You can one-shot this project in any LLM today. So lets diagnose.
The Build
The Magic 8-Ball was my first project on this site. Shake the ball, get a response. While it initially was just a pool of strings and a click handler, this was conceptualized at a time when LLMs were just starting to enter the public eye — my thought was to eventually hook it up to a chatbot and learn how that worked. It was a great learning project: React state, CSS animations, and figuring out how to deploy something real to a URL I could share. I was freshly graduated and still under the impression that coding a lot was going to be a large portion of my upcoming career...
Why It Died
The problem is simple: shake it once and you've seen it all. No depth. No progression. No reason to come back. The interaction loop is "click → read → maybe click again → close tab."
The custom response library was more rewarding to write than to interact with. Besides, we have ChatGPT now. If you want a sarcastic oracle, the bar has moved.
The More Interesting Lesson
Our current ability to rapidly prototype has pointed out how truly half-baked some ideas are. If I could have spun this up as an artifact in five minutes back then, it probably never would have made it to the internet. Shipping it did show me that a clever response library doesn't fix a shallow mechanic. The mechanic itself has to be interesting. Back then, the technical challenge of building something was enough to occupy my full attention. Whether the idea was actually good got lost in the noise. We no longer have that issue.
It's made me respect the planning stage. Put the idea together more thoroughly before jumping into creation.
The Verdict
Graveyard. Every project on this site exists because this one came first. Seeing where something falls short is learning. It's also made me wish I'd taken more notes along the way. Now I'm blowing through my notebook at a pace I didn't imagine.
Stack: Vite, React, Flask, CSS