[brandonbloom]

Except I really don't have that choice because the language and library design matters. If I chose to use ByteString for everything, I'd first have to implement Tcl in order to get anything done.

But, yes, you're right, most dynamic languages lack good tools for stating invariants and checking them early. I would like to see that change. However, I'd rather the solution account for runtime dynamism, extensibility, and partiality. We're _slowly_ getting there with more and more advanced type system features. It's time to take that knowledge and repackage it at the foundational level of typed languages.