If you want to use a single metric, why not use people per mile of road?
This correlates well with the only network metric that matters, linear route miles of cable per subscriber needed to provide service. Every (on-grid) dwelling and business is on a street or a road, so if you put a fiber cable on every road or street, then you will pass every inhabitant in the country.
Sweden has 10 people per mile of road. The US has 30 people per mile of road.
Sweden simply does better, despite having to deploy more infrastructure to provide service.
People still live in the countryside. Rural areas aren't some uninhabited wilderness.
If your aim is to provide broadband to all, you'll have to build there too. Sweden aims to do 100 Mbps to 90% of the population by 2020. Can't do that by just covering urban areas.
Try driving through Wyoming some time (on a highway!). You can easy drive for an hour or two and there is nothing but BLM (gov't) land all around you. No one lives there.
That metric works great if you have 100% penetration.
But what if a place doesn't have service on a large portion of roads? I doubt that is particularly true for this comparison, but the metric probably shouldn't reward a place with more unserved areas.
This correlates well with the only network metric that matters, linear route miles of cable per subscriber needed to provide service. Every (on-grid) dwelling and business is on a street or a road, so if you put a fiber cable on every road or street, then you will pass every inhabitant in the country.
Sweden has 10 people per mile of road. The US has 30 people per mile of road.
Sweden simply does better, despite having to deploy more infrastructure to provide service.