About

Slopguesser was conceived as a simple game to test how easily you can identify AI generated code. In reality, it’s also a small tribute to the old days of Software Engineering, before AI.

I still remember writing my first programs in C at university, back in Madrid. Since then, I’ve been lucky enough to work with extraordinary engineers and learn a lot from them. And like many developers, I had my own heroes along the way: John Resig, Nikita Popov, Linus Torvalds, and many others. I’ve tried to include snippets of their code in the game as well.

I had a lot of fun creating Slopguesser, and I hope you enjoy playing it too.

If you have any suggestions for making it better, I’d love to hear from you.

How it works

Slopguesser currently supports 7 languages: JavaScript, TypeScript, Python, PHP, Go, Rust, and C.

To generate the challenges, I aimed to find as many chunks of code “in the wild” on GitHub as possible. To reliably find human code, I filtered on code committed before 2021. For AI code, I searched for commits signed by coding agents: Claude Code, Codex, Cursor, GitHub Copilot, Devin, Jules, Amp, OpenHands, OpenCode, Gemini CLI and DeepSeek.

Additionally, and especially for languages like C, where there is not much AI-authored code in the wild, I created pairs with 8 models from 7 AI labs: Claude Sonnet 5.5, Claude Haiku 4.5, GPT-6.1 Sol, Gemini 3.8 Flash, DeepSeek V4.1 Flash, Qwen3 Coder Next, GLM-5.3 FlashX and Kimi K2.7 Code. Claude describes a piece of human code (an Advent of Code solution, a function from a well-known library or an npm package) in plain words, and one of these models writes its own version from that description alone, without ever seeing the original.

Attributions

Code: every snippet comes from a public repository with an open-source license (MIT, Apache, BSD, GPL and others) and keeps that license. After each round, the reveal links the project, file and commit it came from, and credits its author, or the agent and model that wrote it. Thank you to everyone whose code appears here.

Avatars: Miniavs is a remix of “Miniavs - Free Avatar Creator” by Webpixels, licensed under CC BY 4.0, made with DiceBear. Bottts is a remix of “Bottts” by Pablo Stanley, made with DiceBear. Bottts Neutral is a remix of “Bottts” by Pablo Stanley, made with DiceBear. Open Peeps is a remix of “Open Peeps” by Pablo Stanley, licensed under CC0 1.0, made with DiceBear. DiceBear is MIT licensed.