it's very easy in academic science to end up working on projects that are just
  little extensions of previously known stuff, and that's sort of a waste of time.