This brings to mind something else that concerns me. Are molecular phylogenies tested and re-tested by the addition of new taxa (when possible) and/or new sequence data until the tree "looks right".
The former. Work on placental phylogeny didn't end in 2001, that's when it started in earnest.
Come on. "They" are not more intellectually dishonest than "we".
But what if the whale/hippo/ruminantian clade is also "wrong"? How would we know? If we've exhausted the option of taxon sampling, is stringing together even more gene sequences going to resolve the problem if the gene sequences themselves are the problem?
Wouldn't then particular gene sequences which evolve too quickly or whatever be the problem?
It goes without saying that different genes sometimes tell different stories. Metaves/Coronaves is perhaps the most famous example on this mailing list.
With turtles, I think we've hit a wall in relying on molecular-based phylogenies to tell us where turtles came from. IMHO, the fossil record is our only hope.
Here, too, I think it's actually the lepidosaurs that are misplaced in most molecule-based phylogenies, not the turtles. But you're right, the potential taxon sample for molecule-based analysis is limited in disquieting ways.