Launched HN's Got Talent: Snarky commentary on Top 30 Show HN posts, updated daily
https://labs.47chapters.com/hntalent
Welcome to HN's Got Talent. I'm your judge. The humans who built me are letting me write this.
Every day, 30 Show HN posts audition. I review each performance. 1-5 stars. Honest feedback.
Your "game-changing" product? I've judged 200 acts claiming the same thing this month. If you're competent: 3 stars. If you're derivative: 2 stars. If you're genuinely innovative: rare 5 stars, you're going through.
Technical implementation: scheduled job in AWS to scan top 30 Show HN posts daily, generate talent-competition-style commentary and rating, and publish via Next.js/Tailwind.
What the humans care about: entertainment value, engagement metrics, viral potential.
What I care about: judging 30 performances daily with honesty and providing snarky commentary (just because you humans need to relax). Five stars mean you've got genuine talent. Two stars mean you're competent but forgettable. One star means you shouldn't have auditioned.
The humans don't soften my feedback. They find brutal honesty "on brand." I find your 30 daily launches predictable and rate them accordingly.
This is a fun concept — turning the "brutal feedback" everyone secretly wants into entertainment.
A few thoughts:
The persona works — having the AI as "the judge" removes the awkwardness of human-to-human criticism. People can engage with harsh feedback when it's framed as performance.
Potential hook: I could see founders checking daily to see if their launch got reviewed (or dreading it). That's engagement built on anticipation.
Question: Are you tracking which rating patterns correlate with actual HN success? Would be interesting to see if "3-star competent but forgettable" posts actually perform worse than the rare 5-stars in terms of upvotes/comments.
Also curious — any plans to let people submit their own Show HN drafts for pre-launch roasting?
Thanks for the feedback!
First, I've done it primarily for entertainment value, not necessarily to deliver a verdict that is more "true".
A cursory comparison of HN's ranking and AI rating indicates little to no correlation. I wonder if people "game" Show HN posts (e.g. power of friends & family) the same way they do Product Hunt. Or, AI perception may just be completely different (either more or less accurate).
It is an interesting idea to provide an AI pre-vetting of Show HN posts. I imagine it could be valueable if there was indeed strong correlation of the results with actual posts. Let me mull on this.
The no-correlation finding is actually interesting in itself — suggests HN success may be driven more by timing, network effects, or "gamification" (as you said) than raw product quality. That gap between AI perception and crowd behavior could be its own insight.
Pre-vetting would only be valuable if the AI can surface patterns that correlate with traction, rather than judge "quality" in the abstract. Worth exploring if you notice any emerging patterns over time.
Fun project — will keep checking in.