That time a company proposed fast trains to New York - via elevated tracks in the Back Bay

It's also a function of the Shoreline being where all the population is. From 1900 when fragmented NY-BOS railroad ownership became consolidated under one roof until about 1955 there used to be four NY-BOS routes you could do between Grand Central Terminal and South Station from a single ticketing window on the New York, New Haven & Boston RR.

-- Shoreline Route via New Haven, New London, Providence (i.e. today's Acela + Northeast Regionals)
-- Inland Route via New Haven, Hartford, Springfield, Worcester (i.e. Amtrak Inland Service through 2004, and proposed upgraded revival being studied right now)
-- Midland Route via New Haven/Hartford, then east to Willimantic and Providence (i.e. longshot Alternative for the NEC FUTURE super-duper federal HSR studies)
-- Air Line Route via New Haven, Middletown, Putnam, and today's Franklin Line.

None were orders of magnitude faster than the others, but there was a decent spread of travel times and schedules were optimized accordingly local vs. skip-stop to accentuate the 4 routes' specialties. Each route had enough patronage going for it to last postwar into the Interstate highway era until the NYNH&H's bankrupt finances forced brutal cuts. Where the Shoreline had curves and congestion the Inland traded better straightness overall for more roundabout routing through Springfield. The Air Line was very straight, but climbed some very steep grades. And the Midland sort of split the difference between the Inland and Air on grades, curves, and straightness.

Technically the Air Line took the prize on the clock for fastest trip, and the Shoreline was--by slight margin, mostly owing to higher number of local stops--the slowest on the clock. But guess which ones were most and least popular of the four: Shoreline the #1 hands-down most popular, and Air Line dragging up the rear. Because the Shoreline stayed almost entirely inside the density wall of the very oldest swath of East Coast megalopolis along U.S. 1 from end-to-end, while the Air Line was almost entirely rural hamlets and skipped every single population center between New Haven and Boston to spend its longest stretch in CT's quiet corner. Second-most popular?...the Inland because of the Hartford-Springfield-Worcester triple-whammy. Next-to-last...the Midland, which had Hartford + Providence going for it but also an even higher cows-to-humans ratio than the Air Line all points in-between through the CT quiet corner.

So unfortunately NY-BOS high-speed rail is always going to have to be some sort of less-than-pristine 'mongrel' construction through a whole lot of wall-to-wall density that has little give for dedicated rights-of-way. The NEC FUTURE commission found this out the hard way as their bypass-a-thon studies started getting chopped up bit by bit. The Westchester County, NY through Danbury/Waterbury, CT HSR bypass had great performance going for it, but you simply could never skip Stamford, Bridgeport, and New Haven on an Acela train and still end up serving the top-tier business travel market that makes the Acela net-profitable. Ditto their asinine scheme to skip Wilmington, DE--the 11th busiest Amtrak station in the entire country--for a straighter alignment through a swamp miles outside of the city. These centuries built-up CBD's that take curves and lots of commuter rail congestion to reach are exactly where the people most need to go, and unfortunately we learned that fact in olden times where the fastest-but-most-rural route had the poorest ridership vs. the slowest one that stuck closest to the unbroken density.

We pretty much know what we are in New England demographically, and that's an unbroken megalopolis of constant demand between NY and BOS with a lot of spikes in the CBD's of the secondary cities. The demand calculus really doesn't end up much different on probable HSR routes post-2020 vs. what we learned from ticket receipts when their 1920 equivalents were still running. So unlike California HSR where the build-up of cities was a lot more historically recent and less-chaotic than the East Coast and thus easier to link with fresh rights-of-way built to perfect geometry...here demand vs. feasible alignments are a lot messier and less satisfying.

But ultimately HSR is about moving people, so compromise has to break along lines of where the people most need to go. Where people most need to go long-distance on an Acela tends to map 1:1 where they most need to go short-distance on a Metro-North train, so there doesn't end up being a whole lot of "Eureka!" moments where bypasses can innovate our way to higher speeds without the downside of unmet demand by skipped CBD's (the Stamfords, Bridgeports, etc.) dimming the upside too much. That means, like it or not, doing business with the congested New Haven Line and troubleshooting at least 90% of it where it lays instead of hoping against hope for miracle new alignments to drop in our laps. NEC FUTURE's already had cold water poured over all its new alignments that miss any major Acela-schedule CBD station catchments, and is having its other construction-disruptive ones that tear gashes through wall-to-wall suburban density stacked against wall-to-wall shoreline wetlands be torn to shreds one-by-one through community opposition (much of it legit oppo and not typical NIMBY whining because of the sheer clumsiness of the fed commission's impact studies).

At the end of the day, the question is going to be "What's the best state-of-the New Haven Line we could possibly run on?" if all state-of-repair were achieved and all inefficiency pounded out of commuter operations. Then figuring out a balance of ramping up the speeds where the people ain't to bank the schedule savings for the slower stretches of built-up areas where the people are. For example...make the arrow-straight stretch through the Jersey swamp an unbroken 165 MPH between Newark and Trenton to bank the maximum schedule savings for the meander from Stamford to Bridgeport to New Haven. Then zero in on the pu-pu platter of bypass options for CT's quiet corner to try to ramp the speeds back up between New Haven and Providence without skewing too far into Air Line territory on missed catchments.

Design perfection?...hardly. But plotting it that way is as close as you're ever realistically going to get to on demand-serving perfection.

Amtrak adds about half an hour of schedule padding between New York and Boston. That says something about something.

They don't need to boost top speeds. They just need to tighten up the timekeeping.