[Alon]

I think you should focus more on inner lines, and not on additional suburban extensions. The only people who you're going to get on the trains if everything just expresses to one or two Manhattan terminals without any inter-agency integration are peak-hour suburban commuters, and those will drive 10 kilometers to the better park-and-ride.

So, since we're talking long-term, here's what needs to happen:

1. Fares and timetables between the agencies should be integrated, including through-running whenever reasonably possible. Maybe for scheduling reasons it's not the best to have trains run through from New London to Newark, DE, but at least run through from New Haven to Trenton, from New London to New York, from New York to Philadelphia, etc. This means through-running even when it's claimed to be impossible now - we're talking about future rolling stock, and honestly even existing trains can be modified at much lower cost than it costs to build a Manhattan cavern.

1a. Integrated fares and tickets include urban rail and the buses. Best industry practice is to let people ride from Oyster Bay to Carteret, involving a bus, a train, a connecting train, and another bus, on one ticket, with all connections timed. This includes the subway, too - i.e. there should be a unified Far Rockaway station with timed LIRR-subway transfers.

(By the way, most rail scheduling innovations that Americans think are impossible to adhere to happen every day thousands of times in Germany, Switzerland, and Japan.)

2. Signaling on the central segments should allow high capacity. Something like SACEM on the RER A is a good example: moving-block signaling in the congested core, standard fixed-block signaling on the branches. Given that Penn Station has more than just one track per direction funneling into the NJT tunnel, 30-32 tph should be achievable.

3. All trains should be EMUs, and future orders should also reduce the weight. FRA regulations have no relationship to actual safety needs and can and should be ignored in favor of more modern rules.

4. Either subsequent orders should allow multi-voltage trains, so that there's rolling stock for any reasonable through-running, or everything should be reelectrified at 25 kV 60 Hz (and the NEC definitely should be so reelectrified, for the benefit of intercity trains).

5. Major suburban stations should not be park-and-ride hell, but instead there should be TOD within a pleasant walking distance. Hicksville is quite close to a large mall, but both the station and the mall are surrounded by parking, and so pretty much nobody takes the LIRR into Hicksville, just from Hicksville.

6. Off-peak frequency should be at worst 10-15 minutes at the urban stations and 30 minutes on the branches. To reduce operating costs, conductors should be replaced with a light rail-style honor system, and off-peak trains could also be shorter.

7. Few trains should be stabled in New York itself. Land for yards is cheaper at the outer terminals of the lines.

8. Physical extensions should be based on the needs of inner suburb-to-inner suburb travel, and also on those of urban travel. Lower Montauk is useless without a connection to Manhattan; stick a fork in it. But frequent-stop through-service between Newark and Jamaica is more useful. Based on funding, additional lines should be built in Manhattan connecting stub-end terminals like Hoboken and Flatbush, and even Grand Central and St. George.

Note, by the way, that the extensions are last on the list. There's a reason for this: they're less important than the concept of running commuter rail more like longer-range rapid transit and less like a 1930s-era suburban shuttle to the CBD. If you have just the extensions, you get SEPTA today, rather than what SEPTA was trying to become in the early 1980s.