Hmm. You are making me think hard, which I like. Given that any human description of a system is going to be an approximation of reality, I suppose the one of the key difficulties is determining which aspects of the system are more important for our model to more closely match.

Even if the information theory metric of complexity is somewhat flawed, given our limitations, I think it is a useful tool. I'm not sure what would be a better way of comparing complexity.

At the very least, I think I can make statements such as: "a closed container of hydrogen gas at room temperature and pressure is a much simpler system than a Swiss watch", which we could roughly quantify in a somewhat robust way. The hydrogen part is easy, as the atoms are indistinguishable, and their movements can be closely approximated with simple formulas.

P.S. I just remembered a good (and short) minutephysics video on entropy & complexity, perhaps it is worth linking: https://youtu.be/MTFY0H4EZx4