What our Brains can Teach us About Software Engineering
summary
Argues software fails because code is an infinitely precise generative model while humans work with noisy, tolerant perception; demonstrates a one-character bug (assignment in an if) that would make a robot kill people. Recommends property testing (generate random inputs to surround bugs), Remedy-first design (group errors by retry/redirect/re-authenticate), and chaos experiments like Netflix’s Simian Army for resilience. Shows Netflix’s Vizceral-style viseral visualization and proposes curiosity-driven test agents and self-healing actions; tools exist and these practices reduce surprises.
AI-written summary · about
chapters (11)
- 0:00Software unpredictability
- 2:51The analogy of life
- 6:18Code as a generative model
- 11:24Property based testing
- 14:46Testing as training data
- 18:37Resilient system design
- 22:58Deliberate failure testing
- 26:35Visualizing system telemetry
- 30:27Intuitive data insights
- 33:58Future of AI in engineering
- 37:43Conclusion and Q&A
related talks
44mBehavioral Software Engineering
41mComputer Science, Software Engineering, and the Scientific Method
34mHow do our ideas about coding affect the software we create?
38mIntro to Empirical Software Engineering: What We Know We Don't Know
38mWhy software developers should build a resilience culture based on chaos engineering
47mLearning from Nature or what biological systems can teach software engineers
35mCraft, Engineering, and the Essence of Programming
41mHow We Tell Stories In Code