It blows my mind that most HN users are still impressed by algebraic types but haven't heard about set theoretic types.
I write a lot in other language that uses ADTs, and I really miss STTs for the ease by which I can construct a "Partial" data type with significantly less boilerplate.
Of course, it also enables laziness, as instead of considerig whether it's proper to have a specific Partial product type, I just create it and move on. Easier to break loosey-goosey on modeling the domain.