Universal Patterns of Evolution
summary
Uses the Fitness Landscapes model and extreme-value theory to show why large-code systems (hundreds of features to millions of parameters) hit hard limits: refactors succeed at GitHub but the Denver Airport luggage project fails after 11 years and $560M, and a 5,000‑line Python 2→3 migration drags on for years. Demonstrates machine learning tradeoffs like catastrophic forgetting during fine‑tuning, cites Goodhart’s law, the curse of dimensionality, and examples such as COBOL’s longevity and a 480‑million‑year stable horseshoe crab.
AI-written summary · about
chapters (14)
- 0:00Bridging biology and code
- 3:05Evolutionary theory patterns
- 5:58Scaling and complexity
- 8:06The curse of dimensions
- 10:45Limits of machine learning
- 12:51Optimization and trade-offs
- 16:29Neutral evolution models
- 19:19Punctuated equilibrium
- 22:39Technology and obsolescence
- 25:46Adapting to rapid change
- 29:08The critical role of context
- 31:47Lessons from evolution
- 35:06Future of software
- 39:00Blue plane innovation
related talks
41mOrganising and Governing Evolutionary Architectures (Fitness Functions)
49mThe Principles of Evolutionary Architecture
39mPrinciples of Evolutionary Architecture
01h 00mEvolutionary Architectures
53mEvolutionary Architecture: Principles and Patterns
50mBuilding Evolutionary Architectures
42mBuilding Evolutionary Architectures: Principles and Practices
35mPrinciples of Evolutionary Architecture