DEV TOOLS Signal 589 2 feeds carried it
Thoughtworks CTO proposes replacing comprehensive code review with pair programming, automated fitness functions, and review-by-exception
↗
Illustration only Photo by Community Archives of Belleville and Hastings County on Unsplash
As AI-generated code volume outpaces human review capacity, Thoughtworks CTO Rachel Laycock argues that teams should shift feedback left by replacing comprehensive code review with pair programming, team design sessions, and automated fitness functions, reserving human review only for exceptional, high-risk changes.
Pull request sizes and code volumes are increasing rapidly, making traditional code review unsustainable. Shifting to pair programming and automated fitness functions preserves knowledge sharing and architectural alignment without relying on post-implementation review cycles that create merge conflicts and delays.
Written by elseif from the cluster below · every claim links back to a sourceThe three things worth knowing
Significant lines of code per human-landed diff at Meta reportedly increased 106% in a year, while median pull request size increased 64%.
Teams should shift feedback left by using pair programming, team design sessions, and automated fitness functions instead of relying on code review for knowledge transfer and architectural alignment.
Human code review should be reserved for exceptions like fundamental architectural changes, security boundaries, or changes with a huge blast radius.
THE CLUSTER