← The Icebox
Killed Personal Project · 2025

YouTube Escape & Explore Chrome Extension

A browser extension designed to counter algorithmic content fatigue by giving users an intent-based discovery layer for YouTube: pick a topic, get a curated escape hatch away from your feed. During discovery I evaluated three technical implementation paths. All three had fatal flaws. I killed the project before writing a line of extension code.

Approach comparison

Approach Maintainability User experience Verdict
API Proxy Low: requires live server, ongoing API costs High: direct, clean navigation Rejected
Too much overhead for a side project
Search Hacking High: automatic, no maintenance Low: noisy, uncontrolled results Rejected
Couldn't guarantee a reliable experience
Manual Curation Medium: requires ongoing content updates High: curated, high-quality videos Rejected
Made me the permanent bottleneck

Approach breakdown

API Proxy

Connect directly to YouTube's data API

The cleanest UX: pull high-quality, structured video data directly from YouTube's official API and surface it inside the extension. No noise, no guesswork.

The problem: the API key has to live somewhere secure. Bundling it inside the extension exposes it to anyone who downloads it. The only safe option is a proxy server, a backend middleman that holds the key and handles requests on behalf of users. That turns a browser extension into a hosted service with ongoing infrastructure costs, API rate limits, and security monitoring. I didn't want to build a startup; I wanted a side project.

Search Hacking

Pre-fill YouTube search queries by topic

Skip the API entirely. On click, open a pre-constructed YouTube search URL ("best history documentaries", "long-form nature footage") and let YouTube's own search do the work. Zero backend, zero maintenance.

The problem: I have no control over what YouTube returns. Search results change constantly, get polluted by SEO spam, and vary by region. The extension would sometimes surface great content, sometimes garbage. Shipping something I can't stand behind is a non-starter.

Manual Curation

Maintain a hand-picked library of videos inside the extension

Curate 50 high-quality videos across topics, hardcode them into the extension, ship it. No server, no API. Clean UX, high quality.

The problem: videos get deleted, go private, or go stale. Someone has to monitor, update, and push new extension versions to keep the library alive. That someone is me. Taking on a recurring maintenance obligation against a full-time job and a product portfolio isn't a side project: it's a second job with no upside. This violates my core constraint: build once, run without me.

Why I killed it

Decision
API Proxy Turns a side project into a hosted service with ongoing costs and security surface.
Search Hacking Produces unreliable results I can't control: ships a product I can't stand behind.
Manual Curation Makes me the permanent bottleneck. Any product that requires my ongoing intervention to survive is not a product: it's a chore.

What I carried forward

Lessons

Operational debt is a silent killer of side projects. The cost to keep a product alive (not to build it, but to run it) is often invisible during ideation. Making it visible early is the job.

Killing an idea is a skill, not a failure. Running a structured analysis and walking away before writing a line of code saved weeks of effort. The discipline to say "the math doesn't work" is exactly what separates senior PM judgment from building for the sake of building.

Back to
The Icebox