,
19 tweets,
2 min read
Read on Twitter
*It actually is that the subway isn't crowded, rush hour crowding was higher in the 1970s and rush hour ridership peaked in ~1989.
In 1990, the subway had 900-ish million riders. In 2015, it had 1.7 billion. The entire difference was off-peak, esp. for non-work trips.
Amidst a near-doubling of subway ridership and hefty increases in commuter rail ridership, metro area mode share has barely risen.
The MSA went from 27% transit commute mode share in 2000 to 30% in the 2010s; subway ridership rose 1.3 -> 1.7 billion.
(And bus ridership was flat until the early 2010s; it's not about people abandoning the bus or anything like that.)
That the entire difference is off-peak suggests there's a real problem with frequency-ridership spirals, which amplify small shocks.
NYCT's rule: off-peak frequency is based on 1.25*seated capacity crowding at the most crowded point, or at worst every 10 minutes.
So a small increase in demand (say, because the middle class felt safer in the 1990s) leads to a positive frequency-ridership spiral.
A small decrease (say, because train reliability is decreasing) leads to a negative one, which is what we see today.
The 10 minute standard is weak. Here the U-Bahn comes every 5 minutes. In Paris the off-peak frequency is 3-6 minutes depending on line.
Moreover, German frequency guidelines are explicitly not based on trying to micromanage peak crowding. They're based on takts.
I haven't seen the timed transfer on the U-Bahn here in action yet, but I imagine I will soon. And the S-Bahn is complexly scheduled.
This is not me yelling into the darkness. Serious planners at NYCT understand that a complexly branched system needs regular schedules.
A document that got leaked to @danrivoli and me criticizes the schedule keeping (though not the frequency guidelines) on those grounds.
The London Underground (esp. subsurface) is a good model for NYCT to follow, as is the S-Bahn here, with its painful reverse-branching.
Also of note, both systems - London and Berlin - have very low peak-to-base ratios. The U-Bahn lines run every 4-5 minutes peak, 5 off-peak.
London's peak-to-base ratios are in the vicinity of 1.3, and Tube trunks get a train every ~3 minutes off-peak.
NYCT should aim at deinterlining, but meanwhile, it should guarantee a minimum of a train every 6 minutes on each branching route.
Instead of looking for excuses to run fewer trains, it should aim to run more. The planners get this. The political appointees don't. /end
