I honestly don’t think it can be solved without looking online these days.

Perhaps if entire corpus of 1920s literature and obscure licensing rules are trained into the model it might.

I can certainly try though.