Fragmented Infrastructures, Fragmented Learners: The Case for Integrated, Longitudinal Learning Ecosystems in Computing Education

Undergraduate computing education is uniquely shaped by heterogeneous technical infrastructures: students move across programming languages, operating systems, IDEs, auto-graders, version-control platforms, and increasingly large language model (LLM)–mediated tools. These systems generate rich but fragmented learning traces that remain isolated within course-bound platforms, limiting observability of how debugging practices, abstraction strategies, and conceptual understanding evolve across the curriculum. Although intelligent tutoring systems (ITS) and AI-enabled assessment tools show promise within individual contexts, their impact is typically localized, lacking semantic interoperability and continuity across languages and courses. Drawing on research in self-regulated learning, conceptual transfer, and identity formation in computing, this article argues that infrastructural fragmentation constrains longitudinal development of computational expertise. We advance a computing-specific vision of integrated learning ecosystems in which execution histories, feedback artifacts, repository data, and learner models are preserved and interoperable across contexts, enabling trajectory-level inquiry into expertise development over time. We conclude by outlining discipline-centered research challenges for building coherent computing education infrastructures.