This is less clear at this point, but I think if these divergences between traditional and relative continue or grow (a fairly recent development), I think it will become more evident in certain datasets (precipitation in particular) that we are missing some ENSO-related variability when we use the traditional.  Over most of the historical record, using either traditional or relative index will get you ballpark where you want to be in describing ENSO impacts.  It is really in the last several years when these differences have become large enough to become concerning.