Tradeoffs in the Software Workflow
summary
Argues that shipping small changes matters quantitatively: DORA shows 46x more frequent deployments, 440x faster commit→deploy, 170x faster MTTR, and 5x lower change failure rate. Gives concrete examples: OSS-Fuzz finds ~10k bugs across ~500 projects, and a single routing mis-push could cost millions. Proposes minimizing aggregate defect costs by reducing detection latency with automation (CI, fuzzing, auto-rollback) and models defect cost as cost ≈ humans × time, plus evidence that better documentation causally improves outcomes.
AI-written summary · about
chapters (15)
- 0:00Start
- 0:14How To Build Reliable Software at a Reasonable Cost
- 6:02When Does Software Have Value
- 8:33Work in Progress
- 9:08Reduce Work in Progress
- 10:33The State of Devops Report
- 21:39The Cost of a Defect
- 28:49Pre-Commit
- 33:06Would You Rather Catch a Bug in Code Review or in Production
- 48:42Technical Debt and Latent Bugs
- 56:52Auto Rollback
- 1:00:48Do We Need Integration Tests At All
- 1:04:18What's the Value in Writing a Book
- 1:05:12Value of Having Good Documentation
- 1:10:27Software Engineering Is Not the Same Thing as Programming
related talks
29mEconomies of Speed
39mDev Team Metrics that Matter
45mAccelerating Software Delivery: Why Your Team Is Slowing Down
32mMaking Sotware Better, Despite Having Users
45mThe Goldilocks Problem of Software Delivery
23mThe Cost of Speed: Rethinking "Move Fast" in DevOps
27mTradeoffs on the road to Observability
10mBeyond “try harder”: Effective strategies to tackle bugs