Vibecoding isn't as fun as writing code by hand

The piece argues that AI-assisted "vibecoding" delivers instant dopamine but undermines the deeper satisfaction of writing code by hand. Vibecoding frontloads the fun: it produces working prototypes quickly, lets novices assemble apps, and enables projects that would otherwise never exist. The writer describes building gratifying small hand-written Python scrapers, then contrasts that with several vibecoded wins - a Hacker News bookmarklet, a Party Market prediction-market demo for a lecture, a Pyronear-based wildfire early-warning demo running on a laptop via Lumo, a browser Kobo migration assistant, and a realtime pitch-training app. These examples show vibecoding’s power to solve problems rapidly and to make playful, otherwise-impractical builds possible.

Those practical gains come with steady costs. Refactoring and debugging vibecoded output lack the "hard-won accomplishment" and learning that make manual coding rewarding; the code often feels low-quality, opaque, and alienating, leaving the coder feeling superfluous or complicit. Experimentation across local and cloud models produced the same pattern: high short-term productivity, low long-term mastery. The writer frames hand-coding’s struggle as Type II fun that builds skill, pride, and attachment to projects, arguing that vibecoding trades those durable satisfactions for fleeting thrills and a detached relationship to one’s creations.