A city map leaves out alleys, elevation, and nearly every detail except roads — and that is exactly why it works. Simulations are maps of reality: they keep the few ingredients that matter for a question and deliberately ignore the rest. A projectile sim keeps gravity, drag, and launch angle; it ignores spin, wind gusts, the curvature of the Earth, and the barrel temperature.
What a model keeps — and throws away
- State: where things are now (position, velocity, temperature…).
- Rules: how the state changes (gravity pulls down, drag pushes back).
- Assumptions: what we agree to ignore (wind, spin, air density, frame size).
- Outputs: the answers we wanted (range, apex, impact speed).
This is the same skill as machine learning from Module 2 — choosing what data and what features represent the world — and the same ethics as Module 4: a model of a loan applicant or a climate system carries its omissions into real decisions.
Engineers, doctors, and game developers all simulate before they act: crash tests, drug trials, rocket landings. The craft is knowing precisely where the model is not.