I take your point(s), although from the point of view of someone who is mostly interested in a local average of the face sizes (like me), the requirement that the node-based sum equals the face-based sum is not that important. In case it's interesting to anyone, if you calculate actual averages of the areas of the triangles to which the node belongs, you get numbers that can be nearly an order of magnitude different from the numbers that SurfMeasures gives, since a densely-sampled region from a FreeSurfer brain may have nodes that participate in as many as 25 or so faces.