Quantifying Competitive Relationships Among Open-Source Software Projects

Throughout the history of software, evolution has occurred in cycles of rise and fall driven by competition, and open-source software (OSS) is no exception. This cycle is accelerating, particularly in rapidly evolving domains such as web development and deep learning. However, the impact of competitive relationships among OSS projects on their survival remains unclear, and there are risks of losing a competitive edge to rivals. To address this, this study proposes a new automated method called ``Mutual Impact Analysis of OSS (MIAO)’’ to quantify these competitive relationships. The proposed method employs a structural vector autoregressive model and impulse response functions, normally used in macroeconomic analysis, to analyze the interactions among OSS projects. In an empirical analysis involving mining and analyzing 187 OSS project groups, MIAO identified projects that were forced to cease development owing to competitive influences with up to 81% accuracy, and the resulting features supported predictive experiments that anticipate cessation one year ahead with up to 77% accuracy. This suggests that MIAO could be a valuable tool for OSS project maintainers to understand the dynamics of OSS ecosystems and predict the rise and fall of OSS projects.

💡 Research Summary

The paper addresses a gap in the literature on open‑source software (OSS) sustainability: while many studies have examined internal factors such as code quality, developer activity, and governance, the external pressure exerted by competing OSS projects has been difficult to quantify. To fill this gap, the authors introduce a novel automated framework called Mutual Impact Analysis of OSS (MIAO). MIAO adapts techniques traditionally used in macro‑economics—specifically structural vector autoregression (SVAR) and impulse response functions (IRF)—to model the dynamic, bidirectional influence among multiple OSS projects over time.

Data collection begins by extracting activity time series for each OSS project, such as monthly commit counts, download statistics, or other measurable signals that reflect project vitality. Because raw software activity data often exhibit non‑stationarity, the pipeline first applies Augmented Dickey‑Fuller (ADF) unit‑root tests. When a series fails the stationarity test, fractional differencing is employed to preserve long‑memory characteristics while achieving stationarity, a step that is crucial for reliable VAR modeling.

Next, the authors estimate a SVAR model for the set of n projects. The SVAR differs from a standard VAR by imposing a structural matrix B0 that encodes contemporaneous causal constraints among the variables. Lag order selection is performed automatically using information criteria (AIC, BIC, HQIC) to balance model complexity and fit. Once the SVAR coefficients are obtained, the model is transformed into an infinite‑order vector moving‑average (VMA) representation, from which the IRF is derived. The IRFij(k) quantifies the effect of a one‑unit shock to project j at time t on project i after k periods, thereby providing a concrete measure of competitive impact and its temporal decay.

Recognizing that OSS projects evolve through distinct life‑cycle stages (birth, childhood, adolescence, adulthood, obsolescence), the authors split the overall observation window into multiple intervals (e.g., yearly slices). For each interval, a separate IRF is computed, and the resulting impulse‑response tensors are aggregated (e.g., averaged or weighted) to produce a final MIAO score for each ordered pair of projects. This design captures structural changes in competitive dynamics that may occur as projects mature or decline.

The empirical evaluation uses a curated dataset of 187 OSS project groups mined from GitHub. Projects are manually labeled as “Rising Event (REV)” if their development ceased primarily due to competition, or “non‑REV” otherwise. For each group, the MIAO pipeline generates a set of features (e.g., average magnitude of forward‑impact, asymmetry of influence, decay rate). These features feed into standard classifiers (logistic regression, random forest, support vector machine). Using 10‑fold cross‑validation, the authors achieve 81 % accuracy in retrospectively identifying REV cases, and 77 % accuracy when predicting cessation one year ahead.

A decision‑tree analysis further reveals that “unidirectional influence patterns” – where one project exerts a strong, persistent impact on another while receiving little reciprocal effect – are the most salient predictor of project death. This aligns with the intuitive narrative that a dominant competitor can render a rival obsolete, as illustrated by the real‑world example of Chainer versus PyTorch.

The contributions of the paper are threefold: (1) it introduces a cross‑disciplinary methodological bridge, applying SVAR‑IRF to software ecosystems; (2) it demonstrates that competitive pressure can be quantified with high predictive power on real OSS data; (3) it uncovers specific structural signatures (asymmetric influence) that signal impending project failure, offering actionable intelligence for maintainers, contributors, and organizations that depend on OSS.

Limitations are acknowledged. The approach relies heavily on the quality and completeness of activity logs; missing commits or mis‑identified forks can bias the IRF estimates. The structural constraints in B0 are based on researcher assumptions, which may not capture all latent causal pathways. Moreover, only quantitative activity metrics are used; qualitative factors such as licensing changes, community sentiment, or external technology trends are omitted.

Future work is suggested in several directions: (a) extending the model to Bayesian SVAR or non‑linear VAR to capture uncertainty and potential regime shifts; (b) integrating additional data sources (issue tracker sentiment, social media mentions, security advisories) to enrich the competitive signal; (c) applying the framework to other OSS domains (e.g., cloud orchestration tools, front‑end frameworks) to test generalizability; and (d) exploring intervention strategies, such as early warning dashboards for project maintainers, based on real‑time MIAO scores.

In summary, the paper provides a rigorous, data‑driven method for measuring and predicting the impact of inter‑project competition in open‑source ecosystems, advancing both the theory of software evolution and practical risk‑management tools for the OSS community.