Vitaly Sharovatov: From Astrology to Code Reviews
summary
Argues that blanket adoption of code review is magical thinking and advocates the scientific method to choose practices, with a concrete three-week experiment: two pairs tried pair programming (mob SSH/Zoom for remote), showing perceived knowledge sharing roughly twice as good, pairs catch about 8 of 10 defects, feature cycle drops from ~4 to ~2.5 days, but cost roughly doubles and sessions feel tiring (sustained about 6 hours vs 10 solo); recommends measuring criteria and iterating.
AI-written summary · about
chapters (16)
- 0:00Magical thinking in software
- 1:34Handwashing revolution
- 5:12Pair programming vs code reviews
- 8:20Scientific method for work
- 10:49Code review problems
- 12:38Brainstorming solutions
- 15:10Evaluating solutions
- 19:47Semmelweis experiment
- 20:31Pair programming results
- 22:39Data-driven decisions
- 24:28Why guesses fail
- 25:33Scientific estimations
- 27:19Estimation alternatives
- 29:29Scientific method for decisions
- 31:17Magical thinking bias
- 33:11Q&A
related talks
45mHow to Get More Out of Your Code Reviews.
01h 01mBetter Code Reviews FTW!
45mCracking the Code Review
44mCracking the Code Review
41mCracking the Code Review from Guesswork to Automated Guidance
01h 03mHow Sigmund Freud would perform a code review?
18mHow To Review Someone's Code While Not Being A Jerk!
48mPair Programming, TDD and Other Impractical Things