Well, it seems logical to me that the quality of the preservation of a
sound recording is effectively determined by its content! Thus, there
is no reason to preserve stereo versions of mono originals, or to use
more bandwidth than was recorded originally (i.e. preservations of
acoustic 78's flat to 20KHz). As well, if the recording is "spoken
word," the frequency range could be limited without loss of
information and/or intelligibility. This also covers the question
of "higher quality for material deemed more...important"...keep in
mind that a preservation can never be anymore accurate than was
the original recording (unless computer algorithms become much
more powerful!)