While the CBTC already existed, I worked on the ultra-wideband positioning system originally demoed on the 7 line (and spent a lot of sleepless nights riding empty 7 line trains, testing out the system).

I wrote the original version of the system that would initiate the time-of-flight signals to the trackside UWB beacons and use multilateration to determine the train position (of both extreme ends of the full train) based on knowing the ECEF coordinates of those beacons, all written in C (with MISRA C restrictions) on what were basically Raspberry Pi compute modules.

Also wrote the app you can see a screenshot of at the top of this post:

https://www.masstransitmag.com/rail/railroad-signals-ptc-con...

That app was C++/QML/Qt-based, using OpenStreetMaps to visualize the position. Originally written just as a visualization tool to consume data that was part of a test/mockup system I made while writing the code (mostly in San Diego, so very far from the NYC tracks we did final testing on), but the visualization of how it found the train position based on time-of-flight distance calculations from the UWB beacons looked pretty cool so they ended up putting it up (receiving data from the real, live system) on monitors while doing the media demos back in early 2020.

I didn't see a published date, but taking another look, I suppose that "Copyright © 1995-2012 nycsubway.org." is a giveaway that some of this information might be stale.

Doesn't feel much like it day-to-day in the NYC subway, but glad things are indeed continuing to change

I think I recall reading that this "on sight" operation continued for quite a while at least on some of the elevated track including the downtown loop where most of the lines converge. I would assume a few bad accidents prompted the upgrades to automated signaling and stop trips.