Does This PR Lower the Energy State?
summary
Claims that software follows an Energy State Principle: systems drift to higher-entropy configurations and good design lowers the system’s accessible state space. Demonstrates concrete rules—use types to make illegal states unrepresentable (three valid plans vs eight boolean combinations), prefer parsing not validating to preserve learned invariants, and prefer duplication over the wrong abstraction. Builds a static-analysis plugin that computes cyclomatic (example: 4) and cognitive (example: 5) complexity, emits SARIF, and works in demos; an LLM (Claude) recognizes a 6-line pattern in ~2s.
AI-written summary · about
chapters (13)
- 0:00The quality of code
- 2:52Entropy in software
- 8:08Information and decoders
- 13:16Complexity in relationships
- 16:09Measuring complexity
- 21:48Decoders and abstractions
- 26:30Managing abstractions
- 31:49State space and types
- 37:42Entropy dumps and structure
- 41:38Boundaries and stability
- 45:38Timing and principles
- 48:35Tooling and analysis
- 51:16Conclusion and takeaways
