Kayla and Celeste were inseparable in high school in Davis, California. With Kayla's analytical mind and Celeste's ability to work a room, they made the perfect team. Years later, Kayla reaches out again. She’s been investigating the town’s long-standing alien rumors, building a mathematical model to explain them. But the model alone isn’t enough. She needs Celeste, the only one she trusts, to step back into their old dynamic and uncover the truth buried beneath the rumors.

Celestial Anomalies is a 2-5 hour paranormal investigation game where you use machine learning tools to separate fact from fiction. Interview eccentric townsfolk, weigh their testimonies, and use a Bayesian belief network to uncover how even tiny clues can tip the scales between truth and fabrication. Piece by piece, you’ll learn to reason under uncertainty and answer the question: are the aliens friendly, hostile, or just a story after all? Uncovering the truth could change the way humanity sees the universe, and how two people see each other.

Features

• Apply real-world AI algorithms to a narrative mystery

• A unique blend of narrative mystery and analytical reasoning

• Bayesian networks made intuitive and playable

• Eccentric characters with unreliable testimonies

========================

A note from the developer: I first built Celestial Anomalies as a teaching tool for my university students. It’s a game about reasoning with probability. Math is at the heart of it. But I’ve worked to make it approachable and fun, even if you don’t come from a technical background. If you enjoy puzzles, systems, or seeing how small details ripple into big consequences, this game might be for you.