Is this just a guess or do you have anything to back this up with?
> You could probably cover 98.5% of the population without covering much of the road network.
Here's a map of xDSL coverage: http://www.europarl.europa.eu/document/activities/cont/20110...
Let's see how you propose to cover that land area without also covering much of the road network.
Looking at the map, the northernmost county has barely any HSPA coverage. Nevertheless that county has a population of a quarter of a million inhabitants. To get 99.6% coverage, you'd still have to cover over 200 000 of those and everybody else in the country. That's clearly impossible as 100 000 inhabitants in that county live outside the ten largest "metro" areas.
There's just no way of not covering a lot of roads if if you aim for coverage in the high nineties with such a dispersed population.
Firstly, Sweden pretty much intends to do so. The official goal is to bring at least 100 Mbps service to at least 90% of the population by 2020. Many municipalities/states have programs to build out fiber to everybody.
Sweden also already has a cable running to almost every premise, so in practice every road and every street is already passed. In 2011 xDSL covered 98,5% of the population and only 1100 households or businesses had no service or less than 1 Mbps service.
Secondly, it's an estimate, but it's a far better estimate than population density or any other single metric presented so far. It correlates with the linear miles of cable needed to pass all premises, which is the deciding factor in determining costs. It is also readily available and it measures compratively how hard it would be wire up a nation.
By this measure Sweden does far better, even having a need for more infrastructure per inhabitant to reach the same end result, i.e. broadband service for everyone.