The inhibiting effects of information overload on the behavior of online social media users, can affect the population-level characteristics of information dissemination through online conversations. We introduce a mechanistic, agent-based model of information overload and investigate the effects of information overload threshold and rate of information loss on observed online phenomena. We find that conversation volume and participation are lowest under high information overload thresholds and mid-range rates of information loss. Calibrating the model to user responsiveness data on Twitter, we replicate and explain several observed phenomena: (1) Responsiveness is sensitive to information overload threshold at high rates of information loss; (2) Information overload threshold and rate of information loss are Pareto-optimal and users may experience overload at inflows exceeding 30 notifications per hour; (3) Local abundance of small cascades of modest global popularity and local scarcity of larger cascades of high global popularity explains why overloaded users receive, but do not respond to large, highly popular cascades; 4) Users typically work with 7 notifications per hour; 5) Over-exposure to information can suppress the likelihood of response by overloading users, contrary to analogies to biologically-inspired viral spread. Reconceptualizing information spread with the mechanisms of information overload creates a richer representation of online conversation dynamics, enabling a deeper understanding of how (dis)information is transmitted over social media.

Influence cascades are typically analyzed using a single metric approach, i.e., all influence is measured using one number. However, social influence is not monolithic; different users exercise different influences in different ways, and influence is correlated with the user and content-specific attributes. One such attribute could be whether the action is an initiation of a new post, a contribution to a post, or a sharing of an existing post. In this paper, we present a novel method for tracking these influence relationships over time, which we call influence cascades, and present a visualization technique to better understand these cascades. We investigate these influence patterns within and across online social media platforms using empirical data and comparing to a scale-free network as a null model. Our results show that characteristics of influence cascades and patterns of influence are, in fact, affected by the platform and the community of the users.

The dynamics of information diffusion on social media platforms vary significantly between individual communities and the broader population. This study explores and compares the differences between community-based interventions and population-wide approaches in adjusting the spread of information. We first examine the temporal dynamics of social media groups, assessing their behavior through metrics such as time-dependent posts and retweets. Using functional data analysis, we investigate Twitter activities related to incidents such as the Skripal/Novichok case. We present three ways to quantify disparities between communities and uncover the strategies used by each group to promote specific narratives. We then compare the impact of targeted, community-based interventions with that of broader, population-wide responses in shaping the diffusion of information. Through this analysis, we identify key differences in how communities engage with and amplify information, revealing distinct patterns in the diffusion process. Our findings provide a comparative framework for understanding the relative consequences of different intervention strategies, offering insights into how targeted and broad approaches influence public discourse across social media platforms.

Implements methods for estimating latent Gaussian processes from categorical functional data using binomial, probit, or multinomial GAM-based approaches, and for clustering individuals via multivariate functional principal component scores. Methods are described in Champon et al. (2026) .