Study: 'AI-DDoS' cuts first-time contributor merge rates 18.18% in OSS

A new academic study coins the term 'AI-DDoS' to describe the pressure generative AI tools are placing on open source communities: plausible-looking but low-quality AI-generated contributions overwhelming reviewer capacity. Researchers first analyzed practitioner accounts from Reddit, OSS mentor mailing lists, and blogs to identify six recurring themes, then tested resulting hypotheses using Bayesian Structural Time Series analysis across 294 repositories and over 2 million pull requests and issues.

The results show PR volume rose in 2025, but merge rates fell overall—one-time contributors in particular saw an 18.18% drop in merge rates relative to a counterfactual baseline, indicating first-time contributions are being rejected far more often than expected.

The study also identified 11 remediation strategies through practitioner interviews, validated via a survey of 229 OSS practitioners, and grouped them into preservative, adaptive, and transformative approaches. The authors conclude AI-DDoS is not just a volume problem but a sustainability trap: communities tend to default to low-effort defensive tactics that protect short-term review capacity while making long-term openness harder to sustain.