#writerdeck

[Video Description: hands typing on a rainbow-lit clicky clacky keyboard, resting on the bottom half of an open black faux-leather clutch, the top half of which has a 4 inch-ish white e-paper screen and a Raspberry Pi module plugged into it. The keys clack and flash as they're typed, and the camera zooms in on the screen where it says "hello happy pi day!!! :D", as an excited and silly-deep voice offscreen says "It works!" End V.D]

Look what I did!

Tech in 2025 is in crisis. Over 94,000 tech workers were laid off in 2024 alone.

For junior developers, the game is borderline impossible. U.S. software job openings have dropped more than 70%. New grads can’t just apply and get lucky anymore.

Now comes the coping mechanism: vibe coding. A way to “build apps” without ever learning to code. Describe what you want in plain English, let the AI build it.

But it’s a trap. Vibe coding doesn’t create developers; it creates fragile intermediaries. People who can generate code but cannot read, debug, or maintain it.

If you’re a new developer, this isn’t a shortcut. It’s a cliff. And you’re being encouraged to run straight toward it.

What is “Vibe Coding”? A Diagnosis

Vibe coding is the practice of feeding prompts into an AI until an application “works.” You don’t write. You don’t debug. You just regenerate and hope.

An indie developer called it “coding without coding.”

The problem is that it allows you to skip the struggle: the very process that forges a real developer. Fundamentals like problem-solving, debugging, and system design are replaced with prompt engineering.

The result is a dangerous illusion of competence. You can generate output, but you cannot command it. When it breaks (and it will), you have no map, no tools, and no idea why. This isn’t coding: it’s outsourcing your own understanding.

The Illusion of Speed is a Quantifiable Lie

Don’t take my word for it. The numbers are in, and they are brutal.

A recent METR study threw AI into the trenches with experienced developers across 246 real-world coding tasks. The developers felt 20% faster with AI. They felt productive. They felt empowered.

The reality? Their actual productivity decreased by 19%.

That’s a staggering 39-point gap between feeling fast and being effective. Why? Because the AI-generated code was a minefield. Only 39% of it was usable without being fixed. The time they “saved” by generating code was burned ten times over in reviewing, debugging, and untangling the AI’s confident mistakes.

When the Vibe Breaks: A Post-Mortem

When vibe coding fails, it doesn’t just bug out; it fails catastrophically.

Consider the indie developer who built his entire SaaS product with AI. He was celebrated on X for his speed. For a few weeks, he was a hero of the “no-code” movement.

Then the foundation cracked.

Users started bypassing subscriptions. The database began corrupting records. And a security researcher quietly pointed out that his API keys were exposed client-side.

The code was a black box. He couldn’t read it. He couldn’t debug it. He couldn’t fix it. Faced with a system he had generated but did not understand, he had only one option: he shut the entire product down.

This isn’t a fluke. It’s the inevitable outcome. Security experts confirm that AI-generated code is a minefield of classic vulnerabilities: SQL injections, poor access controls, the works.