Keynote: The Architecture of Understanding
summary
Practical recipe for understanding a messy Ruby codebase: start by zoom in then zoom out on a single behavior (example: Circle’s email “send later”), write small diagnostic questions, draw flow diagrams, follow call sites, add breakpoints, validate with tests, and capture results in lightweight docs like architecture.md. Uses AI as a memory-enabled rubber duck and concrete tooling (mentions Claude MCP) to summarize history and prod data so understanding compounds instead of evaporating.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 1:29Beyond reading code
- 3:18Complex system comprehension
- 4:43Foundational understanding
- 6:32Flow diagrams for causality
- 8:17Validating with tests and AI
- 9:20Identifying system patterns
- 13:59When to stop zooming
- 15:03Familiar patterns in code
- 15:44Ruby patterns implications
- 18:35Architecture and boundaries
- 22:49Team ownership
- 23:11API organizing principles
- 27:46Data models as intent
- 32:25Context beyond code
- 35:32Preserving understanding
- 38:26Keeping docs useful
- 39:09Conclusion
