> The aforementioned barrier-separated box is high in entropy but simple.
This isn't true in any meaningful objective sense; representing the exact state of the system in some basis (e.g. the position or momentum states of all particles in the system) requires a huge amount of information, corresponding to the entropy. The GP only said it's a "simple" system because we (for arbitrary reasons) don't really care very much about the position of each gas molecule. We as humans are satisfied with describing the system in terms of bulk statistical characteristics like temperature and pressure. However, we would be unsatisfied doing the same thing with a similarly information-rich system such as a microprocessor, because now the bulk characteristics of the system aren't sufficient information to describe the characteristics that humans care about.
"Simplicity" or "complexity" as described by the GP is not a physical quantity; it's a reflection of which precise dynamical behavior we happen to care about. You could probably find all sorts of heuristics that generally match with human intuition, but in the end it's up to opinion.
