Thanks for the interesting presentation on Causal Loop Diagrams which I was not familiar before. It reminded me a bit of Simulink block diagrams to describe dynamic systems. Then several aspects come to my mind:

- any node has usually control inputs, internal state variables and observable variables; any node output can become the control input to another node. The entirety of state variables describes the system state, usually called system variables.

- dynamics is only possible through "inertia" in one form or the other, expressible in differential equations, linear or non-linear. The important part is the dominant time constant for every node. This seems to be missing in the CLD.

- when loops form, there are two cases: first, algebraic loops when nodes in the loop do not have inertia; dynamic loops with an overall loop gain which determines whether the loop is unstable or not.

Another way to model compex systems is using Bond Graphs you may be familiar with, e.g. https://journals.sagepub.com/doi/10.1177/003754979506400303 . They have an interesting aspect that causality can be derived from the graph.