Synergy, not size: How collaboration architecture shapes scientific disruption
Abstract
The mechanisms driving different types of scientific innovation through collaboration remain poorly understood. Here we develop a comprehensive framework analyzing over 14 million papers across 19 disciplines from 1960 to 2020 to unpack how collaborative synergy shapes research disruption. We introduce the synergy factor to quantify collaboration cost-benefit dynamics, revealing discipline-specific architectures where Physics peaks at medium team sizes while humanities achieve maximal synergy through individual scholarship. Our mediation analysis demonstrates that collaborative synergy, not team size alone, mediates 75% of the relationship between team composition and disruption. Key authors play a catalytic role, with papers featuring exceptional researchers showing 561% higher disruption indices. Surprisingly, high-citation authors reduce disruptive potential while those with breakthrough track records enhance it, challenging traditional evaluation metrics. We identify four distinct knowledge production modes: elite-driven, baseline, heterogeneity-driven, and low-cost. These findings reveal substantial heterogeneity in optimal collaboration strategies across disciplines and provide evidence-based guidance for research organization, with implications for science policy and the design of research institutions in an increasingly collaborative scientific landscape.