Study of 100,000 developers finds AI coding gains shrink to about 30% at release stage
amcafee · x · 2026-07-28
A podcast episode discusses a new empirical paper on how AI coding tools affect software productivity across generations of tools.
The study combines public GitHub records from more than 100,000 developers with confidential Microsoft data to measure the impact of:
- autocomplete tools
- synchronous agents
- asynchronous agents
The key finding is an attenuation effect: huge gains at the code-writing level do not translate one-for-one into shipped software. The paper suggests very large improvements in lines of code or related upstream metrics can shrink to roughly a 30% increase in released software once you move up the production chain.
The hosts then debate what this means for broader labor and economic conclusions, including a related estimate in the abstract about substitution effects.
More from coding & agent
- Linear-attention hybrids may need finer caching for long prompts and workflows — stochasticchasm · 2026-07-28
- A research-agent ranking of 8 stock-data MCP servers puts Equibles first — DanielAPO · 2026-07-28
- Agent quality now depends more on context and harness design than on prompting — verrsane · 2026-07-28
- Building Self-Learning Pipelines and Multi-Turn Evals for Counselling Agents — Technical-Sort-8643 · 2026-07-28
- A new guide says most codebases are not ready for cloud coding agents — vinvan · 2026-07-28
- Meme asks whether to write code now or wait for a model that one-shots it — Darpinian · 2026-07-28