Should You Learn to Code in 2026?
The debate is loud but the answer is simple. Yes. Not because AI can't code. Because you need to understand what it builds. Here's how I learned and how I still learn today.
Thoughts on AI-assisted development, software engineering education, and building tools that teach.
The debate is loud but the answer is simple. Yes. Not because AI can't code. Because you need to understand what it builds. Here's how I learned and how I still learn today.
Addy Osmani called it comprehension debt. The gap between code you ship and code you understand. It compounds silently, and no AI tool is designed to fix it. Here's why I'm building one.
How I run 5 parallel Claude Code agents with an orchestrator, work trees, and CLAUDE.md to ship production code solo. And why shipping fast without understanding is still dangerous.
An honest comparison of the top AI coding tools in 2026. What Claude Code, Cursor, and OpenCode do well, what they don't, and where pear fits in.
The hidden cost of shipping code you can't explain. 66% of developers spend more time fixing almost-right AI code. Here's why understanding matters more than speed.
Why context-switched learning fails for working developers. Courses, docs, and tutorials break your flow. Learning should happen where you code.
AI can generate code, write tests, and explain syntax. But five core engineering skills require human experience to develop. Here is what they are and how to build them.
Vibe coding feels productive. But when juniors skip understanding and ship prompts instead of code, the long-term cost is steep.
The data is clear: AI tools increase output while degrading understanding. Here is why that matters and what we can do about it.