Migrating an Ecosystem: An Empirical Study of How React Native Libraries Adopted the New Architecture

Abstract Context: When a platform replaces its core architecture, every library built on it must migrate. React Native replaced its asynchronous bridge with a New Architecture in three steps: opt-in in version 0.68 (2022), default in 0.76 (2024), and mandatory in 0.82 (2025). Objective: We study how its library ecosystem responded: how far and how fast it migrated, by which routes, which libraries were left behind, and how much usage depends on them. Method: We analyze all 2,782 React Native Directory libraries matched to npm, combining full release histories, static scans of published packages, popularity data, and 70 monthly snapshots of declared compatibility. A blind second classification of 140 libraries agreed with ours (Cohen's κ=0.89). Results: Of 1,444 native libraries, 62.0% have migrated and carry 95% of downloads; 35.9% still rely on interop layers. The monthly share of remaining libraries that migrated rose from 0.43% (optional) to 0.78% (default) and 1.55% (mandatory), although segmented regression could not attribute the increases to the milestones. Only an estimated 23% of libraries unmigrated at the default change migrated within two years. Unmigrated libraries are mostly old and dormant, fewer than one in ten documents its status, yet 115 still receive at least 10,000 weekly downloads. Of migrated libraries created in 2026, 62% use Expo Modules or Nitro rather than the official Codegen route. Conclusions: Milestones were followed by faster migration but did not reach the dormant tail; compatibility layers became long-term infrastructure, and the migration target fragmented.