Code Reviews are Mostly Useless
summary
Claim: code reviews are mostly useless because cultural failures turn them into five-line review nitpicks while a 10,000-line no-review sails through. Evidence: Python examples (PEP8 edge cases, numpy truth-value errors), a subtle bug from non-position-only forwarded args, and better fixes like using generators and itertools.islice/takewhile or itertools.pairwise. Concrete pattern: many disputes trace to inband vs out-of-band encoding mistakes (find/xargs null-byte solution) and disappear once foundational theory guides the review.
AI-written summary · about
chapters (10)
- 0:00Code reviews introduction
- 3:15Why reviews feel like a burden
- 6:44Cultural factors in reviews
- 14:24Culture of training in reviews
- 15:47Learning in effective reviews
- 21:00Culture of learning in reviews
- 23:06Knowledge culture in reviews
- 37:35Avoiding shortcuts in reviews
- 40:26Code quality importance
- 44:39Conclusion on reviews
related talks
18mHow To Review Someone's Code While Not Being A Jerk!
40mCode Review Skills for Pythonistas
21mMaking the Most Out of Code Reviews
22mThe truth about code reviews
18mCode Review: That Thing That Serves No Purpose
48mEffective Code Reviews
45mHow to Get More Out of Your Code Reviews.
45mCracking the Code Review