Well, now I feel like I should apologize for causing such a stir. I didn't actually intend my post to be a demand for a feature request, so sorry, Rick, but thanks for adding that on my behalf.
I would say that the concept of assigning an area to a node is going to be artificial no matter how you decide to do it. My particular interest is that I want to be able to estimate local Jacobians of the transformation from subject surfaces to warped atlas surfaces, which is why I'm using the particular artificial method I described.
Peace,
Shane