> It's also home to one of the biggest transportation deserts in the city, which will be functionally unlocked in the next five years or so once self-driving cars are a thing.
Self-driving cars don't magically make other cars, self-driving or not, vanish from the roads and they certainly can't match the raw throughput of moving thousands of people in a single train on a dedicated right of way.
> Self-driving cars don't magically make other self-driving cars vanish
I mean that's what they're supposed to do, if they can all drive one behind another at a constant speed without crashing or creating traffic jams. But more importantly, if you can be doing work while your car is driving itself then your effective commute is zero. And if your car can park itself in the suburbs, then that creates yet more space in the city that can be used to create housing. Self-driving cars, if they actually work, should radically reduce housing costs -- both by making formerly unlivable places suitable for residential buildings, and by freeing up significant amounts of land.
> if you can be doing work while your car is driving itself then your effective commute is zero
That’s a big if. I’ve commuted in an Uber pool many times and hardly ever see any of the other passengers working. Most of them seem to be going to/from the office like me.
Working on a rail commute seems more common. The smoother ride lends itself better to laptop work.
Another big if here is that your company considers that time to be on the clock. A commute that takes away family time without it counting towards your working time is still a commute. Getting more work done =/= not having a commute.
My point being that they don't make other cars vanish and they certainly don't make human-driven cars vanish in any sensible timeframe.
On a person-kilometre-minute basis, cars are absurdly inefficient. Stupendously so.
A typical subway train typically runs to about 157 metres. A 2018 Camry runs to about 5 metres. That gives about 31 car lengths for the subway car.
Assume that every Camry is fully loaded with 5 passengers. That's 155 people. In the ideal condition, driving bumper-to-bumper with no other cars, 155 people in 157 metres -- about a person per linear metre.
Here's the fun bit: the passenger capacity of an R142A carriage is 176.
Not the whole train.
One subway car.
One subway car, approximately 16 metres long, holds more people than an entire train's length of self-driving Camrys driving with zero tolerance.
Now add the fact that the trains have dedicated rights of way and note that even with the profound multi-decadal mess made of the MTA you simply cannot replace the subway with any other mode of transportation, not now, not ever. Self-driving cars don't repeal the laws of physics and they don't override Little's Law either.
> On a person-kilometre-minute basis, cars are absurdly inefficient. Stupendously so.
The problem is the subway comes maybe once every 11 minutes when it's working properly, and if someone tries to kill themselves by jumping in front of a train in south Brooklyn (or trips or gets pushed) then trains stop working all the way up through The Bronx. And potentially on both the east and west sides of both Manhattan and The Bronx if it happens at one of the stations where the 2/3 and 4/5 overlap. And this isn't an unusual occurrence, it happens once or twice per week during rush hour alone. Every time some minor thing goes wrong ('sick passenger', signal malfunction, track fire, etc.) then the entire city's transportation basically gets shut down. I agree that cars are absurdly inefficient, but at least they're somewhat more resilient. If we could use self-driving technology to at the very least increase the bus coverage then that would be a pretty good outcome.
You may have noticed that traffic jams occurs in NYC at rush hour. There are accidents, there are road closures, there are UN and presidential visits, there are special events, sometimes steam pipes blow up or a truck breaks down or it's garbage day and they're moving slowly or ...
The problems of the MTA aren't solved by switching to something that won't fix the problems of the MTA. They're solved by fixing the MTA. And even its currently-degraded state it can't be replaced by any other mode at rush hour. There is literally no physical way to fit enough vehicles on the road to replace even shitty subway service.
> On a person-kilometre-minute basis, cars are absurdly inefficient. Stupendously so.
You have forgotten about the human actually getting to the subway station in the first place. You need to include the distance from every person to the subway station they want in your length calculation.
You also need to include how long it takes the person to get there in your minute calculations.
The calculations work the same way. A sidewalk can hold far more people in the same space than a road full of cars can, no matter how intelligent the cars.
Your point is an argument in favour of extending subway coverage and doesn't change the basic physics of how cars and trains utilise space.
Yes, rail offers unparalleled efficiency at transporting people from one place where they don't want to be to another place where they don't want to be.
That's pedantic nonsense. If I took that argument to its logical conclusion, I'd point out that I also don't want to be in my garage or the city car park.
If you don't want to be in place A or place B, how you get from A to B won't change things.
For my own self, everywhere I want to go is served by the subway system. Often poorly, but that's a rip on the political history of the MTA, not on the nature of subways in general.
If you closed the subways, New York would seize up and die. In fact we're going to get a taste of what a minor arterial clot is like when the L shuts down next year.
If you don't want to be in place A or place B, how you get from A to B won't change things.
You do that by not going there in the first place.
A properly-designed network of self-driving cars would be a train. One that is always there when you need it, one that inherits the vast scale of existing infrastructure, and one that isn't vulnerable to paralysis through single points of failure.
But as the posts on here tend to emphasize over and over again, no one has the vision to make this happen. HN's love affair with fixed rail is nothing less than one of history's great irrational obsessions.
Public transportation is the one engineering discipline where solutions based on technology are "pedantic nonsense," while literally binding and constraining entire cities with hardware straight out of the Civil War era is considered the height of progress.
> A properly-designed network of self-driving cars would be a train.
No, it wouldn't, for the reason I describe. Cars occupy too much space per person by a factor of about 11x times on the figures I gave further up. Self-driving cars will be part of the mix but they physically cannot replace the subway.
But those 11x as many people are only there because that's where the train is.
Useful mental exercise: next time you're out of the house, imagine that every automobile parking spot, parking lot, and parking garage you see is actually a transit depot. That's what optimal engineering and regulation of self-driving cars would buy us. Of course, being Americans, I'm sure we'll try everything else first.
If you have an infinite spreading plane of cars moving all in the same direction at the same speed, and if nobody wants to join or leave this infinite plane of cars, and if none of the cars ever breaks down or runs out of energy, then this works perfectly. Back here on Earth it's not going to solve anything.
Indeed, cars trained on data gathered in California and Nevada are going to need a fair amount of tweaking to understand that pedestrians are a bit pushier hereabouts.
Self-driving cars don't magically make other cars, self-driving or not, vanish from the roads and they certainly can't match the raw throughput of moving thousands of people in a single train on a dedicated right of way.