About p(Doom)1
One developer, a strategy game about running an AI-safety lab, and the story of why it exists.
Why this exists
I built p(Doom)1 because AI safety mostly lives in papers and blog posts, and I figured losing a strategy game to it might teach the ideas faster. You run an AI-safety lab — hiring, funding, compute — and try to lower p(doom) while the paperwork does its best to stop you.
The tone owes a lot to Papers, Please and Pandemic: satirical on the surface, serious underneath. You manage resources, make policy calls, and try to prevent catastrophic outcomes — mostly via spreadsheet.
Educational First
Every game mechanic is designed to teach real AI Safety concepts in an engaging way.
Source available
All of it on GitHub — warts included — to read, evaluate and contribute to. Not open source yet: an engine licence is planned around 1.0. Contributions welcome.
Accessible
Free to play and designed to be understood by both experts and newcomers to AI Safety.
Research-Based
Grounded in real AI Safety research and current discussions in the field.
Development Story
p(Doom)1 started as an experiment in whether a game could carry ideas about AI safety and existential risk. It's grown from crude prototypes into an actual strategy game — early alpha, but a real one.
Initial Concept
First prototype focusing on resource management and p(doom) mechanics.
Core Systems
Implementation of competing AI labs, policy systems, and economic models.
Enhanced Features
Added leaderboard system, improved UI, and comprehensive type annotations.
Early Alpha
Alpha builds for Windows, macOS and Linux — Windows is the tested one, Mac and Linux are fresh and largely untested. Current version is in the stats below, pulled live from the release data. Rough on purpose, held together with optimism in places.
Who's Building It
p(Doom)1 is built by Pip Foweraker, working solo, in the belief that games can educate and inspire. It is developed in the open, so anyone is welcome to pitch in. (Source-available rather than open source for now — see the licence.)
Pip Foweraker
Design, code, AI systems, and overall direction — the whole thing, so far.
Open to Contributors
Code, ideas, and feedback are welcome through GitHub. If you'd like to help, the door's open.
Interested in contributing? Check out our GitHub repository to get involved!
Technology & Approach
p(Doom)1 is built with Godot 4.5.1, a modern open-source game engine, chosen for native cross-platform builds and a strong open-source ecosystem. This stack allows for:
Godot Engine
Built with Godot 4.5.1 for a fast, contributor-friendly workflow and native builds across platforms.
Native Cross-Platform
Built from a single Godot codebase and run natively, with no runtime to install. Windows and Linux are available now — Windows is the tested build; Linux is new and largely untested. macOS — no build in the current release; it broke and a fix is written.
Data-Driven Design
Game mechanics are configurable through data files, making it easy to balance and expand content.
Modular Architecture
Modular GDScript, typed where it earns its keep — and held together with optimism in the remaining places.
Community & Contact
There's no forum yet — it's one developer and a GitHub repo. Feedback goes through the in-game bug reporter or GitHub issues, and both genuinely get read.
Ready to Play?
Free, source-available, early alpha. Try to lower p(doom) — the paperwork usually wins.