Harvard researchers Fiona Chen and James Stratton used aggregated Jellyfish analytics covering 300 million engineering “work events” across more than 700,000 employees at over 700 software firms (2021-Mar 2026) to measure how AI coding assistants and AI coding agents changed engineering output. Using difference-in-differences regressions and GitHub activity to time tool introductions, they show agentic coding produces far more raw code - about 30% more lines, 20% more commits, and 23% more pull requests - yet yields no statistically significant increase in resolved Jira-tracked Issues or Epics and no compositional shift in their size or complexity.
The reason is a downstream bottleneck in human review: average time from pull request submission to merge rises 49%, the share of PRs requiring changes nearly doubles, and comments per PR increase about 35%. Firms respond by shifting 14% more workers into code review, but overall employment doesn’t change significantly. AI-assisted review remains limited - by March 2026 agents generated roughly 23.3% of review comments and touched 10.8% of PRs - so efficiency gains in code generation are largely absorbed by increased human review effort, leaving net software output unchanged for most firms.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
Lukáš Lalinský developed zt, a type-safe HTML templating library for Zig, addressing the need for safe, fast, and HTML-like syntax in low-level web development. Inspired by JSX and Templ for Go, zt compiles templates into Zig code, enabling safe and efficient web output without runtime errors or allocations. (lalinsky.com)
Linus Torvalds is developing a significant update to the Linux kernel using Claude Code. The project aims to enhance kernel performance and functionality. (phoronix.com)
The Reddit post discusses why the bun JavaScript runtime faces dislike among developers. It highlights concerns or issues that lead to negative perceptions of bun. (reddit.com)
Preloop is an agent-native tool that allows GitHub Actions to run locally or on self-hosted microvms with features like debug, pause-on-failure, and step-level retries. It aims to improve local development and troubleshooting for CI workflows. (github.com)
TEX SIDE PREVIEW provides an Overleaf-style side-by-side LaTeX editing experience within VS Code, featuring live recompile, error highlighting, and clickable links. It supports two-way sync with Overleaf and works with local LaTeX installations on various operating systems. (marketplace.visualstudio.com)
Externalized proofs of cyclic trait implementations are incompatible with Rust's design, making modular proofs the only viable approach. Modular proofs require each implementation to explicitly demonstrate that supertrait obligations are satisfied, ensuring program correctness. (smallcultfollowing.com)
Today's best Hacker News stories, summarized and screenshotted, one email a day.