Simulation

Runs a model forward from parameters you set, so the consequence of a setting shows up instead of being argued about.

You have a diagram of your afternoon slump taped to the wall. Lunch feeds the crash, the crash feeds the late coffee, the late coffee feeds the short night, and the short night feeds tomorrow's lunch. Everyone who sees it nods. Nobody can tell you what happens if you move your bedtime by 90 minutes, because a picture of a loop does not tell you what the loop does over two weeks.

Simulation is the carrier that runs the model forward. You set the parameters, the thing steps ahead on its own rules, and you read a consequence off the far end. The move you make is not clicking to reveal something already drawn. It is choosing a starting condition and watching a future get built out of it. That is why the same picture can be run twice: two bedtimes, two curves, and a gap between them that no static drawing contained.

What it makes visible is timing. Delay, overshoot, compounding, and the moment a slow drift crosses a line. Loops and stocks in particular behave in ways their diagrams do not show, because the behavior lives in the arithmetic of repeating them 100 times. What it hides is where the numbers came from. A parameter you invented looks exactly like a parameter you measured once both are sliders, and a run built out of guesses produces a curve as smooth and as sure of itself as one built out of records.

Use it when the argument is about what happens next, and when people disagree about size and timing rather than about direction. Skip it when nobody can say what a parameter would even mean. And write the assumptions where the reader will meet them, next to the controls that use them, not in a note underneath.