Maybe I'm misunderstanding, but as far as I've gathered, the main source of unreliability is old and shitty signaling systems. Found it odd that signaling wasn't mentioned.
Japan Chuou line mentioned in the article is still on the old ATS-P system and it runs 28 train per hour. Even older ATS-S system in Osaka runs 16 train per hour on the inner track of the Tokaido line.
The NY Subway made a big fuss of CBTC and that it would allow them to run more train more reliably than their current old ATS system, which I find funny looking at various line in Japan.
In fact, CBTC in Japan is mostly planned for rural line with infrequent service so the amount of infrastructure (track circuit, etc) are reduced
The current signals being replaced by CBTC date from the 1930s, and are mechanically operated. They are literally historical relics. https://www.youtube.com/watch?v=Mjx3S3UjmnA
London runs 36 TPH on the Victoria Line: https://www.londonreconnections.com/2017/ninety-second-railw...
I just checked Line 14 schedule and the most I can find is 34 train per hour, which is approx 105 second headway.
I am not saying that CBTC isn't capable of running higher interval than regular ATS and wayside signal (which is actually a very old technology), but in most case it doesn't matter. Even without moving block of CBTC, a very short static block (say, 100m) and in-cap signally should theoretically have the came capacity as CBTC.
Meanwhile in Germany: a quarter [1] of the network's signal controller stations still runs on levers, wires and semaphore signals - some of these are from the 19th century and I wish I were joking. [2]
[1] https://www.zeit.de/2019/45/deutsches-schienennetz-signalwer...
In Germany, most freight trains run at night because at daytime many tracks are at capacity with passenger trains.
[1] https://www.sfgate.com/opinion/article/Amtrak-has-priority-o...
[2] https://specialty-freight.com/2021/06/us-requires-railroads-...
That doesn't generally apply to the Northeast Corridor though which I assume was being referenced by the mention of the MTA. That said, while it's certainly imperfect, Amtrak is actually pretty reasonable for the most part on that route. Certainly the epic delays that are fairly common on Amtrak long distance trains elsewhere are rare on that corridor outside of major infrastructure problems, such as power failures.
For light-rail transit, shared routes (LRV + automobiles) and traffic signals mean that arrivals are not predictable. Where shared-roadway and dedicated-route segments exist (say, with San Francisco's Muni Metro), hold times must be built in for the dedicated-route travel segments. Train arrivals are highly unpredictable. A level below Muni, on Bart, rush-hour headways through the San Francisco subway are 4 minutes. That's close to the theoretical maximum Alon gives of two minutes.
For conventional heavy rail, boarding / debarking delays, equipment failures, medical emergencies, police activity, level-grade collisions, and other factors, may all lead to service delays.
Signalling matters mostly in that it both doesn't introduce new delays into the system, and that it can compensate for delays originating elsewhere.
Reliability overall, as discussed in the article, is largely based around signalling to manage flows.
That should probably be doable on some american light rail systems, with signal priority and avoiding street running on roads with a lot of traffic congestion.
BART operates up to 80 mph (128 kph), though I'm not sure it hits those speeds within San Francisco.
> though I'm not sure it hits those speeds within San Francisco.
due to the shorter station spacing I'd assume that, too.
As for the stations, the limiting factor in turn consists of the sum of dwell time (wheel stop to wheel start) and platform reoccupation time (governed by the signalling system).
Arguably fixing that by building a fly-under (a tunnel under the existing tracks) fixed a huge source of problems for that part of the UK rail network, probably more so than the signalling.