What I find is that you commit to the data model either way. Your code has to be written against something. It's just that with a type system it is explicit s

What I find is that you commit to the data model either way. Your code has to be written against something. It's just that with a type system it is explicit so that if a change is made which would break something the compiler along with tests where needed can tell you. However with a dynamic language you are much more reliant on hand written tests as there is very little the compiler can verify.