Computer, Program Theyself!
summary
Claim: use Ruby plus testing, stats and DSLs to get programs that surprise instead of just automate. Implementation: a tiny language called shift that mutates number grids and runs hundreds to thousands of variants as a genetic Sudoku solver prototype. Method: write scenarios, iterate small experiments, inspect surprising behaviors (not blackbox millions) and then refine simulations. Result: the DSL kind of works now, yields unexpected hunches, and remains under active evolution.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 0:57Learning Funnel
- 7:18Risk Management
- 9:13Risk Mitigation
- 13:40Machine Learning
- 14:35Eliza
- 15:40Genetic Programming
- 18:10Whats the Difference
- 20:28Genetic vs Generator Programming
- 21:33The Knowledge File
- 22:56Risk Management Curve
- 25:28Side Project
- 27:10Why Ruby
- 29:27Shift
