Wild to me that pieces like this ignore Wolfram and specifically his computational irreducibility. Much more precise rendition of the equivalent but less robust concept undecidability. The key insight he arrives at is that predicting/modeling a system is itself a computational process and the ability of a modeler to make computationally less expensive predictions is dependent both on their own computational capability and the nature of the computation. It is very neat and precise and systematic. "Next level" chaos is just more poor imprecise metaphor.