Yes it does. But other factors divide it more efficiently Hair colour, eye colour for example. And of course while the normal position is XX and XY, there are others, like XXY.

This 'categorisation strength' is quite subtle. In some cases a characteristic will divide things into very unequal sets: a two state characteristics could divide the people such as 95% - 5% so it is great for narrowing to the 5% but pretty useless if the person concerned is in the 95%. So sometimes you want to split things into roughly equal sets and sometimes in very unequal sets.

I am wracking my brains a bit, but unfortunately I can't remember the details. There was an AI researcher in Australia whose name started with - I think - H some 30 years who did a lot of work with determining efficient characterisation rules on large datasets.

However, all that is by the by. In this case, it boils down to if the international rules allow an ungenedered declaration, is there any reason to disallow it apart from the one given, which is essentially that we didn't allow for that elsewhere?