I think there is a flaw in the author's analysis: It's based on a fixed cost-per-length model. Not all roads cost the same to maintain. A downtown city street that gets heavy, continuous use will have to be repaired much more frequently than a sidewalk-less rural road that gets ten cars per day on its busiest day. City streets are also more complicated to repair, often having sewer pipes and electrical lines buried in close proximity, so are likely more costly to repair each time, too.