I would have assumed browsers had one internal implementation and exposed it through both APIs. They wouldn't want to have two separate implementations of a large amount of complex code.