Online network topology shapes personal narratives and hashtag generation

TL;DR

This study experimentally examines how online network topology influences narrative coordination and hashtag generation during a disaster narrative. Using a controlled networked setup with Fukushima-content materials, participants operate on spatially-embedded versus homogeneously-mixed graphs across multiple network sizes, performing 40 hashtag coordination trials per run and analyzing both social dynamics and narrative causality. Key findings show that global connections accelerate the emergence of a dominant hashtag, while local neighborhoods promote intra-cluster coordination and can sustain multiple local hashtags; networked interactions also shift causal language in participant narratives, with stronger causal-chain amplification in homogeneous networks. The results offer insights into designing online spaces to mitigate echo chambers and promote cross-network understanding, highlighting how topology shapes belief coordination, entropy of responses, and narrative causality.

Abstract

While narratives have shaped cognition and cultures for centuries, digital media and online social networks have introduced new narrative phenomena. With increased narrative agency, networked groups of individuals can directly contribute and steer narratives that center our collective discussions of politics, science, and morality. We report the results of an online network experiment on narrative and hashtag generation, in which networked groups of participants interpreted a text-based narrative of a disaster event, and were incentivized to produce matching hashtags with their network neighbors. We found that network structure not only influences the emergence of dominant beliefs through coordination with network neighbors, but also impacts participants' use of causal language in their personal narratives.

Figures (7)

• Figure 1: Causal model communicated in the nuclear disaster narrative. This diagram is just for illustration purposes, participants did not see this diagram. They read a four-paragraph narrative describing how the T$\overline{\mathrm{o}}$hoku earthquake triggered a massive tidal wave that damaged the Fukushima Nuclear Power Plant, resulting in electricity outages, radiation leaks and poisoning, human displacement, and Setsuden, a national energy-saving holiday.
• Figure 2: Two neighborhood structures tested in this experiment: Spatially-embedded (left) and homogeneously-mixed (right) networks with $N = 10$ nodes. Red edges represent the neighborhoods for a hypothetical node 0 in both networks. As a network's size grows, the diameter of spatial networks grow whereas homogeneous networks maintain a diameter of $1$.
• Figure 3: Proportion of participants generating a dominant hashtag as a function of trials across network structures. Curves represent marginal effects of the linear model, and error bars represent $95\%$ Bayesian credible intervals. Homogeneously-mixed networks show faster emergence of dominant responses due to information aggregation across network connections.
• Figure 4: Entropy of hashtag responses decrease over time across network structures. Lower values imply greater coherence of responses across participants because generated hashtags are more similar. However, entropy of responses decrease more rapidly in homogeneously-mixed networks due wider communication of information.
• Figure 5: Emergence of coordinated behavior across network structures.