1. 5
  1.  

  2. 1

    This sounds like advocating accidental ad-hoc a/b testing. ;)