A Replication Study on Measuring the Growth of Open Source

TL;DR

This study replicates previous measurements of open source growth over 30 years using Open Hub data, revealing a peak in 2013 followed by a decline, challenging existing growth models and raising questions about open source evolution.

Contribution

It provides a long-term replication of open source growth metrics, highlighting the decline post-2013 and questioning the applicability of prior growth models.

Findings

Open source growth peaked in 2013 and has been declining since.

Existing growth models do not fit the long-term data.

Open Hub may not fully represent the open source ecosystem.

Abstract

Context: Over the last decades, open-source software has pervaded the software industry and has become one of the key pillars in software engineering. The incomparable growth of open source reflected that pervasion: Prior work described open source as a whole to be growing linearly, polynomially, or even exponentially. Objective: In this study, we explore the long-term growth of open source and corroborating previous findings by replicating previous studies on measuring the growth of open source projects. Method: We replicate four existing measurements on the growth of open source on a sample of 172,833 open-source projects using Open Hub as the measurement system: We analyzed lines of code, commits, new projects, and the number of open-source contributors over the last 30 years in the known open-source universe. Results: We found growth of open source to be exhausted: After an initial exponential growth, all measurements show a monotonic downwards trend since its peak in 2013. None of the existing growth models could stand the test of time. Conclusion: Our results raise more questions on the growth of open source and the representativeness of Open Hub as a proxy for describing open source. We discuss multiple interpretations for our observations and encourage further research using alternative data sets.