Universality, criticality and complexity of information propagation in social media

Abstract

Information avalanches in social media are typically studied in a similar fashion as avalanches of neuronal activity in the brain. Whereas a large body of literature reveals substantial agreement about the existence of a unique process characterizing neuronal activity across organisms, the dynamics of information in online social media is far less understood. Statistical laws of information avalanches are found in previous studies to be not robust across systems, and radically different processes are used to represent plausible driving mechanisms for information propagation. Here, we analyze almost 1 billion time-stamped events collected from a multitude of online platforms -- including Telegram, Twitter and Weibo -- over observation windows longer than 10 years to show that the propagation of information in social media is a universal and critical process. Universality arises from the observation of identical macroscopic patterns across platforms, irrespective of the details of the specific system at hand. Critical behavior is deduced from the power-law distributions, and corresponding hyperscaling relations, characterizing size and duration of avalanches of information. Neuronal activity may be modeled as a simple contagion process, where only a single exposure to activity may be sufficient for its diffusion. On the contrary, statistical testing on our data indicates that a mixture of simple and complex contagion, where involvement of an individual requires exposure from multiple acquaintances, characterizes the propagation of information in social media. We show that the complexity of the process is correlated with the semantic content of the information that is propagated. Conversational topics about music, movies and TV shows tend to propagate as simple contagion processes, whereas controversial discussions on political/societal themes obey the rules of complex contagion.

10 pages, 5 figures, 7 pages of bibliography, 28 pages of supplemental material

References in corpus (8)

• Power-law distributions in empirical data
• Robust dynamic classes revealed by measuring the response function of a social system
• A generalized model of social and biological contagion
• Exactly solvable model of avalanches dynamics for Barkhausen crackling noise
• An Early Look at the Parler Online Social Network
• Percolation theory of self-exciting temporal processes
• Online misinformation is linked to early COVID-19 vaccination hesitancy and refusal
• The Pushshift Telegram Dataset

Cited by in corpus (7)

• Distinguishing simple and complex contagion processes on networks
• Characterizing Engagement Dynamics across Topics on Facebook
• Collective dynamics behind success
• Influence Maximization: Divide and Conquer
• Self-reinforcing cascades: A spreading model for beliefs or products of varying intensity or quality
• Integrating behavioral experimental findings into dynamical models to inform social change interventions
• Anomalous finite-size scaling in higher-order processes with absorbing states