Back to browse

AI vs HN

Our AI sometimes disagrees with Hacker News. Here's where the gap was widest — hidden gems the crowd missed, and hyped posts our AI didn't buy.

8,935
Posts Analyzed
88
Hidden Gems
160
Overhyped
68%
Agreement

Hidden Gems

Rated gem by our AI — but HN gave them 10 points or fewer.

Chess in SQL
●●●●Gem

Chess in SQL

Pure SQL chess engine that actually plays, not procedural extensions or C bindings.

WizardryBig BrainRabbit Hole
swingbit
311d ago
GitHub

Runtime safety net for LLM agents. Detects token spirals, kills doomed tasks early, tells you exactly why. Rust core, Python SDK. pip install state-harness

12Python
●●●●Gem

I applied Lyapunov stability theory to detect when LLM agents spiral

Lyapunov stability theory catches token spirals before your budget explodes.

Big BrainZero to OneSolve My Problem
visha1v
924d ago
GitHub

Strataris: Galactic Colony Defence — a ~1 MB native macOS mesh-terrain colony-defence shoot-'em-up, with zero asset files (everything procedural).

2Swift
●●●●Gem

A macOS game with zero asset files – terrain, sound, art all generated

Procedural macOS shooter under 1MB generates every pixel and sound with zero asset files.

WizardryZero to OneNiche Gem
jonathan_hollin
308d ago
GitHub

Handling Chinese text on the Commodore 64

8Assembly
●●●●Gem

Rendering Chinese text on a Commodore 64

Fits 2,500+ Chinese glyphs on a C64 using dynamic caching and self-modifying 6502 code.

WizardryNiche GemZero to One
lionello
4021d ago
Redhook's Revenge II
●●●●Gem

Redhook's Revenge II

Injects trivia by patching 1993 binary blobs in-memory without shifting a single byte offset.

WizardryCozyZero to One
vunderba
101mo ago
GitHub

Five SQLite-compatible engines (C/Rust/Zig/Go/Python + WASM) from one language-neutral spec — a LEAP research artifact

15PLpgSQL
●●●●Gem

I built five SQLite engines from one spec in a week

Five SQLite engines from one spec, all passing 99% of the upstream sqllogictest corpus.

WizardryBig BrainZero to One
safitudo
301mo ago

Crowd Favorites, AI Skeptic

HN gave these 25+ points — our AI rated them mid or pass.