Stylometric categorization would end up being computationally limited right?