Intelligence notes

◉Dashboard↗

Politics and Propaganda on Social Media: How Twitter and Meta Moderate State-Linked Information Operations

3 min read

Politics and Propaganda on Social Media: How Twitter and Meta Moderate State-Linked Information Operations

Authors: Nihat Mugurtay, Umut Duygu, Onur Varol
Published: 2024-01-04
arXiv: 2401.02095
Source: arXiv (cs.SI)

Abstract

Why do Social Media Corporations (SMCs) engage in state-linked information operations? This research investigates the two SMCs’ policy orientation to explain which factors affect their reaction against state-linked information operations. Twitter and Meta carried out take-down operations against propaganda networks, accusing them of interfering foreign elections, organizing disinformation campaigns, and manipulating political debates. We find that good governance indicators such as democracy are significant elements of SMCs’ country-focus. This article also examines whether Meta and Twitter’s attention to political regime characteristics is influenced by international political alignments. The research illuminates recent trends in SMCs’ take-down operations and the interplay between geopolitics and domestic regime characteristics.

Why This Work Matters

Platform governance of state-linked IO is not neutral: Twitter and Meta’s takedown operations are themselves political acts, and selection of which campaigns to action is systematically biased. This paper provides the first rigorous empirical map of which state actors’ operations receive platform enforcement attention and what structural factors predict that attention.

The finding that democracy scores and geopolitical alignment predict takedown focus — not campaign size or documented harm — is analytically significant for understanding the limits of platform self-governance as an IO counter-measure. For vault analysis of IO campaigns: absence of takedown action is not evidence of absence of state-linked IO. Actors whose campaigns were not taken down are not cleaner; they may be geopolitically aligned with the platforms’ home political environment.

Core Concepts and Contributions

Platform enforcement as political act: Takedown decisions predicted by: (1) democracy scores of targeted countries (autocracies targeted more), (2) geopolitical alignment between the targeted country and the platforms’ home political alignment (US adversaries targeted more), (3) campaign visibility/public attribution (third-party reports accelerate takedown). Platform enforcement is a filtered counter-IO mechanism, not a comprehensive one.

Comparative Twitter vs. Meta: Twitter’s takedowns (pre-Musk) were more responsive to external researcher attribution (Stanford Internet Observatory, DFRLab); Meta’s were more proactive at scale. Post-Musk Twitter governance collapse represents a structural change the dataset predates but contextualizes.

Geopolitics → governance gap: States with authoritarian governance geopolitically aligned with Western interests (Gulf states, some Southeast Asian governments) exhibit significantly lower takedown rates despite documented IO activity. Platform governance reproduces geopolitical alignment hierarchies.

Dataset: Meta’s and Twitter’s public CIB takedown disclosures (2017–2023) — the most comprehensive public record of platform-level IO enforcement.

Connections

Graph View

Backlinks