Just a point of perspective from an intermediate-level modeler ... From a technical standpoint, there would be many disadvantages of creating a Poser model directly from human scan data. First of all, the mesh would need a considerable amount of cleaning up, as the human data would not create a very clean mesh. Secondly, symmetry would be off, as the human body is not perfectly symmetrical. So you would have a problem with creating morphs that would move both sides of the face the same; I would suspect that it might also create problems with grouping and symmetrical posing as well, but I'm not sure. So, I would suspect that when DAZ used human scans for their models, there was a considerable amount of cleanup, rework, or (yes) even reforming a hand-built mesh around the scan data. I don't see the similarities in the meshes as a bad thing ... it's the results that you get with them that count.