"This is an aspect of the modeler's skill and inventiveness, not some objective concept of "perfect topology"."

Nope. It is an objective mathematical problem.

Let's take a single muscle. Edgelooping around this muscle takes a certain amount of vertices. You could use six, but then the muscle would look blocky. You could use eight, but then you waste a vertice you could use elsewhere in the mesh. So seven is the ideal.

Two different modellers of similar skill would both come to the conclusion that seven is the ideal amount of vertices to sculpt that muscle and therefore both would actually use seven vertices.

Now, there is also only one "perfect" way to connect that seven vertex muscle to the rest of the mesh.

And suddenly we have two figures that share the same mesh topology.

Did one modeller copy from the other ? No. But both were confronted with the same problem and both found the ideal solution for that problem, resulting in identical topology.

In the end sculpting a human body with the least amount of vertices possible while still properly edgelooping every major muscle isn't much different from creating a cube with the least amount of vertices possible.

Don't think of a figures mesh as a whole. Think of it as a large group of tiny little problems for each of which is only one perfect solution. (like the cube)

The more stringent the rules get, the less choices are left.

Building a mesh efficiently isn't about creativity. It's about finding the most efficient (perfect) meshflow for the shape.

If you don't have to worry about mesh weight, if "everything's fine as long as it's somewhere below 80.000 polygons", you can think like a figure artist who works with clay or wood or stone and just worry about the shape.

But once you start fighting about every single polygon, once you enter "19.999 vertices are fantastic but 20.000 are a completely unuseable bloated waste" territory, you have to start thinking like a mathematician.

What I'm saying is: Once a mesh topology is (near) perfect, there really aren't much (if any) options left to change that topology without either loosing efficiency or loosing "perfectness".

Looking at the DAZ meshes, (and especially the 3rd gen meshes as I think they are built cleaner as they don't rely on subdivision (which always adds a certain amount of waste), I'd be hard pressed finding vertices that could be changed or re-routed without negative effect, simply because the topology is that perfect.