PM Flight Simulator

The Problem

People who plan projects assume clean linear phases — A to B, done, on time, on budget — and real projects near-universally do not behave that way. Existing project-management education transmits the vocabulary (Gantt, CPM, EVM) without transmitting the felt experience of feedback dynamics: undiscovered rework, path convergence, the compounding cost of schedule pressure. Management flight simulators exist in academia but are dated and not engaging.

What I'm Building

A browser game where rigor is the substrate and play is the surface. Every mechanic traces to a published model or an empirical dataset rather than invented numbers:

1. Rework cycle (system dynamics) — To Do / In Process / Done stocks with a hidden undiscovered-rework stock; the mathematical heart of why projects slip
2. Monte Carlo + CPM/PERT — probabilistic task durations over a dependency network, producing P50/P80 outcome distributions instead of single false-precision dates
3. Reference-class priors — empirical overrun distributions calibrate the engine so a "typical" run lands where real projects land

The player can win — deliver on time and on budget — but rarely, because the win rate is tuned toward the empirical base rate. The scarcity of winning is the calibration, not artificial difficulty.

Current State

• Tick-based difference-equation engine in TypeScript, React front end, no runtime data dependencies
• 116 passing tests, build gate green at every commit
• Checkpoint time model shipped: runs pause at kickoff, weekly status meetings, and the deadline; an unattended verification run reproduces the engine harness's exact day-86 outcome
• In active development — gameplay workshop phase (negotiation mechanics next)

Tech Stack

TypeScript, React 19, Vite, Vitest. Engine is pure difference equations — every mechanic is a swappable module against a spec'd interface.