the dynamical theories of arms races and results
on detection capability in a noncooperative environment
will be fruitful.
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 . Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
Subject: Comments on the Geometry of the Arms
Race by Brams, Davis and Straffin
Distribution:
Orig. --
DCI
1
--
DDCI
1
--
ER
2
--
DDI
2
--
D/OSR
STAT OD/OSR (19 July 1977)
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved ForBelease 2004/03/23 : CIA-RDP80M00165A00O800080001-8
STAT
STAT
STAT
CHECK CLASSIFICATION
TOP AN
SENDER WILL
D BOTTOM
L' VCLASSIdED CONFIDENTIAL SECIdET
OFFICIAL
ROUTING
SLIP
Executive Req
TO
NAME AND ADDRESS
DATE
INITIALS
i
D/OSR
a
DDI
3
ADDCI
4
DCI
s
6
ACTION
DIRECT REPLY
PREPARE REPLY
APPROVAL
DISPATCH
RECOMMENDATION
COMMENT
FILE
RETURN
CONCURRENCE
INFORMATION
SIGNATURE
Remarks:
3-4: The attached comments were
drafted by of the
Analytical Methodology Research
Division of the Office of
Research and Development. She
is less sanguine than Professor
Brams about the likelihood that
the US and USSR will adopt a
policy of conditional cooperation
in arms control--a critical
assumption of Brams methodology.
FOLD HERE TO RETURN TO SENDER
FROM: NAME. ADDRESS AND PHONE NO.
DATE
S. N. Gra beat 3000
19 July
. UNCLASSIFIED CONFIDENTIAL SECRET
FOUL MO. 237 Use previous editions (40)
1-67
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 25X1 Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
Next 1 Page(s) In Document Exempt
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For Release 2004/03/23 : CIA-RDP80M0016500800080001-8
The paper, "The Geometry of the Arms Race," by
Brams, Davis and Straffin, is-interesting and--is-ind-icative
of"an important subject.;Unfortunately it is of limited
practical. -vale - as - it stands: r Fowever; it does seem. -to~ be
a useful- vehicle :,to. focus and stimulate.-interest-that could -
ultimately lead-to-results of practical-significance.
The following discussion indicates the deficiencies in
the results as presented in the paper.
Criticism of the paper falls naturally into three
categories: mathematics, assumptions, and conclusions.
The categories are discussed in that order. The mathematics
in the paper is correct. With the single exception of the
game theoretic payoff matrix presented. in Figure 1, there
are no errors in the mathematical concepts or the derivations.
In Figure 1 there is an interchange of subscripts, and the
correct matrix should be:
Disarm
A
Arm
A21B2
A4, B1
Al,B4
A3,B3
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For Release 2004/03/23 : CIA-RDP80M001654000800080001-8
In the body of the paper, the subscripts always appear
correctly. Hence, the derivations are correct.
Certain assumptions are presented which link the
physical. reality of arms races to the mathematical formal-
ism of Prisoners Dilemma game theory. Some of these
assumptions are made for convenience, and can be easily-
modified without changing the fundamental results. Other
assumptions are a necessary part of the main theme of the
paper. The assumption that both parties have the same
detection probability, p, is a matter of convenience. Un-
equal probabilities, pA and pB' could have been retained
throughout with only an increase in complexity. The. assump-
tion that each is greater than 50o,-however, is more funda-
mental to the paper and deserves some attention. While it
is certainly the case that a flip of a coin will result in
p = .5, it does-not follow that this lower bounds the d.e-
tection probability. A rational player will only flip a
coin if he knows that his detection probability is less
than 50%. In a covert armament program, however, player
B might force player A's probability well below .5 while
player A still believes it to be well above .5. Thus,
the game becomes one not only of imperfect information but
one of misinformation. This changes player A's views on
the security level achieved by a policy decision.
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For RcKease 2004/03/23 : CIA-RDP80M001600800080001-8
Another assumption which is central to the paper is--
that each side is willing to accept an expected value
criterion of the worth of a policy. While this assumption
can be changed to, say, aminimax criterion, such a change
makes the mathematical conclusions of the paper inapplicable.
Perhaps of most importance, however, is the main as-
sumption of conditional cooperation of the two players,
the "I will disarm if I do not detect a violation-by my
opponent" policy. In view of the way many analysts consider
the arms race, and international conflict in general, this
appears to be an unlikely assumption. Unfortunately,
it is central to the paper and is thus intimately related
to the results.. While possibly a valid assumption,._.the
results of the paper need to be considered within the con-
text of this assumption. Discarding this assumption pro-
duces radically-different conclusions from this methodology.
The conclusions of the paper, which tie the mathematical
formalism back to the reality of arms races are less strong
than the authors assert, as they rely heavily on the previous
assumptions. Thus, while a pooling of verification intelli-
gence or technology might be a good idea in its ownTight,
that:co3zclusion from this paper is conditioned on the assump-
tion of conditional cooperation. Otherwise, it, itself,
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For Rase 2004/03/23 : CIA-RDP80M00165"00800080001-8
becomes an uncertainty which must also be monitored. This
is directly related to the issue of a player's knowledge
of his own misclassification probability.
In_,.summary, this paper should be-taken-as ?-an-inter-t
esting attempt,>,to bring.'-.same' game theoretic.:_ide&S toy=-the
arms, con-tro_l,-ver-if ication< ,problem,;-. The interest. ofr -the
paper ...lives :in -its- .methodology rather than.-its-,results.
Game theoretic analogies to arms control are not new,
rather it is the introduction of probabilisitc knowledge
that renders the paper novel and is an important step.
The fact that previous game theoretic models of arms control
have not had a dramatic inpact on reality should indicate
the difficulty of the problem, not the futility of.,.the
approach. Perhaps the ideas generated in this paper coupled
with both the dynamical. theories of arms races and results
on detection capability in a noncooperative environment
will be fruitful.
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For Release 2004/03/23 : CIA-RDP80M00163f4000800080001-8
Subject: Comments on the Geometry of the Arms
Race by Brams, Davis and Straffin
Distribution:
Orig. -- DCI
1 -- DDCI
1 -- ER
2 -- DDI
2 -- D/OSR
OD/0SR
(19 July 1977)
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 F-VyfL
r UNCLASSIFIED CONFIDENTIAL SECRET
OFFICIAL ROUTING SLIP
~O
NAME AND ADDRESS
DATE
1NITIAI S
DDI
2
s
ORD
4
6
ACTION
DIRECT REPLY
PRYP_ARE REPLY
APPROVAL
DISPATCH
RECOMMENDATION
COMMENT
FILE
RETURN
CONCURRENCE
INFORMATION
SIGNATURE
Remarks :
See DCI's note inside.
Ce-
FOLD HERE TO RETURN TO SENDER
FROM: NAME. ADDRESS AND PHONE NO. DATE
DCI 8/19/77
vNCLASSIIE F L
cao>
FOOM 814- 237 use grorious editions
1-67 237
 STAT Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
Next 3 Page(s) In Document Exempt
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For'Release 2004/03/23 : CIA7RDP80M00165A000800080001-8
STAT
SENDER WILL CHECK CLASS FICATI()N TOP AND BOTTOM
U1rCI,.ASSxf~IE13 _-~CCi1f'f;?%_ti`I'fr.I, =-3-k;----
OIL FICIAL R..OUTING SLIP
OFFICIAL
TO
NAME AND ADDRESS
DATE
INITIALS
D/OSR
E
DDI
3
DDCI
4
DCI
5
6
ACTION
DIRECT REPLY
PREPARE R#PLY
APPROYP,L
DISPATCH
RECOf:I1riEIlDATION
CO~:~MEI:T
FILE
RETURN
CONCURRENCE
II{FORMATION
SIGNATURE
Remarks:
2-4: OSR and ORD are reviewing the Brams'
article for possible applicability
to our analysis of verification
issues.
Sayre Stevens
FOLD HERE TO RETURN TO SENDER
FROM: NAME, ADDRESS AND PHONE NO.
DATE
UNCLASSIFIED
SECRET
roatr Mo. ')~)7 Use previous editions
I-67 LJl
Approved For Release 2004/03/23: CIA-RDP80M00165A000800080001-8
 ?, Approved For.galease 2Q04103123 : CIA'RDP$040016W00800080001-8
the Director of Central Intelligence
Washington. D. C. 20505
77
Dear Professor Brains:
Thank you for your draft article on "The
Geometry of the Arms Race." I have sent it
to several members of my staff for study.
The evaluation of new methodologies and
their application to difficult problems are
continuing priority tasks for our analysts.
Your thoughtfulness in bringing your research
to our attention as soon as practicable is
appreciated. It strengthens my conviction
that there are many academicians who recognize
the importance of intelligence and are willing
and anxious to contribute to our analytic
effort.
Yours sincerely,
STANSFIELD TURNER
Professor Steven J. Brains
Professor of Politics
New. York University
25 Waverly Place
New York, N.- Y. 10003
Approved For Relgase 2004/03/23 :.CIA-RDP80M00165A000800080001-8
 Approved For,itelease 2004/03/23 : CIA-RDP80M0016X000800080001-8
SUBJECT: Response to Professor Steven J. Brams Letter
eI
Once Date
Distribution:
Original
Prof.
1 -
DCI
1 -
DDCI
1 -
ER
2 -
DDI
2 -
D/OSR
D/OSR~
(1 July 1977)
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved ForIease 2004/03/23: CIA-RDP80M001,*000800080Q01-8
MEMORANDUM FOR: D/OSR
I :A
All? %1 6
Please prepare by L0i-, ~8- July, an appropriate
response from the Director back to Professor
Brams. Perhaps it would be better if some
knowledgeable individual could include some
critical remarks in the response.
STAT
5 R7M
101 EDIT I,NgI0Us
Date 6/29/77
Approved For Release 2004/03/23 CIA-RDP80MOOl65A000800080001-8
 r '
Approved For.Qelease 200 /03/23 CIA D~80M001 000800039001-8
~= 11NC1A551FIFD -1-~ CONFIDENTIAL SECRET
EXECUTIVE SECRETARIAT
Routing Slip
ACTION
INFO
DATE
INITIAL
1
DCI
2
, DDCI
3
D/DCI/IC
DDS&T
5
) D D I
1
6
DDA
7
DDO
8
D/DCI/NI
9
GC
10
LC
11
IG
-
12
Compt
13
D/Pers
14
D/S
15
DTR
16
Asst/ DCI
17
AO/DCI
C/IPS
DCI/SS
M
D/EEO
-
2
.
22
Approved For Release.2004/03/23: CIA-RDP80M00165A000800080001-8
 7:x(iciviva Hcgi,;rp
New York U ~FS~ked For4eelease 2004/03/23: CIA-RDP80M0016 c0008000$QQO
Department of Politics
25 Waverly Place
New York, N.Y. 10003
Telephone: [phone]
The Honorable Frederick Turner
Director
Central Intelligence Agency
Washington, D.C. 20505
Dear Director Turner:
June 23, 1977
DDl /#rtrs- /
I have enclosed a copy of a paper entitled "The Geometry
of the Arms Race" that I thought might interest you because of
:its'focus on intelligence and verification capabilities.
Please note especially the section of our paper on "Policy
Irq-)lications" and our major conclusion on p. 17: "It is in the
interest of the United States not only to improve its own detection
[verification] capabilities but also to abet those of the Soviet
Union."
This conclusion seems to fly in the face of current policy,
though recently some interest has been expressed in negotiations
concerning a new SALT agreement that both sides develop a common
data base. Our. analysis strongly supports this development and
also supports heavy investments in the research and development
of expensive new weapons systems. We find less support for measures
that make the benefits of an arms-control agreement more attractive.
These are, in my opinion, significant policy conclusions. To
be?sure, if,the underlying assumption of our analysis-=that both
sides will cooperate if they are reasonably- assured that the other
.side will follow suit--is untrue, then our analysis is no longer
applicable However, I think that both sides recognize that it
is in their mutual interest to cooperate, at least on a conditional
basis.' .
1-hope you find our analysis helpful.
Siinerely,
%
Steven J. Brams
Professor of '1~nlitics
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For. elease 2004/03/23: CIA-RDP80M0016 0008000$0001-8
DRAFT -Not for Quotation
TILE GEOMETRY OF THE ARMS RACE
Steven J. Brams
Professor of Politics
New York University
New York, N.Y. 10003
Associate Professor of Mathematics
City College of New York
New York, N.Y. 10031
Philip D. Straffin Jr..
Associate Professor of Mathematics
Beloit College
Beloit, Wis. 53511
Approved For Release 2004/03/23 CIA-RDP80M00165A000800080001-8
 Approved Forlease 2004/03/23 : CIA-RDP80M0016i0008000B0001-8
THE GEOMETRY OF THE ARMS RACE
Of all the significant problems that confront the world, the
nuclear arms race between the United States and the Soviet Union
has proved one of the most intractable. Its intractability, however,
stems not from the awesome amounts both sides have expended on arms,
nor even in the millions of-lives at stake should the arms race
culminate in a nuclear war. While these facts help to explain why
the arms race looms so large in our lives, they do not explain why
this race has proved so difficult to slow down.
To be sure, a variety of explanations for the persistence of
the arms race has been advanced. The military--industrial complex
in each country holds sway over major policy decisions Q). The
economies of the United States, and perhaps the Soviet Union as
well, require major military expenditures to avoid recessions or
even depressions (2). The dynamic nature of an arms race requires
that each side match or exceed the expenditures of the other side
(3) Or, where moves toward disarmament are observed, they are
no more than an elaborate fraud by which the superpowers deceive
the rest of the world so that they can maintain their hegemony (4).
It is not our purpose to criticize these and other purported
explanations of the arms race, though we believe all are seriously
flawed. For the most part, they are ad hoc, single-factor explana-
tions--sometimes colored by ideological considerations--that are not
embedded in a general model that disciplines. the weighing of benefits
A ro ed For Relea a 2004103/23 : CIA-RDP80M00165A000800080001-8
and costs to decision makers in the arms race.
 Approved For lease 20b4/03/23 : CIA=RDP80MO0161 0080?g80001-8
Our purpose in this article is to construct a model of the arms
race that rests primarily on rationalistic calculations. We do not
mean to imply that national decision makers go exactly through the
calculations we set forth or that they are unmoved by nonrational
considerations. Rather, we believe that where the stakes are high,
as they are in the nuclear arms race, decision makers, at least in
a rough way, take account of benefits and costs in the manner postu-
lated in our model.
These benefits and costs are dependent.on what both sides do,
and for this reason we believe that.the proper representation of
the arms race is as a game. The game we begin with, called Prisoners'
Dilemma, is well-known in game theory and has been used by others as
a model of the arms race-(5). In our view, it shows up in a
strikingly simple way why the arms race is as intractable as it
'.is, which is the note on which we introduced this article.
But wo are also concerned with possible solutions to the arms
race, and for this purpose we posit a sequence of moves by the super-
powers that we believe may lay the basis for future cooperation that
leads to arms-control agreements. (There is already some evidence
to support this.sequence, as we indicate later.) Consequences of
this sequence are'investigated when each side (i) possesses an ability
to detect what the other side is doing with a specified probability,.
and (ii) pursues a "tit-for-tat" policy--cooperates if the other
side does, otherwise does not.. Given the detection probabilities,
and the reciprocity norm, we show, geometrically, when cooperation
between the suveer owers is rational and therefore likely to occur.
pprod Fpor Release 2004/03/23 : CIA-FADP80M00165A000800080001-8
 Approved Forelease 2004/03/23 : CIA-RDP80M001i0008000;809.01-8 .3
Prisoneiis' Dilemma and the Arms Race
Prisoners' Dilemma is a two-person game that is illustrated in
Fig. 1. We shall not describe the original story that gives Prisoners'
Dilemma its name but shall instead interpret it in t}ie context of
the arms race between the superpowers, whom we call A and B.
The superpowers each have a choice between two strategies,
"Disarm" and "Arm," as shown in Fig. 1. The choice of a strategy
by both superpowers results'in one of the four possible outcomes
shown in the payoff matrix of Fig. 1. An outcome is. defined by an
ordered pair of numbers (Ai, Bj), where Al is the payoff to A (row
player), Bi the payoff to B (column player).
For player A we assume that Al is his best payoff, A2 next best,
A3 next worst, and A4 worst; a similar ordering obtains for B. Thus,
for example, (A2, B2) is a better outcome for both players than (A3, B3)'
The dilemma in this game is that both players have an unconditionally
best, or dominant,, strategy of Arm: whatever the other player does
(Arm or Disarm), each player obtains a higher payoff if he chooses
Arm. Yet, if both players choose Arm,.the outcome is (A3, B3), which
is worse than if both players choose Disarm and thereby obtain (A2, B2).
If this is.the case, should not both players choose Disarm? The
problem here.is that (A2, B2) is not in equilibrium: given the choice
of (A2, B2), each player has an incentive unilaterally to switch to.
Arm and thereby obtain his best payoff (A1 or B1), inflicting on the
other player his worst payoff (B4 or A4). This temptation for each
player to doublecross the other makes (A2, B2) unstable and, we
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved ForRIease 2004/03/23 CIA-RDP80M0016 C000800Q8Q001-8 .4
F IGIJRE 1
THE ARMS RACE AS A PRISONERS' DILEMMA GAME
(A2) _B2) (A1, B11)
B1)
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For eIease 2004/03/23 : CIA-RDP80M0019ii 0008000x0001-8.
believe, points up the fragility of cooperation (when both players
choose Disarm) in the arms race. It is precisely this temptation
to doublecross that induces each player to "play it safe" and choose
his dominant strategy of Arm, even though the resultant outcome,
(A3, B3), is the next worst for both players.
The outcome (A3, B3), which is circled in rig. 1, is in fact
the unique equilibrium outcome in Prisoners' Dilemma--once chosen,
neither player can do better by unilaterally switching to his
Disarm strategy. The fact that both players prefer (A2, B2) leads
us to ask how movement from (A3. B3) to (A2, B2)---as indicated by
the arrow in rig. 1--can be induced, given that (A29. B2), once
reached, is unstable.
Introducing Detection Probabilities
Assume that A and B begin the game by both announcing a tit-
for-tat policy of conditional cooperation: "I'll cooperate (i.e.,
choose Disarm) if I detect you do; otherwise, I won't." Then, to
show.their good intentions, assume both players initially cooperate
and choose Disarm. This is the-first?stage of the game O .
The second stage begins when each player makes a second strategy
choice, depending on what he detected his opponent did in the first
stage. Assume that A can detect with a certain probability the
strategy choice of B; and B can likewise detect A's strategy choice.
Specifically, let
pA = probability that A can detect B's strategy choice in the first stage;
PB probaupirloiity Ftnate acan0cietec 3 AC~Is -strateg0y Choi 0800 in8thel-first stage.
 Approved For Ri9pase 2004/03/23: CIA-RDP80M00165A@29800Q80001-8 6
We assume 0 t= PA' PB..~- a-? ?
Presumably, the better intelligence one player has of the other's
capabilities and intentions, the higher his detection probability
will. be. Although we assume that PA and pB may vary between 0 and 1,
practically speaking it seems reasonable to suppose that these proba-
bilities will never assume values less than 1/2. Otherwise, a
player can better guess his opponent's strategy choice by flipping
a coin.
Consistent with a policy of conditional cooperation, assume
that a player chooses Disarm if he detects that his opponent chose
Disarm in'the first stage; otherwise, he chooses Arm. The question
is: does a policy of conditional cooperation benefit the players
in the second--and perhaps later--stages of the game?
The expected payoff a player derives in the second stage is the
sum of the payoffs he obtains from each of the four possible outcomes
times-the probability that each occurs. (The expected-payoff in
the first stage is A2 for A and B2?for B,'because by assumption
the "cooperative" outcome (A2, B2) is chosen with probability 1.)
For A, his expected payoff in the second stage will be
E (A) = A2PAPB + Al (1-PA) pB + AL PA(1-PB) + A3 (1-PA) (1-PB)
assuming A and B make independent strategy choices based solely on
their probabilities of detection. Thus, for example, the first term
on the right-hand side of (1) says that A and B will correctly
detect th~i A~RDP86M00165A0008D0080001 paroba-
 Approved For Rase 2004/03/23 : CIA-RDP80M00165AW8800'0$pO01-8
bility?pApB, so A will obtain a payoff of A2 with this probability.
Rearranging terms in (1), we obtain
E(A) = pB[A2PA + Al(1-pA)] + (1-pB)[AypA + A3(1-p.A)] (2)
Whatever the value of pA, we know that the first term in brackets.
on the right-hand side of (2) is always greater that the second term
in brackets since A2 > A4 and Al > A3. Therefore, it is in A's
interest that pB be as high as possible (so.B will correctly detect
cooperation and thereby cooperate himself), and similarly for B with
respect to PA'
This is not a surprising conclusion. Rearranging terms in
again, we obtain a more curious result:
E (A) = pA[A2pB + A4 (1-pB)] + (1-pA) [AlpB + A3 (1-pB)]
(1)
(3)
Now the second terms in brackets on the right-hand side of (3) is
.always greater than the first term in brackets, so it. is in A's
interest that (1-pA) be as high as passible, or pA be as low as
.possible. This-is because A, if he incorrectly detects that B
chooses Arm in the first stage and thereby chooses Arm himself in
the'second stage, obtains a higher expected payoff than if he
correctly detects cooperation on the part of B.
But surely B could anticipate this consequence if he knew PA
were low. Hence, B should not mechanically subscribe to a policy'
of conditional cooperation in the second stage unless he is assured
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For Rgase 2004/03/23 : CIA=RDP80M00165AQ6800Q80Q01-8
that A can predict with a high.probability his cooperative choice
in the first stage and thereby respond accordingly. A similar
conclusion applies to B. Therefore, it is in the interest of A
and B that both PA and PB be as high as possible (-7)
Equalizing the Detection Probabilities
glow can both players ensure that PA and pB are as high as
possible? One way, which has been proposed in recent negotiations
on a new SALT agreement (8J , is to pool their information so that
they both operate from a common (and enlarged) data base. A common
data base, presumably, would have the effect of setting the detection
probabilities equal- to each other. Alternatively, if "national
technical means for verification"--in the terminology. of current
.arms-limitation talks--of both players were equally good, their
..detection probabilities would also be equal.
To investigate the consequences of_equal detection probabilities,
assume that PA PB = p. The expression for E(A) given-by (1) then
becomes
E (A) = A2p2 + (A1+A4) (1-P) P + A3 (1-P) 2. (4)
An analogous expression can be obtained for B, but henceforth we.
shall make only calculations for A'.since the conclusions we derive
apply to B as well.
Without loss of generality, we may assume that the payoffs
associated with the best and worst outcomes are one and zero, respee-
Approved For Release 2004/03/23 :.CIA-RD $0M00165ApQ08000gpp0 8
tively, i.e., Al = 1 and A4 = U. Given this assumption, l14) becomes
 `Approved For RRiease 2,004/03/23 : CIA-RDP80M00165A'9'08000180QO1-8
.0
E (A)
= Alp 2 + (1--p) p + A3 (-I-"p) 2 1).
(A2+13-1) p2 + (1-2A3) P + A3 ,
(5)
which is a parabola in p.
What is of interest is the shape of the parabola in the four
regions of the A2-A3 coordinate system shown in Fig. 2. This tells
us how beneficial a policy of conditional cooperation is as a
function of p, assuming (for now) that A2 and A3 are fixed.
Since by assumption 0 < A3 < A2 < 1, we need not consider the
area on or above the diagonal A2 = A3. . If- (A2+A3-l) > 0, which
defines regions I and II, the parabola is concave up; if (A2+A3-1) < 0,
which defines regions III and IV, the parabola is concave down.
In the interval 01.4 p, 1, graphs of E (A) as a function of p
are shown in Fig. 2 for each of the four regions. The vertex of
the parabola in all regions is at
2A3-1
2 (A2 + A3 -- 1)
(A3 -- 1/2)
-----------------------
(A3 - 1/2) + (A2 - 1/2) .
J Q
When substituted into (5), the vertex gives the minimum value of
E(A) in regions I and II, the maximum value of E(A) in regions III
and 'N.
In regions I and II, where the denominator is positive, the
minimum is at p > 0 if and on1 if the numerator is also ositive,
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 'Approved For R J ase 2004/03/23 : CIA-RDP80M00165AQ90800080001-8 10
EXPECTED PAYOFFS IN FOUR REGIONS
I
E (A)
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For FlW ase 2004/03/23 : CIA-RDP80M00165A` N800000001-8 ?11
i.e., A3 > 1/2. This occurs in region I; in 'region II the minimum
occurs at p 4 0, but in the interval 0. p - 1, the minimum of E(A)
is at the boundary p = 0, as shown in Fi.g. 2.
In regions III and IV, both the numerator and denominator of
(6) are negative, so the maximum is always at p 0. Rewriting (6),
(A2 - 1./2)
p ----..--_------..--------,
(A3 - 1/2) + (A2 - 1/2)
(7)
we see that the maximum is at p 4 1 if and only if the numerator in
the second term on the right-hand side of (7) is negative, i.e.,
A2 < 1/2. This occurs in region IV (2) in region III, the maximum
occurs at p 1, but in the interval 0 p 1, the maximum of E(A)
is at the boundary p = 1, as shown in Fig. 2.
When Is Conditional Cooperation Rational?
The graphs of E (A) in Fig. 2 show that E(A)Z: A. for all values
of p in regions II, III, and IV. Thus, a policy of conditional
cooperation in these regions ensures at least the security level of
A--the minimum payoff he can ensure for himself, A3., whatever B
does. In fact, this policy will always yield an expected payoff
greater than the security level A3 except when p = 0, which occurs
when A always detects the choice of Arm by B, the opposite of what
B does.
No such assurance can be offered A if he is in region I. This
is the region in which A271- A3> 3/2, i.e., where both the cooperative
payoff A2 avt-f~goi5cgt~, ~f ~.~1 .W23a} 1Rt389 O gQq$0W -~l = 1
 Approved For Rase 2004/03/23 : CIA-RDP80M00165AWp?800080001-8 12
than A,, = 0. In this case, the loss A suffers from being double-
crossed is significantly below all his other payoffs.
For this reason, it may be advantageous for A to accept his
security level A3 rather than commit himself to a policy of condi-
tional cooperation. After all, conditional cooperation could
result in the payoff AL = 0, which is much worse than A3 > 1/2
in region I.
In region I, the advantage of A3 over. E(A) is greatest when
E(A) is at a minimum, which occurs when p < 1/2, as shown in Fig. 2.
Even for p 1/2, however, E(A) may be less than A3. To determine
how high p must be in order that E(A) exceed A3, we solve
E (A) = (A2 FA3-l) p2 + (1-2A3) P + A3
for p, and get
p = 0 or.p = (2A3-1)/(A2+A3-1) .
(8)
(9)
We already know E(A) ) A3 if p'> 0 in regions II, II[, and IV. In
region I, E (A) > A3 if .
2A3 - 1 2(A3 - 1/2)
P > ----------- _ -----------------------
A 2 + A3 1 (A3 - 112) + (A2 - 1/2).
Algebraic manipulation gives
? (10)
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 -Approved For R ase 2004/03/23: CIA-RDP80M00165AQ00800080001-8
1
(A3 - 2) < 2_p (A2 2
Thus, in region I, a policy of conditional cooperation is better
than security level A3 if the point (A2, A3) lies below the line
which passes through (3/2, 1/2) and has slope in = p/(2-p) . For
several representative values of p between 0 and 1, these isolines
are illustrated in Fig. 3 and show that as the detection probability
approaches 1,-the possibility that conditional cooperation yields
less than one's security level vanishes.
Because the slope m of the isolines is convex in p (d2rldp2 > 0) ,
raising p will make conditional cooperation more advantageous if p
is already high (cf. representative values of p and m in Fig. 3).
Moreover, since m is always less than 1 except when p = 1, raising
A2 [see (11)] is in general less effective in encouraging conditional
.1
cooperation than lowering A3.
Policy Implications
We.have shown that a policy of conditional cooperation always
yields. an expected payoff that is at least equal to, and generally
exceeds, one's security level in three of the four regions that are
feasible for Prisoners' Dilemma when both sides have the same-detec-
tion probability. In these regions, therefore, this policy will-
generally work to the players' mutual advantage, even if the detec-
tion probability is low.
Unfortunately, the arms race between the two superpowers
probably occurs in region I. Here the consequence of being double-
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 'Approved For Rase 2004/03/23: CIA-RDP80M00165AA9U80008.0001-8
14
TSOLINES BELOW WHICH E(A) > A3 IN REGION I
A2
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 -Approved For Rase 2004/03/23 CIA-kDP80M00165AQQ0800080Q01-8 15
crossed (A4 = 0) is very unsatisfactory compared to accepting one's
security level. (A3 > 112). Yet, our analysis indicates that condi-
tional cooperation even in region I may be beneficial, depending on
the detection probability p of both sides. The area: in this region
where conditional. cooperation leads to a higher expected payoff than
one's security level increases as (i) p increases, (ii) A2 increases,
or (iii) A3 decreases. Indeed, the effects of (i) seem already to
have been felt in the limited agreements so far achieved in SALT I
and SALT II (10)
If p continues to increase as technology improves, conditional
cooperation should become even more attractive. This is because the
slope m increases faster than p when
dp
(2-p) 2
p> 2 --V2 V0.586.
(l2)
Thus,, technological improvements that raise p above 0.586 will even .
more rapidly expand the area in which conditional cooperation is
rational for both sides.
We indicated earlier that the effects of (iii) in encouraging
conditional cooperation are greater than the effects of (ii). This
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 'Approved For RiWease 2004/03/23 : CIA-RDP80M00165AB0080008Q001-8
means that developments that increase the costs of, a continuing
arms race (decrease A3) do more to encourage conditional cooperation
than developments that increase the benefits of an arms-control
agreement (increase A2)-
Of course, raising the benefits of an agreement and raising
the costs of no agreement are two sides of the same coin. But if
there is a lesson to be derived from our model, it is that they have
unequal trade-offs. Since the multiplier effect is on the cost side
of the equation, behavior that raises the costs of an arms race
provides the greater incentive for making reciprocal concessions.
Probably the best way to make an arms race more costly is to
invest heavily in research and development. This investment increases
the probability of technological breakthroughs that create the need
for expensive new weapons systems. Paradoxically, perhaps, by making
present weapons systems more vulnerable to-technological breakthroughs,
and hence less cost effective, we may better foster a future policy
conducive to arms-control agreements.
Since the early 1960s, one of the most significant qualitative
chaiiges in?the nuclear arms race has been the dramatic rise in the
detection capabilities of both sides' which has been principally.due
to the use of reconnaissance satellites Indeed, President
Johnson once stated that space reconnaissance had saved enough in
military expenditures to pay for the entire military and space
programs (12) .
IF this detection capability of either side is destroyed or
even thre,06 &I Ftor el~~snedl ~~/~ rz3 ~~> SI~0 In6 i~ t '~0 8~dd -8 once
 ? Approved For Lease 2004/03/23 : CIA-RDP80M00165A$8080Q080001.-8 17
.again be rendered unappealing and the prospects of a continuing arms
.race will be high. On the other hand, if each side's detection
capabilities can be ensured or even strengthened--especially through
the sharing of data that helps render PA = pD = p--alien further
agreements in SALT would appear not only desirable but also rational
for both sides.
Just as stability in the arms race has depended up to now on
the ability of each side to respond to a possible first strike by
the other side, a diminution in the arms race now seems to depend on
the ability of each side to detect cooperation on the part of the
other side and to respond to it in kind. Unfortunately, "probably
nothing the United States does is more closely held than the tech-
niques and performance of its verification machinery" 13 . To
promote movement toward an arms-control agreement, we believe it is
clearly in the interest of the United States not only to improve its
own detection capabilities but also to abet those of the Soviet Union.
Summary
The arms race between the two superpowers was conceptualized as
a Prisoners' Dilemma game, with the additional property that each
player can detect initial cooperation or. noncooperation on the part
of the other player with a specified probability. Consequences of
the. following scenario were investigated: both players initially
cooperate; each player knows. the other player's detection probability
.and follows a policy of conditional cooperation--cooperates if he
detects cooperation on the part of the other player, otherwise does
not coopeAp rjved For Release 2004/03/23 : CIA-RDP80.M00165A000800080001-8
 -Approved For RJease 2004/03/23 : CIA-RDP80M00165AW08000>i,0001-8
For the case in which the detection probabilities of the two
players are equal, conditional cooperation by both players yields
the following conclusions:
i. Each player's expected payoff as a function of the
detection probability is a parabola, which may assume
four different forms depending on the payoff each player
assigns to the cooperative versus noncooperative outcomes
in Prisoners' Dilemma.
ii. The different payoffs can be represented geometrically
by four different regions; in only one of the four
regions does conditional cooperation not guarantee a
player at least his security level.
iii. Even in this region, as the detection probability
approaches one, the possibility that conditional coop-
eration yields less than one's security level vanishes.
Policy implications of this analysis for SALT are discussed, and a
?suggestion for the sharing of intelligence data is advanced.
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 -Approved For Rase 2004/03/23 : CIA-RDP80M00165AP8080Q0$.0001-8
1.9
1. The Military-Industrial Complex and Foreign Policy, 0. L. Carey,
Ed. (Wash. State Univ. Press, Pullman, Wash., 1969); S. Lens,
The Military-Industrial Complex (Pilgrim Press, Philadelphia,
1970) ; R. E. Kaufman, The War Profiteers (Boobs--Merrill,
Indianapolis, 1970); The Military-Industrial Complex, C. W.
Pursell, Jr., Ed. (Harper & Row, New York, 1972); The Military-
Industrial Complex: A Reassessment, S. C. Sarkesian, Ed. (Sage,
Beverly Wills, Calif., 1973); Testing the Theories of the
Military-Industrial Complex, S. Rosen, Ed. (Lexington, Lexington,
Mass., 1973).
2. See the semi.fanciful-semiserious Report from Iron Mountain on the
Possibility and Desirability of Peace (Dell, New York, 1967).;
for reactions to this report, see "Comments on Report from Iron
Mountain," in Peace and the War Industry, 2d ed., K. E. Boulding,
Ed. (Transaction, New Brunswick, N.J., 1973), pp. 55-83. For a
..comprehensive review of the literature, see The Economic Con-
sequences of Reduced Military Spending, B. Udis, Ed. (Lexington,
Lexington, Mass., 1973). _
3. -L. F. Richardson, Arms and Insecurity;' A Mathematical- Study of
the Causes and Origins of War (Boxwood, Pittsburgh, 1960); for
recent work on "Richardson-type process models," see the articles
in Part III of Mathematical Models in International Relations,
D. A. Zinnes and J. V. Gillespie, Eds. ((Praegger, New York 1976).
Approved For Release 2004/03123 : CIA-RDPEOMOOI65A00080008000'1-8
 Approved For Rase 2004/03/23 : CIA-RDP80M00165Rifi80800080Q01-8
20
For a critique of the assumptions underlying U.S.-Soviet arms
race models, see A. Wohlstetter, Foreign Policy, 15, 3 and 16,
48 (1974).
It. A. Myrdal, The Game of Disarmament: How the United States and
Russia Run the Arms Race (Pantheon, New York, 1.976); J. W.
Spanier and J. L. Nogee, The Politics of Disarmament: A Study
in Soviet--American Gamesmanshi (Praeger, New York, 1962).
5. A. Rapoport and A. M. Chammah, Prisoner's Dilemma: A Study in
Conflict and Cooperation (Univ. of Mich. Press,, Ann Arbor, Mich.,
1965); for a recent review of the literature on Prisoners'
Dilemma, see S. J. Brams, Paradoxes in Politics:
Introduction
to the Nonobvious in Political Science (Free Press, New York,
1976), chs. 4 and 8.
6. Other scenarios are, of course possible, but these moves seem
the most plausible to assume if both players are seriously
interested in slowing down the arms race. For evidence that
this assumption has become reality in the. recent period of-detente,
see W. A. Gamson and A. Modigliani, Untangling the Cold War: A
Strategy for Testing Rival Theories (Little, Brown, Boston, 1971).
7. For further details, see S. J. Brams, J. Conflict Resolution, 19,
.596 (1975). Cf. N. Howard, Behavioral Sci., 21, 524 (1976), for
a general metagames analysis of Prisoners' Dilemma.
8. New York Times, April 27, 1977, p. A7.-- For an argument that data
be collected and verified under United Nations auspices, see A.
Myrdal, Sci. Amer., 231, 21 (1974).
RegicA vjd Fb1Pg2dQ6b M 6030n Gf)#3P fi01 5AW08bt090 > mum
 v . . .-Approved For R ease 2004/03/23 : CIA-RDP80M00165 0800p80001-8 21:
when p = 1 (in the interval 0 -4 p ce- 1). This is because
2A2 `4, AI + A4 = 1
in this region, so an alternation of the players between their
strategies associated with outcomes (Al, Bid) and (A4., Bl) yields
A a higher expected payoff than does outcome (A2, B2). For
this reason, Prisoners' 'Dilemma is sometimes defined so as to
preclude payoffs in region IV. See Rapoport and,Chammah, pp.
34-35.
10. J. M. - Lodal, Foreign Policy, 24, 40 (1976).
.11. F. A. Long, in Arms, Defense Policy, and-Arms Control, F. A.
Long and G. W. Rathjens, Eds. (Norton, New York, 1975), p. 10;
T. Greenwood, Sci. Amer., 228, 14 (1973). For a history of
aerial reconnaissance programs since the early 1950s, see H.-F.
York and.G. A..Greb, Bull. of Atomic Scientists, April 1977,
pp. 33-42.
12. W. F.. Diddle, Weapons, Technology, and Arms Control (Praeger,
New York, 1972), p. 252,
13. J. Newhouse,. Cold Dawn: The Story of SALT (Holt, Rinehart, and
Winston, New York, 1973), p. 14; security aspects of reconnais-
sance programs are discussed in Greenwood, and York and Greb.
14. 'S. J. Brams gratefully acknowledges the financial support of
Mathematica, Inc.
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For Release 2004/03/23 : CIA- viturtotlm -
PATY
'TRANSMITTAL SLIP
TO:
Executive Registry
-it
ROOM NO.
7E12
-
BUILDING
Hq
REMARKS:
FROM:
ROOM NO.
BUILDING
EXTENSION
01m No EB 55 24 I REPLACES FORM 36-8
WHICH MAY BE USED.
I F
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 CON' ENTIiAL
Approved For Release 2001 23 -
OFFICIAL ROUTING SLIP
DIRECT REPLY
DISPATCH
FILE
INFORMATION
FROM: NAME_ADORES3 AND TNONC NO.
fame at 237 Uss pr,riovs = itiaes
1-3)
CONFIDENTIAL
OTTOM
DATZ
SECRET
(40)
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
p/ DC I
 . Approved For Release
OFFICIAL ROUTING SLIT"
TO
NAME AND ADDRESS
DATE
INITIALS
1
D/OSR
2
DD-
3
DDCI
,--2`
4
DC I 6
JUL 1977
6
ACTION
DIRECT REPLY
PREPARE REPLY
APPROVAL
DISPATCH
RECOMMENDt.110ft
COMMENT
FILE
RETURN
CONCURRENCE
INFORMATION
SIGNATURE
Remarks:
2- 4: , OSR and. ORD: arm- revi&w?ino . the-..Brans'
article for :pass?ible apRlricabil.ity
--to--our'-a_nai.ysis: of verific_atiioii
i s sue s`a Fkecutlva R ....
771771,
Sayre Stevens JUL 7977
FOLD HERE TO RETURN TO SENDER
FROM: NAME, ADO ESS AND PHONE NO.
DATE
S. N. Gra b
7
UNCLAS FIE
L
SECRET
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 ?Approved~iF&r r2W4MV3ct: CIA-R
Washington. D. C 20505
6 July 1977
Thank you for your draft article on "The
Geometry of the Arms Race." I have sent it
to several members of my staff for study.
The evaluation of new methodologies and
their application to difficult problems are
continuing priority tasks for our analysts.
Your thoughtfulness in bringing your research
to our attention as soon as practicable is
appreciated. It strengthens my conviction
that there are many academicians who recognize
the importance of intelligence and are willing
and anxious to contribute to our analytic
effort.
Professor Steven J. Brams
Professor of Politics
New York University
25 Waverly Place
New York, N. Y. 10003
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved-For R ase 2004/03/23 CIA-RDP80MO0165 0800080001-8
'DDI? -7711
~~----1 ':ENDE.RWILL CHECK GI ASSIFICATtON TOP AN QI
CJ?tCLASSIFIF: Otr+i'IiDEN~'tAl ~- SECItF.T^
O FICIAL ROUTING SLIP
TO
NAME AND ADDRESS
DATE
INITIALS
i
D/OSR
z
DDI
3
DDCI
4
DCI
s
6
ACTION
DIRECT REPLY
PREPARE REPLY
APPROVAL
DISPATCH
RECOMMENDATION
COMMENT
FILE
RETURN
CONCURRENCE
INFORMATION
SIGNATURE
Remarks:
2-4: OSR and ORD are. reviewing the Brams'
article for possible applicability
to our analysis of verification
issues. -
Sayre Stevens
FOLD HERE TO RETURN TO SENDER
FROM: NAME. ADDRESS AND PHONE NO. DATE
S. N. Cra beal_- 3G0 T
iIN('I.ASStFIF.D-- I C SECRET
FOIM NO. 237 Use pnriOUS editions
1-67 237
Approved For Release 2004/0312 : CIA-RDP80M00165A000800080001-8
 Approved For Ruse 20.04103/23: Cl A-RDP80M00165080008
The Director of Ccntral Intelligence
Washington, D. C. 20505
6 JUL
Dear Professor Brams:
Thank you for your draft article on "The
Geometry of the Arms Race." I have sent it
to several members of my staff for study.
The evaluation of new methodologies and
their application to difficult problems are
continuing priority tasks for our analysts.
Your thoughtfulness in bringing your research
to our attention as soon as practicable is
appreciated. It strengthens my conviction
that there are many academicians who recognize
the importance of intelligence and are willing
and anxious to contribute to our analytic
effort.
Yours sincerely,
Let Stanfield Turner
STANSFIELD TURNER
Professor Steven J. Bram?s
Professor of Politics
New York University
25 Waverly Place
New York, N. Y. 10003.
Approved For Release 2004/0.3/2 :.CIA-RDP80M00165A000800080001-8
 Approved For R@Wase 2004/03/23 : CIA-RDP80M00165A 0800080001-8
SUBJECT: Response to Professor Steven J. Brains Letter
Deputy Director for Intelligence
Distribution:
Original - Prof. Steven J. Brains
1 - DCI
1 - DDCI
1 - ER
2 - DDI
2 - D/OSR
D/OSR:
(1 July 1977)
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For RaWase 2004/03/23: CIA-RDP80M00165AQ90800080001-8
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
 Approved For Release 2604/03/23 : CIA-RDPtOM00165A000800080001-8
MEMORANDUM FOR: D/OSR
Please prepare by {a?B-, ,8 July, an appropriate
response from the Director back to Professor
Prams. Perhaps it would be better if some
knowledgeable individual could include some
critical remarks in the response.
FORM
5-75 lOi UESDE
Approved For Release 2004/03123 GIA-RDP80M00165A000800080001-8
 Approved For Re'ILase 2004%03/23: CIA=IDP80M001650800080001-8
i UNClASS9FIED
CONFIDENTIAL ~l- SECP.ET
EXECUTIVE SECRETARIAT
Routing Slip
9
ACTION
INFO
DATE
INITIAL
1
DC1
2
DDCI
3
D/DCI/IC
4
DDS&T
5
) DDI
6
DDA
7
DDO
8
D/DCI/N1
9
GC
10
LC
11
IG
12
Compt
13
D/Pers
14
D/S
15
DTR
16
Asst/ DCI
17
AO/DCI
18
C/IPS
19
DCI/SS
20
D/EEO
21
22
Approved For Release 2004/0312 ,CIA-RDP80M00165A000800080001-8
 STAT Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8
Next 22 Page(s) In Document Exempt
Approved For Release 2004/03/23 : CIA-RDP80M00165A000800080001-8