Problem spaces can be simple, complex and chaotic.
Simple
you can clearly identify actor, input and output
Complex
There are too many factors in play. Complex systems rely a lot on situational data for their outcome and are rarely (if ever reproducible). That said, reason out the output afterwards. An example of a complex system are elections.
One can navigate these system by tweaking aspects of them and seeing how it plays out (In many ways, it sounds a lot like machine learning).
Chaotic
Impossible to predict or understand an outcome. An eldritch horror of sorts...
Disorder is not knowing how to fit a problem space into one of the aforementioned categories.