Level 3: Reading and Judging Code

Writing and reading are different skills.

A film critic doesn't need to direct movies. A food critic doesn't need to cook professionally. An editor doesn't need to write the books they edit. Their job is evaluation, not creation.

Your job is the same. AI creates. You evaluate.

This works because most code problems aren't hidden in obscure technical details. They're visible in the structure, the organization, the obvious patterns. Does this look organized or chaotic? Are things named clearly? Does the flow make sense?

You can answer these questions without understanding every line.

The goal isn't to understand code perfectly. The goal is to understand it well enough to: - Know if it does what you asked - Spot obvious problems before they become bigger problems - Ask intelligent questions when something seems off - Decide when to trust AI's work and when to push back

That's a lower bar than writing code from scratch. It's achievable. And it's enough.

When you look at code, you're looking for signs of organization—or signs of its absence.

You don't need to understand what the code does to see how it's organized. It's like looking at a bookshelf: you can tell if books are sorted by topic or thrown on randomly, even if you haven't read any of them.

One of the most valuable skills in building software is restraint.

If something works, and you don't need to change it, leave it alone.

This sounds obvious, but it's surprisingly hard to follow. You'll see code that looks messy or old-fashioned, and you'll want to 'clean it up.' Resist this urge.

When you don't understand code, don't pretend you do. Ask AI to explain it.

But the quality of explanation depends on the quality of your question. Vague questions get vague answers. Specific questions get useful answers.

Let's be clear about what this level asks of you—and what it doesn't.

You are learning to read code, not write it. These are fundamentally different skills.

Reading requires recognition. You see patterns, notice organization, spot problems. You don't need to produce anything from scratch.

Writing requires generation. You start with nothing and create working code from your understanding of syntax, logic, and systems.

The gap between these skills is enormous. A person can learn to read music without ever composing a symphony. A person can learn to read blueprints without ever designing a building. A person can learn to read code without ever writing a program.

That's you. And that's enough.

When you finish this level, you won't be a programmer. You'll be something more useful for your purposes: someone who can look at what AI produces, understand whether it makes sense, and make informed decisions about what to do next.

You'll know when to trust AI's work and when to question it. You'll know when to ask for changes and when to leave things alone. You'll know when something looks right and when something feels wrong.

That's not a consolation prize. That's the skill. That's what makes vibe coding work.

Trust your growing intuition. Use it. And keep building.