Decoding paradoxes: why are many good ideas in software delivery counter intuitive
summary
Argues that more frequent deployment improves quality because it shortens feedback loops and reduces batch size, demonstrated by a causal model where larger releases raise change failure rate, increase time to restore service, and grow the next release into a reinforcing loop that worsens outcomes. Uses causal loop diagrams and the iceberg model to trace events to mental models, tests interventions like WIP limits that cap new changes and thereby lower future failure rates, and ties findings to measurable behaviors rather than vague best practices.
AI-written summary · about
chapters (19)
- 0:00Counterintuitive software delivery ideas
- 1:05Why delivery feels complex
- 4:24Software delivery as wicked problem
- 7:38Heuristics and their flaws
- 9:02Paradox of frequent deployments
- 11:13Feedback loops with causal diagrams
- 13:50Simple causal model for releases
- 18:31Reinforcing loops in delivery
- 21:24Why interventions fail
- 25:41Iceberg model for leverage
- 28:13System structure shapes delivery
- 30:41High-leverage interventions
- 32:29Collaborative modeling
- 34:40Nonlinear strategies
- 36:03Understand system as-is
- 37:29Iterate and improve models
- 38:57Living with complex systems
- 40:23Recommended resources
- 41:45Q&A
related talks
41mA Defence of Technical Excellence
52mUnderstanding UnHealthy Syndromes of Software Delivery
51mA Defence of Technical Excellence
29mFrom big to small and back again
01h 04mPredictably Unpredictable
40mContinuous Delivery Anti-Patterns
45mThe Goldilocks Problem of Software Delivery
47mOvercoming Obstacles to DevOps