Long copy v/s short "Try Now" approach for webapps would certainly be a an interesting A/B test.

(PS: we have been trying to do such a split test for Visual Website Optimizer homepage. Will post results once we do it)