Security Studies

ISSN: 0963-6412 (Print) 1556-1852 (Online) Journal homepage: http://www.tandfonline.com/loi/fsst20

Covert Communication: The Intelligibility and

Credibility of Signaling in Secret

Austin Carson & Keren Yarhi-Milo

To cite this article: Austin Carson & Keren Yarhi-Milo (2017) Covert Communication: The
Intelligibility and Credibility of Signaling in Secret, Security Studies, 26:1, 124-156, DOI:
10.1080/09636412.2017.1243921

To link to this article: http://dx.doi.org/10.1080/09636412.2017.1243921

Published online: 23 Nov 2017.

Submit your article to this journal

Article views: 213

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=fsst20

Download by: [Austin Carson] Date: 21 December 2016, At: 07:54

SECURITY STUDIES
2017, VOL. 26, NO. 1, 124–156
http://dx.doi.org/10.1080/09636412.2017.1243921

Covert Communication: The Intelligibility and Credibility

of Signaling in Secret
Austin Carson and Keren Yarhi-Milo

ABSTRACT
Can states credibly communicate their intentions through covert
policy tools, despite the absence of credibility-enhancing publicity?
Most extant research suggests covert action and secrecy in general
are uniquely uninformative and often used as an alternative to
signaling. Yet episodes such as Richard Nixon’s secret bombing of
Cambodia suggest that leaders have used covert action to convey
intentions and coerce adversaries. This article builds a theoretical
framework for understanding signaling in the covert sphere,
developing reasons why states find covert communication both
intelligible (that is, the basic intended message is understandable)
and credible (that is, the message is believable). We argue that two
target audiences—local allies and strategic adversaries—tend to
observe covert action and that the costs and risks incurred by
initiating and expanding covert action credibly convey resolve. We
assess our arguments empirically through careful process tracing of
a set of nested covert interventions by Soviet and American leaders
in conflicts in Angola and Afghanistan. Drawing on a trove of
recently declassified material, we assess intentions and inferences
related to covert signaling. We find that both strategic adversaries
and local partners observed and drew inferences about resolve.
Covert lethal aid programs thereby served as a credible indicator of
resolve through three mechanisms we identify in the paper: sunk
costs, counter-escalation risks, and domestic political risks. These
findings have important implications for the study of coercive
bargaining, secrecy, and reputation. They also shed light on an
important policy tool contemporary policymakers will likely use,
suggesting the kinds of effects covert action has and elucidating
the basic interpretive framework needed to communicate
messages with new methods like covert cyber attacks.

In March 1969, US B-52 Stratofortress bombers conducted secret raids on Viet-

cong targets in Cambodia.1 These massive bombers, capable of carrying seventy
thousand pounds of explosives, conducted raids that could not feasibly be kept

Austin Carson is Assistant Professor of Political Science at the University of Chicago. Keren Yarhi-Milo is
Assistant Professor of Politics and International Affairs at the Department of Politics and the Woodrow
Wilson School of Public and International Affairs at Princeton University. The authors contributed equally.
Author order is alphabetical.
1
A description of the Cambodian raids can be found in Seymour Hersh, The Price of Power: Kissinger in the Nixon White
House (New York: Summit Book, 1983), 54–61.
© 2017 Taylor & Francis Group, LLC
 SECURITY STUDIES 125

secret from local leaders in Cambodia and North Vietnam. In fact, they were
meant to know. According to one of the principal planners, the raids were primar-
ily intended to “signal to North Vietnam that we meant business. [Henry] Kis-
singer wanted them to know that we were serious about possible escalation.”2 Yet
extraordinary measures, including the falsification and destruction of records,
were taken to ensure that the bombing raids on Cambodia were kept secret from
other audiences, including the American public.3
The use of secret bombing in Cambodia to signal resolve is but one example of a
larger phenomenon almost entirely unexplored by scholars of international rela-
tions (IR): that is, whether and how leaders can use covert policy activity to com-
municate political messages. We define covert action as policy action undertaken
by a government outside its own territory without official acknowledgment that
most observing audiences do not know about or cannot attribute to the actor. The
existing literature has largely been silent on the specific issue of whether and how
states might communicate intentions through covert action. A careful reading of
signaling literature, moreover, provides reason to be skeptical of the signaling effi-
cacy of covert action. Secrecy is often seen as a tool useful for preventing informa-
tion from reaching domestic or international audiences, and to the extent that any
information is communicated, the conventional wisdom is that audiences will at
best infer weakness rather than resolve from those activities. After all, the argu-
ment goes, words and actions that lack publicity lack accountability and thus can
be issued without incurring domestic or international reputational costs.4 A recent
wave of studies on secrecy in international relations has advanced our understand-
ing of the strategic, operational, and domestic political benefits of using secrecy.
Once again, however, the value of covert actions as policy tools intended to signal
resolve to external actors (either allies or adversaries) has been largely unstudied.
This paper thus puts forth the first systematic treatment of the conditions under
which covert actions communicate resolve, the mechanisms by which such actions
generate credibility, and the inferences adversaries and allies draw from them. The
empirical analysis we conduct to test our theory indicates that this unexplored fea-
ture of covert actions should receive more attention in the burgeoning literature
on secrecy and in international relations more broadly. At the core of our approach
to understanding the role of secrecy is the idea that states share a basic communi-
cative grammar regarding activity in the covert realm that allows leaders to send
targeted messages to external actors. Covert action is rarely completely secret; the

2
Air Force Colonel Ray Sitton, as quoted in ibid., 54.
3
Ibid., 54–61; see also Stephen Wrage, “Major Knight and Cambodia,” in Case Studies in Ethics for Military Leaders (Bos-
ton, MA: Pearson, 2011), 105–7, 221–25.
4
For example, James D. Fearon, “Domestic Political Audiences and the Escalation of International Disputes,” American
Political Science Review 88, no. 3 (September 1994): 577–92, DOI:10.2307/2944796; James D. Fearon, “Signaling For-
eign Policy Interests: Tying Hands versus Sinking Costs,” Journal of Conflict Resolution 41, no. 1 (February 1997): 68–
90, DOI:10.2307/174487; Kenneth A. Schultz, “Domestic Opposition and Signaling in International Crises,” American
Political Science Review 92, no. 4 (December 1998): 829–44, DOI:10.2307/2586306. For a recent analysis of the impor-
tance of reputation, see Alex Weisiger and Keren Yarhi-Milo, “Revisiting Reputation: How Past Actions Matter in Inter-
national Politics,” International Organization 69, no. 2 (March 2015): 473–95, DOI:10.1017/S0020818314000393.
126 A. CARSON AND K. YARHI-MILO

basic contours and many details are often observed by local partners and strategic
adversaries with sophisticated intelligence capabilities. We contend states have
developed a basic interpretive framework that assigns meaning to observed covert
behavior. New covert initiatives and qualitative expansions thereof provide valu-
able insight into a sponsor’s intentions.
This article builds a theoretical framework for understanding signaling in the covert
sphere that develops reasons why states find covert communication both intelligible
(the basic intended message is understood by perceivers) and credible (the message is
believable). We focus on signals of resolve sent via a number of secret policy actions,
including secret military mobilizations, covert aid programs, and discreet military
strikes, which we refer to generally as “covert action.” 5 To shed light on the intelligibil-
ity of covert signaling, we draw on insights about covert or “backstage” communication
dynamics from the sociologist Erving Goffman and insights about how states make
sense of observable deeds from Thomas C. Schelling and Alexander L. George. To shed
light on credibility, we theorize the costs and risks incurred by sponsors of covert action,
much of which is influenced by the secret nature of the activity itself.
We evaluate our theory through careful process tracing of two important conflicts:
the civil wars in Angola (1970–76) and Afghanistan (1979–89) during the Cold War.
Both conflicts hosted a set of nested covert interventions by Soviet and American lead-
ers that provide multiple opportunities for evaluation of the theory. Both conflicts also
benefit from a robust documentary record due to recent declassification. We find that,
consistent with the theory, leaders on both sides used covert action to demonstrate
resolve to the other superpower and to reassure local allies. We also see that leaders
assessed expenditure of resources, domestic risks, and escalatory dynamics of these con-
tests in ways that support our specific theoretical mechanisms.
To be clear, while our case analysis reveals that decision-makers launched covert
interventions in order to signal their resolve, we do not claim leaders use covert action
exclusively to signal in any given case. On-the-ground effects are almost always part of
the story as well. We also do not contend signaling explains all cases of covert action.
Sometimes a secret is just a secret with no signaling purpose. Concealed tunnels dug by
Hamas fighters in Gaza, China’s secrecy-enabled surprise entry into the Korean War,
and the clandestine raid on Osama bin Laden’s compound in Pakistan are examples of
secret policy activity without a signaling component.6
The article features three primary contributions to debates in IR. First, we advance
theoretical debates about how states signal and for whom those signals are intended.
Credible communication under anarchy has been a central research area in IR for deca-
des. We introduce a previously overlooked signaling tool—covert action—and provide

5
Covert action can be defined as “the effort of one government to influence politics, opinions, and events in another
state through means which are not attributable to the sponsoring state.” Elizabeth E. Anderson, “The Security
Dilemma and Covert Action: The Truman Years,” International Journal of Intelligence and CounterIntelligence 11, no. 4
(1998/99): 423. Note the term has specific bureaucratic and legal connotations in the American context that we do
not address.
6
The distinction between secret operations that are clandestine (complete concealment) and covert (identity of spon-
sor is concealed or at least plausibly deniable) in the US foreign policy bureaucracy reflects this.
 SECURITY STUDIES 127

a novel theoretical explanation for why secret policy activity by states can intelligibly
and credibly convey resolve. Our theory and empirics specifically highlight the role of
multiple audiences. Existing studies of signaling resolve tend to focus on dyadic messag-
ing between adversaries; we add the considerations of local allies. Our cases show the
value added, as the appeal of superpower covert action in Angola and Afghanistan is
difficult to understand without knowing the importance of demonstrating resolve to
local partners such as Zaire and Pakistan to Moscow and Washington. The article fur-
thers efforts to expand the aperture of scholarship on signaling to incorporate resolve’s
effects on other audiences like allies in addition to adversaries.7 More broadly, our prop-
ositions about secrecy’s unique signaling effects expand on insights in extant studies of
private threats, offering new ideas that can be adapted and tested by scholars interested
in messages other than resolve and secret activity in settings other than conflict.8
Second, the article makes an important methodological contribution. We test
our claims with declassified primary materials that allow us to carefully trace inten-
tions and inferences and address questions, such as whether a particular policy
action was intended to be a signal of resolve and whether it was interpreted as
such, which are often overlooked in signaling studies. These are not trivial ques-
tions; indeed, policymakers often struggle with understanding a message as much
as diagnosing its credibility.9 This issue of the intelligibility of signals is also essen-
tial when introducing a new domain for signaling. By empirically analyzing the
communicative grammar of both covert signal senders and receivers, we under-
score the value of qualitative archival materials for studies of signaling.
Third, the article has implications for larger debates in IR on signaling and repu-
tation.10 Demonstrating the efficacy of covert signaling contributes to recent
debates about how signals credibly communicate—such as tying hands, sinking
costs, and raising risks—and which mechanisms work best. We explain the appli-
cability of these mechanisms in the covert realm and how that realm alters them.
Moreover, extant studies on reputation have not addressed the impact of secret
policy action and instead focus on how public actions build or hurt a state’s reputa-
tion.11 Yet we find clear evidence that, behind closed doors, leaders consistently

7
See also Scott Wolford, “Showing Restraint, Signaling Resolve: Coalitions, Cooperation, and Crisis Bargaining,” Ameri-
can Journal of Political Science 58, no. 1 (January 2014): 144–56, DOI:10.1111/ajps.12049.
8
For example, see Anne E. Sartori, “The Might of the Pen: A Reputational Theory of Communication in International
Disputes,” International Organization 56, no. 1 (February 2002): 121–49, DOI:10.1162/002081802753485151; Shuhei
Kurizaki, “Efficient Secrecy: Public versus Private Threats in Crisis Diplomacy,” American Political Science Review 101,
no. 3 (August 2007): 543–58; Robert F. Trager, “Diplomatic Calculus in Anarchy: How Communication Matters,” Amer-
ican Political Science Review 104, no. 2 (May 2010): 347–68, DOI:10.1017/S0003055410000158.
9
On assessing intentions, see Keren Yarhi-Milo, Knowing the Adversary: Leaders, Intelligence, and Assessment of Inten-
tions in International Relations (Princeton, NJ: Princeton University Press, 2014). For an example of failed intelligibility,
see Scott D. Sagan and Jeremi Suri, “The Madman Nuclear Alert: Secrecy, Signaling, and Safety in October 1969,”
International Security 27, no. 4 (Spring 2003): 150–83.
10
Matthew Fuhrmann and Todd S. Sechser, “Signaling Alliance Commitments: Hand-Tying and Sunk Costs in Extended
Nuclear Deterrence,” American Journal of Political Science 58, no. 4 (October 2014): 919–35, DOI:10.1111/ajps.12082.
For an early and influential contribution, see Fearon, “Signaling Foreign Policy Interests.”
11
For example, see Jonathan Mercer, Reputation and International Politics (Ithaca, NY: Cornell University Press, 1996);
Daryl G. Press, Calculating Credibility: How Leaders Assess Military Threats (Ithaca, NY: Cornell University Press, 2007);
Weisiger and Yarhi-Milo, “Revisiting Reputation.”
128 A. CARSON AND K. YARHI-MILO

worry about loss of reputation as a result of their actions in the covert sphere and
possible reputation recovery via covert action.
The article proceeds as follows. The first section reviews existing research on signal-
ing and secrecy. The second section unpacks our theory of signaling via covert action
and derives conditions and hypotheses for the empirical analysis. We then discuss our
research design in section three and in section four assess the role of signaling in a set of
nested covert actions during conflicts in Angola and Afghanistan. We conclude by
reflecting on generalizability, policy implications, and future research.

Muted Messages: Extant Research On Secrecy and Signaling

Our interest lies in the signaling potential of “covert military action” (or simply
“covert action”). We use this term to refer to a variety of secret foreign policy
actions that may be administered by military or intelligence bureaucracies. Their
unifying feature is any attempt by a state to use intelligence or military policy
instruments to influence outcomes abroad in a way that conceals and renders deni-
able the role of the sponsoring state for most audiences.
At the core of IR research on interstate communication is the concept of costly
signals, or how words and behavior by states convey messages in ways that help
observers distinguish between different kinds of “types.”12 For example, leaders
can demonstrate they are a more resolved type by engaging in expensive actions
(sinking costs), deliberately increasing dangers such as the risk of crisis escalation
(raising risks), and/or engaging constraining institutions or political dynamics to
reduce future flexibility (tying hands). Following Fearon’s influential work on audi-
ence costs, most studies in the last two decades have implicitly or explicitly
addressed the signaling power of publicity.13 Making verbal and behavioral coer-
cive threats public can generate credibility by activating political constraints that
tie hands and/or generate greater risks and sink costs.14 James D. Morrow’s sum-
mary of costly signaling, for example, mentions reassurance in the Arab–Israeli

12
The literature on signaling resolve is vast; key reference points are Thomas Schelling, The Strategy of Conflict (Cam-
bridge, MA: Harvard University Press, 1960); Fearon, “Signaling Foreign Policy Interests.” For signaling restraint, see
Andrew Kydd, “Trust, Reassurance, and Cooperation,” International Organization 54, no. 2 (Spring 2000): 325–57;
Charles L. Glaser, Rational Theory of International Politics: The Logic of Competition and Cooperation (Princeton, NJ:
Princeton University Press, 2010).
13
Fearon, “Domestic Political Audiences and the Escalation of International Disputes.” For an earlier analysis that also
includes mechanisms not limited to domestic audiences, see Thomas C. Schelling, Arms and Influence, new ed. (New
Haven, CT: Yale University Press, 1966). Studies testing, extending, refining, and critiquing the use of publicity to tie
hands include Kenneth A. Schultz, “Looking for Audience Costs,” Journal of Conflict Resolution 45, no. 1 (February
2001): 32–60; Michael Tomz, “Domestic Audience Costs in International Relations: An Experimental Approach,” Inter-
national Organization 61, no. 4 (Fall 2007): 821–40, DOI:10.1017/S0020818307070282; Jessica L. Weeks, “Autocratic
Audience Costs: Regime Type and Signaling Resolve,” International Organization 62, no. 1 (January 2008): 35–64,
DOI:10.1017/S0020818308080028; Jack Snyder and Erica D. Borghard, “The Cost of Empty Threats: A Penny, Not a
Pound,” American Political Science Review 105, no. 3 (August 2011): 437–56, DOI:10.1017/S000305541100027X; Marc
Trachtenberg, “Audience Costs: An Historical Analysis,” Security Studies 21, no. 1 (January 2012): 3–42, DOI:10.1080/
09636412.2012.650590.
14
An example of linking publicity to risk raising is Jessica Chen Weiss, “Authoritarian Signaling, Mass Audiences, and
Nationalist Protest in China,” International Organization 67, no. 01 (January 2013): 1–35, DOI:10.1017/
S0020818312000380.
 SECURITY STUDIES 129

peace process and argues that Anwar Sadat’s public bid to reassure Israel was costly
and therefore credible while Arab leaders’ secret offers of peace “did not carry the
same credibility as Sadat’s signal precisely because they were private.”15 While the
rationalist literature on signaling has not eliminated the possibility of costly
“covert” action, most of the mechanisms for costly signaling discussed in the con-
text of this literature (for example, significant arms reductions, membership in
binding international institutions, or invasion of third countries) are inherently
“public” in nature. This leaves open, both theoretically and empirically, our funda-
mental question of when and how covert actions might communicate resolve.
A growing literature on secrecy in IR, on the other hand, has advanced various
important arguments as to when and why states use covert military operations.16
The majority of these studies, however, is either silent about the potential signaling
effects of covert operations or gives us reasons to be pessimistic about it. For exam-
ple, Scott D. Sagan and Jeremi Suri’s study of the effectiveness of a secret American
nuclear alert by President Richard Nixon to signal resolve to Soviet leaders con-
cludes that Nixon’s attempt to signal resolve to Soviet leaders is an example of
“cheap signals” that “avoid public commitments” and “can be explained away if
discovered as military exercises, and therefore do not raise the stakes that leaders
face if they back down in a crisis.”17 Still, some scholars have advanced a more pos-
itive link between secrecy and signaling. Beginning with Anne E. Sartori’s study of
reputation for honesty, a cluster of research on private diplomacy suggests verbal
or written threats issued behind closed doors can still credibly express resolve or
reassurance.18 Unlike the focus of our paper, however, those studies explore secret
but explicit and direct communication among leaders.
Finally, other studies have looked at secrecy as an alternative to signaling
resolve. Brian Lai’s analysis of military mobilization finds private mobilization
lacks risk-raising properties and therefore cannot signal.19 Perceiving private
mobilization as a military move that is entirely unobservable to external audiences,
he argues that “private” mobilization produces a tactical advantage if an opponent

15
James D. Morrow, “The Strategic Setting of Choices: Signaling, Commitment, and Negotiation in International Poli-
tics,” in Strategic Choice and International Relations, ed. David A. Lake and Robert Powell (Princeton, NJ: Princeton
University Press, 1999), 87–88.
16
Ward Thomas, “Norms and Security: The Case of International Assassination,” International Security 25, no. 1 (July
2000): 105–33, DOI:10.1162/016228800560408; Alexander B. Downes and Mary Lauren Lilley, “Overt Peace, Covert
War?: Covert Intervention and the Democratic Peace,” Security Studies 19, no. 2 (April 2010): 266; Erik Gartzke and
Jon R. Lindsay, “Weaving Tangled Webs: Offense, Defense, and Deception in Cyberspace,” Security Studies 24, no. 2
(April 2015): 316–48, DOI:10.1080/09636412.2015.1038188; Austin Carson, “Facing Off and Saving Face: Covert Inter-
vention and Escalation Management in the Korean War,” International Organization 70, no. 1 (December 2016): 103–
31, DOI:10.1017/S0020818315000284.
17
Sagan and Suri, “Madman Nuclear Alert,” 155. See also William Burr and Jeffrey Kimball, “Nixon’s Secret Nuclear Alert:
Vietnam War Diplomacy and the Joint Chiefs of Staff Readiness Test, October 1969,” Cold War History 3, no. 2 (Janu-
ary 2003): 113–56.
18
Sartori, “Might of the Pen”; Anne E. Sartori, Deterrence by Diplomacy (Princeton, NJ: Princeton University Press, 2005);
Kurizaki, “Efficient Secrecy”; Trager, “Diplomatic Calculus in Anarchy”; Kristopher W. Ramsay, “Cheap Talk Diplomacy,
Voluntary Negotiations, and Variable Bargaining Power,” International Studies Quarterly 55, no. 4 (December 2011):
1003–23, DOI:10.1111/j.1468-2478.2011.00687.x.
19
Brian Lai, “The Effects of Different Types of Military Mobilization on the Outcome of International Crises,” Journal of
Conflict Resolution 48, no. 2 (April 2004): 211–29.
130 A. CARSON AND K. YARHI-MILO

is caught off guard with a surprise attack. Branislav L. Slantchev models misrepre-
sentation in war and China’s secret entry into the Korean War. He links secrecy to
operational military effects (that is, surprising one’s adversary) and finds the “per-
verse incentives” from surprise attacks can make leaders “unwilling to send costly
signals even when they could have done so.”20 Older studies similarly link the use
of secrecy to surprising one’s adversary and deception rather than any role in sig-
naling.21 A similar theme of covert action as an alternative signaling is present in
theories of secrecy’s value as a way to relieve leaders from domestic criticism.22
While both of these literatures do not take head-on the question of whether states
might also signal with covert tools, they have left the question open by focusing on
the noncommunicative properties of secrecy.
In sum, the existing literature for the most part is either silent or pessimistic
about the ability of covert military actions to signal resolve. Our emphasis on
covert military action as a signaling tool leads us to two important points of depar-
ture: (1) theorizing how covert action can intelligibly signify resolve among leaders
despite the use of secrecy and the ambiguity of interpreting state behavior; and (2)
focusing on the unique properties of secret military action rather than the binding
force of diplomatic messages. Our theory therefore addresses both the intelligibility
(ability to be understood) and potential credibility (worthiness of belief) of covert
action as a signal of resolve.

Theory: Intelligibility and Credibility in Backstage Communication

We argue states can share a basic communicative grammar regarding activity in
the covert realm that allows leaders to send targeted messages. This is possible
because covert activity conducted over time and in a sensitive geostrategic area is
rarely a complete secret. Other states with sophisticated intelligence capabilities
and/or those that are located close to the action often know or are informed of the
basic contours.23 We believe states can use this partial observability as a signaling
opportunity rather than merely regard it as an inconvenience. Our theory thus

20
Branislav L Slantchev, “Feigning Weakness,” International Organization 64, no. 3 (Summer 2010): 384.
21
Robert Axelrod, “The Rational Timing of Surprise,” World Politics 31, no. 2 (January 1979): 228–46; Richard K. Betts,
“Surprise Despite Warning: Why Sudden Attacks Succeed,” Political Science Quarterly 95, no. 4 (Winter 1980–81):
551–72.
22
David N. Gibbs, “Secrecy and International Relations,” Journal of Peace Research 32, no. 2 (May 1995): 213–28; Mat-
thew A. Baum, “Going Private: Public Opinion, Presidential Rhetoric, and the Domestic Politics of Audience Costs in
U.S. Foreign Policy Crises,” Journal of Conflict Resolution 48, no. 5 (October 2004): 603–31; John M. Schuessler, “The
Deception Dividend: FDR’s Undeclared War,” International Security 34, no. 4 (Spring 2010): 133–65, DOI:10.1162/
isec.2010.34.4.133; Jonathan N. Brown and Anthony S. Marcum, “Avoiding Audience Costs: Domestic Political
Accountability and Concessions in Crisis Diplomacy,” Security Studies 20, no. 2 (April 2011): 141–70; Dan Reiter,
“Democracy, Deception, and Entry into War,” Security Studies 21, no. 4 (October 2012): 594–623. For democratic
peace theory, see David P. Forsythe, “Democracy, War, and Covert Action,” Journal of Peace Research 29, no. 4
(November 1992): 385–95; Downes and Lilley, “Overt Peace, Covert War?”
23
Robert Jervis, The Logic of Images in International Relations (New York: Columbia University Press, 1970), 27–29, 35.
Jervis analyzes observed secret activity that was not anticipated to be observable. We theorize covert activity that is
likely to be detected by at least some states. For (dis)incentives for revealing this detected activity, see Carson, “Fac-
ing Off and Saving Face”; Austin Carson and Allison Carnegie, “The Spotlight’s Harsh Glare: Rethinking Publicity and
International Order,” n.d, available from authors upon request.
 SECURITY STUDIES 131

proceeds in two steps. First we describe how observed covert action can convey
intelligible messages about intentions to other leaders, such as “we intend to defend
our ally” or “we intend to prevent the emergence of a new nuclear neighbor.” We
then address why any message of resolve should be credible in the eyes of targeted
audiences.
Sociologist Erving Goffman analyzed everyday social encounters to understand
how we tailor our impressions to be socially appropriate and to avoid embarrass-
ment. Invoking an analogy from theater performance, Goffman argued social life
reflected dynamics on both the “frontstage” and “backstage.” The backstage is “a
place, relative to a given performance, where the impression fostered by the perfor-
mance is knowingly contradicted as a matter of course” and that safeguards the
coherence and consistency of the visible (frontstage) performance. 24 Yet Goffman
also notes there is partial visibility among others on the backstage because “illu-
sions and impressions are openly constructed” in that space.25 Psychologists draw-
ing on similar insights note communicators that confront multiple audiences can
use messages that contain “covert hidden content” to target one and avoid others.26
These clever communications draw on familiarity and experience, allowing “shared
knowledge” to create “successful covert communication” through a shared inter-
pretive frame.27
We view state behavior vis-a-vis the covert sphere in similar terms. State-spon-
sored covert action is often visible to other states with access to the “backstage” via
their intelligence capabilities and/or proximity to the site of activity. This permits
sponsors to covertly communicate messages to others with access to the backstage.
But how do other leaders know what the intended message is? This is a question of
intelligibility, conceptually prior to the issue of credibility that is addressed in exist-
ing literature and theorized below.28 The intelligibility of covert signaling was first
suggested by Schelling and George. Schelling’s bargaining framework introduces
“focal points” and “salient thresholds” as ways adversaries can communicate
through deeds rather than words. Focal points and salient thresholds are coordi-
nating features of a situation that allows observable behavior to express intentions
like “they are resolved” or “they seek to fight limited war.” Schelling therefore sug-
gests that qualitative changes in how states act in a crisis or war can send intelligi-
ble messages about intentions.29 George applies these concepts to crisis behavior
including covert action.30 Analyzing the Cold War, he argues American and Soviet
leaders developed “an ad hoc set of ground rules” regarding involvement in outside

24
Erving Goffman, The Presentation of Self in Everyday Life (Garden City, NY: Doubleday, 1959), 112–13.
25
Ibid. For adapting Goffman’s backstage metaphor to IR, see Carson, “Facing Off and Saving Face.”
26
John H. Fleming et al., “Multiple Audience Problem: A Strategic Communication Perspective on Social Perception,”
Journal of Personality and Social Psychology 58, no. 4 (April 1990): 593–609, DOI:10.1037/0022-3514.58.4.593.
27
Ibid., 599.
28
On intelligibility in signaling, see Emanuel Adler and Vincent Pouliot, “International Practices,” International Theory 3,
no. 1 (February 2011): 1–36.
29
Schelling, Arms and Influence, 135.
30
Alexander L. George, “Crisis Prevention Reexamined,” in Managing U.S.-Soviet Rivalry: Problems of Crisis Prevention,
ed. Alexander L. George (Boulder, CO: Westview Press, 1983).
132 A. CARSON AND K. YARHI-MILO

conflicts.31 This framework meant “qualitative changes [that] represent the cross-
ing of a new threshold,” such as “the introduction of a new weapon into combat,
initiation of attacks against a type of target heretofore not under fire, intervention
by external forces and/or a shift in their participation from an advisory or training
role to a more important supporting or combat role,” came to signify readily inter-
pretable messages about their intentions in the local contest.32
Collecting these insights, we argue using covert action is akin to states choosing
to retreat to a metaphorical backstage; this permits communication with select
states that share access to that backstage. Drawing on Schelling and George, we
further argue covert action is intelligible because it contains a range of salient, qual-
itative thresholds that are mutually meaningful as symbols of a sponsor’s resolve.
But who receives these covert signals? And why are these covertly conveyed
messages believable? We now turn to the issue of credibility.

Who Witnesses? Strategic Adversaries and Local Allies

We specifically postulate that covert military action tends to be visible to two spe-
cific audiences that therefore constitute the targeted audience for signaling. The
first consists of strategic adversaries. Covert action is often visible to a state’s main
strategic rival due to the robust data provided by modern intelligence systems.
Intelligence bureaucracies draw on a wide variety of intelligence sources (for exam-
ple, human, signals, technical) to monitor the public and private activities of other
states. As a result, the basic contours of covert behavior are often visible to other
states. The long-term rivalry and large resources of Soviet and American govern-
ments, for example, led them to develop sophisticated intelligence collection capa-
bilities to monitor one another’s activities. This was no idiosyncrasy of bipolarity;
historians of the interwar period, for example, have documented a strikingly simi-
lar level of transparency among rival adversaries in the covert activity that accom-
panied the Spanish Civil War.33 More recently, covert involvement by various
outside powers in the Syrian Civil War and in Ukraine suggest states continue
carefully monitoring covert activity.34
The second audience for whom covert action tends to be visible is leaders of
local allies. The implementation of covert action often relies on cooperation with
local partners. A covert airstrike may use the airspace of an ally; secret aid ship-
ments on their journey to a distant rebel group may need to enter ports of partners;
even military mobilization can alter military routines at bases in an ally’s territory

31
Ibid., 389.
32
Ibid., 390.
33
Willard C. Frank Jr., “Politico-Military Deception at Sea in the Spanish Civil War, 1936–1939,” Intelligence and National
Security 5, no. 3 (Autumn1990): 84–112.
34
See, for example, a report on the “shadow war” featuring Iran, Israel, and the United States in Nicholas Kulish and
Jodi Rudoren, “Murky Plots and Attacks Tied to Shadow War of Iran and Israel,” New York Times, 8 August 2012,
http://www.nytimes.com/2012/08/09/world/middleeast/murky-plots-and-attacks-tied-to-shadow-war-of-iran-and-
israel.html.
 SECURITY STUDIES 133

tipping off local leaders. While extra concealment efforts can allow deception of
these local partners, states may need or prefer to keep friendly leaders informed. A
revealing remark from the American secretary of state Dean Acheson demon-
strates that as part of their mobilization efforts early in the Korean War, American
leaders secretly repositioned atomic weapons and bombers to bases in both Europe
and the Pacific. Knowing that assets at military bases on British territory would be
used, Acheson argued one benefit of this mobilization was to “impress the British
with America’s determination to prevail in Korea” in a “demonstration of
resolve.”35

Credibility: Mechanisms of Covertly Signaling Resolve

What makes a message of resolve credible? To be clear, we do not argue that signaling
properties are independent from on-the-ground effects. Consider, for example, two dif-
ferent interpretations of a recent covert action: Israel’s secret strike against Syria’s
nuclear reactor in 2007. In both versions, Israel is constrained from overt military
action by concerns for regional stability but motivated to preemptively shut down
Syrian nuclear development. In one interpretation, Israeli leaders proceed with the
strike solely because it will set back Syria’s nuclear program. In a second version, Israeli
leaders proceed both because it damages Syria’s nuclear program and sends a message
to other nuclear proliferators in the region (that is, Iran), confident that intelligence
monitoring will reveal its demonstration of resolve to both rivals and friends. We
embrace this second view. (The second view of motivation for covert action is also rele-
vant for the Stuxnet worm that damaged Iran’s industrial and nuclear facilities in
2010). We posit that only analyzing operational effects, or the political benefits of hiding
activity from particular audiences, misses an important message-sending component.
Our intuition of the relevance of signaling is supported by subsequent reporting on the
signaling motive in the 2007 Israel–Syria episode.36
We identify three mechanisms that give credibility to signals of resolve in the covert
sphere. In doing so we note the important distinction between costs/risks that derive
from the action itself versus from the unique features of acting in secrecy.
Sinking Costs. One reason why states using covert action distinguish themselves
from more cautious “types” is because such activity often carries significant expen-
diture of non-recoverable resources. A program to covertly provide weaponry to a
local ally, for example, can involve the expenditure of millions of dollars on sophis-
ticated weaponry, deplete national stocks, and potentially spread technical knowl-
edge about such weaponry to other states. Similarly, a partial military mobilization
or deployment of a specialized brigade abroad can create a strategically significant

Roger Dingman, “Atomic Diplomacy during the Korean War,” International Security 13, no. 3 (Winter 1988–1989): 58.
35

Erich Follath and Holger Stark, “The Story of ‘Operation Orchard’: How Israel Destroyed Syria’s Al Kibar Nuclear Reac-
36

tor,” Spiegel Online, February 11, 2009, http://www.spiegel.de/international/world/the-story-of-operation-orchard-

how-israel-destroyed-syria-s-al-kibar-nuclear-reactor-a-658663.html; David Makovsky, “The Silent Strike,” New Yorker,
17 September 2012, http://www.newyorker.com/reporting/2012/09/17/120917fa_fact_makovsky.
134 A. CARSON AND K. YARHI-MILO

opportunity cost by diverting those resources from normal strategic purposes.

Sunk costs can also grow over time when a given covert action swells to include
more expansive and expensive operations. A state’s covert military interventions,
which we focus on in empirical testing, may expand in terms of lethality, scope,
and the directness of involvement.37 Greater resolve should be communicated as a
state moves toward the extreme end of the intervention spectrum, as in greater
lethality (for example, from small arms to surface-to-air missile systems), expand-
ing geographic scope, and more direct combat involvement.38
The use of secrecy itself can generate unique sunk costs. One observable cost is
logistical: the act of concealing is effortful and can require circuitous planning.
More significant, however, are the costs a sponsor can incur from acting secretly
when domestic and other constraints remove overt alternatives from the menu of
options. Covert action can permit leaders to employ a more “creative” way to
address a given class of security threats. For example, leaders using covert means
can work with cooperative partners that would be otherwise off limits. Legal con-
cerns, both domestic and international, can be circumvented if states employ
covert signaling methods. A democratic leader, for example, might covertly partner
with a ruthless dictator; an autocratic leader might use secrecy to work with a part-
ner from a rival ideology.39 While the existence and kinds of constraints vary for
different regime types, most democratic leaders and many autocrats can find acting
covertly to be the only practical method for proactively addressing a crisis, there-
fore offering unique sunk cost opportunities.
Counter-Escalation Risks. Using covert action to signal resolve can also appear
credible because of its impact on the risk of crisis escalation. Other states often
react to covert action in ways that can scale up to a serious confrontation. Such
reactions can risk an action-reaction sequence that intensifies or widens a crisis or
war.40 A covert military strike like that used by Israel, for example, can significantly
raise the risk of crisis escalation by inviting reprisal by the target or additional
covert or overt action by other third-party powers.
Like sunk costs, secrecy itself creates some unique escalation risks. As we note
above, secrecy reduces domestic and international constraints on leaders. This per-
mits riskier as well as costlier activity; leaders using covert signaling tools can be
free to engage in more aggressive behavior. For example, a state might be more
willing to tolerate an operation with collateral damage and civilian casualties, or
one involving cooperation with a rival that shares an interest in confronting a third

37
On this variation, see George, “Crisis Prevention Reexamined,” 389–93.
38
A report on foreign involvement in Syria notes “the distinction between lethal and non-lethal weapons is a crucial
one for the governments involved in their supply.” See Louisa Loveluck, “What’s Non-Lethal about Aid to the Syrian
Opposition?” Foreign Policy Blogs, 20 September 2012, http://mideastafrica.foreignpolicy.com/posts/2012/09/20/
whats_non_lethal_about_aid_to_the_syrian_opposition.
39
An example of the latter is postrevolutionary Iran using secrecy to quietly work with Israel and the United States to
oppose Iraq under Saddam Hussein. Trita Parsi, Treacherous Alliance: The Secret Dealings of Israel, Iran, and the United
States, 1st ed. (New Haven, CT: Yale University Press, 2007).
40
On horizontal versus vertical escalation, see Richard Smoke, War: Controlling Escalation (Cambridge, MA: Harvard Uni-
versity Press, 1977).
 SECURITY STUDIES 135

party. Moreover, secrecy often entails much greater reliance on other states—allies
and local intermediaries—for implementation. This can add to risks of unautho-
rized or unintended escalation.41 As with the sunk cost mechanism, larger and
more expansive covert action will tend to create large counter-escalation risks.
Domestic Risks. Lastly, in the presence of significant domestic opposition, covert
action allows leaders to observably demonstrate to foreign audiences they are more
hawkish and risk-acceptant than domestic political rivals. As existing secrecy
research shows, states often use secrecy for military activity to avoid domestic
political complications. When observers are aware of domestic constraints this cre-
ates an overlooked signaling opportunity. A state observing a new or intensified
covert action despite domestic constraints sees an up-front demonstration that the
sponsor is willing to act creatively and incur their own domestic risks to accom-
plish a policy goal. Acting covertly itself underscores the presence of domestic con-
straints but simultaneously shows the capacity to circumvent those. This dynamic
does not depend on leaks; whether or not exposure is likely or actually takes place,
observing covert action provides tangible evidence that more aggressive leadership
controls decision-making.42
Additional domestic risk-generation is created by the potential for and actual
exposure of covert action. Unwanted exposure can result from third party disclo-
sures by journalists, nongovernmental organizations, dissatisfied bureaucrats, or
other states who detect the program. A leaked covert action can trigger domestic
political backlash and prompt policy interference by political rivals. This is addi-
tional basis for an inference of resolve.43 Lastly, the domestic risk mechanism may
initially appear limited to democratic leaders with free presses. However, studies of
autocracies suggest many leaders are simply accountable to specific elite factions
(for example, other party members, the military) who may be more or less dovish
than the leader.44 Thus, it is plausible covert action by at least some autocratic lead-
ers can be a domestically risky proposition.

Conditions Favoring Covert Signaling

While these mechanisms provide a plausible logic for why covert action can signal
resolve credibly, one can imagine a very different inference: weakness. Leaders
choosing to signal covertly do not make that choice in a vacuum; they have overt
signaling methods as well. Moreover, while overt alternatives lack secrecy-specific
costs and risks, they can have some similar effects. Choosing a public military

41
Schelling, Arms and Influence, 89–99.
42
Put differently, a regime with total media control and a known faction of “doves” signals resolve with covert action
because it is an observable indication that the “hawks” control policy.
43
This is a corollary to the mechanism of reassurance in Keren Yarhi-Milo, “Tying Hands behind Closed Doors: The Logic
and Practice of Secret Reassurance,” Security Studies 22, no. 3 (July 2013): 405–35, DOI:10.1080/
09636412.2013.816126.
44
For example, see Weeks, “Autocratic Audience Costs”; however, party insiders in an autocracy may be better placed
to discover such covert action. See Brown and Marcum, “Avoiding Audience Costs.”
136 A. CARSON AND K. YARHI-MILO

mobilization or lethal aid program, for example, can sink costs and raise risks in
many of the ways a covert version would; in fact, an overt version may involve
more robust costs and risks in some ways. Moreover, existing coercive diplomacy
studies suggest the publicity of these alternatives could make their message more
credible by tying leaders’ hands. These overt alternatives provide a logic for the
opposite inference about covert action. Even if intelligible, observers might infer
restraint or even weakness from covert action. A sponsor opting to use covert tools
might be seen as rejecting the boldest way of expressing resolve in favor of a more
cautious and flexibility-preserving option.
These comparisons beg an important question: under what conditions does covert
action generate a signal of resolve rather than weakness?45 Answering this question
both clarifies the logic of our theory and identifies the kinds of situations and historical
eras in which leaders should be tempted to use covert signaling for resolve. Our answer,
foreshadowed in the previous section, focuses on the degree of leader constraints. In
short, observably constrained leaders can use covert action to signal resolve; in contrast,
unconstrained leaders invite inferences of weakness. Constraints influence what
observing audiences compare covert action against. When leaders are constrained from
using more robust overt options by domestic or other sources, target audiences will
tend to discount the practicality of overt alternatives and compare observed covert
action to the more plausible alternatives of inaction or diplomacy. The cost and risk
properties we describe above therefore tend to produce an inference of resolve. If a
leader is unconstrained, however, observers are free to compare covert action to more
robust and binding overt alternatives. An inference of weakness is more likely as a
result. Lastly, we also consider an alternative inference. Constrained leaders resorting to
covert action may be seen as unable to escalate or maintain their program in light of
those constraints. This would reverse the relationship: more constrained leaders using
covert action would be seen as weak.
What are the constraints that can remove overt options and, as a result, produce
an inference of resolve rather than weakness from covert action? We specify con-
straints of two basic kinds: domestic and international.46 The domestic environ-
ment can create legal and practical barriers to leaders using overt alternatives. In
the American context, for example, congressional restrictions on foreign aid can
prevent presidents from overtly providing aid to certain groups and governments;
such aid can violate an embargo or be illegal. Similarly, a dominant domestic anti-
war political climate can functionally eliminate the option to overtly use military
force. In the American context, the periods following the Vietnam War and the
recent wars in Iraq and Afghanistan have been in conditions in which other

45
Note that a necessary condition for a covert signal of any kind is visibility to at least one target audience. Partial
observability to an adversary is most likely within long-term rivalries in which intelligence capabilities are regularly
directed toward one another. Similarly, partial observability to allies is most likely when the sponsor is part of a net-
work of partners who can be logistically involved in implementation. Covert actions by states with few partners and
outside a long-term rivalry are therefore least likely to have possible covert signaling effects. Small, one-off covert
operations implemented unilaterally are also unlikely to enable signaling.
46
Our thanks to two anonymous reviewers for help clarifying these issues.
 SECURITY STUDIES 137

audiences likely perceived a domestic constraint on overt uses of force. Note that
while domestic constraints are most likely in consolidated democracies, they can
arise in some autocratic regimes as well.47
Second, several kinds of international constraints can render overt options infeasi-
ble. Some originate in the signaler’s relationship with its adversary. Overtly signaling
resolve may be impractical, for example, if adversaries are engaged in sensitive peace
negotiations or have cooperative initiatives that both seek to protect and would be
endangered by a public affront. Crises in the shadow of extreme risks of conflict
escalation may also functionally eliminate overt options.48 For example, covert sig-
nals sent by Iran or the United States during their sensitive nuclear negotiations
were likely compared to doing nothing rather than to overt alternatives. Interna-
tional constraints can also originate from partner state considerations. A sponsor
may not have the option to openly provide aid or use force if partnering states nec-
essary for implementation face acute domestic political risks in their own countries.

Hypotheses
Our empirical testing focuses specifically on covert interventions during ongoing
civil or interstate conflicts. Any external intervention features an outside power
sending lethal military assistance (for example, weapons and related equipment)
and/or combat units (for example, personnel performing combat or combat sup-
port roles) to a local ally.49 Such interventions can be public, or “overt”; an outside
power can publicly announce and acknowledge the lethal aid and any combat per-
sonnel they send to a local ally. Interventions can also be secret, or “covert”; a state
can send aid and/or personnel without an official announcement and take meas-
ures to conceal its role.
The discussion thus far suggests several hypotheses about signaling and covert
interventions listed below. First, we assess a null hypothesis that reflects a skeptical
view based on existing studies linking credibility in signaling resolve and publicity.

H1 (Null). Strategic adversaries and local allies that detect any covert activity will
disregard its significance as an indicator of resolve due to its absence of pub-
licity and therefore political constraint.

We also assess a set of hypotheses that summarize our own claims about the
intelligibility and credibility of covert signals of resolve. We decompose the infer-
ence hypothesis to reflect the importance of both initiating and expanding covert
action.

47
Jessica L. Weeks, “Autocratic Audience Costs: Regime Type and Signaling Resolve,” International Organization 62, no.
1 (January 2008): 35–64.
48
Carson, “Facing Off and Saving Face.”
49
On defining intervention, see Patrick M. Regan, “Conditions of Successful Third-Party Intervention in Intrastate Con-
flicts,” Journal of Conflict Resolution 40, no. 2 (June 1996): 342; 339 n. 1, DOI:10.1177/0022002796040002006.
138 A. CARSON AND K. YARHI-MILO

H2 (Intention to signal resolve). Intervening states should expect covert military

assistance to convey resolve to local allies and strategic adversaries.
H3 (Inferring credible resolve). Strategic adversaries and local allies may draw
inferences of resolve when witnessing covert interventions and should find
them credible.
H3a. Initiation of a new intervention will lead to updated inferences of
resolve based on the willingness to invest resources and tolerate both
escalation and domestic risks.
H3b. Changes toward greater (lower) lethality, larger (smaller) scope, and
more (less) direct involvement notwithstanding domestic risks should
be the basis of inferring greater (lesser) resolve.

Finally, we consider a fourth hypothesis that reflects the potential inference

weakness. As we note, under some conditions observers might discount the
costs and risks incurred by a covert military intervention if an overt alterna-
tive was feasible and more comparatively costly/risky. The basis for this infer-
ence is a second order insight about signaling tools another leader chose not
to utilize.

H4 (Inferring weakness). Strategic adversaries and local allies may draw infer-
ences of irresolution or weakness when observing covert intervention.

Case Studies and Research Design

Our hypotheses focus on both states’ intentions to communicate resolve
through covert signaling and the inferences observers should draw therefrom.
Our dependent variable is therefore not the conflict outcome but rather the
beliefs leaders hold about how they can communicate resolve and inferences
others draw given the kind of covert signal issued.50 We define resolve as
determination to stand firm to protect one’s interest. In intervention scenarios,
standing firm may be further linked to specific behavioral goals that vary in
two ways. First, signaling states may seek to display their resolve along a con-
tinuum that ranges from narrowly situational to broadly dispositional. That is,
a state engaging in covert intervention may want to display their willingness
to stand firm specifically in the crisis/theater in which the intervention is tak-
ing place. This was Kissinger’s goal in the opening anecdote: covertly bombing
Cambodia was intended to show resolve to escalate further in the Vietnam
War. More ambitiously, a signal sender may also hope a covert action shows
their willingness to oppose an adversary in a class of other similar crises in

50
Whether the target of an intelligible and credible signal of resolve ultimately backs down is an important but distinct
question. It is also not straightforwardly determined by the signaling process. The ultimate outcome of covertly
waged geopolitical struggle can be overdetermined such that many factors besides assessments about resolve influ-
ence whether and how a given state’s goals are reached.
 SECURITY STUDIES 139

the future. Both constitute attempts to signal resolve. In addition to situa-

tional/dispositional varieties, states signaling resolve may link their effort to
specific behavioral goals regarding their adversary. To simplify, a signal sender
may intend covert action to demonstrate their willingness to undermine a
rival’s exercise of power; a more ambitious goal would be to use covert action
to signal resolve to reverse a rival’s extant progress, force their backing down,
and even expel them from a given region. Importantly, whether the goal is sit-
uational or dispositional resolve and modest or significant policy changes,
leaders hope covert action provides tangible proof to local allies and/or strate-
gic rivals that they are determined to stand firm in some way.51 To empirically
evaluate these arguments we present primary evidence on both signal sender
and perceiver to assess the intention to signal, the intelligible receipt of that
signal, and the ultimate inference of resolve (that is, credibility) from that sig-
nal. Admittedly, gaps in the evidentiary record preclude us from closely track-
ing the inference process of perceivers at places. Nevertheless, there is
considerable documentary support to assess our hypotheses.
We test our claims in a set of nested cases of external military interventions
within two conflicts during the Cold War: Angola and Afghanistan. Both conflicts
fall within the scope of the theory. Angola and Afghanistan involved long-term
rivals—the United States and the Soviet Union—regularly monitoring each other’s
activities. The use of covert interventions by both sides also involved local allies as
an important potential audience. Both sets of cases feature multiple observations
including within-case variation over time in our main independent variable (that
is, the scope and nature of the covert military action) as well as between-case varia-
tion in domestic risks. Importantly, analyzing changes over time allows us to assess
shifts in resolve, intentions, and inferences while holding larger contextual factors
constant. The relatively robust declassified documentary record for each conflict
also gives unique access to internal intentions and perceptions, though the avail-
ability of fewer records on the Soviet side limits our ability to make equally defini-
tive claims at times. Analyzing conflicts in which American and Soviet leaders play
both acting and observing/inferring roles allows us to draw on the more robust
American record to shed light on specific questions like inferences of resolve versus
weakness.
Lastly, these two conflicts host cases that are by no means “easy” tests of the the-
ory. Most obviously, Afghanistan and Angola were considered on the extreme
periphery of the Cold War rivalry until the interventions themselves transformed
perceived stakes. This makes the conflicts unlikely places for attempts to send
covert signals and inferences of credible resolve for two reasons. Logistically, detec-
tion of covert activity by American and Soviet intelligence was comparatively less
likely in these areas; strategic adversary signaling was therefore less likely.

51
On situational versus dispositional resolve, see Joshua Kertzer, Resolve in International Politics (Princeton, NJ: Prince-
ton University Press, 2016).
140 A. CARSON AND K. YARHI-MILO

Politically, the reduced reputational implications of these conflicts should reduce

the inferential value of any signals to both one another and to local allies. Thus,
any findings that covert action was monitored effectively, interpreted correctly,
and seen as credible is especially strong support.

Angola, 1970–76
As Angola’s struggle for self-government gained momentum, three national inde-
pendence movements emerged: the People’s Movement for the Liberation of
Angola (MPLA), the National Union for the Total Independence of Angola
(UNITA), and the National Liberation Front of Angola (FNLA). In 1974, Portugal
considered independence for Angola, but the country soon dissolved into a civil
war as the MPLA, led by Agostinho Neto and supported by the Soviet Union and
Cuba, waged war against Holden Roberto’s FNLA and Jonas Savimbi’s UNITA, a
political alliance that enjoyed the support of the United States.
The longitudinal pattern of superpower assistance in this case strongly corrobo-
rates hypotheses H2 and H3. Both superpowers used covert aid primarily to signal
resolve to local allies and each other, and inferred seriousness of purpose from
observed changes in the magnitude and lethality of their opponent’s aid.

Covert Signaling (1970–74)

Although the Soviet leadership publicly disavowed any involvement, the US intelli-
gence community detected Soviet covert involvement in the Angolan civil war and
monitored changes over time. As early as 11 August 1970, intelligence reports
noted “a team of four to six Soviets” entering Angola and an “extraordinarily large
Soviet arms shipment to [the MPLA].” The United States viewed this aid as an
attempt to send a message beyond the specific crisis about the Soviet disposition
regarding region, noting the Soviets “would see stepped-up aid as a means to
increase their own influence” with African liberation movements and “progressive
leaders of independent African states who sympathize with them.” Although the
modesty of early Soviet aid led US officials to opt against their own involvement,52
Washington agreed to provide symbolic assistance to Zaire (whose leader Mobutu
Sese Seko was seen as “erratic” but according to Kissinger was “ruthless and [will-
ing to] get the job done”)53 and the FNLA. William Colby, director of Central
Intelligence, informed Kissinger in September 1974, “We intend to keep these pay-
ments fairly low but high enough to assure President Mobutu that we are sympa-
thetic to his concern.”54 On 22 January 1975, the National Security Council’s

52
“Soviet and Cuban Intervention in the Angolan Civil War,” Intelligence Memorandum, March 1977, Digital National
Security Archive (DNSA), CT00970.
53
Memorandum for the record, Foreign Relations of the United States [FRUS], 1969–76, Southern Africa, 5 June 1975,
doc. 106, vol. 28 (Washington, DC: GPO, 1976).
54
Memorandum from Colby to Kissinger, September 1974. doc. 100, ibid.
 SECURITY STUDIES 141

(NSC) Forty Committee further allocated three hundred thousand dollars in covert
assistance to the FNLA.

American Covert Weapons

From April to June of 1975, the CIA detected a considerable increase in Soviet
arms deliveries to the MPLA.55 Administration officials saw this as a substantial
Soviet escalation taking place prior to the US decision to increase its own covert
funds.56 The creation of a Cuban Military Mission in Angola helped the MPLA
employ the Soviet arms and make gains against the FNLA.57
Consistent with H2, US decision-makers recognized that one benefit of their
own lethal covert assistance was a more credible signal of their commitment to the
defense of the region. They were also acutely aware of domestic exposure risks.58
With Congress already angry over past CIA “rogue” activities, an NSC report on
Angola in June argued, “there would be strong Congressional opposition to any
U.S. involvement in support of one of the contending factions.”59 Officials in the
State Department actually objected to covert military aid in Angola for these and
other reasons, noting “(1) we have no vital interests; (2) the risks of exposure are
extreme; (3) our clients will be discredited; and (4) the results will be indecisive.”60
Under Secretary of State for Political Affairs Joseph Sisco believed Soviet leaders
would match American aid and felt the United States could not keep up. Kissinger
and President Gerald Ford disagreed and specifically invoked a signaling logic for
local partners. Kissinger explained, “It is a question of [Mobutu’s] perception of
who is behind him…. What he wants to know is if the U.S. is politically inter-
ested.”61 Indeed, Mobutu was urging the United States to show more commitment
by expanding the scope and lethality of its covert action. In private meetings,
“Mobutu reiterated need for urgent action. ‘The Soviets are continuing to send
arms into Angola,’” he said, “‘while we are sitting here talking.’”62
The CIA’s wariness about congressional backlash led Director Colby to suggest,
“Let’s give dollars … and this will keep Congress off our backs.” Unconvinced, Kis-
singer pushed for covert arms supply as a means of signaling commitment, saying,
“I’m scared of losing. Is anyone else? Why would Zaire break with the USSR and
Yugoslavs if the U.S. will not give political support?”63 Kissinger worried about the
broader lesson about American involvement in the region. He saw a failure to
covertly aid American friends as dangerous, asking, “If Angola is taken by the Com-
munists, what conclusions can the African leaders draw about the United States? …

55
CIA intelligence information cable, May 1975, ibid.
56
Memorandum of conversation,” 19 December 1975, doc. 156, ibid.
57
CIA Intelligence Information Cable, 28 May 1975, ibid.
58
Memorandum, 11 June 1975, doc. 108, ibid.
59
Paper prepared by the National Security Council Interdepartmental Group for Africa, 13 June 1975, doc. 109, ibid.
60
Memorandum for the record, 14 July 1975, doc. 115, ibid.
61
Ibid.
62
Telegram from the Embassy in Zaire to the Department of State, 24 July 1975, doc. 120, ibid.
63
Memorandum for the record, 14 July 1975, doc. 115, ibid.
142 A. CARSON AND K. YARHI-MILO

They can only conclude that we don’t care.”64 The advocates of covert assistance
therefore sold the initiative as a tool for demonstrating American commitment to
African leaders in general and Mobutu in particular in response to clear signs of
Soviet determination (itself reflected in their escalating aid, consistent with H3).
By early July, the MPLA’s military position began to improve due to the increas-
ing flow of Soviet arms and Cuban military advisers. Consequently, on 18 July
1975, President Ford authorized the CIA to supply arms to the FNLA and UNITA
covertly through Zaire. It represented a qualitative shift in US covert involvement
that significantly increased US risks of exposure. The rationale for using Zaire was
not just because of domestic political constraints; “Such an arms flow to Angola
would be quickly detected and publicized … with damage to the international
standing and political prospects of the FNLA and UNITA. Similar side effects
argue against the hiring of mercenaries or the provision of [American] aircraft.”65
The CIA clearly recognized the risk of exposure of American arms through
Mobutu not just for its own domestic standing (“if it became public knowledge
that we were sending American arms in…. There would be a great uproar about
CIA getting involved in a war”) but also because of concerns for the ability of their
partners to work together if this became public (it “would tend to spoil political
efforts to get African leaders such as [Kenneth] Kaunda, [Julius] Nyerere and
[Yakubu] Gowon behind efforts to stop the fighting.”)66

Monitoring Escalation and Inferring Intentions

In September 1975, the United States detected a further increase in the scope and
lethality of Soviet covert assistance, including armored vehicles, heavy artillery, air
defense weapons, and possibly even MIG aircraft.67 The introduction of roughly
1,200 Cuban troops68—“the first time that Cuba or any other Soviet ally had sent
an expeditionary force to intervene in another country’s civil war”69—was seen as
a Soviet surrogate force that would avoid troop deployments from the United
States and China.70 After an increase in airlifts of Cuban troops in November, by
December it was no longer the MPLA forces but Cuban troops who “were bearing
the brunt of the fighting.”71
This crucial point did not pass unnoticed as predicted by H3b. “The November
escalation,” the intelligence community reasoned in a retrospective report, “shows
the length to which the Soviets were willing to go and the risks they were willing to
run to achieve their basic objective—imposition of the MPLA as the sole ruling

64
Memorandum of conversation, 20 June 1975, doc. 111, ibid.
65
Memorandum, 11 June 1975, doc. 108, ibid.
66
Ibid.
67
Message from President Ford to French President Giscard d’Estaing, 25 November 1975, doc. 141, ibid.
68
Report prepared by the Working Group on Angola, 22 October 1975, doc. 132, ibid.
69
“Soviet and Cuban Intervention in the Angolan Civil War,” DNSA, 3.
70
Ibid., 14.
71
Ibid., 4.
 SECURITY STUDIES 143

group in Angola.”72 Confirming hypothesis H2, the Americans also believed Soviet
leaders inferred the level of US resolve in Angola from the amount of American
covert aid. “Since the Soviets had probably concluded that weapons alone could
not win the war,” one report argued, “the amount of U.S. aid was probably of inter-
est to them mainly for its relevance to the question of whether the U.S. attached
sufficient importance to Angola to send troops.”73
The November airlift represented the point “at which the Soviets made a full
commitment to secure an MPLA victory…. [It] symbolized the Soviets’ seriousness
of purpose. Never before had the Soviet Union undertaken a military airlift over
such long distances.”74 President Ford specifically linked Soviet escalation to its
broader reputation in the region, reasoning that “Angola would enable the Soviets
to exert a major influence on the liberation drive in southern Africa…. It contrib-
utes to Soviet credibility and influence with other clients in the region…. For the
time being … we believe that their primary goals are political rather than
strategic.”75
Importantly, American decision-makers and the intelligence community did
not believe the Soviets escalated because they inferred US weakness or irresolution.
Rather, they believed that three factors in particular shaped the Soviet decision to
escalate: First, the Soviets viewed the airlift as a necessary rescue operation since
“MPLA military defeat was a distinct possibility had aid not been stepped up.”
Thus, the Soviets were faced with an “either/or situation” of winning or losing
Angola. Second, the US intelligence community assessed that Fidel Castro proba-
bly urged the initially reluctant Soviets to escalate and offer him a guarantee that
“there would be no repetition of the October crisis of 1962” in that there would be
no agreement on Angola between the two superpowers, which would place the
Cubans in a difficult position. As a result, the Soviets could not collude with the
Americans and reach a settlement on Angola. Third, the Soviets, according to
American assessments, had information at the time of a likely South African inter-
vention and movement northward, increasing the risks of Soviet loss of Angola.76
Meanwhile, fearful that Congress would soon terminate its support for the oper-
ation, Kissinger and Ford grasped for other sources of coercive pressure. They
chose to privately threaten to link Angola to detente, a form of cooperation they
knew the Soviets hoped to protect. In an NSC meeting on 22 December, Kissinger
reasoned: “If we keep going and the Soviets do not think there is a terminal date
on our efforts and we threaten them with the loss of detente, we can have an
effect.”77 Ford concurred. And in a meeting with Andrei Gromyko, Kissinger
accordingly reasoned: “We can’t defend to our people your massive airlift and the

72
Ibid., 4.
73
Ibid., 17.
74
Ibid., 25.
75
Message from President Ford to French President Giscard d’Estaing, FRUS, 25 November 1975, doc. 141 (emphasis
added).
76
“Soviet and Cuban Intervention in the Angolan Civil War,” DNSA.
77
Minutes of a National Security Council meeting, FRUS, 22 December 1975, doc. 163.
144 A. CARSON AND K. YARHI-MILO

Cuban troops. It can’t go on without raising serious questions here. We will have to
find ways either to insulate it or match it.”78 Thus, the Soviet choice to escalate
covertly gave American leaders a kind of leverage to threaten to escalate into the
overt realm if necessary and to use their domestic constraints to make the adverse
consequences of such a linkage credible.
To be sure, however, the US intelligence community estimated that the Soviet
leadership at the time probably already understood that there would be some rami-
fications for their actions in Angola on bilateral relations. Thus, they inferred from
their actions in Angola that they were determined to win there notwithstanding
their domestic constraints and debates about the strategic bilateral desirability of
these actions. As one intelligence memorandum concluded, “In the final analysis
Soviet actions there [in Angola], at least in the last stage of the conflict, must be
seen as taking place not in ignorance of the damage of detente, but in spite of that
damage.”79 American decision-makers shared this assessment, viewing Soviet
covert actions in Angola as conveying resolve despite, or especially in light of,
being somewhat constrained by the desire of some members of their leadership to
preserve detente. The president’s assistant for National Security Affairs, Brent
Scowcroft, and deputy national security advisor, William Hyland, for example,
noted in a memo to Kissinger that the Soviets were split among themselves over
Angola because of the ramifications of their actions on US–Soviet relations. Yet,
they and others in the NSC drew the important inference that their decision to act
in Angola, even covertly, was a signal of their commitment to winning Angola and
of the determination to suffer the cost of disturbing detente. Thus, for example,
Scowcroft and Hyland explain, “The Soviets are committed to the MPLA; and
have resumed the airlift and given us no encouragement…. Linkage of SALT and
Angola might shore up position of Kremlin hardliners who pressed Angolan oper-
ation…. They must have argued that Angola was worth some risk in any case.
Even if they had argued they could score up a quick cheap victory and are now
faced with the consequences of miscalculation, we are not sure that moderates in
Kremlin, including [Leonid] Brezhnev, would have political punch to repudiate
Angolan adventure just before a Party Congress.”80
Such assessments consequently led them to predict that even “threat of cancella-
tion [of his visit to discuss SALT with the Soviets] might have some sobering effect
in [the] Kremlin, but will not alter the situation in Angola or cause Soviets to ter-
minate this adventure.”81 That Kissinger and Ford nonetheless attempted to
threaten the Soviets with such a linkage reflects their limited options at the time.
In addition to the threat of escalation through strategic linkage, Kissinger and Ford

78
Memorandum of conversation, 9 December 1975, doc. 145, ibid.
79
“Soviet and Cuban Intervention in the Angolan Civil War,” DNSA, 34.
80
Message from the President’s Assistant for National Security Affairs (Scowcroft) and the President’s Deputy Assistant
for National Security Affairs (Hyland) to Secretary of State Kissinger in Jamaica, FRUS, 1969–1976, 31 December 1975,
doc. 235, vol. 16.
81
Ibid.
 SECURITY STUDIES 145

also sought help from US allies and other third parties (reportedly even apartheid
South Africa) to join the fighting in Angola.82
In sum, contrary to the weakness hypothesis (H4), American decision-makers
did not view Soviet covertness as indicating irresolution because of the quantitative
and qualitative shifts in their involvement discussed above, and the risk their
actions generated could jeopardize detente with the United States (and arms con-
trol negotiations). Moreover, Soviet covert activities in Angola were especially
meaningful because they ran against the familiar grammar the two rivals had come
to share when intervening in the Third World. In a candid remark after the opera-
tion ended, Kissinger explains how the deviation from Soviet standard covert prac-
tices was a significant signal by itself, noting that “if the MPLA won only with
normal Soviet covert assistance, I wouldn’t like it but I would accept it as is the
case of FRELIMO [Mozambique Liberation Front] in Mozambique…. We could
have accepted $20 million in Soviet assistance in Angola but not $285 million,
which was more than the whole world had put into black Africa.”83
Fears of exposure, as it turned out, were warranted. Media reports in November
that the United States had joined forces with the apartheid government of South
Africa in Angola caused outrage in Congress, particularly in light of the recently
concluded war in Vietnam.84 On 19 December, the Senate passed an amendment
sponsored by John V. Tunney that effectively blocked any further covert support
to Angola. This amendment was seen by advocates of covert aid in the Ford
administration as a blow to America’s reputation.85

The Soviet Side

Assessing Soviet intentions and inferences from the available record, we find solid
support for our hypotheses, especially H2 and H3. On the intention side, direct evi-
dence is not available but, as one historian argues, Moscow enhanced its own
covert aid to Neto in part as a way to send a broad signal of resolve to China and
the United States and “believed that the Soviet Union had to come to the aid of its
ally.”86

82
For example, see backchannel message from Secretary of State Kissinger to the Ambassador to Iran (Helms), 20
December 1975, doc. 159, vol. 28. Kissinger notes continued American covert aid shows “the Executive Branch is
resolved to pursue the Angola matter vigorously and with full determination” and that “a Soviet power play in the
heart of Africa not be permitted to succeed unchallenged. … We intend to do everything possible to continue our
support of the FNLA and UNITA through Zaire, and we will also in our discussions with the Soviets, making clear to
them that their involvement in Angola will inevitably raise questions about detente.”
83
MemCon, 1 April 1976, DNSA, doc. 01924.
84
Piero Gleijeses, Conflicting Missions: Havana, Washington, and Africa, 1959–1976 (Chapel Hill, N.C.: University of North
Carolina Press, 2002), 362–65.
85
MemCon, FRUS, 25 January 1976, vol. 76; MemCon, 18 December 1975, ibid.
86
Odd Arne Westad, The Global Cold War: Third World Interventions and the Making of Our Times (Cambridge: Cam-
bridge University Press, 2007), 234; Jiri Valenta, “The Soviet-Cuban Intervention in Angola, 1975,” Studies in Compara-
tive Communism 11, no. 1 (March 1978): 15, DOI:10.1016/0039-3592(78)90029-7; Westad, “Moscow and the Angolan
Crisis, 1974–1976: A New Pattern of Intervention,” Cold War International History Project Bulletin nos. 8–9 (Winter
1997/1996): 24, http://www.wilsoncenter.org/sites/default/files/CWIHP_Bulletin_8-9.pdf.
146 A. CARSON AND K. YARHI-MILO

In terms of Soviet inferences, three aspects of the American covert aid effort in
Angola prompted the Soviet Union to view the United States as resolved and hence
to dramatically increase its own covert involvement in mid-November. One was
intelligence about an imminent South African plan to invade Angola, which Mos-
cow believed was in coordination with Washington. This is consistent with the
counter-escalation risks the Soviets believed the United States was willing to take.
The second aspect of US covert aid, consistent with the sinking costs mechanism,
was the perceived increase in US covert military assistance that appeared to signal
US determination to win in Angola.87 As Odd Arne Westad points out, “the Soviet
perception of the widening role of the CIA … played a role in Moscow’s reevalua-
tion of its Angolan policy. The KGB station in Brazzaville supplied vital informa-
tion on the dramatic increases in US assistance, and Iurii Andropov believed that
the Americans had a long-term strategy of equipping large groups of Angolan,
Zairean, and Western mercenary troops, to be sent to Angola.”88 The third aspect
of American covert aid that led Moscow to infer resolve was the light it shed on
domestic politics in Washington. Rather than inferring weakness from the con-
gressional constraints American leaders were clearly laboring under, Moscow saw
covert action as an expression of resourcefulness. As Ted Hopf’s analysis of Soviet
perceptions of the United States notes, the “efforts of the United States to intervene
covertly in the Angolan civil war caused a reversal in Soviet perceptions of Ameri-
can and allied capabilities,” thereby “enhancing American credibility.” Soviet offi-
cials appeared to learn a broad lesson from a covert action taken despite anti-
intervention domestic sentiment; Hopf concludes Angola taught Soviet leaders
“there remain ‘extremely persistent efforts in the leading circles for the United
States to … leave loopholes of the possible application of force in that or another
concrete situation.’”89
Lastly, it is worth noting one specific point about the Cuban role in Soviet efforts
in Angola that demonstrates unique escalatory dangers in covert signaling. During
the summer of 1975, the Soviets refused repeated Cuban requests to send Soviet
officers to transport Cuban troops out of concern that more direct Soviet involve-
ment would affect US–Soviet relations.90 Now available primary evidence on
Cuba’s role suggests that the November airlift was decided in Havana without con-
sulting the Politburo in Moscow.91 Yet this contradicts Ford and Kissinger’s per-
ceptions of decision-making power. Kissinger admits in his memoirs, “At the time,
we thought he [Castro] was operating as a Soviet surrogate. We could not imagine
that he would act so provocatively so far from home unless he was pressured by
Moscow to repay the Soviet Union for its military and economic support. Evidence

87
Gleijeses, Conflicting Missions, 306.
88
Westad, Global Cold War, 234.
89
Ted Hopf, “Soviet Inferences from Their Victories in the Periphery: Visions of Resistance or Cumulating Gains?” in
Dominoes and Bandwagons: Strategic Beliefs and Great Power Competition in the Eurasian Rimland, ed. Robert Jervis
and Jack Snyder (New York: Oxford University Press, 1991), 170.
90
Westad, Global Cold War, 233.
91
Westad, “Moscow and the Angolan Crisis, 1974–1976,” 31.
 SECURITY STUDIES 147

now available suggests that the opposite was the case.”92 The Cubans were thus far
less beholden to their Soviet patrons than American decision-makers had sup-
posed.93 American escalation in response to Cuba’s entry, including soliciting sup-
port from others such as South Africa, was therefore based on a misunderstanding
of the political dynamics between client and patron. This underscores how covert
signaling has unique escalatory dangers not present in overt forms of external
involvement; ambiguity about responsibility for specific decisions, even among
adversaries carefully tracking one another, can ratchet up conflict intensity.

Discussion
To summarize, the evidence generally supports H2: American and Soviet deci-
sion-makers sought to signal resolve to each other as well as local allies and
third parties through the use of covert military action that they knew would
be detected. Moreover, sinking cost and escalation risks, consistent with H3b,
constituted the primary mechanisms US observers used to infer Soviet resolve.
The way in which Soviet activity in Angola deviated from past covert inter-
ventions registered clearly in the eyes of American policymakers and under-
scores the development of a shared interpretive frame giving meaning to
qualitative shifts in covert action. Notwithstanding the eventual defeat of the
United States, Soviet leaders also drew important inferences about US resolve
from covert activities. Domestic risks appear to have played a prominent role
in their inference process: the Soviets saw Kissinger’s and Ford’s attempts to
act decisively despite domestic anti-intervention constraints as evidence of its
determination to rebuild the US reputation for resolve and resist Soviet expan-
sion. We do not find support for H4: covert actions were not seen as signaling
weakness, although they were seen as indicating a measure of restraint during
the early stages when scope of involvement was modest. Finally, it seems that
Soviet and American willingness to inflict harm covertly through sinking cost
and generating risks added credibility to their activities. But the fact that these
actions were not public gave both sides the opportunity to signal in a setting
that otherwise would have forced them to choose between two unpalatable
options: publicly signaling resolve but wrecking detente or doing nothing,
appearing weak to local allies and one another but safeguarding detente.

Afghanistan, 1979–89
A coup in Kabul in 1978 catapulted Afghanistan into the very center of superpower
relations. Soviet leaders embraced the communist People’s Democratic Party of
Afghanistan (PDPA) but found its new ally mired in a growing insurgency. Mos-
cow first sent secret combat support units to help quell the insurgency. When this

92
Henry Kissinger, Years of Renewal (London: Simon and Schuster, 1999), 816.
93
Gleijeses, Conflicting Missions.
148 A. CARSON AND K. YARHI-MILO

failed, Soviet leaders authorized what they hoped was a short-term invasion. The
nine-year conflict that ensued hosted a dense network of covert activity; we focus
on three specific episodes. First, we analyze American and local ally inferences in
1979 about resolve based on observed activity prior to the Soviet invasion. Second,
we analyze the role of signaling in the initiation and escalation of US covert aid to
Afghan rebels. Lastly, we analyze covert cross-border military incursions into
Pakistan by the Soviet military in the mid-1980s and the lessons American and
Pakistani observers drew.

Soviet Pre-Invasion Covert Involvement

Uprisings in rural Afghanistan in March 1979 threatened the duumvirate of
Nur Muhammad Taraki and Hafizullah Amin and prompted Soviet leaders to
initially send covert combat support six months before the more well-known
Soviet invasion.94 Soviet leaders added additional combat units in November
and early December after Amin assassinated Taraki to seize sole control of
power in Kabul.95 American intelligence carefully tracked changes in Soviet
involvement and detected this covert activity by the summer. Embassy cables
in May and June reported the presence of covertly deployed Soviet helicopter
crews.96 An interagency assessment in September concluded, “The Soviets
increased the numbers and expanded the counterinsurgency role there of what
now are at least 2,500 of their military personnel.”97 Evidence of a covert
Soviet role was also referenced in interagency warnings from CIA director
Stansfield Turner and the NSC.98
Consistent with H3, US leaders inferred greater Soviet resolve to protect its
interests in Afghanistan and escalating further to do so. Reflecting the role of
Schelling’s “salient thresholds” in signaling, intelligence memos analyzing
Soviet intentions in mid-September used covert combat support as “key evi-
dence” that Moscow was serious about saving the faltering regime, including

94
Raymond L. Garthoff, Detente and Confrontation: American-Soviet Relations from Nixon to Reagan, rev. ed. (Washing-
ton, DC: Brookings Institution Press, 1994), 991; Odd Arne Westad, “Prelude to Invasion: The Soviet Union and the
Afghan Communists, 1978-1979,” International History Review 16, no. 1 (February 994): 57–58; Steve Coll, Ghost
Wars: The Secret History of the CIA, Afghanistan, and Bin Laden, from the Soviet Invasion to September 10, 2001 (New
York: Penguin, 2004), 39–40.
95
Excerpt from Minutes of CC CPSU Politburo, 31 October 1979, Vol. 2: Russian and East European Documents,
“Towards an International History of the War in Afghanistan, 1979–1989,” CWIHP Document Reader (CWIHP Vol. 2),
29–30 April 2002, http://www.wilsoncenter.org/sites/default/files/AfghanistanV2_1974-1989.pdf. “The Soviet subunits
located in Afghanistan (communications centers, the parachute battalion, the fixed-wing and helicopter transport
squadrons) and also the Soviet institutions’ security detachment are to continue to perform the assigned missions.”
Aleksandr Antonovich Lyakhovskiy, “Inside the Soviet Invasion of Afghanistan and the Seizure of Kabul, December
1979,” January 2007, Working Paper No. 51, Cold War International History Project, 10, http://www.wilsoncenter.org/
sites/default/files/WP51_Web_Final.pdf.
96
Cable from Embassy Kabul to Secretary of State, 18 July 1979, from “Towards an International History of the War in
Afghanistan, 1979–1989,” CWIHP Document Reader, Volume I: US Documents, 29–30 April 2002, http://www.wilson
center.org/sites/default/files/AfghanistanV1_1978-1979.pdf (hereafter CWIHP Vol. 1). Cable from Embassy Kabul to
Secretary of State, 8 May 1979, CWIHP Vol. 1.
97
Interagency Intelligence Memorandum, “Soviet Options in Afghanistan,” 28 September 1979, Central Intelligence
Agency Electronic Reading Room (CIAERR), Document 0000267105.
98
Memorandum from Stanfield Turner for the National Security Council, 14 September 1979, CWIHP Vol. 1; National
Security Council Paper, August 1979, “Soviet Position in the Third World,” CWIHP Vol. 1.
 SECURITY STUDIES 149

potentially via invasion.99 Subsequent changes in Soviet covert involvement

that seemed to indicate a broadening role in defending the regime led to
updated American inferences. Detection of additional covert Soviet combat
battalions deployed to Bagram Airfield in early December led a defense intelli-
gence assessment to conclude that “this significant escalation … demonstrates
Moscow’s resolve in pursuing its interests in Afghanistan despite the obvious
pitfalls and at a time when the Kremlin might consider the U.S. preoccupied
with events in Tehran.”100 These “obvious pitfalls” were noted in previous
American analysis, as when the Kabul embassy cabled headquarters noting the
role of detente in constraining Soviet activity in Afghanistan.101 Another noted
a “qualitative increase in their military presence” whose arrival “signals greater
Soviet concern for developments in Afghanistan than previously noted.”102 A
National Intelligence Daily update interpreted airborne battalions capable of
performing counterinsurgency operations on behalf of the regime as “indica-
tive of a decision by the Soviets to increase their military commitment in
Afghanistan substantially.”103 This led President Jimmy Carter to privately
express concern about the Russian presence in Afghanistan to both Ambassa-
dor Anatoly Dobrynin and Leonid Brezhnev.104
Soviet private meetings with American leaders in December did include oblique
references to deepening Russian involvement, suggesting that Soviet leaders knew
US leaders were generally aware of their activities. Available evidence on the Soviet
covert action in 1979 does not provide conclusive support that Moscow hoped to
signal to Washington covertly. However, a signaling motive is especially plausible
regarding their local ally. A high-level March 1979 telephone conversation between
Moscow and Kabul, for example, features Afghanistan’s leader pleading for Soviet
assistance to address a growing rebellion. After Kosygin expresses reluctance, Tar-
aki suggests the Soviets send personnel covertly because “no one will be any the
wiser…. They will think these are government troops.”105 Following the Polit-
buro’s approval of a small covert personnel deployment, Soviet documents show
regular requests for additional covert aid from Soviet representatives in Kabul in

99
The final judgment at this time was that such an intervention was unlikely. See Interagency Intelligence Memoran-
dum, October 1980, CIAERR, Document 0000278538, 21.
100
Ibid., 26–36 (quotation, 26). As Douglas MacEachin summarizes, “there were reports that additional military advisors
were being assigned to Afghan units engaging in combat, and that some Soviet military personnel were piloting
helicopters in ground strikes and operating tanks in combat.” MacEachin, “Predicting the Soviet Invasion of Afghani-
stan: The Intelligence Community’s Record,” Books and Monographs, Center for the Study of Intelligence, Central
Intelligence Agency, 2002.
101
An overt Soviet role was judged unlikely because it “would deal a severe blow to detente with West at a time when
Moscow is increasingly pre-occupied with the growing Chinese threat in the east. Such a move would almost cer-
tainly doom SALT.” Cable from Embassy Moscow to Secretary of State, 24 May 1979, CWIHP Vol. 1.
102
Quoted in Interagency Intelligence Memorandum, October 1980, CIAERR, 26.
103
Ibid.
104
Garthoff, Detente and Confrontation, 1051–52. Note that while Moscow’s covert role was perceived as a sign of
greater willingness to run risks and tolerate costs to succeed, American intelligence did not specifically predict the
Soviets would escalate to a full invasion. This led some policymakers to express disappointment with the perfor-
mance of US intelligence despite detection of covert action once the invasion was apparent. See Interagency Intelli-
gence Memorandum, October 1980, CIAERR; MacEachin, “Predicting the Soviet Invasion of Afghanistan.”
105
Kosygin/Taraki, telephone transcript, 18 March 1979, CWIHP Vol. 2.
150 A. CARSON AND K. YARHI-MILO

light of continued rebellion growth.106 The consistent requests from Afghanistan

suggest an appreciation for the constraints on Moscow’s ability to provide aid and
acceptance of covert forms of support as a critical indicator of Soviet backing.

Initiating American Covert Lethal Aid

Covert Soviet assistance proved insufficient and, in late December 1979, Soviet
troops invaded the country to replace Taraki and restore stability. American poli-
cymakers viewed the Soviet invasion as “outrageous and unprecedented.”107 A
NSC memo invokes shared, tacit rules about superpower involvement and Soviet
transgression of them, arguing “Soviet action can only be taken as representing a
qualitatively new, dangerous stage of Soviet assertiveness. In the past, the Soviet
leadership has moved cautiously and deliberately in undertaking moves of such
consequence. It is clear that the West should now reassess Soviet policy in terms of
how likely it is that the 1980s will see greater Soviet willingness to achieve its for-
eign policy aims through the overt intervention of its military forces, particularly
in the Third World.”108 National Security Advisor Zbigniew Brzezinski warned
that “if the US is perceived as passive in the face of this blatant transgression of civ-
ilized norms, our international credibility and prestige will be seriously eroded,
particularly in the eyes of those countries most vulnerable to Soviet interven-
tion.”109 An intelligence report noted the Soviets themselves were watching the
West’s response and an aid to Brzezinski noted it was “imperative that we not only
act to counter what the Soviets have done in Afghanistan, but that we are perceived
as having done so.”110
Consistent with H2, American leaders believed their covert aid was a critical part
of a targeted message it hoped to send to Moscow’s friends in the region. President
Carter quickly endorsed a package of public and private punishments to do just
that. This included authorization for $30 million to expand an existing covert non-
lethal aid program to include weaponry and ammunition, sent to Afghan rebels
via Pakistan and in coordination with Egypt, China, and Saudi Arabia.111 Secrecy
was helpful for Pakistan; American military aid was domestically controversial in
some quarters for President Muhammad Zia-ul-Haq.112 Advisors argued the
United States had to make “a concerted effort to teach Moscow that aggression
does not pay.” To do so, the United States had to “make this a costly effort for the

106
See, for example, Puzanov’s relaying request for three special battalions of Soviet troops in Kabul and transport heli-
copters with crews; Memorandum, 12 August 1979, CWIHP Vol. 2.
107
Memorandum from Marshall Brement for Zbigniew Brzezinski and David Aaron, 27 December 1979, CWIHP Vol. 1.
108
Memorandum from Matthews, 29 December 1979, CWIHP Vol. 1.
109
Memorandum from Brzezinski to the President, 2 January 1980, CWIHP Vol. 1.
110
Memorandum from Matthews, 29 December 1979, 5; Memorandum from Matthews, 29 December 1979, 5; Memo-
randum from Brement for Brzezinski and Aaron, 27 December 1979, 1.
111
Coll, Ghost Wars, 58–59. The preexisting CIA-run covert aid program was explicitly nonlethal and had begun in sum-
mer 1979. See Robert M. Gates, From the Shadows: The Ultimate Insider’s Story of Five Presidents and How They Won
the Cold War (New York: Simon and Schuster, 2007), 144–46.
112
Coll, Ghost Wars, 59–65; Gates, From the Shadows, 251–52, 319–21.
 SECURITY STUDIES 151

Soviets” through “covert arms supply to Afghan insurgents.”113 Covert measures

would be a key part of a “strong and vivid US response.” The goal was a broad
message beyond Afghanistan itself. The Soviets would be made to understand “this
was a very expensive invasion and that it should not set a precedent for future
action.”114 A pre-invasion NSC memo suggested covert aid then only under con-
sideration could show “our determination to become involved in Gulf security”
and serve “as a global signal to the Soviets.”115
The American covert aid program also provided regular evidence of US resolve
to Pakistan and showcases the unique signaling opportunities that can arise when
using covert rather than direct supply chains. Pakistan’s role was critical for keep-
ing the American program plausibly deniable. Evidence of Islamabad’s view of aid
to Afghan rebels as an indicator of American resolve can be found in its increas-
ingly urgent pleas for American aid prior to the Soviet invasion. A June 1979
American diplomatic cable from Islamabad, for example, describes the “perception
of the strength of the U.S. commitment to Pakistan” and concerns that some
believe “the U.S. has deserted her” in part because of the American “failure to stand
up to the Soviets in Angola, Ethiopia, and Afghanistan.”116 An NSC memo on
American South Asia strategy from September, moreover, notes, “There seems to
be increasing Pakistani interest in assistance to the insurgents” in reaction to grow-
ing signs of instability and Soviet activity in Afghanistan.117 Pakistan’s was enthusi-
astic even after covert aid was initiated. Pakistan’s Zia reportedly developed a close
relationship with CIA director William Casey and often emphasized Pakistan’s
pessimistic view of Soviet ambitions in the region. Zia saw low-visibility covert aid
as an ideal instrument—superior to overt aid—because it allowed costly pressure
to be put on the Soviet Union without triggering retaliation against Pakistan.118
Domestic risks played an interesting role and contrast with those in the Angola
case. The domestic risks for American leaders incurred by providing aid to Afghan
rebels were minimal because the political mood had changed in Washington.
American covert aid to fight a Soviet invasion was not a potential political liability
at home; in fact, it proved a political windfall for the Reagan administration once
its details began to leak in 1981.119 Yet domestic risks did play a role in the covert

113
Memorandum from Oksenberg for Brzezinski, 28 December 1979, CWIHP Vol. 1.
114
Memorandum from Brement for Brzezinski and Aaron, 27 December 1979.
115
Note the memo ultimately opposed covertly arming the rebellion at this stage for other reasons. Memorandum
from Thornton to Brzezinski, 24 September 1979, CWIHP Vol. 1.
116
Cable from Embassy Islamabad to Secretary of State, 24 June 1979, CWIHP Vol. 1. Note that the “failure” in Angola is
referencing the cutoff of the American aid program
117
Memorandum from Thomas Thornton to Zbigniew Brzezinski, 24 September 1979, CWIHP Vol. 1.
118
As one analyst summarizes, “Zia expressed his intent—in words which became his Afghan policy mantra—that the
Afghan support program ‘keep the pot boiling, turn up the heat, but don’t let the pot boil over.’ In other words, irri-
tate and undermine the Soviets, but don’t provoke them to massively escalate the level and intensity of their com-
mitment. Make them pay for their incursion, but don’t drive them to retaliate against the Pakistanis.” Kirsten
Lundberg, “Politics of a Covert Action: The US, the ‘Mujahideen,’ and the Stinger Missile,” Kennedy School of Gov-
ernment Case Study Program, 9 November 1999, 13.
119
Carl Bernstein, “U.S. Weapons for Afghanistan,” Chicago Tribune, 22 July 1981. Note that other covert aid programs
in response not to Soviet invasion but alleged Soviet patronage (that is, Nicaragua) remained controversial during
the Reagan administration.
152 A. CARSON AND K. YARHI-MILO

aid program—domestic risks in Pakistan. As noted above, Islamabad’s partnership

with Washington in a covert aid program was divisive within Pakistan. The need
for keeping aid limited and at least plausibly deniable was in large measure due to
domestic Pakistani considerations.120 Diplomatic reports from Pakistan reflect this
constraint, noting that Islamabad “must walk a tightrope in dealing publicly with
the insurgency in Afghanistan.”121 Thus, part of what demonstrated American and
Pakistani resolve to Soviet leaders was a local ally’s willingness to tolerate domestic
political risks in order to impose costs on a Soviet occupation.
Direct evidence regarding inferences by Soviet leaders about American covert
aid is unfortunately not available. Soviet leaders certainly anticipated and moni-
tored the covert aid program. Some intelligence reports to Moscow from Kabul
warned of a substantial increase in covert aid in January 1980 and described spe-
cific training and equipment coming via Pakistan in September 1980.122 Reports in
1981 predicted increases in covert aid by the new Reagan administration and care-
fully described weapons flows.123 The best evidence regarding inferences comes
from a former CIA analyst whose history of detente noted American covert aid
“contributed to Soviet fears of the desire and readiness of the United States … to
attempt to influence the situation in Afghanistan.”124

Soviet Secret Cross-Border Activity

As President Reagan expanded the scope of covert aid, Soviet leaders faced an
increasingly protracted counterinsurgency and sought to cut off the swelling flow
of weapons via Pakistan. Soviet leaders launched secret cross-border air strikes
into rebel camps in Pakistani territory as part of this effort. Consistent with our
hypotheses on inferences and expanded covert action (H3a and H3b), Washington
and Islamabad interpreted escalating unpublicized cross-border attacks as a signal
of Soviet resolve and willingness to raise risks in Afghanistan/Pakistan. A mid-
1982 CIA report noted new Soviet border violations that “seemed deliberate …
and appeared aimed at intensifying pressure on Islamabad to come to terms with
the Soviet-dominated government in Kabul.”125 A Pakistani governor complained
to American representatives that Soviet cross-border air strikes were intended “to
demonstrate to Pakistanis that the [Government of Pakistan’s] Afghan policy and
continued harboring of the refugees entailed great dangers.”126 The message

120
An Intelligence Assessment, July 1982, CIAERR, Document 0000534961. Congressional advocacy for more expansive
and aggressive weaponry (for example, Stinger missile) initially exceeded what the Pakistani government would tol-
erate and was only approved after Pakistan gave its consent. See discussion of Pakistan as the “predominant over-
riding concern” for opponents of the Stinger missile in Lundberg, “Politics of a Covert Action,” 34–35.
121
Cable from Embassy Islamabad to Secretary of State, 24 June 1979.
122
Excerpt from Bogomolov, 20 January 1980, CWIHP Vol. 2; Unsourced document [presumably from intelligence sour-
ces], September 1980, CWIHP Vol. 2.
123
Report from Gen. Mayarov et al., 10 May 1981, CWIHP Vol. 2; Report by Military Intelligence Representatives, Sep-
tember 1981, CWIHP Vol. 2.
124
Garthoff, Detente and Confrontation, 1031, emphasis added.
125
An Intelligence Assessment, July 1982, CIAERR, 5–6.
126
Cable from Secretary Shultz to U.S. Embassy Afghanistan, 23 November 1982, DNSA, Document AF01403.
 SECURITY STUDIES 153

became even clearer in 1984 and 1985. An intelligence assessment in early 1985
noted “the Soviets … have already signaled that more outside support will not dis-
suade them from pursuing their objective of full control over the country” in part
by “continuing cross-border incidents.”127
These inferences specifically suggest American and Pakistani observers viewed
the pace and intensity of Soviet cross-border activity as a bellwether regarding
Soviet ambitions in Afghanistan and vis-a-vis Pakistan. Moreover, the cross-border
raids specifically appear to involve the counter-escalatory risks mechanism; such
strikes into Pakistani territory substantially increased the chances of an incident
with Pakistani border defense units and threatened to destabilize Islamabad’s con-
trol of its northwestern frontier provinces. While cross-border actions had opera-
tional benefits as well, the primary documents show that American and Pakistani
leaders inferred a larger coercive political message regarding Afghanistan. The tim-
ing of these increases and the Soviet preference for activity that was visible to Pak-
istani and American leaders, but not wider audiences, suggests a signaling motive.
Definitive answers regarding H2 must await further document releases on this
aspect of the Afghanistan campaign.

Discussion
To summarize, the conflict in Afghanistan features three instances in which covert
actions were used to signal resolve, interpreted as signals of resolve, or both. Con-
sistent with H2 and H3, pre-invasion covert combat support by the Soviet Union
was detected by American intelligence, was critical to judgments of Soviet resolve,
and likely helped express Moscow’s support to its local ally. Regarding the weak-
ness hypothesis (H4), American leaders in 1979 did not interpret the constraint on
Moscow created by detente as weakness; instead, expanded covert involvement
despite the risks to detente was seen as a sign of resolve. American covert aid post-
invasion was intended to convey to Washington’s regional partners and Moscow a
willingness to punish aggression, was tracked by Soviet intelligence, and had a reas-
suring effect on a key American local partner (Pakistan). Later Soviet secret cross-
border strikes were consistently interpreted as sending a coercive message to Pakis-
tani leaders. Overall, reactions to developments in the covert sphere support our
basic contention that leaders share an understanding about qualitative changes in
covert interventions and the symbolic significance of crossing thresholds. We find
specific support for the unique escalatory and domestic political risks incurred by
covert aid and covert cross-border raids. We find less specific evidence regarding
sunk costs. Finally, Afghanistan provides little evidence in favor of the null (H1)
and weakness (H4) hypotheses. While observing leaders could have interpreted
covert aid and secret cross-border raids as weak half-measures, we found

127
The report also cites evidence Soviet military leaders were doubling down on defeating the counterinsurgency
within Afghanistan. Special National Intelligence Assessment, March 1985, CIAERR, Document 0000518057.
154 A. CARSON AND K. YARHI-MILO

consistent evidence of an appreciation for limitations on overt alternatives and for

resolve expressed through covert action.128

Implications
Do leaders use covert action to send signals of resolve and are such messages intel-
ligible and credible to signal receivers? We argue leaders share a basic interpretive
framework that allows covert action to express intentions like resolve. Covert
action is especially useful as a method for sending targeted signals to leaders of
local allies and strategic adversaries. Moreover, these messages are credible because
of the sunk costs and escalation and domestic risks they incur. As we note theoreti-
cally and identify in the cases, much of the costs and risks in covert action are
unique to this form of signaling.
Readers may wonder whether this phenomenon is a curious artifact of the Cold
War. This observation would suggest that a unique feature of the Soviet-American
rivalry—unusually dense interactions, advanced intelligence collection capabilities,
alliance networks, or the presence of nuclear weapons—was necessary for covert
signaling to be detected, understood, and taken seriously. In fact, these features are
quite common even outside the superpower rivalry, differing if at all in degree
rather than kind. Dense and repeated interactions are common in regional rivalries
including Israel and the Arab states, India and Pakistan, and the two Koreas. While
Soviet and American intelligence organizations were large, other major powers also
possess extensive intelligence coverage. Moreover, regional rivals often develop
their own regional intelligence capacities to track local covert activity. Alliance
considerations are hardly restricted to the Cold War era and have recently played
an important role in Western debates over policy regarding Crimea/Eastern
Ukraine. Shared nuclear weapons are featured in current rivalries like India and
Pakistan. In any case, while nuclear weapons may push more activity to the covert
sphere due to fears of direct war, their presence is not necessary for leaders to use
covert action and find targeted signaling useful. Covert signaling therefore remains
an attractive option for leaders in a range of regional and global geopolitical rival-
ries. While anecdotal, this point is underscored by the lessons from the Israeli
covert strike in 2007 discussed above and reporting on covert activity regarding
the Ukraine crisis.129
As we note in the article’s opening, we do not claim leaders use covert action
exclusively to signal in any given case. On-the-ground effects are often part of the

128
Though space constraints do not allow a full exploration, hints of the same dynamics regarding Chinese covert aid
are available as well. An American intelligence report on a Chinese military delegation’s visit to Islamabad in Octo-
ber 1979 (before the Soviet invasion) noted President Zia’s request for Chinese weaponry for Afghan insurgents,
Chinese opposition to public aid in light of sensitive Sino-Soviet relations, and Chinese and Pakistani satisfaction
with covert arms arrangements instead. See Intelligence Report, October 30, 1979, DNSA, CIAERR, Document
AF00716.
129
For example, Michael R. Gordon, “Russia Displays a New Military Prowess in Ukraine’s East,” New York Times, 21 April
2014, http://www.nytimes.com/2014/04/22/world/europe/new-prowess-for-russians.html.
 SECURITY STUDIES 155

story as well, and our case evidence confirms as much in several examples. How-
ever, the evidence regarding cases in both Angola and Afghanistan makes clear
that a primary motivation for these covert interventions was to signal resolve to
local allies and strategic adversaries. In other words, while it is possible a state
could use covert action primarily for operational reasons and any signaling func-
tion is relatively insignificant, the empirical record in the cases examined here
shows that a primary function was signal sending. In general, establishing the rep-
utation of being a global power determined to contest and even extend influence in
the Third World influenced the use of covert action by both sides during this
period. Geostrategic stakes of the Persian Gulf, coupled with the collapse of
detente, magnified the importance of appearing resolved for the superpowers.
These conditions are hardly unique. American leaders acting in the post-Iraq cli-
mate, and other states that use covert action, can face their own challenge of con-
straints on overt signaling and geostrategic demands for demonstrating resolve.
As we note, introducing and theorizing a novel signaling tool contributes to
ongoing debates about how and under what conditions states can communicate
intentions like resolve. Several specific aspects of the theory—the unique costs and
risks of acting secretly, an emphasis on multiple audiences (for example, local
allies), a general framework for and unique evidence regarding the intelligibility of
covert actions—may prove especially useful for scholars interested in signaling,
reputation, crisis escalation, and related research areas. There are useful policy
implications as well for contemporary leaders seeking ways in which to respond to
crises like those in Syria, Yemen, and Ukraine.130 Moreover, findings about specific
mechanisms such as sunk costs and risk generation provide critical lessons for
leaders choosing among possible covert signaling tools. Cyber technology, for
example, is a growing arena for covert action. Our theoretical framework and
empirical findings suggest certain kinds of covert cyber attacks are more likely to
convey resolve than others, with implications for the kinds of attacks to expect
from rivals, the kinds of effects (that is, sinking costs, generating escalation, and
domestic risk) to look for, and the necessity of an evolved interpretive framework
if cyber attacks are to create intelligible messages of resolve in the first place.131
Finally, our theory and findings raise important questions for future research.
Scholars could explore the larger universe of signaling messages to address whether
and how states use secret policy tools to reassure rivals, court new allies, express
restraint to adversaries, etc. Future work could systematically address conditions
under which covert action signals restraint or reassurance to allies, and investigate

130
See, for example, proposals for military aid to Ukraine in Michael R. Gordon and Eric Schmitt, “U.S. Considers Supply-
ing Arms to Ukraine Forces, Officials Say,” New York Times, 1 February 2015, http://www.nytimes.com/2015/02/02/
world/us-taking-a-fresh-look-at-arming-kiev-forces.html.
131
For example, Martin C. Libicki, Cyberdeterrence and Cyberwar (Santa Monica, CA: Rand Corporation, 2009); Martin C.
Libicki, Crisis and Escalation in Cyberspace, MG-1215-AF (Santa Monica, CA: RAND Corporation, 2012), http://www.
rand.org/pubs/monographs/MG1215.html.
156 A. CARSON AND K. YARHI-MILO

how psychological biases and patterns of misperception influence the effectiveness

of covert signaling games.

Acknowledgments
For especially helpful comments, we thank Jon Brown, Ahsan Butt, Charlie Glaser, Jessica Got-
tleib, Ryan Grauer, Robert Jervis, Michael Joseph, Danielle Lupton, Julia McDonald, Josh Shi-
frinson, Michael Poznansky, and Brian Rathbun. We also thank the editors at Security Studies
and two anonymous reviewers for extremely thoughtful commentary. We thank Doyle Hodges,
Anatoly Levshin, Rohan Mukherjee, and Joshua Zuckerman for excellent research assistance.
All remaining errors are our own.