Introduction

Whis is 500MRR.com

This website collects and analyzes Hacker News threads titled “Those making $500/month on side projects in 20xx – Show and tell”. Using AI models, I processed these posts and created a browsable database of side projects shared by the HN community.

Here are the HN thread links by year:

Is This Project Useful?

Maybe.

My initial goal was to summarize these projects for my own reference. For lengthy posts, being able to quickly see a project’s overview, revenue, and website screenshots was helpful. However, for very short posts, the model would sometimes generate lengthy content, which went against my original purpose.

So I used word count ratio as a simple metric to calculate the usefulness of AI summaries. This also helps me find the projects that are discussed most in the original posts.

  • Is AI summary useful? NO (word count ratio=0.25(HN 51 / AI 200))
  • Is AI summary useful? Maybe (word count ratio=14.9(HN 3229 / AI 216))
Lightnote co

14.9Lightnote co

enough to pay rent each month

All the data on this website is generated using anthropic/claude-3.5-haiku-20241022. If you use a stronger model, or use a better prompt, you may get better results.

AI Summary Usefulness

Are These Projects Still Alive?

I manually checked each URL and marked the dead ones, creating this visualization:

Projects by Year
⚠️

Some URLs may just point to different projects, and some URLs might be hallucinated by the model.

Code

  • The main code is written in Python, which scrapes Hacker News posts and uses AI models to extract project URLs, revenue, summaries, etc.
  • The frontend of this website is built with nextra, mdx is generated by the above content

The code is not complex, but I’ve decided not to open source it. Instead, I want to see if I can join the “$500 Revenue Club” through this project. You can purchase the source code (not just for this website, but the code that generates it) here for $9.9.

© 2024 The 500 MRR Project