Evolution Patterns of Software-Architecture Smells

We report about recent research on evolutionary analysis of architecture smells, originally published in the Journal of Systems and Software (JSS) [GSK24]. Long-living software systems tend to degrade over time due to continuous modifications, leading to a decrease in maintainability, modularity, and code quality. Architecture smells — such as cyclic dependencies, hub-like dependencies, and unstable dependencies — are symptoms of this degradation and contribute to the accumulation of architectural technical debt. While previous research has focused primarily on detecting these smells, limited work has examined how they evolve throughout a system’s lifecycle. This paper contributes a longitudinal analysis of architecture-smell evolution using 485 releases from 14 open-source software systems, introducing new visualization and tracking techniques to understand growth patterns, merging and splitting dynamics, and the linkage between architecture smells and technical debt.