Chapter 24: Further Reading — Computational Propaganda and Bot Detection

Sources are annotated and grouped thematically. All sources are cited in the chapter text.

Foundational Frameworks

Woolley, S. C., & Howard, P. N. (Eds.). (2019). Computational Propaganda: Political Parties, Politicians, and Political Manipulation on Social Media. Oxford University Press.

The definitive academic reference on computational propaganda. Woolley and Howard edited this volume synthesizing research from 28 countries, documenting the global spread of bot-assisted political manipulation. Individual chapters cover elections in the United States, UK, France, Germany, Brazil, Taiwan, Russia, and many other countries, providing the broadest comparative empirical picture available. The introductory chapters establish the theoretical framework — automation, algorithm exploitation, micro-targeting — used throughout Chapter 24. Open access version available through the Oxford Internet Institute.

Howard, P. N., & Kreiss, D. (2010). Political parties and voter privacy: Australia, Canada, the United Kingdom, and the United States in comparative perspective. First Monday, 15(12).

An earlier Howard work that contextualizes computational propaganda within the broader history of political data practices. The paper traces how political campaigns began using commercial data mining long before social media, establishing that computational targeting is not new but has been dramatically amplified by social media platforms. Useful historical context for understanding computational propaganda as an evolution of existing practices rather than an entirely novel phenomenon.

Bot Detection: Systems and Methods

Varol, O., Ferrara, E., Davis, C. A., Menczer, F., & Flammini, A. (2017). Online human-bot interactions: Detection, estimation, and characterization. Proceedings of the International AAAI Conference on Web and Social Media, 11(1), 280–289.

The original Botometer (BotOrNot) paper. Describes the feature engineering pipeline, the machine learning architecture, and the first large-scale validation of automated bot detection on Twitter. Reports an AUC of approximately 0.95 on the benchmark dataset, demonstrating the potential of machine learning approaches for bot detection. Essential reading for understanding the Botometer system; should be read in conjunction with Yang et al. (2022) for the updated architecture.

Yang, K. C., Ferrara, E., & Menczer, F. (2022). Botometer 101: Social bot practicum for computational social scientists. Journal of Computational Social Science, 5, 1511–1528.

An updated, accessible description of the Botometer system for social scientists rather than computer scientists. Covers the evolution of Botometer from version 1 to version 4, discusses limitations (dataset shift, false positive disparities, API dependencies), and provides practical guidance on how to use and interpret Botometer scores responsibly. The paper explicitly addresses common misinterpretations of Botometer output, making it essential reading before using the system.

Ferrara, E., Varol, O., Davis, C., Menczer, F., & Flammini, A. (2016). The rise of social bots. Communications of the ACM, 59(7), 96–104.

An influential and accessible overview of social bots and their ecosystem, written for a broad computer science audience. Covers the taxonomy of bots (spam bots, social bots, political bots), detection approaches, and the broader implications for online discourse. Less technical than the Botometer paper and a better starting point for students new to the field.

State-Sponsored Information Operations

King, G., Pan, J., & Roberts, M. E. (2017). How the Chinese government fabricates social media posts for strategic distraction, not engaged argument. American Political Science Review, 111(3), 484–501.

The landmark study analyzed in Case Study 24.2. King, Pan, and Roberts combine leaked government document analysis with large-scale Weibo data to characterize the Chinese government's domestic online opinion manipulation operation. The finding that the operation uses strategic distraction rather than engaged argument remains the single most surprising and influential empirical result in computational propaganda research. The paper is also methodologically exemplary — the natural experiment design is a model for causal inference with observational social media data.

Roberts, M. E. (2018). Censored: Distraction and Diversion inside China's Great Firewall. Princeton University Press.

Roberts' monograph extends the King et al. paper into a book-length treatment of Chinese internet censorship and online opinion management. She argues that the Chinese government uses a combination of "fear" (harsh punishment of the most sensitive content), "friction" (making access to sensitive content difficult but not impossible), and "flooding" (distraction) rather than relying solely on censorship. This three-part model is theoretically illuminating beyond the China case and provides a framework for understanding repressive information management more generally.

United States Senate Select Committee on Intelligence. (2019). Report of the Select Committee on Intelligence, United States Senate, on Russian Active Measures Campaigns and Interference in the 2016 U.S. Election, Volume 2: Russia's Use of Social Media. US Government Printing Office.

The definitive official US government account of the IRA's social media operations. Volume 2 covers Facebook, Twitter, Instagram, YouTube, and Reddit activities in detail, with extensive appendices documenting specific accounts, content, and targeting strategies. While written for a legislative audience rather than an academic one, it provides the most comprehensive official documentation of the IRA operation and is essential reading for anyone working on Russian influence operations. Freely available online.

Starbird, K., Arif, A., & Wilson, T. (2019). Disinformation as collaborative work: Surfacing the participatory nature of strategic information operations. Proceedings of the ACM on Human-Computer Interaction, 3(CSCW), 1–26.

An important qualitative study of how disinformation operations function through the collaborative amplification of strategic narratives, rather than through top-down broadcasting. Starbird et al. show that state-sponsored operations succeed in part because they resonate with existing communities' beliefs and are actively amplified by those communities, making attribution and counter-messaging genuinely difficult. This paper productively complicates simple "state actor as broadcaster" narratives.

Coordinated Inauthentic Behavior

Sharma, K., Ferrara, E., & Liu, Y. (2021). Identifying coordinated accounts on social media through hidden influence and covert coordination. Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, 1501–1509.

Introduces an efficient algorithmic approach to detecting coordinated inauthentic behavior using MinHash-based content similarity and temporal co-occurrence analysis. The FastCoordination method can scale to large datasets by avoiding expensive all-pairs comparisons. Essential technical reading for students implementing CIB detection systems.

Nizzoli, L., Tardelli, S., Avvenuti, M., Cresci, S., Tesconi, M., & Ferrara, E. (2021). Coordinated behavior on social media in 2019 UK general election. Proceedings of the International AAAI Conference on Web and Social Media, 15(1), 443–454.

Applies coordinated behavior detection to a major electoral context. The paper examines the 2019 UK general election, identifying coordinated amplification networks for multiple political parties and analyzing the content strategies of each. Demonstrates the application of detection methods beyond the US electoral context and provides a model for election integrity research.

Astroturfing and Influence Operations

Keller, F. B., Schoch, D., Stier, S., & Yang, J. (2020). Political astroturfing on Twitter: How to theorize and operationalize it. Political Communication, 37(2), 317–336.

A careful conceptual and methodological paper that distinguishes astroturfing from related phenomena (genuine grassroots campaigns, transparent organized campaigns) and proposes operational criteria for detection. Keller et al. develop a multi-criteria framework including coordination signals, concealment of organizational affiliation, and strategic intent. This paper is essential for anyone who wants to define their dependent variable precisely before beginning astroturfing research.

Platform Transparency and Research Access

Arnaudo, D. (2017). Computational Propaganda in Brazil: Social Bots during Elections. Oxford Internet Institute Computational Propaganda Project Working Paper.

One of the Oxford Computational Propaganda Project's 28-country reports. The Brazil case is particularly informative because Brazilian political operations have been well-documented and demonstrate that computational propaganda is not a phenomenon of only Western democracies or authoritarian states. The report demonstrates how computational propaganda methods translate across political systems.

Lukito, J. (2020). Coordinating a multi-platform disinformation campaign: Internet Research Agency activity on three U.S. social media platforms, 2015 to 2017. Political Communication, 37(2), 238–255.

One of the first studies to systematically examine IRA activity across multiple platforms (Facebook, Twitter, and YouTube simultaneously). Lukito documents how IRA content migrated and was adapted across platforms, with different content types finding different homes. This cross-platform perspective is essential for understanding how influence operations exploit the multi-platform information ecosystem.

Freelon, D., Bossetta, M., Wells, C., Lukito, J., Xia, Y., & Adams, K. (2020). Black trolls matter: Racial and ideological asymmetries in social media disinformation. Social Science Computer Review, 40(3), 560–578.

An important critical analysis of the IRA dataset that examines how the IRA's targeting of Black American communities relates to existing racial justice discourse on social media. Freelon et al. demonstrate significant asymmetry in how researchers have analyzed IRA content targeting different political communities and argue that the disproportionate targeting of Black communities has been underemphasized in mainstream computational propaganda research. Essential reading for understanding the racial politics of influence operations.