Posts tagged “code-review”

A study of 718 firms found coding agents added 30% more code and no detectable gain in finished work. The review data suggests the extra output piled up in a queue nobody resized.

Linus Torvalds built decades of code review on one proxy: a coherent explanation proved understanding. LLMs made that proxy free, and he just replaced it with a provenance requirement.

A LinkedIn post circulating this week made an observation worth sitting with. AI write-throughput in a serious engineering team can hit fifty pull requests a day. The team's merge-throughput is…