# A Longitudinal Analysis of Bloated Java Dependencies

arXiv: https://arxiv.org/abs/2105.14226 · HTML: https://mlsanity.com/paper/2105.14226/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 29 May 2021.

## Built on by (most cited first, 2 of 2)

- [Speed at the Cost of Quality: How Cursor AI Increases Short-Term Velocity and Long-Term Complexity in Open-Source Projects](https://mlsanity.com/paper/2511.04427/) (2025)
- [Machine Learning Systems are Bloated and Vulnerable](https://mlsanity.com/paper/2212.09437/) (2022)

## Abstract (arXiv)

We study the evolution and impact of bloated dependencies in a single software ecosystem: Java/Maven. Bloated dependencies are third-party libraries that are packaged in the application binary but are not needed to run the application. We analyze the history of 435 Java projects. This historical data includes 48,469 distinct dependencies, which we study across a total of 31,515 versions of Maven dependency trees. Bloated dependencies steadily increase over time, and 89.02% of the direct dependencies that are bloated remain bloated in all subsequent versions of the studied projects. This empirical evidence suggests that developers can safely remove a bloated dependency. We further report novel insights regarding the unnecessary maintenance efforts induced by bloat. We find that 22% of dependency updates performed by developers are made on bloated dependencies and that Dependabot suggests a similar ratio of updates on bloated dependencies.