Part of the issue is that the randomization has not been crafted carefully, and instead seems totally uniform. If 50% of the haircuts / outfits / whatever are

Part of the issue is that the randomization has not been crafted carefully, and instead seems totally uniform. If 50% of the haircuts / outfits / whatever are feminine vs masculine, maybe you get a 50/50 split. But if then 3/4 of the time, the "Other" category adds facial hair, you're getting masculine people 7/8th of the time.