I think the real thing with the nonstop is that it represents the first increase in peak-hour frequency and capacity in a long time.
I also think that if you were thinking about this as "premium runs are supposed to make premium profits" then you are thinking about this wrong.
Think just for a moment: this is a utilization run with resultingly low new revenue expectations.
Do we really know why it is a Nonstop? Could be any mix of:
1) MAYBE It was good for PR, but not meant to be sustainable. They could make some news while waiting for the Avelias to arrive and could generally get people talking and thinking about the speed and convenience of the train, for which even ongoing operating losses would be made up by marketing value for the whole service (besides, few would ride the Nonstop both ways...any new pax on the Nonstop would probably run a regular Acela on the other leg)
2) MAYBE It was a utilization run paid for by squeezing two extra frequencies out of the same aged fleet. By scheduling when they did, they got themselves a "free trainset" at peak times. As such the train did NOT need to win normal load factors in order to be an economic win--they just needed to pay for the extra electricity and grease, not the extra rolling stock ownership costs (besides, the Acelas are reverting to the lessor...why not run them extra hard knowing that a replacment is coming)
3) MAYBE Departure and Arrival were dictated by peak-time platform slots at NYP and WAS--it wasn't that the trip needed to be fast, it was that it needed to be at its platforms at certain available times at either end. Similar to the utilization run theory, at (2) above, the real story here is that the nonstop does a better job of sqeezing marginal new fare revenues out of a very fixed asset set.
4) MAYBE They actually thought skipping PHL and BAL was worth it to NYC and WAS area patrons. I personally find this hard to believe. Given how big PHL is as a market, and how the train has to crawl through WIL and BAL anyway, it is hard to argue that the costs of stopping exceeded marginal revenue from selling seats that are OBVIOUSLY going unsold.
5) MAYBE it was done to hurt the DL and AA shuttles-to encourage Delta and American to continue reducing capacity in the market, by biting them at what'd otherwise be peak flight times. Given that the train is at something like 3/4 market share, one could easily believe that Acela wooed 100 customers off the airlines--easily enough to tip some AA and DL flights into losses. If it means that either down-gauges a couple of planes or drops a frequency, it could be well worth it from a market-share and pricing power standpoint.
But, I would like readers here to consider that Reasons 1, 2, & 3 may have been enough to make the Nonstop a win on PR & utilization alone, and so they may be (a) real reasons for calling half-full trains a success and (b) lots of tolerance for not needing to add back PHL, WIL, or BAL.
Personally, I do think they should add back PHL--a onestop is probably closer to optimal, unless only the most precise scheduling at 2 & 3 above can make the train and slots available.
"Trying to solve congestion by making roadways wider is like trying to solve obesity by buying bigger pants."--Charles Marohn