Would Hadoop Text have simply extended String and implement Writable if String weren't final?

Would Hadoop Text have simply extended String and implement Writable if
String weren't final?

Hadoop Text is often used in lieu of Java's builtin String in order to
transfer data over the network more efficiently. But String has a lot of
functionality missing from Text; would it make more sense to simply extend
String and implement Writable, to define the custom serialization methods?
And then I noticed that String is final, not allowing subclassing ("for
security"). Would Text have been implemented differently if this weren't
the case?