I have wondered about this. If I have a plant in hand, what if I could
type in a suite of distinctive characters (which would immediately
eliminate many genera and species). The computer could then consider the
remaining possible taxa. It could then pose a series of queries about the
unidentified specimen. These queries would reflect the most parsimonious
way to separate the remaining possible taxa, and might bypass that
tedious mucking around with the numbers of teensy little carpels, and
deciding if those nutlets are spreading or not, or whether those floral
glands are REALLY "petalloid" or not... Is this blasphemy? :-)
-Michael