Autonomous Testing and the Future of Software Development
summary
Claims that testing eats half of engineering time and trillions in lost productivity, cites industry estimate of ~30% and argues real cost is higher, then describes concrete tools: deterministic simulation used to make FoundationDB tests nondeterministic and reliable, plus fuzzing and property-based testing to generate inputs automatically, extensive tuning of random seeds and failure injection to expose bugs, and a working prototype company product aimed at making autonomous testing easy and broadly applicable.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 0:39Moores Law
- 2:56Software is Early
- 7:49Why Testing is Miserable
- 9:14Testing is Fragile
- 11:33Not Testing the Right Stuff
- 14:21Automated Testing
- 16:22Autonomous Testing
- 20:10Its insanely difficult
- 22:02Its super scary
- 25:13You should still be using these techniques
- 29:47Conclusion
related talks
44mHow Testability Supports Your Agility
53mTest architecture is a real thing
55mThe Science of Unit Tests
06mBeyond the Hype Cycle
42mWhat our Brains can Teach us About Software Engineering
01h 07mHappiness U-Curve: Navigating the AI Validation Bottleneck with Conformance Testing and(…)
59mAgile Engineering: The Hard Parts
28mWhy Software Quality Is the Key to Scaling AI Development