[vidarh]

While there are distances that are certainly "too long", there are very few regions in the US where the majority of the population is so spread out that more trains wouldn't enable a lot of travel to nearby towns and cities even if it's not attractive for long distance travel.

E.g. I got in an argument with someone a while back who insisted on using Montana as an example, but while Montana is sparsely populated, Southern Montana had Amtrak passenger service until 1979 (ironically the only Amtrak service currently serving Montana goes through some of the sparsest populated Northern parts of the state as part of the Empire Builder route), and has a density and distance between the main towns comparable to places in Norway with a perfectly viable regular passenger service.

It's not like you need to have good train links everywhere to improve things, but often the point is not to connect big cities far away with each other but to serve as extra life lines which makes it viable to move further out of the big cities for those in between. That means the relationship needs to be reversed from what Americans are used to: You can't look at where there is demand for transport now. You need to look at where people might want to live or want to work, or where you want to encourage growth, and where better transport might encourage that. And then you need to take a 20+ year horizon and commit to it, because only once the links are there will it start factoring into peoples decisions about where to move, and it will take many years before demand shifts. This is less obvious with cars because as long as there's a road people can forgive a lot. It doesn't help to have rails if there are no trains running on them.

And then you need to consider that it's ok to lose money on a per passenger basis on infrastructure if it promotes economic growth or positively affects other factors (e.g. reduces the amount of car congestion and environmental effects). You should still look at the economy of course, but take all the factors into account, because it often changes the calculations drastically.