Journal Archive — Page 16 of 165

The Cost of Forgetting: The Epistemic Price List and the Art of the Lean Codebase Explore the epistemic price list: why caches are induction with a TTL, probes are falsification, and forgetting code is the most expensive operation in AI.
Pinball, Pachinko, and the Art of Replayable AI Workflows An important look at software workflows in the Age of AI, contrasting agentic pachinko with checkable pinball control loops.
The Three-Folder Boundary: Ergonomic Workspaces and the Blank Stare in the Age of AI Streamlining the local AI workspace into three clean folders—corporate, personal, and shared—while fixing stale Jupyter roots and resealing ASCII checksums.
Closing the Loop: Fixing Patch Parser Failures and Actuator Surprises An important look at refining our actuator tooling in the age of AI. Discover how parser adjustments and clearer feedback make development more reproducible.
The Three Plain Words: Unifying Context and Building Verifiable Receipts Merging adhoc files into context.txt and replacing obscure commands with context, prompt, and compile creates verifiable, audit-ready AI workflows.
When the Power Returns: Reconciling Diverged Git Branches and Single-Point Failures Recovering from a four-day blackout by reconciling diverged git branches, fixing misleading shell hooks, and auditing single points of failure in the Age of AI.
Local Chrome Profile Inventory and Typed Matrix Resolution An important technical exploration of resolving ambiguous local Chrome profile collisions using reproducible typed selectors and robust metadata validation.
Local-First Development and Resilience in the Age of AI How a midnight blackout tested a local-first development workflow, proving that git, vim, and reproducible environments survive network partitions.
The Letting-Go Moment: Building Vim Training Wheels for the Age of AI How thoughtful onboarding, beginner empathy, and verifiable prompt routers turn a terrifying first terminal encounter into reproducible confidence.
Disciplined Forgetting: Scaling AI Context Through a Two-Tier Memory Hierarchy Shrink hot AI context without losing accumulated wisdom. Explore the forgetting kata, copy-before-cut refactoring, and two-tier memory hierarchies.