I think strong/static typing with inference is the sweet spot. Complex types can change, so it helps to at least have your boundaries well defined. Within the scope of a function, if you need explicit typing on everything then your function might be getting too complex.