Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Beyond Interaction Patterns: Assessing Claims of Coordinated Inter-State Information Operations on Twitter/X

Valeria Pantè, David Axelrod, Alessandro Flammini, Filippo Menczer, Emilio Ferrara, Luca Luceri

TL;DR

The paper investigates inter-state coordinated influence campaigns on Twitter/X, addressing claims of state-level collaboration. It employs multiple behavioral traces and state-of-the-art coordination detection models, incorporating a control dataset to differentiate organic activity from coordinated efforts and to avoid spurious signals, with analyses encompassing both similarity- and interaction-based indicators and 197 experiments. Using Mann-Whitney $U$ tests (with Bonferroni correction) to compare IO and control distributions, the study finds no statistically significant evidence of inter-state coordination, even for aggregates that appeared suspicious at the state level. The results emphasize the importance of robust methodologies and control data when detecting online coordination and caution against inferring coordination from interaction patterns alone.

Abstract

Social media platforms have become key tools for coordinated influence operations, enabling state actors to manipulate public opinion through strategic, collective actions. While previous research has suggested collaboration between states, such research failed to leverage state-of-the-art coordination indicators or control datasets. In this study, we investigate inter-state coordination by analyzing multiple online behavioral traces and using sophisticated coordination detection models. By incorporating a control dataset to differentiate organic user activity from coordinated efforts, our findings reveal no evidence of inter-state coordination. These results challenge earlier claims and underscore the importance of robust methodologies and control datasets in accurately detecting online coordination.

Beyond Interaction Patterns: Assessing Claims of Coordinated Inter-State Information Operations on Twitter/X

TL;DR

The paper investigates inter-state coordinated influence campaigns on Twitter/X, addressing claims of state-level collaboration. It employs multiple behavioral traces and state-of-the-art coordination detection models, incorporating a control dataset to differentiate organic activity from coordinated efforts and to avoid spurious signals, with analyses encompassing both similarity- and interaction-based indicators and 197 experiments. Using Mann-Whitney tests (with Bonferroni correction) to compare IO and control distributions, the study finds no statistically significant evidence of inter-state coordination, even for aggregates that appeared suspicious at the state level. The results emphasize the importance of robust methodologies and control data when detecting online coordination and caution against inferring coordination from interaction patterns alone.

Abstract

Social media platforms have become key tools for coordinated influence operations, enabling state actors to manipulate public opinion through strategic, collective actions. While previous research has suggested collaboration between states, such research failed to leverage state-of-the-art coordination indicators or control datasets. In this study, we investigate inter-state coordination by analyzing multiple online behavioral traces and using sophisticated coordination detection models. By incorporating a control dataset to differentiate organic user activity from coordinated efforts, our findings reveal no evidence of inter-state coordination. These results challenge earlier claims and underscore the importance of robust methodologies and control datasets in accurately detecting online coordination.

Paper Structure

This paper contains 1 section, 1 figure, 2 tables.

Table of Contents

  1. Introduction

Figures (1)

  • Figure 1: Inter-country connections based on aggregated interactions (top row) and similarity features (bottom row) among IO users. For interactions, edges are weighted by the out-strength of each node. For similarity networks, edges are weighted by the Jaccard coefficient, which is computed as the ratio of shared features to the total number of unique features between the two countries.