[zamfi]

I think you missed the central point of this paper. It's true that there is an offsetting factor of building energy use vs. VMT, but the reality remains that denser environments, with jobs near residents, emit less per capita -- as that graph you linked to shows.

What's surprising is that at the other end of the spectrum, in rural (not sub-urban) areas, other factors (forests, etc.) combine to cause lower per-capita emissions than suburbs (and in some cases, negative per-capita emissions because of sequestration).

Basically, from a carbon-emissions perspective, post-war suburbs are unsurprisingly worst, rural areas get some help from sequestration in forests (this fact being the contribution of the paper), and dense urban environments are the least-emitting.