In reference to J. Whites comment:
"Homology requires that genes or structures (fosil structures) be similar by
descent. Similarity alone cannot provide precise information on ancestral
relationships, even though we use similarity for that purpose all the time.
We infer homology from similarity."
It seems that to infer homology from similarity might be unreliable with
anatomical comparisons, but with DNA sequence or aa sequence
similarity of 70% or higher with a sequence of something as large
as a gene you would have very strong basis to say that two genes were
homologous, ie descended from a common ancestral gene. This term is
broader than orthologous. I don't see the need for fossil record, nor
identical function of the two gene products.
I have the impression that "the similarity is not enough" is an
idea that more meaning in the pre-DNA sequence era than today.