Sticks, a developer who writes about technology and software, has a confession: he loves building things, especially software. But he’s not sure AI-assisted coding, or “vibecoding,” delivers the same joy. His argument, published on Hacker News, is that vibecoding frontloads the fun — giving you a quick thrill at the start — but leaves behind a nagging emptiness that hand-built code avoids.
“I can, and have, built things with AI that I either never could have, or never would have, built otherwise.”
That admission sits at the heart of his piece. He’s not dismissing AI entirely. He’s arguing that the kind of satisfaction it offers is different from the kind he gets from writing code by hand.
The Joy Of Hand-Built Software
Sticks describes a simple moment that defined his relationship with coding. He built two small Python scrapers that did something useful he’d wanted for a long time. They were basic, but they gave him so much irrational pleasure.
“It was the most fun I’d had programming in years,” he writes.
That joy comes from several sources, he says. There’s the thrill of the idea, the excitement of making it real, the hard-won accomplishment of working through difficulty, the satisfaction of a job well done, the thrill of learning, and the pleasure of solving a problem you have.
Vibecoding can give you the first three and the last one. But the middle ones — the hard work, the sense of accomplishment, the mastery that comes from struggle — aren’t natural consequences of letting an AI do the heavy lifting.
Why Vibecoding Feels Different
The problem, Sticks argues, is that vibecoding introduces a layer between you and reality. When you write code by hand, you engage with the problem directly. You fight the compiler, you debug the logic, you trace the flow of execution. That engagement builds a relationship with the project.
With vibecoding, you skip that middle layer. You describe what you want, and the AI produces something that works. The result is functional, often quickly. But the process itself lacks friction. There’s no wrestling with syntax, no staring at a blank screen for hours, no moments of pure frustration followed by sudden clarity.
He compares vibecoding to a series of one-night stands. You get the thrill, but you don’t develop a relationship with the project. You don’t get the same kind of warm fuzzies, the oxytocin, the attachment. The problem with a dopamine-only life is that thrills become dull with repetition.
What Vibecoding Actually Delivers
Sticks lists several examples of things he’s built with AI that he either couldn’t or wouldn’t have built otherwise:
- A has-this-already-been-submitted-to-hackernews bookmarklet, based on Taylor Troesh’s Discuss on HN snippet, which he uses regularly on his Wallabag favourites page.
- A prediction market simulator party game called “Party Market,” which he ran live during a lecture on prediction markets, complete with a market on “how long will I drone on for.”
- A wildfire early warning system using Pyronear, built in two afternoon sessions with free plan Lumo, which detected the first frame of smoke on ignition footage.
- A browser-based Kobo database migration assistant that helps people unfamiliar with SQLite move their books and annotations to a new device.
- A realtime pitch training app for his singing lessons, which he’d planned to build for about a decade.
In most of these cases, he was pleased with the results. But the process wasn’t fun the way writing his dumb little scrapers was fun.
The Cost Of Instant Gratification
There’s a pattern here. Vibecoding gives you instant dopamine. If you have an idea, you can get a working prototype very quickly, with very little effort. Someone with no coding experience can “code” an app that probably works. Instant gratification.
But Sticks has a personal policy against frontloading the fun. He doesn’t like getting paid for work he hasn’t done, going into debt, or arrangements where someone else does his work today in trade for him doing their work tomorrow. Vibecoding feels like that kind of arrangement. It delivers the payoff up front, but the work feels hollow.
He’s also blunt about the quality of the output. He’s tried everything from local models running in llama-server to copy-and-pasting from free, no-sign-up services like Phind and Lumo, to full-on remote server plans with custom-written auto-approval hooks that were worryingly close to dangerously-skip-permissions. It’s all the same: productive, but unfulfilling.
“I don’t like the way it makes me feel, and I don’t like the quality of the output,” he writes. “It manages to make me feel dumb and superfluous (and, sometimes, dirty for even using it), without earning my respect.”
The Case For Hand-Built Code
Sticks acknowledges that handwritten software development isn’t always fun. His first Golang project was, for large sections of time, with a deadline, certified Not Fun™. Fighting the Rust compiler the first time he tried to write twenty lines of Rust code was Not Fun™.
But it was, in the long run, Type II fun. It gave him the sense of accomplishment and the glorious I did it! feeling. And next time, it would be easier.
That’s the core of his argument. Vibecoding delivers a different kind of fun, and that kind of fun has limits. It’s thrilling at first, but the thrill fades fast. The work itself becomes mechanical, and the sense of ownership over the project is weaker.
His conclusion is modest. He’s not saying everyone should stop using AI. He’s saying he personally prefers the slower, harder path. The joy of building something by hand is a different kind of joy, and it’s one he finds more sustaining.
The piece is a personal one, not a data-driven study. But it captures a tension many developers feel. AI tools are powerful, but they change the texture of the work. Some of that texture is pleasant to lose. Some of it is not.
Sticks is betting on the latter. He’ll keep vibecoding when it makes sense, but he’ll keep writing code by hand when it matters. The dopamine rush is nice, but the warm fuzzies last longer.
Source material: “Vibecoding isn't as fun as writing code by hand,” autodidacts.io.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.