People "overcomplicate" the rule because they find counterexamples to the simple rule.
It's a fool's errand because the way human language works is that people happily accept odd exceptions by rote memory. So the rule simply says that there exist these exceptions. Also, there is something called euphony: speakers find utterances questionable if they are not in some canonical form they are used to hearing. For instance "black & white" is preferred over "white & black".
The rules boil down to "what people are used to hearing, regardless of the underlying grammar offering other possibilities".
It's a fool's errand because the way human language works is that people happily accept odd exceptions by rote memory. So the rule simply says that there exist these exceptions. Also, there is something called euphony: speakers find utterances questionable if they are not in some canonical form they are used to hearing. For instance "black & white" is preferred over "white & black".
The rules boil down to "what people are used to hearing, regardless of the underlying grammar offering other possibilities".