The main idea of the article is that people think gas savings go linear with MPG when "upgrading" their vehicles, for example, going from a 10 MPG to 15 MPG would be considered a worse option than going from 40 to 45 MPG. Or, even going from a 20 MPG to a 30 MPG being worse than going from a 20 MPG to a 40 MPG. It turns out it's not so easy, because gas consumption is not linear with MPG,
If one also considers that high MPG vehicles can cost significantly more these days, it might be wise to reconsider how to shop for a car.
From this graph, which I reproduced it from the paper and added the two linear regimes, it easy to see that somewhere around the 30 MPG the gas consumption difference slows significantly for a given change in MPG. This calculation was done for total travel of 10,000 miles a year which I think seems a pretty good average.