Think of complex systems as a pinball game.
As the ball bounces, zig-zags, and ricochets around the surface – lighting up features and driving up the score – you might be tempted to think to yourself, “There’s surely a mathematical model that could predict where the ball will end up, based on it’s trajectory off the paddle.” This would be approaching the pinball game as a closed system.
As a closed system, you could explain and model the physics of how the game behaviour would work with some complicated math equations. You could think of the pinball game like a car engine – a mechanistic sequence of cause and effect. But this would make for a very boring game of pinball, because nothing would actually happen.
As soon as you add a player to the game, you suddenly have an infinite number of new variables to add to your model. How will you account for the reflexes, focus, and attentiveness of the player – or their current caffeine dosage? Are they the sort that try to tilt the table? What about the barometric pressure and humidity of the room? In the real world, it is possess any certainty about the outcome of a pinball game, let alone predict a map for the ball’s trajectory at every turn.
Working in systems is similar to playing pinball in that you have very little prophetic knowledge of where the ball will finally end up every time you whack it with the paddle. However, ‘success’ in both cases comes down to responsiveness: closing the gap as much as possible between seeing changes and responding to them.