Show HN: Searching for optimal strategies in a puzzle game with chance (Drop7)

Category: ai-ml

Tags: game-ai, drop7, expectimax, research, c++

Score: 5.3/10 (Innovation: 5, Technical: 6, Documentation: 7, Utility: 3)

This project investigates optimal strategies for the puzzle game Drop7, focusing on fair expectimax and other policy families to maximize average score in Hardcore mode. It includes a playable web version, a research console, reproducible benchmarks, and agent-assisted research workflows, making it a thorough but niche exploration of game AI.

Target audience: game AI researchers, reinforcement learning enthusiasts, puzzle game players

Repository: https://drop7.dev · C++ · MIT · 1 stars

View on Hacker News