Three Tweets to Midnight: Effects of the Global Information Ecosystem on the Risk of Nuclear Conflict

By Independent Publishers Group

Disinformation and misinformation have always been part of conflict. But as the essays in this volume outline, the rise of social media and the new global information ecosystem have created conditions for the spread of propaganda like never before—with potentially disastrous results.

In our "post-truth" era of bots, trolls, and intemperate presidential tweets, popular social platforms like Twitter and Facebook provide a growing medium for manipulation of information directed to individuals, institutions, and global leaders. A new type of warfare is being fought online each day, often in 280 characters or fewer. Targeted influence campaigns have been waged in at least forty-eight countries so far. We've entered an age where stability during an international crisis can be deliberately manipulated at greater speed, on a larger scale, and at a lower cost than at any previous time in history.

This volume examines the current reality from a variety of angles, considering how digital misinformation might affect the likelihood of international conflict and how it might influence the perceptions and actions of leaders and their publics before and during a crisis. It sounds the alarm about how social media increases information overload and promotes "fast thinking," with potentially catastrophic results for nuclear powers.

• Language: English
• Publisher: Independent Publishers Group
• Release date: Mar 15, 2020
• ISBN: 9780817923365

Book preview

ADVANCE PRAISE FOR

THREE TWEETS TO MIDNIGHT

Effects of the Global Information Ecosystem on the Risk of Nuclear Conflict

A gripping story of how social media can result in a nuclear catastrophe, either through a blunder or through the actions of a malignant provocateur. No issue could be timelier or more important, considering the profligate use of tweets today by the president and other government officials, and the need for deliberation in dealing with national security crises.

—WILLIAM J. PERRY, 19th US Secretary of Defense

Highlights new and rising dangers that social media pose to managing any future great power crisis, and in the extreme to avoiding nuclear war. It is a must-read for policy makers, legislators, foreign policy experts, nuclear strategists, and indeed for any serious student of national security.

—JAMES N. MILLER, former US Under Secretary of Defense for Policy

The next nuclear crisis will be tweeted. How decision makers cope with the increasing speed and volume of information during that crisis will weigh heavily on whether the world can avoid nuclear catastrophe. The authors in this volume brilliantly help us understand—and get ahead of—the challenges from today’s information ecosystem.

—KEITH PORTER, President, the Stanley Center for Peace and Security

We know that the new media environment has an impact on nuclear crises, but how and when does it matter? This pathbreaking volume assembles an impressive interdisciplinary lineup to explore these questions with new frameworks, new evidence, and new arguments. An important opening contribution to what is clearly a phenomenon that is here to stay.

—VIPIN NARANG, Associate Professor of Political Science at MIT and a member of MIT’s Security Studies Program

Three Tweets to Midnight:

Effects of the Global Information Ecosystem on the Risk of Nuclear Conflict

The Hoover Institution gratefully acknowledges the following individuals and foundations for their significant support of this publication.

HANK J. HOLLAND

LYNDE AND HARRY BRADLEY FOUNDATION

THE DAVIES FAMILY

ROBERT AND MARION OSTER

The Stanley Center for Peace and Security, the Hoover Institution, and the Center for International Security and Cooperation at Stanford are pleased to collaborate in supporting this scholarship and bringing forward our collective understanding of its profound implications for the future of journalism, nuclear weapons policy, and international peace and security.

Three Tweets to Midnight

Effects of the Global Information Ecosystem on the Risk of Nuclear Conflict

EDITORS

Harold A. Trinkunas

Herbert S. Lin

Benjamin Loehrke

CONTRIBUTING AUTHORS

Kelly M. Greenhill

Danielle Jablanski

Jaclyn A. Kerr

Mark Kumleben

Jeffrey Lewis

Herbert S. Lin

Benjamin Loehrke

Rose McDermott

Ben O’Loughlin

Paul Slovic

Kate Starbird

Harold A. Trinkunas

Kristin Ven Bruusgaard

Samuel C. Woolley

HOOVER INSTITUTION PRESS

Stanford University

Stanford, California

With its eminent scholars and world-renowned library and archives, the Hoover Institution seeks to improve the human condition by advancing ideas that promote economic opportunity and prosperity, while securing and safeguarding peace for America and all mankind. The views expressed in its publications are entirely those of the authors and do not necessarily reflect the views of the staff, officers, or Board of Overseers of the Hoover Institution.

hoover.org

Hoover Institution Press Publication No. 707

Hoover Institution at Leland Stanford Junior University, Stanford, California, 94305-6003

Copyright © 2020 by the Board of Trustees of the Leland Stanford Junior University

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without written permission of the publisher and copyright holders.

For permission to reuse material from Three Tweets to Midnight: Effects of the Global Information Ecosystem on the Risk of Nuclear Conflict, edited by Harold A. Trinkunas, Herbert S. Lin, and Benjamin Loehrke, ISBN 978-0-8179-2335-8, please access www.copyright.com or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and registration for a variety of uses.

Efforts have been made to locate the original sources, determine the current rights holders, and, if needed, obtain reproduction permissions. On verification of any such claims to rights in the articles reproduced in this book, any required corrections or clarifications will be made in subsequent printings/editions.

Cataloging-in-Publication Data is available from the Library of Congress.

ISBN 978-0-8179-2335-8 (paperback)

ISBN 978-0-8179-2336-5 (ePub)

ISBN 978-0-8179-2337-2 (Mobipocket)

ISBN 978-0-8179-2338-9 (PDF)

Contents

LIST OF FIGURES AND TABLES

ACKNOWLEDGMENTS

1. Retweets to Midnight: Assessing the Effects of the Information Ecosystem on Crisis Decision Making between Nuclear Weapons States

Danielle Jablanski, Herbert S. Lin, and Harold A. Trinkunas

2. Psychological Underpinnings of Post-truth in Political Beliefs

Rose McDermott

3. The Caveman and the Bomb in the Digital Age

Paul Slovic and Herbert S. Lin

4. Gaming Communication on the Global Stage: Social Media Disinformation in Crisis Situations

Mark Kumleben and Samuel C. Woolley

5. Information Operations and Online Activism within NATO Discourse

Kate Starbird

Appendix A: Temporal Patterns (Tweets per Hour) by Cluster

Appendix B: Top Most Retweeted Accounts by Cluster

Appendix C: Top Most Tweeted Domains by Cluster

Appendix D: Most Frequent Terms in Account Profiles (User Descriptions)

Appendix E: Most Frequent Terms in (Unique) Tweets by Cluster

Appendix F: Percentage of Retweets from Retweeter Cluster (rows) to Retweeted Cluster (columns)

6. Of Wars and Rumors of Wars: Extra-factual Information and (In)Advertent Escalation

Kelly M. Greenhill

7. Crisis Stability and the Impact of the Information Ecosystem

Kristin Ven Bruusgaard and Jaclyn A. Kerr

8. Bum Dope, Blowback, and the Bomb: The Effect of Bad Information on Policy-Maker Beliefs and Crisis Stability

Jeffrey Lewis

9. The Impact of the Information Ecosystem on Public Opinion during Nuclear Crises: Lifting the Lid on the Role of Identity Narratives

Ben O’Loughlin

10. What Can be Done to Minimize the Effects of the Global Information Ecosystem on the Risk of Nuclear War?

Harold A. Trinkunas, Herbert S. Lin, and Benjamin Loehrke

ABOUT THE EDITORS AND CONTRIBUTORS

INDEX

List of Figures and Tables

Figure 1.1. Hypothetical tweet by President John F. Kennedy during Cuban Missile Crisis.

Figure 3.1. Two normative models for valuing noncombatant lives as the number at risk increases.

Figure 3.2. Psychic numbing: a descriptive model where the value of a life depends on how many lives are at risk.

Figure 3.3. Compassion Collapse: value sometimes decreases when many lives are at risk.

Figure 3.4. World War II poster: Enemies Both! It’s Your Job to Help Eliminate Them.

Figure 5.1. Retweet network graph of NATO (spring 2018) tweets (without edges).

Figure 5.2. Retweet network graph of NATO (spring 2018) tweets (with edges).

Table 5.1. Descriptive Statistics for the Clusters of Accounts Shown in Figures 5.1 and 5.2.

Acknowledgments

This project began offline. Domestic and global political events in 2016 and 2017 shook loose much complacency about social media and their role in reshaping how the world interacts. While platforms such as Facebook and Twitter were once hailed as liberation technologies, scholars increasingly began to worry about how the rapid changes wrought by social media might also affect international security. An early example of the kinds of interactions made possible by the emerging global information ecosystem was the unprecedented and combative use of social media by the US president to directly talk about North Korea and its nuclear and missile programs. This raised the immediate question of how such communications would be interpreted in Pyongyang and whether direct, instantaneous, globally broadcast, and potentially escalatory rhetoric could make the outbreak of a catastrophic war with North Korea more likely.

Over an informal conversation in mid-2017 at Stanford University’s Center for International Security and Cooperation (CISAC), the organizers of this project began to ask if nuclear crises, like the one brewing with the United States and North Korea, might be particularly susceptible to destabilizing effects from information driven through social media. Decisions on the use of nuclear weapons would be made under extreme stress, with imperfect information, potentially in just minutes, by a small group of advisers or a single decision maker. No decision would be more fraught, with millions of lives in the balance. Yet the psychology of it would be prone to leading decision makers to engage in heuristic thinking, leaving them vulnerable to emotional responses and misinterpretation.

How would 280-character missives be read during a nuclear crisis? How might the new information ecosystem, as reshaped by social media, affect leaders and publics before and during crises? And how might digital misinformation and disinformation affect the likelihood of international conflict?

This volume, and the two multidisciplinary workshops that informed it, were the product of a partnership between the Stanley Center for Peace and Security, the Hoover Institution at Stanford University, and CISAC.

The Stanley Center for Peace and Security hosted an exploratory workshop at its Strategy for Peace Conference in October 2017, the observations from which were then summarized in a briefing paper titled Three Tweets to Midnight: Nuclear Crisis Stability and the Information Ecosystem.¹ For more in-depth study of the arguments surfaced at that workshop, the organizers commissioned a set of working papers and hosted a second workshop at the Hoover Institution in September 2018. This volume is the culmination of the two events and features the contributions of authors for the 2018 workshop.

Participants in those workshops provided invaluable insights that shaped the trajectory of this project. For their contributions and thoughtful arguments, the editors thank James Acton, Jeffrey Berejikian, Malfrid Braut-Hegghammer, Paul Edwards, Jennifer Erickson, Henry Farrell, Anya Fink, Matt Fuhrmann, Tom Glaisyer, Deborah Gordon, Robert Gorwa, Andy Grotto, Rosanna Guadagno, Brian Hanson, Peter Hayes, Colin Kahl, Jennifer Kavanagh, Rupal Mehta, Anna Péczeli, Steve Pifer, Keith Porter, Philip Reiner, Scott Sagan, John Scott-Railton, Lior Tabansky, Phil Taubman, Devon Terrill, Ben Valentino, Heather Williams, and Amy Zegart.

The editors give special thanks to Amy Zegart for her early thought contributions and support throughout this project. As a professor of political science, codirector of CISAC, and Davies Family Senior Fellow at the Hoover Institution, Zegart was uniquely well positioned to recognize an opportunity to invest in and contribute to an innovative research project that aimed to address highly important policy questions that were of immediate relevance. Throughout the writing of this volume, the editors benefited from her questions, insights, and participation. In addition, we are grateful for the support of the Hoover Institution and, in particular, the Robert and Marion Oster National Security Affairs Fellows Program and the Lakeside Foundation in making this work possible by providing critical funding and meeting space. We also thank CISAC for providing the staff support that kept this whole enterprise moving. Herb Lin’s work at CISAC on this project was also supported in part by the Carnegie Corporation of New York grant G-17–55292 (International Program), For research, teaching, international engagement, and outreach in international security.

In addition to contributing as an author, Danielle Jablanski was an essential partner in every aspect of this project from ideation to implementation, both while she was at the Stanley Center for Peace and Security and as part of CISAC. The editors also thank Patty Papke, Cayte Connell, and Caitlin Lutsch at the Stanley Center for Peace and Security, Matt Ellison at the Hoover Institution, and Alida Haworth at CISAC for their work that made this project possible.

We would also like to acknowledge the work of the Bulletin of the Atomic Scientists, whose iconic Doomsday Clock provided an inspiration for the front cover of our book. The clock was conceived in 1947 by the Bulletin as a means of conveying to humanity the organization’s collective expert assessment of the risks posed by nuclear weapons. Each year, the Science and Security Board of the Bulletin (of which one of the editors, Herb Lin, is a member) updates the Doomsday Clock. As of the time of writing in 2019, the clock stands at two minutes to midnight, the closest it has been since 1953. We share the concern of the experts and staff of the Bulletin over the growing risk of nuclear conflict, which was a major motivation for the writing of this volume. That its title is Three Tweets to Midnight reflects creative license rather than any disagreement by the editors with the assessment of the Science and Security Board of the Bulletin of the Atomic Scientists.

Finally, we would like to be clear that the views expressed in this book are those of the editors and authors, and any remaining errors are ours alone.

HAROLD A. TRINKUNAS

HERBERT S. LIN

BENJAMIN LOEHRKE

Notes

1. Stanley Center for Peace and Security, Three Tweets to Midnight: Nuclear Crisis Stability and the Information Ecosystem, 58th Strategy for Peace Conference, January 2018, https://www.stanleyfoundation.org/resources.cfm?id=1646&title=Three-Tweets-to-Midnight:-Nuclear-Crisis-Stability-and-the-Information-Ecosystem.

Chapter 1

Retweets to Midnight:

Assessing the Effects of the Information Ecosystem on Crisis Decision Making between Nuclear Weapons States

Danielle Jablanski, Herbert S. Lin, and Harold A. Trinkunas

What if the Cuban Missile Crisis had taken place in today’s global information environment, characterized by the emergence of social media as a major force amplifying the effects of information on both leaders and citizens? President Kennedy might not have had days to deliberate with the Executive Committee of the National Security Council before delivering a measured speech announcing to the world the discovery of Soviet medium- and intermediate-range nuclear-armed missiles in Cuba.¹

Nongovernmental open source intelligence organizations like Bellingcat could have used commercially available satellite imagery to detect the presence of these missiles and publicize them to the world on October 12, 1962, four days earlier than the president did. Imagine pictures of the missile sites going viral on social media, alarming millions around the world. Imagine that these real-time images were accompanied by deliberate information operations from adversaries seeking to cast doubt on the facts to sow confusion and cause paralysis among domestic populations and between NATO leaders, as well as by internet trolls promoting misinformation and reposting and propagating tailored information leaks.

The shooting down of a U-2 spy plane over Cuba might have been news within the hour, becoming the subject of numerous tweets and relentless commentary on Facebook and other platforms. When the Joint Chiefs of Staff’s recommendation to invade Cuba was overruled by President Kennedy, alt social media accounts that served as fronts for disgruntled Pentagon officials might have leaked the proposed invasion plan to induce the administration to reverse course on the chosen alternative—a blockade. Pressured by public opinion and the Joint Chiefs of Staff, President Kennedy might not have had the luxury of picking which of Premier Khrushchev’s letters to respond to, which the historical record shows helped to de-escalate the crisis. In this situation, which former secretary of defense William J. Perry has characterized as the closest the world has come to nuclear catastrophe, the current global information ecosystem could have magnified the risk of the conflict’s escalating into all-out nuclear war.²

Figure 1.1. Hypothetical tweet by President John F. Kennedy during Cuban Missile Crisis.

Source: Scott Sagan, The Cuban Missile Crisis in the Age of Twitter, lecture at Stanford’s Center for International Security and Cooperation, April 3, 2018.

What’s New? Characteristics of the Modern Information Ecosystem

Social media and the resulting dynamics for interpersonal interconnectivity have increased the volume and velocity of communication by orders of magnitude in the past decade. More information reaches more people in more places than ever before. Algorithms and business models based on advertising principles utilize troves of user data to draw aggregate inferences, which allow for microsegmentation of audiences and direct targeting of disinformation or misinformation. Mainstream-media outlets no longer serve their traditional role as gatekeepers with near-universal credibility.³ In this ecosystem, propaganda can rapidly spread far and wide, while efforts to correct false information are more expensive, often fall short, and frequently fail altogether.

Nor are all of the voices on social media authentic. Some inauthentic voices are those of paid human trolls, for example from the Internet Research Agency, revealed to have created and spread false information on behalf of the Russian government prior to the 2016 US presidential election.⁴ Others are Macedonian entrepreneurs who at one point discovered ways to monetize an affinity among some voters for fake news critical of Hillary Clinton.⁵ Some voices are not even human, as demonstrated by the introduction of bots—automated social media accounts designed to mimic human behavior online that further complicate our ability to discern fact from fiction within the ecosystem.

Rapid transmission of content and curated affinity networks polarize citizens around divisive issues and create waves of public opinion that can pressure leaders.⁶ So many different narratives emerge around complex events that polities splinter into their disparate informational universes, unable to agree on an underlying reality. Does this unprecedented availability of information and connectivity amplify the ability of actors to sow discord in the minds of the domestic publics and even the leadership of adversaries? Could these dynamics affect leaders and citizens to the degree that miscalculation or misperception can produce crisis instability ultimately leading to a nuclear exchange? Can governance mechanisms be designed and implemented that are capable of countering and combating the manipulation of information in this ecosystem?

This volume argues that the present information ecosystem increasingly poses risks for crisis stability. Manipulated information, either artificially constructed or adopted by a strong grassroots base, can be used by interested actors to generate pressure from various constituencies on leaders to act. At the same time, these leaders themselves face information overload and their ability to distinguish between true and false information may be impaired, especially if they are receiving information simultaneously from their own sources and other sources from within their constituencies. Such confusion can ultimately lead to inaction or bad decisions. Or, this environment might produce an accelerated reaction based on slanted or unanalyzed information. Most worrisome is the possibility that the rapid spread of disinformation or misinformation via social media may in the end distort the decision-making calculus of leaders during a crisis and thereby contribute to crisis instability in future conflicts, the effects of which could be most severe for nuclear weapons states.

The Psychology of Complex Decision Making and Nuclear Crisis

Many theories of deterrence rely on the rationality assumption, namely that a rational actor can be convinced that the cost-benefit ratio associated with initiating an attack is unfavorable due to a credible threat of retaliation by the adversary. The risk of a nuclear exchange during the Cold War led theorists to focus on how leaders might approach crises and what could be done to avert deterrence failure. This prompted debates about a range of putatively rational actions that nuclear states might engage in to build a reliable framework for deterrence: reassurances to allies by extending the nuclear umbrella, force postures designed to ensure a survivable retaliatory capability, credible signaling to convince adversaries that any attack would meet with massive retaliation, etc.⁷

But human decision makers are just that—human—and a great deal of psychological research in the past few decades has demonstrated the limits of rational thinking and decision making. Paul Slovic has written extensively about the human brain, decision making, and limits for comprehending the weight of decisions that could imperil large numbers of human lives. Various psychological processes come into play when considering a cognitive calculation on the value of lives lost in large numbers, including psychic numbing, tribalism, the prominence effect, imperative thinking, and victim blaming. As Slovic and Herbert Lin argue in chapter 3, this implies that leaders facing the task of making a decision on whether to order the use of nuclear weapons find it difficult to operate rationally.

Psychology also tells us that—more often than not—fast, intuitive judgements take precedence over slower, more analytical thinking. Fast thinking (also identified as System 1 thinking by the originator of the concept, Daniel Kahneman) is intuitive and heuristic, generating rapid, reflexive responses to various situations and—more often than not—useful in daily life. Slow thinking (also known by cognitive psychologists as System 2 thinking) is more conceptual and deliberative.⁸ Although both are useful in their appropriate roles, their operation in today’s information ecosystem can be problematic. Fast thinking is problematic when when we are trying to understand how to respond to large-scale human crises, with catastrophic consequences, Slovic and Lin write. Slow thinking, too, can be incoherent in the sense that subtle influences—such as unstated, unconscious, or implicitly held attitudes—can lead to considered decisions that violate one’s strongly held values. The prevalence of heuristic and imperative thinking among humans suggests that an overarching important goal, such as national defense in the face of a nuclear crisis, would likely eclipse consideration of second-order effects and consequences, such as the likelihood of massive loss of life on all sides or catastrophic effects on the global environment, to the extent that such discussion is actively, if not subconsciously, avoided.⁹

Observers have always anticipated that leaders would be under severe time pressures when deciding whether or not to use nuclear weapons, the most important of which is launch on warning, the pressure to launch fixed land-based ICBMs before they can be destroyed on the ground by incoming enemy warheads. Fast, reflexive thinking (i.e., System 1 thinking) is more likely to be used under the kind of pressure this scenario highlights. Against a ticking clock, combined with the difficulty of comprehending the consequences of nuclear conflict, the argument that rational and deliberate decision making and deterrence will likely prevail, particularly under the added weight of the misinformation and disinformation that might propagate through the global information ecosystem during a crisis, is a highly debatable proposition.

The possibility that decision makers may rely on incorrect perceptions of potential adversaries has long been an important critique of rational deterrence theory. International relations theorists such as Robert Jervis have argued that the failure of deterrence can frequently be attributed to misperception among leaders: of intentions, of capabilities, of the consequences of conflict, etc. This misperception can have its roots in leaders’ psychology, in lack of information, and in leaders’ assumptions about what information the other side has or how they in turn perceive the situation.¹⁰

In the 1980s, Jervis had already argued that misperception was a quite common cause for deterrence failure. In today’s global