Radical Object Orientation in the Age of AI Code Generation
summary
Argues that two 1967 Alan Kay principles — unidirectional messaging and integration/operation segregation — make AI code reliable by forcing tiny, testable functions and clear composition. Demonstrates a TypeScript forecasting app (histogram of cycle times and probabilistic multi-issue forecasts): a one-shot prompt produces working but tightly coupled files, while a 71-prompt iterative workflow yields modular code with unit tests at the leaves, no mocks or DI, and a single integration test — code that runs and is easy to read, test, and reuse.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 17:31Motivation for radical OO
- 19:21Alan Kay’s 1967 OO vision
- 25:26Cells and software objects
- 32:44Functional dependencies problems
- 34:12Radical OO principles
- 41:24Integration and single responsibility
- 51:05Data flow hierarchy
- 1:07:16Functions first, classes containers
- 1:12:20AI code generation small scope
- 1:22:22AI experiments and results
- 1:30:39Summary
related talks
36mEngineering Intent: Solving the AI Productivity Paradox
58mAI, Determinism, Code Generation and Refactoring
39mAI and the Democratization of Software Development
56mFundamentals of Software Engineering In the age of AI
30mThe AI Revolution in Coding
26mAl and Software: The Good, The Bad, The Scary
14mThe 4 Patterns of AI Native Development
14mWhat 80 years of coding history says about AI tools today