What Could Go Wrong? The Subtle Science Of Coding for Failure
summary
Claims a simple model that quantifies danger by likeliness vs badness and uses micromort comparisons to show surprising risk tradeoffs. Demonstrates concrete fixes: guarded URL-fetching handling nil, redirects, and non-200 responses; replaces brittle automated tests for a fragile Salesforce stream with an interactive script; builds a constant-time auth test with statistical sampling to suppress timing leaks and documents when complexity outweighs benefit.
AI-written summary · about
chapters (9)
- 0:00Coding for failure intro
- 1:09Danger: likelihood and harm
- 3:44Rigorously disagreeing risks
- 6:12Social bias in risk perception
- 10:51Code-level risk questions
- 15:12Over-engineering pitfall
- 16:44Testing beyond conventional tests
- 18:06Security vs complexity
- 20:35Summary and takeaways
related talks
58mThreat Modeling Developer Behaviour: The Psychology of Bad Code
53mOptimistic Security
50mCracking the Code to Secure Software
42mCode as Risk
40mOptimistic Security
13mStandard Classification: Back to Basics: Looking for subtle bugs in beginne...
01h 24mCode Red: Navigating the Minefield of Security Programming Blunders
31mDon't Be Afraid of the Dark