Abercrombie spectrum: threat or menace?

Comments on my Dropbox post on the blog and via Twitter offered characterizations of "Dropbox" for cloud storage as suggestive, descriptive, and generic--running the gamut from inherently distinctive to unprotectable.  I can't help wondering if that diversity of opinion is really that unusual.  (The list of terms found descriptive v. list of almost exactly identical terms found suggestive by courts/the TTAB in McCarthy's treatise is either amusing or depressing, depending on your mood.) 

Is the Abercrombie spectrum really doing anyone any good?  The statute does require distinctiveness, but couldn't we get at that in other ways?  We already know that some forms of marks, color and product design trade dress, can only be distinctive through secondary meaning.  Why not look for other contextual clues about whether a term is likely to be perceived as a mark, as Tom Lee has argued, instead of relying on an incredibly manipulable and barely meaningful test that predates most empirical knowledge about how consumers see marks and about how courts evaluate them?