The "teeing" combinator is giving you separation of concerns. There are two separate calculations, potentially defined in separate functions, that are composed into a single operation that is performed in a single pass of the stream. Sure, you could write a monolithic imperative for-loop that does it all, but such an approach will not scale.