Growing-network models that incorporate triadic closure and preferential attachment serve as controlled mechanism endpoints rather than universal empirical laws, according to an analysis of synthetic and empirical systems published in Applied Network Science. Researchers Sergei Sidorov, Timofei Emelianov, and Sergei Mironov evaluated how distinct neighbor-selection rules govern network evolution, finding that exact degree-proportional attachment consistently performs worse than uniform selection when tested against empirical network data.
Triadic closure represents a fundamental mechanism driving the high clustering observed across social, technological, and biological networks. While the established Holme–Kim model couples triadic closure with preferential attachment, it assumes a uniform selection rule when choosing neighbors during triad formation. To assess this assumption, the authors compared four distinct growing-network models that independently vary the mechanisms governing initial attachment and subsequent triad formation between uniform and preferential selection rules.
Using mean-field and master-equation calculations alongside finite-size numerical simulations, the investigators derived conditional predictions for degree growth and degree distributions. The selection rules strongly altered degree dynamics, clustering, hub formation, and network robustness across the four frameworks. In simulated networks, Model 1 exhibited degree-growth slopes close to one-half, whereas Model 2 displayed decreasing slopes across finite observation windows. Furthermore, incorporating preferential choice into triad formation generated significantly stronger hubs in Models 3 and 4.
The study also identified critical limitations in analytical approximations of network behavior. Closure diagnostics revealed that several fitted parameters remained unstable or failed to reach their proposed saturation regimes during finite-size implementations. In addition, duplicate target attempts remained frequent for high-degree hubs. Because of these finite-size effects and diagnostic instabilities, the authors cautioned that the corresponding asymptotic tail exponents represent theoretical predictions of the mathematical approximations rather than established properties of the implemented finite networks.
To test these theoretical models against real-world systems, the team conducted temporal analyses on four empirical networks: DBLP, CollegeMsg, MathOverflow, and Email-EU. The empirical data revealed positive but sublinear degree bias in DBLP, CollegeMsg, and MathOverflow, while the Email-EU network behaved close to uniform choice. Crucially, exact degree-proportional selection underperformed uniform choice on held-out data across all four empirical datasets, demonstrating that idealized selection rules function as controlled baseline mechanism endpoints rather than universal empirical laws.
Subject of Research: Selection mechanisms and degree dynamics in triadic closure network models
Article Title: Selection mechanisms in triadic closure networks: systematic analysis of uniform and preferential rules
Article References: Sidorov, S., Emelianov, T., & Mironov, S. (2026). Selection mechanisms in triadic closure networks: systematic analysis of uniform and preferential rules. Applied Network Science. https://doi.org/10.1007/s41109-026-00847-7
Image Credits: AI Generated
DOI: 10.1007/s41109-026-00847-7
Keywords: triadic closure, preferential attachment, Holme-Kim model, network clustering, growing networks, degree dynamics, hub formation, network robustness, mean-field calculations, master-equation calculations, degree distribution, sublinear degree bias, DBLP network, CollegeMsg network, MathOverflow network, Email-EU network, finite-size networks, uniform selection
News Source: Denise Maddox. (November 21, 2026). Degree Proportional Selection Underperforms Uniform Choice Across Empirical Network Datasets. Scienmag.



