A shorter version of this was originally posted as a comment on The Transport Politic.
One problem with Northstar, and most new systems, is that by running only at rush hour they are aimed strictly at the 8-4 / 9-5 crowd. Anyone who ever has to stay later at their job can't take the train, or wind up a very costly ($80-120) mile cab ride from home. In order to plug a deficit, the MBTA proposed cutting service after 7 p.m. in Boston and there was a ton of outcry. People basically said "having late trains, even ones running every two hours, allows me to take the train every day. If you cut those trains, I have to drive." So there's a whole market which is ignored by limiting transit to commute hours only.
Of course, Boston and other legacy commuter systems (New York, Philly, some of DC, Chicago and San Francisco) have existing trackage rights or own their track outright, so don't have to worry about freight rail's demands. (The Northstar Line shares track with a major BNSF transcon route which sees about 50 freight trains a day.) Of the non-legacy systems, only Miami-West Palm Beach, Utah, Dallas-Fort Worth, and Connecticut's Shore Line East provide any sort of evening service.
This evening service, while not well patronized, helps more people take transit. While there are no definitive numbers, let's make a some assumptions/educated guesses for cities with full-schedule commuter rail. Eighty percent of transit ridership is during traditional rush hours: in by 9, out between 3:30 and 6:30. Ten percent is on midday services, and ten percent on evening services. It seems that you could cut these services, and lose only 20% of your ridership while eliminating 40% of the trains. However, it's not this simple.
One of the reasons people don't ride transit is because of their families. A frequent issue brought up by many potential riders is "what happens if my kid gets sick and I need to pick them up at at school?" (Nevermind that this is an infrequent enough occurrence that the money saved on gas and parking would more than cover a cab fare twice a year.) But, it's a valid concern, especially for systems where suburban transit quits from 9 to 3. Even with hourly or every-two-hour service, it provides some safety net. If your kid gets sick you'll be able to get there at some point in before the school day is out.
The other reason people eschew transit in underserved cities is the "what happens if I have to stay late?" question. As mentioned, in the the legacy commuter rail cities, if you have to stay late, you'll get home. You might not be able to walk down to the station and get on a train which runs every twenty minutes like at rush hour, but if you have to stay until 7:00 you'll at least get home in time to say the proverbial good night to the kids. The late trains provide a sort of safety net—for people who have to occasionally and unexpectedly stay late, it allows to them to come to work without a car and know that they'll get home. For a lot of employees, this makes the difference between taking the train to work and driving.
So let's go back to the 80-20 rush hour–non-rush split. For people who might sometimes have to work late, probably 95% of their trips are during rush hour—ten or twelve times a year they have to stay late at work. So, out of a hypothetical 100 riders, 95 of are crowding trains at rush hour, and 5 are taking trains later on. However, if you cut the late service, you not only lose the five people on these relatively uncrowded services, but the 95 on the earlier trains, too. So you can't cut evening trains and expect to retain your full peak ridership.
There is another element to running late-night service: it allows people who work downtown to stay downtown. Many American cities has 9-5 downtowns: they empty out at night and seem desolate. This is mainly because there is not the critical mass of people to populate the streets and go to restaurants and shows. In a city like Minneapolis, which has a rather high transit mode-split but little late night service, people who want to stay out late, if they are from the suburbs, are forced to drive. Minneapolis has a good number of services downtown, from a full-scale Macys (originally Daytons) to a full-scale Target, as well as baseball and basketball arenas, as well as the Guthrie Theater and Orchestra Hall. Providing later service would allow people to take transit in in the morning and stay downtown to avail themselves of the amenities which are not available in the suburbs.
One means to this end might be shorter trains or, in the very long run, electrification of commuter runs. The Trinity Railway Express between Dallas and Fort Worth uses old Budd RDCs for some off-peak trips, which are more efficient for transporting smaller numbers of passengers. With locomotive-hauled services (or motor-hauled electric services) there is little incentive to shorten trains, as uncoupling a few cars decreases efficiency very little. (This is why off-peak trains in Boston, for example, often operate full-length trains with only two or three cars open.) With DMUs or EMUs, there is a significant energy-use savings with shorter trains.
In the long run, the plan is to run service from Minneapolis to Saint Cloud. While criticisms of the first phase of this project may be apt, it will make much more sense if there are trains running to the proper end of the line there. It would be more of a cross between intercity rail and a commuter service, and would probably necessitate midday and evening trips (these could be run every three hours in each direction with only one train set and crew). Saint Cloud is home to 60,000 people, has a local bus system, and, possibly most importantly, has a 20,000 student state university, with most of the students undergrads, and many from the Twin Cities. The campus is about 1/2 mile from the potential station site, adjacent to downtown, and bus lines currently connect the two.
So while the first phase is probably not cost effective, the overall project—with passenger rail service between two cities' downtowns at highway-competitive speeds—may be quite a bit more useful to quite a few more riders.