Y2K and Other Disappointing Disasters: Risk Reduction and Harm..
summary
Examines concrete outcomes not slogans: Y2K remediation consumes millions of dollars and thousands of person‑hours and thousands of programmers patched legacy COBOL and firmware so most failures happen in staged tests, not on New Year’s Day. Gives hard numbers: Mexico City 1985 kills ~10,000, later quake kills ~228 after stricter building codes; I‑35W bridge collapse kills 13 due to undersized gusset plates plus added resurfacing and heavy construction loads. Recommends kill switches, canary launches, duplicate data centers, and feature flags like LaunchDarkly.
AI-written summary · about
chapters (18)
- 0:00Start
- 6:11No one remembers the crisis averted.
- 6:42Risk Reduction
- 7:32How to reduce risk
- 11:36Harm Mitigation
- 18:16How to mitigate harm
- 18:46Assume failure
- 20:29Fail safe or fail secure?
- 22:50Failure is inevitable.
- 23:20It's complicated and hard
- 26:29What's a disaster?
- 30:41Microservices and loose coupling
- 31:15Fault-tolerant messaging
- 32:04Data duplication
- 33:08Canary launches
- 34:18Kill switches and circuit breakers
- 35:07Automatic recovery
- 36:03Testing for load, stress and outage
