Nonlinear effects of spatial determinants on the popularity of scientific fields: Evidence from 40 years of computer science

Understanding the dynamics of scientific fields is essential for interpreting scientific development and optimizing resource allocation in research policy. The evolution of scientific disciplines is influenced by complex spatiotemporal dynamics; yet the temporal heterogeneity of field life cycles and the role of spatial determinants remain underexplored. To address this research gap, this study leverages the exponential growth of scientific output to conduct a large-scale analysis of over 6 million Computer Science publications from the OpenAlex database spanning 1981 to 2020. To more comprehensively measure the real-time intensity of scientific activity, this study introduces a novel composite popularity metric that integrates publication volume and team size. Applying Dynamic Time Warping (DTW) distance-based clustering, this study identifies three distinct life cycle patterns of sub-field evolution: continuous increase, continuous decrease, and an initial rise followed by a decrease. Further analysis using a two-way fixed effects regression model, which accounts for sub-field and life cycle patterns heterogeneity, reveals that spatial concentration has an increasing positive marginal effect on popularity, while spatial ubiquity exerts an increasing negative marginal effect, and both effects intensify over time. This study presents a comprehensive spatiotemporal analysis framework, offering both theoretical insights and empirical evidence on the mechanisms of scientific development. Additionally, it provides practical guidance for monitoring field trends, informing scientific policy, and optimizing resource allocation.