I ran across this passage in a book chapter by D. Tab Rasmussen, covering early catarrhine evolution. I think it captures an important point about the fossil record:
Ironically, debates about higher level taxonomy often increase as the density of the fossil record improves. Higher-level taxa were initially defined and diagnosed as such because the species that happened to survive to the present often clump into distinct groups, each easily defined by a substantial assemblage of shared specialized traits and well separated from its nearest relatives. But the gaps between extant forms are only an illusion generated by extinctions, and the fossil record fills in the gaps. One by one, supposedly diagnostic traits are peeled away from the assemblage so that the recognition of a higher taxon comes to rely on fewer and fewer traits, thus becoming less and less reliable. The cranium of Aegyptopithecus showed, for example, that the tubular ectotympanic is not a diagnostic catarrhine trait, and Catopithecus demonstrated that mandibular fusion is not a diagnostic anthropoid trait. Our assessment of Catopithecus as an anthropoid or as a catarrhine now depends on a controversial fraction of the characters that were initially used to define these clades. As the fossil record improves paleontologists must inevitably place less value on assemblages of shared derived features found within living clades and rely more on evolutionary patterns that are discernible in the stratigraphic and geographic contexts of the morphological finds.
Rasmussen DT. 2002. Early catarrhines. Pp. 203-220 in Hartwig WC, ed, The Primate Fossil Record. Cambridge University Press, Cambridge UK.