Death by 1000 Cuts: The NYCT Subway Slowdown

 Starting in the 1990's, the New York City Transit Authority (NYCTA) started a process to slow down the largest subway system in North America, ostensibly in the name of safety. Over the next two decades the process, conducted slowly and out of public view, went from costing riders a few minutes here and there to triggering a full on capacity meltdown as the system, despite its decreased performance, benefited from record ridership. Transit Twitter and Blog personality Uday Schultz has recently completed an exhaustive history of the great slowdown and the science of transit speed control in general. It's a great read and starts with a zero based explanation of the NTCYA's trip-stop and timer based ABS signaling system up through the events that triggered the management action and the subsequent slide into dysfunction. Still, while this piece does a great job explaining why, it comes up a bit short explaining "why". So lets dive in a bit.

One shot GT timer signals added to CANAL ST interlocking before re-signaling.

In the early 2000's everyone in the NYC Subway fan community was aware of the performance decreases and would track how the NYCTA seemed to seek out any location where trains could get moving and just find ways to throttle service back to a plod. Even in locations with no infrastructural changes the trains were operated with an appreciable lack of urgency. The community was full of theories as to why this slowdown was happening and, to a lesser degree, why nobody seemed to care. After all other cities, even those with traditional signaling systems like Philly, Chicago and Boston, found ways to achieve brisk acceleration and top speeds of 55-70mph, making the 25mph crawl of the NYC Subway a distinct outlier. 

SEPTA Broad Street Subway Express train @58mph.

As Uday's article covers (read it now to avoid spoilers), the speculation the early 2000's fan community was both right and wrong at the same time. They were right in that most of their theories were correct.  They were wrong in that there was no one reason that bore primary responsibility for the problem. The newer equipment, up through R68, did have slightly better performance than those the signal system was designed for. The new composite brake shoes did have slightly worse performance than the old iron shoes. The system did rely on train operator rules compliance and related management thereof to ensure safety. Then, between 1991 and 1995 all of these factors combined in varying degrees to cause four significant accidents, opening the NYCTA up to both liability and public pressure.

Inbound Williamsburg Bridge ramp with carlength long grade timer blocks.

The response was similarly multi-pronged from slowing rolling stock down in both acceleration and top speed (55 to 40mph), ubiquitous use of intermittent speed control devices, curtailment of restricted speed operation and harsh punishments for trip stop engagement. Much of this action plan was implemented over a period of 20 years so casual riders didn't really notice the decline in performance. The cherry on top was that the intermittent speed control devices were then allowed to drop below posted thresholds making operators wary of even trying to follow the posted speeds. This is what caused the opportunity to get a skilled operator and a "good run" to vanish over the course of the 2000's, especially as the pre-90's workforce that learned to run trains without speedometers, gradually retired. 

All of this background leads to the real question, why did ostensibly high level management decide that such a drastic decrease in performance was acceptable? This is important because in an age when getting the public to *want* to choose public over private transport, the performance of public transport is increasingly throttled by policy leaving private transport as the only option that can attempt to offer speed and convenience. Well, lets put on our 1991 hats and see what management may have been thinking.

1. The most salient factor was the long term plan to equip the NYC Subway with a full time ATC/ATO system, later realized with the selection of CBTC to replace the wayside signals, timers and trip stops. Investment in an end-of-life signal system would be wasteful and performance decreases could be argued as temporary.
   
   
2. After peaking at 2 billion annual riders in 1948, the shift to non-urban living and private transport dropped ridership by over half with the peak of NYC's crime wave coinciding with the subways trough of ridership. The system was running at half capacity so "slightly" increasing trip times was likely not seen to be a big deal.
   
   
3. Decades of disinvestment had caused the NYC Subway to fall into a prolonged state of bad repair. With limited funds compelled trading performance for safety.
   
   
4. The threat of continued accidents was a political liability while small overall changes in performance would be unlikely to generate much if any notice let alone political pushback.
   
   
5. The reliance on operator skill presented not only the continued risk of accident, but would also put up pressure on costs as said skilled workers had to be trained and retained. Uniform operations according to the speed control systems would make operators fungible and require a lower skill floor.
   

These 5 factors could be arranged multiple ways to create a compelling policy proposal to management. It would have taken an extraordinary amount of personal risk for any of the top officials to insist on maintaining performance standards when CBTC was right around the corner anyway. I think the decisive element was the NYC Subway consistently running at half capacity for over two decades. It's not even that signal system capacity *could* be reduced with little impact, but that the 900 million annual ridership seemed to be both a floor and generally baked into the city. If local New Yorkers were willing to risk their life to ride the system, an extra few minutes wasn't likely to deter them either. To some extent management was proven correct, their slowness campaign only became a problem after ridership doubled over the following 20 years.

If you want a take away its that reducing performance has, is and will be the go-to fix for even rare safety problems. We've seen this with PTC and we've seen this on other transit systems like SEPTA and WMATA. The changes are rarely publicized and the public rarely objects even as they unconsciously sour on rail transit and make the switch to private vehicles. After the service meltdown NYCTA did set up a speed improvement task force that has been fixing the mis-calibrated timers and raising speeds that were subjected to overly conservative calculations. Still, while the rollout of CBTC has allowed for increased performance profiles, it would be interesting to calculate if they match what was achievable by human operators working under a system with a slightly greater tolerance for risk.

PS: An interesting comparison can be had with how the UK responded to the Ladbrook Grove Rail Crash of 1999. The crash program to install TPWS at select locations was similar to NYCT's system-wide modifications, however as far as I can tell, it had minimal impact on train operation, perhaps in part of the UK's continued reliance of train drivers' compliance with rules to ensure safety, as opposed to technical mechanisms.