Um, not really my field of expertise, but I can probably summarize succinctly. Since the 70's various scientists have been gathering up samples of critters and subjecting their most primative DNA-produced parts (the very parts the existence of which, we have taken for a reason to put them in related families, phylum & such) to analysis regarding their relative mutational distance. (How many changes must be made to get from, say hummingbird to human, assuming a reasonably constant clicking of the mutation clock). And, as it turns out, this produces a tree of relationships whose branching tree structure matches astonishingly well with the tree the fossil guys have drawn up from the bones, along all the main branches.
however, this mutational distance clock is a mismatch for the geological record in which the fossils are buried, according to the guys at Cambridge who update the geological/fossil clock every 10 years, by about 30% overall. And with some especially acute problems distinct from that at the very beginning of life, as per the latest revision of the tree of life, just 2 years ago. I can give you pointers to a longer discussion if you like.
Science is far to young to have a suficient window of observation to accurately estimate a rate of genetic drift (if there even is a constant one, since evolution doesn't appear at all to be gradual and constant at all times).
Genetic drift theories remind me of linguistic drift theories, whereby the time frame for the differentiation of Latin into the Romance languages becomes the presumed yardstick for extrapolating backwards to some date of a proto-Indo-European language.
Way too incautious an assumption, IMO.