The only thing that kept passenger service going as long as it did on private RRs was mail contracts (does that sound like a subsidy?) and treatment as an advertising expense (on certain railroads). Nobody has done any better than break even "above the rail" since cars became popular in the 1930s-40s, and the economics of long-distance trains evaporated when WW2 ended and subsidized airlines (which they very much were until the 1980s) took over. Nobody has made money on passenger trains, period, since before the 1930s if capital costs even for just the equipment are included.

If passenger trains are to be treated as basic accommodation, then they fall into the category of transit service. THAT is all subsidized too - a really good transit operation recovers 40-50% of its basic operating cost (no capital) from fares, and can't do better than that as long as they are subject to the need to compete with subsidized cars (yes, the roads ARE subsidized as long as General Fund money is spent on them, which is common since nobody wants to raise gas taxes or otherwise arrange things so cars & trucks pay their own way) and to meet various social justice etc. requirements.

Many "corridor" trains use state-owned equipment, and more is being purchased (yes, with some federal money - it's capital not operating); the states pay for rail upgrades needed for passenger service, usually with only a modest federal share; and the cost recovery for those trains (operated by Amtrak under contract in most cases) is usually in the 40-50% range with occasional forays closer to 60. Some CA service has flirted with break-even on occasion. It's also targeted at moderate-distance corridors and is designed with lots of bus connections that keep the trains full. Interestingly, a lot of Amtrak long distance services (not all by any means) are in that cost recovery area too - they're just not the NEC where there has been so much investment that they can run enough trains to be a significant factor in the transportation businesss (1 or 2 trains a day doesn't matter -- 8-10-12 a day provides a real transportation alternative, which is what 2 of the 3 CA corridors provide and the 3rd comes close. NEC is almost like a subway line - trains every hour or less (usually 1/2 hour) and usually near-full.

I don't like wasting money on service that doesn't matter. I'm willing to spend money on service that does (as a taxpayer). So I would agree with your point as long as a train is once (maybe twice) a day or less - cover the coach cost as a public service and contract with somebody like Iowa Pacific for first class (rent 'em the cars and charge for the extra locomotive unless they want to bring their own like some states do). But given the length of most train runs you still need some kind of food service for steerage - and I don't think vending machine cars (yes, I did ride the Daylight with one of those back in college days) cut it. If the service frequence and capacity is more than that, it's attempting at least to be meaningful and should be funded at a higher level (and be subject to cost recovery criteria) consistent with transit service.

Just don't be surprised if that basic coach cost is a very considerable fraction of what the train costs now. Simply showing up costs most of the money because all the overhead (staff, reservation system, maintenance of equipment, lawyers to talk with the railroads, etc.) has to be covered even if the train only has 1 car. That's why adding service (at some added cost) often results in better cost recovery than just trying to squeeze cost out of the current business.