Rapid advances in large language models (LLMs) have empowered autonomous agents to generate social networks, communicate, and form shared and diverging opinions on political issues. However, our understanding of their collective behaviours and underlying mechanisms remains incomplete. In this paper, we simulate networked systems involving thousands of LLM agents across different backbone models (GPT-3.5, GPT-4o, ChatGLM, Llama-3, and DeepSeek-V3), in which agents interact through LLM-guided conversations and update their opinions over time, resulting in the emergence of opinion polarization. We discover that these agents spontaneously develop their own social network with properties characteristic of human social networks, such as homophilic clustering. The collective opinions of these LLM agents evolve in ways that exhibit behavioural patterns consistent with social phenomena and mechanisms widely discussed in empirical studies of human behaviour and opinion-dynamics models. This consistency suggests that LLM agents can serve as a valuable synthetic testbed for exploring hypothetical intervention strategies in networked LLM-agent systems. Using this testbed, we further examine the effects of a range of network- and individual-level interventions, such as promoting more diverse interactions and reducing confirmation bias. Overall, this work not only sheds light on subtle opinion dynamics and collective behaviour of LLM agents from a network perspective, but also demonstrates their potential for testing hypotheses about opinion dynamics and intervention strategies.

Jinghua Piao, Zhihong Lu, Chen Gao, Fengli Xu, Qinghua Hu, Fernando P. Santos, Yong Li & James Evans 

Nature Communications (2026)

Read the full article at: www.nature.com

Antonio Cabrales, Gema Pomares, David Ramos Muñoz & Ángel Sánchez

Journal of Economic Interaction and Coordination

We experimentally study a reduced-form network-formation game inspired by Cabrales et al (2024). A set of network agents chooses (i) a privately profitable number of outside projects and (ii) bilateral links with other network agents. Links generate concave private benefits, whereas a partner’s outside projects impose a cost on each of its network neighbors. Because an agent does not internalize the costs that its own projects impose on its partners, the equilibrium benchmark features more projects and fewer network links than the social optimum. In a repeated laboratory experiment with fixed groups, treatment comparisons are broadly consistent with these comparative-static predictions. Exact choices, however, are systematically pulled toward the interior of the feasible action set, especially when the theoretical benchmark lies at a boundary. We interpret the evidence as support for the model’s core cross-layer externality mechanism, rather than as a direct test of a particular banking or supply-chain institution, and discuss learning, stochastic choice, and coordination as possible sources of the quantitative deviations.

Read the full article at: link.springer.com

Ollin D. Langle-Chimal, Steffen Knoblauch & Marta C. González 
Nature Communications (2026)

Human mobility in urban areas largely depends on the spatial distribution of jobs and who pursues them, yet how workers balance proximity against economic opportunity remains poorly understood. Existing mechanistic models capture physically interpretable effects such as distance decay, but overlook how socioeconomic factors shape commuting decisions. We present WorkReach, a model grounded in discrete choice theory that embeds informality and economic complexity into a utility function and represents commuting as the outcome of workers behaving as if maximizing it. Applied to four cities in the United States, Mexico, and Brazil, it reproduces commuting flows as accurately as widely used benchmarks while making the underlying choices interpretable. Here, we show that workers consistently commute farther to economically sophisticated areas, while the role of informality differs across regions, and that measuring accessibility by the perceived benefit of opportunities, not just physical proximity, highlights disparities conventional metrics overlook.

Read the full article at: www.nature.com

Fabiane de Fatima Carvalho, Ivan Bergier, Silvia Maria Fonseca Silveira Massruhá, Jayme Garcia Arnal Barbedo
Complexities 2026, 2(3), 21

Collaboration networks are frequently studied as empirical instances of complex social systems, yet standardized methodological frameworks for consistently identifying heterogeneous mesoscopic structural regimes remain limited. This study proposes an integrated multi-criteria classification framework and demonstrates its application to the structural characterization of communities extracted from a large-scale scientific collaboration network. The framework combines community detection, classical network metrics, statistical modeling of weighted degree tails, small-world diagnostics, information-entropy measures, and fractal analysis based on the Song–Havlin–Makse box-covering renormalization framework. As an empirical application, the methodology is applied to the giant coauthorship component of Embrapa’s scientific production (1974–2024), derived from the Brazilian Agricultural Research Database (BDPA), comprising 60,636 nodes. The weighted Louvain algorithm partitions the network into 25 major communities, which are evaluated through an integrated classification protocol combining the Akaike Information Criterion model selection, Kolmogorov–Smirnov goodness-of-fit tests, small-worldness diagnostics, and fractal scaling analysis. The proposed framework identifies three network families, namely Barabási–Albert (BA-like)/scale-free small-world, scale-free fractal (non-BA) and small-world (non-scale-free), while explicitly distinguishing supported and ambiguous classifications according to the overall consistency of the statistical and structural evidence. The results demonstrate that distinct mesoscopic structural regimes coexist within the same connected collaboration system, highlighting the usefulness of the proposed reproducible multi-criteria framework for comparative topological characterization across complex collaboration networks.Collaboration networks are frequently studied as empirical instances of complex social systems, yet standardized methodological frameworks for consistently identifying heterogeneous mesoscopic structural regimes remain limited. This study proposes an integrated multi-criteria classification framework and demonstrates its application to the structural characterization of communities extracted from a large-scale scientific collaboration network. The framework combines community detection, classical network metrics, statistical modeling of weighted degree tails, small-world diagnostics, information-entropy measures, and fractal analysis based on the Song–Havlin–Makse box-covering renormalization framework. As an empirical application, the methodology is applied to the giant coauthorship component of Embrapa’s scientific production (1974–2024), derived from the Brazilian Agricultural Research Database (BDPA), comprising 60,636 nodes. The weighted Louvain algorithm partitions the network into 25 major communities, which are evaluated through an integrated classification protocol combining the Akaike Information Criterion model selection, Kolmogorov–Smirnov goodness-of-fit tests, small-worldness diagnostics, and fractal scaling analysis. The proposed framework identifies three network families, namely Barabási–Albert (BA-like)/scale-free small-world, scale-free fractal (non-BA) and small-world (non-scale-free), while explicitly distinguishing supported and ambiguous classifications according to the overall consistency of the statistical and structural evidence. The results demonstrate that distinct mesoscopic structural regimes coexist within the same connected collaboration system, highlighting the usefulness of the proposed reproducible multi-criteria framework for comparative topological characterization across complex collaboration networks.

Read the full article at: www.mdpi.com