Crowding Is a Symptom of What Ails the Subways, Not a Cause

Rising ridership is nothing new and today's passenger volumes were entirely foreseeable. The system could have kept pace if political leaders had prioritized core subway capacity.

Gov. Cuomo. Photo: Governor’s Office/Flickr
Gov. Cuomo. Photo: Governor’s Office/Flickr

The Times came out with a piece on subway delays this morning that’s getting heavy play on Twitter. It has some compelling visualizations of the rise in ridership and decline in reliability, but it starts off by framing poor service in a way that obscures the root of the subway’s troubles.

Here are the problematic paragraphs:

Is the main culprit behind delays the aging equipment in the 112-year-old system — signals and tracks taxed beyond their limits and patched together to eke out a few more creaky years, only to break down again and again?

That is certainly an enormous problem, but it is not the No. 1 reason.

The major cause of subway delays is a factor that basically did not exist 15 years ago: overcrowding. The subway is a victim of its own success and the city’s resurgence. Large crowds slow down trains, which creates more crowding in a vicious circle that takes hours to unwind during every rush.

This is a false dichotomy. Crowding and “aging equipment” are inextricably linked. The subway system wouldn’t get overloaded by today’s passenger volumes if an upgraded signal system, allowing trains to run more frequently during peak hours, were in place. Equipment-related delays can also lead to crowding problems, as more passengers file onto platforms, waiting for trains that have fallen behind schedule. It’s not clear how many crowding delays originate from infrastructure failures.

The importance of modernizing the signal system eventually gets clarified farther down in the Times piece, but a lot of social media takes are running with the MTA-as-victim-of-its-own-success narrative.

It’s more accurate — and more helpful in resolving the current crisis — to frame the situation subway riders now face as a result of poor management and political shortsightedness.

Today’s problems were very foreseeable. The upward trend in subway ridership started decades ago, and the MTA began preparing to upgrade its relic of a signal system in the 1990s. As big and complex as the subway network may be, this should have been enough time to get out ahead of rising ridership and computerize signals on more than just the L line by now.

Instead, a succession of governors, including, for the past six years, Andrew Cuomo, have thrown their weight behind extravagant stations, insanely expensive expansion projects, and other monuments to themselves.

Riders have been victimized not by the success of the subway system but by Cuomo’s failure to steward the transit infrastructure that New York is built on.

  • Where is Andy Cuomo?

    This really speaks to a problem within the MTA — customers are viewed as a problem and not the whole reason for their existence. Even if you do take at face value that crowding is the biggest problem the agency faces, it doesn’t account for the many delays during off peak hours when travel time is the least reliable.

  • Larry Littlefield

    “The upward trend in subway ridership started decades ago, and the MTA began preparing to upgrade its relic of a signal system in the 1990s.”

    The Board of Transportation started upgrading the signal system in the mid-1950s, just as the last of the original system (a few small extensions aside) was built. Starting with the replacement of the original IRT signals, and proceeding from there at a 60 year lifespan pace for decades.

    But with a couple of stalls, on in the 1970s, and perhaps one recently. And as a result of those stalls, virtually the whole IND is over 75 years old now, described by engineers as the drop dead date.

    Moreover the cost has exploded. Given what Local 3 IBEW, Railworks (the only old time signal installer — a monopoly that uses NYCT profits to fund the rest of the firm) and Siemens charge, replacing the signals at the current prices would bankrupt not only the MTA but also the city and state. The boom in private construction compounds the problem. The private jobs always get the electricians when they are scarce. The public sector will always wait and pay up.

    As for many things that are degraded by Generation Greed, a huge leap to a much cheaper future is possible alternative to ongoing decay. That means leaving the entire relay-based signal system behind and living with the consequences (can’t get good relays anymore anyway. NYCT installed some manufactured in Rochester by Alston in the early 2000s and they failed unsafe).

    I’m told by an industry insider that once 5G comes along (whatever that is) ground-based phone, internet and TV will become completely obsolete, which is why Verizon is trying to get out of installing more FIOS. So a lot of those people installing and running fiber-optic networks are going to lose their current jobs and become available. Big opportunity.

    On top of that you have all those people working on driverless cars, a far more complex challenge than signal driven trains in a fixed-space with known parameters.

    Bring in those people, add the signal engineers/maintainers who know about interlockings, bring the whole thing in house and replace all the signaling over five years for a copule of $billion, and maintain it for less that the signals present now. Maybe maintain a block system with each leg between stations as a separate block for times the computerized system needs to be “rebooted.”

    Either that or set aside the money to upgrade the signals as described, then have the legislature pass a 20/50 pension sucking up all the money set aside for improvement, the way they did with the schools.

  • skelter weeks

    I wonder if the transit infrastructure problems tormenting riders in NYC and Washington DC is the result of a ‘brain drain’ caused by massive retirements of knowledgeable personnel. Key employees may have found out the best way to fix equipment and keep it maintained, but didn’t pass this information on, so when they leave, ba-boommmm. Just a theory. These problems aren’t everywhere, just a few places. Why?

  • Larry Littlefield

    The transit system is melting down in Boston too. It’s pretty much everywhere.

  • rubagreta

    As for Cuomo, it looks like the new Tappan Zee Bridge will be named the Mario Cuomo Bridge.

    Not the Mario Cuomo Tappan Zee Bridge (the official name now is the Malcolm Wilson Tappan Zee Bridge, but most people don’t know that and nobody uses Malcolm Wilson). Just the Mario Cuomo Bridge.

    Nothing against Mario Cuomo, but I bet the vast majority of users would prefer keeping Tappan Zee Bridge.

    So along with monuments to himself, there is a monument to his father.

  • rubagreta

    “As big and complex as the subway network may be, this should have been enough time to get out ahead of rising ridership and computerize signals on more than just the L line by now.”

    A chart in today’s New York Times shows that the L line is the only one that runs on time. And I believe prior to its updating, it was the very worst line.

  • Anonymous

    The good performance of the L is great evidence that crowding is not a cause, but a symptom. Crowding on the L is as bad as anywhere, and yet it continues to perform well.

    Because computers run those trains, the major problems that plague the rest of the system are avoided:

    1) Variability in operator performance is not a factor (with the wave of retirements, this is especially problematic on the other lines).

    2) Fewer signal failures (because the system uses CBTC and the traditional block signaling system is only a backup).

    3) Tying the above two reasons together, this is the one that’s really causing the death spiral: on all lines except the L, inexperienced operators are bumping up against a signal system that’s being artificially slowed down in the name of “safety” (with CBTC there’s nothing to mess with – no timers, cars are allowed to run at full speed, etc. There are ways to ruin it, but they require programmers, not signal maintainers).

  • Flakker

    Again this article bears repeating. Poor technical decisions were made but the bottom line right now is that the agency is incompetent and doesn’t prioritize speed in any case.

  • Cuomo is a disgrace. On this matter he is as bad as Trump with his fake news and hiding the truth: He has siphoned off funds that belonged to the MTA to build roads and bridges for car drivers .

  • sbauman

    Starting with the replacement of the original IRT signals, and proceeding from there at a 60 year lifespan pace for decades….Given what Local 3 IBEW, Railworks (the only old time signal installer — a monopoly that uses NYCT profits to fund the rest of the firm) and Siemens charge, replacing the signals at the current prices would bankrupt not only the MTA but also the city and state.

    60 years is 3 to 5 times too long for the lifespan of new signal infrastructure. After 10 years, replacement parts will become increasingly difficult to obtain. The reason for this short lifespan is that new equipment is far less expensive. That 10 year replacement clock began before most systems were installed.

    This has not been the case for railroad signal equipment because they have ignored technology, since the transistor’s invention. They have developed equipment design equipment criteria that are outside the practices for other applications.

    Maybe maintain a block system with each leg between stations as a separate block for times the computerized system needs to be “rebooted.”

    There are two models for computer systems. One is centralized control, the other is the automata model, where each unit is autonomous and connects with its near neighbors. The two are equivalent. The two are functionally equivalent in what they can do. The translation from automata to centralized control and vice-versa is a simple algorithm.

    The device you are describing is a programmable logic controller or PLC. They have been around since the late 1970’s. It’s what automated the factory floor. It’s got watchdog timers for failure detection. Newer enhancements have reduced the conductor count between units. That link can be a single ethernet link or a radio link. The modules are dirt cheap. Something like $1M should replace all the relays in the MTA’s arsenal with plenty of spares.

    The real beauty of PLC’s is that no programming is required. Their input is the ladder logic diagram used to specify the current relay logic bays.

  • Larry Littlefield

    At the 60 year rate and current costs, the subway requires a new $400 million project to start every year.

    At a ten year pace, that’s six $400 million projects.

    As I said to the head of the signal program 15 years ago — here is the total income — wages, benefits, everything — of everyone living in New York City. We can’t afford the signal program at these costs.

    Now perhaps that program has stalled.

    When I was there the Dept. of Subways was afraid of solid state modules failing, and wanted relay redundancy.

  • Hux

    As many of us know, “crowding” delays are really just what’s used when trains are slow. Maybe they are crowded because of an incident or a slow crew. Often, the trains aren’t even crowded – they’re just slow because of all the long-term changes to the subways over the years… flagging rules, signal modifications, etc. We have more crews because we need more trains due to all the slowness, and the newer crews are being trained to go slower.

    So why use crowding? Well, it’s a nice way to blame the problem on the public. It looks better than saying “unknown”. And the union wouldn’t want dispatchers to tell crews to go faster – dispatchers themselves are in the union too. Plus, how do you prove the train is delayed because of a slow crew or something? How do you incentivize crews to go faster? If they’re late, it’s often not their fault. And they make more money with overtime if they’re delayed…

    Delays do cause crowding, but crowding itself is not the cause of the vast majority of “overcrowding” delays. For instance, a car without AC will cause crowding in other cars, causing delays. But crowding is not the cause of the delay. And ridership is not the same as crowding – more riders will lead to more delays (more sick customers, police incidents, etc.), but you can have more crowding and fewer riders, too, if service is really bad.

    So how do we stop all this crowding? Stop the system from slowing down! Delays cost billions in economic losses for New Yorkers. But our leaders don’t seem to know what’s going on underground. Maybe because they don’t know about operating trains. They also don’t even ride the trains.

  • Hux

    Yes. Our number of actual incidents is relatively stable. The problem is that there are more delays for each of them now, because the system is slower and less resilient. Over 80 percent of our “crowding” delays are on the off-peak, on lines that are not at capacity. Clearly the vast majority of these delays are not due to crowding… the system is slowing down. The trains run slower. They enter stations slower. Crews are more cautious. Rules are tightened and signals are slowing service. System is a lot less resilient. Leadership does not understand and/or does not want public/politicians to blame them.

    We are not going to solve this problem with platform controllers, step aside boxes, revised door announcements, nurses, CWR installations, ultrasonic testing and inspections, response personnel, emergency dispatch vehicles, flexing the headways, coordinating planned work, help points, WiFi, track cleaning… these bigger issues seem to be ignored.

    We also won’t solve it by adding running times to schedules or telling crews to just “go faster”. There are no rules against slow crews and it pays to go slow, since they will never get in trouble. Many crews are often given skips, simply because they are always slow. Union rules make it hard to keep the slow crews out of rush hour, off of busy lines, etc. It is all about seniority.

    So we’re left with snowballing delays, especially due to all the interlining. Trains are so slow now that we are running out of extra trains, we can’t even get gap trains anymore.

  • Hux

    Yes. We were actually more crowded during the peak in the 1970s, since there was less service and work hours were less flexible, people did not want to ride off-peak due to fears of safety, etc. So also, ridership is not the same thing as crowding. We had a lot fewer riders in the 70s, but we were more crowded in the peak. Also, ridership has declined this year, but “overcrowding” delays continue to rise… so just looking at that, couldn’t someone see that ridership is not to blame for most delays? Yes, ridership does have to do with delays — more customers means more sick passengers, etc. But most delays are not due to overcrowding. And most of the increase in ridership is in reverse directions or off-peak, so it is not really impacting crowding too much.

    The L line has decent OTP, even though everyone knows it is one of the most crowded lines. This is because the L has CBTC, the trains run automatically, so there are no slow crews or signal modifications on the older signals that are slowing down service. We also have better data, every year the data quality increases with new forms of electronic data. I think around 20% of the increase in delays is simply due to more delays being captured. In the 70s, our system probably was a lot worse than now, but data quality was so poor, it is difficult to compare.

    There are two sets of books. The first is recorded directly at the RCC, and it is a log of incidents. From that book, delays are attributed to incidents and the cause must be determined. Often the overcrowding “incidents” in this book are not really overcrowding; they don’t know what is causing the problem.

    The second set of books is recorded at the terminal by the dispatcher. They record late trains electronically (it used to be paper sheets, called into the RCC every 2 hours, so of course this transition caused a big increase in recorded delays). Sometimes, they do not include notes or reasons, and it is the job of analysts to put a reason for each one of these late trains. Often, they will put overcrowding. For ones that they have absolutely no idea about, they will take the proportion of the ones that they logged, and ration those reasons to the other delays. (So 20% ROW failures, 30% overcrowding, etc.) These are the numbers that get reported publicly, not the incident numbers.

    Either way, it’s FAKE NEWS.


The Boom in Subway Ridership Is Waning. Why?

Transit officials recently reported that 1,763,000,000 subway trips were taken last year, the most since 1948. But the rise in ridership was meager, with only 12 million more trips in 2015 than in 2014. The percentage growth rate was seven-tenths of one percent. Over the same year, employment in New York City rose three times […]

Straphangers: Ancient Train Signals a Prime Culprit of Subway Delays

Has your subway been delayed recently? Blame New York City’s aging transit infrastructure, especially its outdated signal system. Then start fighting to make sure Albany fully funds the MTA’s next capital plan. A new report from the Straphangers Campaign shows just how prevalent signals failures are on the subway system. In 2011, the MTA sent […]