[stephc_int13]

It is a completely different domain, but I think that notion of Computational Irreducibility of Stephen Wolfram is useful to reason about complexity.

Even with perfect knowledge of the laws and initial conditions, it is not always possible to produce a reliable model.

Simple cellular automaton like Conway's Life or Langton's Ant appears chaotic.

Not even talking about simple mechanical apparatus like the double pendulum.

[metaphorhacker]

Absolutely, you can think of a complex weather system or history as computationally irreducible - Wolfram is afterall just reconceptualising complexity with it. You have to walk through all the steps as they happened to be able to find out the outcome.

And the big models are really just trying to find pockets of computational reducibility. But the pockets really only stay at a certain level of magnification. If you zoom in or out too far computational irreducibility swoops in with a vengeance.