The Invisible Art of Refactoring: Why System-Level Practice Escaped Its Own Definition
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
Fowler (1999) defined refactoring as changing the internal structure of software withoutchanging its observable behavior. The definition contains no scope restriction: it is logicallyscale-free, applying with equal validity to extracting a local variable and to replacing anentire service implementation behind a stable facade. Yet after thirty years, the authoritativerefactoring catalog remains locked at the level of code. Strangler Fig — a pattern thatFowler himself named, that satisfies his own definition precisely, and that addressesarguably the most consequential structural improvement challenge organizations face —does not appear in his catalog. This paper analyzes why. We demonstrate that the literatureon refactoring has fragmented along authorial and commercial lines: Opdyke (1992)provided formal foundations, Feathers (2004) developed techniques structurally identical toStrangler Fig without naming them, Birchall (2016) organized the full spectrum withouttheorizing it, and Tornhill (2014, 2018) contributed the only quantitative prioritizationframework — each independently, none synthesizing across contributions. We introducethe concept of refactoring scaffolding: the guarantee mechanisms (type systems, tests,contract verification, canary deployment, dark launch) that enable safe structural change ateach granularity level. We argue that TDD tests and static types are both scaffolding ratherthan quality assurance, and that the second edition's switch from Java to JavaScriptcollapsed scaffolding into a single form, obscuring the concept needed to extend refactoringacross granularity levels. We trace the conceptual ceiling to the incentive structure oftechnical knowledge production, which systematically selects for introductory content overexpert synthesis. Drawing on the capability invisibility framework (Sophia, 2026), we showthat the ceiling is not merely present but actively invisible: practitioners whose framework isthe code-level catalog cannot perceive system-level refactoring as refactoring. We proposea granularity-times-scaffolding matrix as a minimal unifying framework that makes thefragmented literature legible as a single, partially filled research program.
Fields of Study & Subjects
Software Engineering Research
Software Engineering Techniques and Practices
Advanced Software Engineering Methodologies
Publication details
- Journal / source
- Zenodo (CERN European Organization for Nuclear Research)
- Publication date
- 1 April 2026
- Language
- EN
- Document type
- Preprint