Everything is broken and that's Okay!
summary
Claims that software fails everywhere with concrete failures: a 787 that needs a reboot after 248 days (integer overflow), a NASA rover recovered by deleting files when flash filled, a university SMTP misconfigured to a zero timeout causing the “500‑mile” email problem, and an image classifier that labels people as gorillas. Remedies shown include Rust preventing undefined memory behavior, using containers for isolation, WebAssembly as a cross‑platform sandbox, plus linters, CI tests, type systems, code reviews, diverse teams, and never rolling your own crypto.
AI-written summary · about
chapters (19)
- 0:00Introduction and theme
- 2:56Broken bureaucracy and software
- 7:33Introducing the speaker
- 8:16Human and computer factors in brokenness
- 11:09Bias as the biggest problem
- 14:49Assumptions are always wrong
- 18:34Falsehoods programmers believe
- 21:55Geometry and computer science
- 25:31How to avoid problems
- 27:59Building a culture of sharing
- 29:02Why legacy code is broken
- 32:49Introducing Rust as a solution
- 35:26Isolation with Docker and WebAssembly
- 38:16Immediate improvements to your code
- 40:01Broken things as opportunities
- 44:58Speedrunning broken games
- 46:37Jailbreaking and creativity
- 47:20Embrace failure and hacker culture
- 48:05Closing and Q&A
related talks
54mThe Error of our Ways
36mKeynote: The Future of Software 
30mA Series of Unfortunate Bugs
44mThe role of catastrophic failure in software design
53mTechnical Enshittification: Why Everything is Horrible and How to Fix It
14mBugs from 1843 to the present day
27mSoftware won't make noise when it breaks
50mSmashing the Stack: Let’s Make It Less Fun And Unprofitable!