All Posts (66)
2026
The Architecture Orphaning Problem with AI Agents
Spec-Driven Development gets functional correctness right but leaves architecture unanchored. The Suggestible Actor model explains why: specs cannot enforce structural constraints, and LLMs, reasoning locally, orphan the architecture one generation at a time.
6 min
The Guardrail Erosion Problem with AI Agents
9 min
AI vs. Open Source, Part 3: The Constant Eyes
7 min
AI vs. Open Source, Part 2: The Hollow Commons
5 min
AI vs. Open Source, Part 1: The Empty Grant
5 min
SECURE Data Act: The dilution in pseudonymization
7 min
Subsidiarity is not Hayek
3 min
Deliverance from Directive Governance
6 min
The Suggestible Actor: A New Model for AI-Assisted Software Development
10 min
Tech Companies and Directive Governance: A Situationship
7 min
Cargo Cult Governance
9 min
2025
The grand flattening: AI Slop is just the next step
8 min
Defense in Depth vs Locality of Behavior
2 min
When Backward Compatibility Can Rescue a Leaky Abstraction
6 min
Const Refs vs. Raw Pointers: Fixing Shared Pointer Reads
3 min
Changing your Jekyll theme without losing your mind (or your content!)
7 min
Cyclomatic Complexity: How Low Can You Go?
3 min
TDD for Bug Fixes
3 min
Let Sleeping Engineers Lie: Why Your Alerts Should Match Your SEVs
2 min
2022
The Law of Demeter and unit tests
4 min
'Privatize' your classes for better unit testing
3 min
Tests should be isolated from each other; not coupled
7 min
In unit tests, I favor Detroit over London
5 min
Defining unit tests: two schools of thought
3 min
Primary attributes of unit test suites and their tradeoffs
6 min
The big WHY about unit tests
4 min
Unit test the brains and not the nerves
3 min
Mocks, Stubs, and how to use them
6 min
DRY unit tests are bad... mkay
9 min
Do not index on test coverage metrics
4 min
Beware of using patch.object to test your Python code
2 min
When I first realized my privilege
4 min
