Scaling laws of human interaction activity

Even though people in our contemporary, technological society are depending on communication, our understanding of the underlying laws of human communicational behavior continues to be poorly understood. Here we investigate the communication patterns in two social Internet communities in search of statistical laws in human interaction activity. This research reveals that human communication networks dynamically follow scaling laws that may also explain the observed trends in economic growth. Specifically, we identify a generalized version of Gibrat’s law of social activity expressed as a scaling law between the fluctuations in the number of messages sent by members and their level of activity. Gibrat’s law has been essential in understanding economic growth patterns, yet without an underlying general principle for its origin. We attribute this scaling law to long-term correlation patterns in human activity, which surprisingly span from days to the entire period of the available data of more than one year. Further, we provide a mathematical framework that relates the generalized version of Gibrat’s law to the long-term correlated dynamics, which suggests that the same underlying mechanism could be the source of Gibrat’s law in economics, ranging from large firms, research and development expenditures, gross domestic product of countries, to city population growth. These findings are also of importance for designing communication networks and for the understanding of the dynamics of social systems in which communication plays a role, such as economic markets and political systems.

💡 Research Summary

The paper investigates whether human communication activity follows universal statistical laws, focusing on two large online social communities that together provide more than a year of message‑exchange data from hundreds of thousands of users. The authors first quantify each member’s average daily message count ⟨fᵢ⟩ and the corresponding standard deviation σᵢ of daily activity. Plotting σᵢ against ⟨fᵢ⟩ on logarithmic axes reveals a clear power‑law relationship σᵢ ∝ ⟨fᵢ⟩^β with an exponent β≈0.52–0.58 in both communities. This scaling is a generalized version of Gibrat’s law: while larger actors exhibit smaller relative fluctuations, the variance does not become completely size‑independent as the classic law would predict.

To uncover the mechanism behind this scaling, the authors examine the temporal structure of each user’s activity series fᵢ(t) using detrended fluctuation analysis (DFA). The resulting Hurst exponents H are consistently above 0.5 (average ≈0.68–0.71), indicating persistent long‑range correlations that extend from daily to yearly scales. Crucially, the study finds a quantitative link β = 1 – H. In other words, the stronger the long‑term memory (higher H), the smaller the scaling exponent β, and thus the more stable the activity of high‑volume users.

A stochastic growth model formalizes this link. The change in activity Δfᵢ(t) = fᵢ(t+1) – fᵢ(t) is modeled as a deterministic proportional term plus a noise term whose amplitude scales as fᵢ(t)^β and whose temporal correlations are characterized by H. Analytically, this yields ⟨(Δfᵢ)^2⟩ ∝ ⟨fᵢ⟩^{2β}, reproducing the empirically observed σ‑⟨f⟩ scaling and confirming the β = 1 – H relationship.

The authors argue that the same statistical structure underlies many macro‑economic phenomena traditionally described by Gibrat’s law, such as firm size growth, R&D expenditures, GDP evolution, and city population dynamics. In those contexts, measured β values (typically 0.2–0.3) correspond to Hurst exponents around 0.7–0.8, suggesting that persistent, long‑range correlations in underlying activity drive the observed scaling. Hence, human communication dynamics may provide a microscopic foundation for the emergence of Gibrat‑type laws across disparate socioeconomic systems.

Beyond theory, the findings have practical implications. Network engineers can use the σ‑⟨f⟩ scaling to anticipate traffic variability of high‑volume users and provision capacity accordingly, reducing the risk of overloads. Platform designers can exploit the identified long‑range memory to devise interventions (e.g., targeted notifications or incentives) that sustain user engagement, especially during periods when activity tends to decay.

In conclusion, the paper demonstrates that human interaction activity in online communities obeys a robust scaling law rooted in long‑term correlated dynamics. This law not only enriches our understanding of digital communication patterns but also offers a unifying statistical mechanism that may explain the prevalence of Gibrat’s law in economics and other complex social systems. Future work should test the universality of these results across different cultures, communication modalities (likes, comments, shares), and offline interaction datasets.