Right now the average car occupancy during the rush hour is ~1.5 persons per car. By increasing that to 3-4 people, you can easily double the effective road capacity. This will completely resolve the congestion in the majority of metro areas, even with induced traffic.
No car platooning, flock routing, car trains, or anything similarly exotic is needed. Just mild carpooling. People are not doing that now because it's a logistics nightmare that removes most advantages of owning a car.
I'm not buying that a significant number of people would switch from their individual car to a self-driving car pool. One of the main advantages of a Waymo over a Lyft to me is that there is no driver. If we now fill the car with random people I'd rather drive my own car.
It's also not clear to me why routing self-driving cars would be more efficient than what Lyft Line did which ended up super slow. The slowness want just the additional route but also other passengers getting in and out and in some cases waiting for them.
There's significant trouble with routing, delays, and reliability: A self driving car does nothing naturally different that a cheaper Taxi, and you can see the lack of popularity of the carpooling offerings Lyft and Uber had.
Ultimately the argument against lower congestion is even easier. Self driving cars for individuals make traveling less annoying than driving. So what we get is more tolerance for long commutes, and ultimately more miles traveled, with ups congestion. Now Timmy can go to his activities by himself, so mom goes elsewhere. Again, when we make something cheaper, we increase demand.
Uber Pool in SF was awesome when it was truly 50% of the cost. But they started hiking up the price (because they could) until it was maybe ~20% less than the full ride. And then COVID happened.
> Self driving cars for individuals make traveling less annoying than driving. So what we get is more tolerance for long commutes, and ultimately more miles traveled, with ups congestion.
That's fine. I did some modeling for that :)
Most of the optional commutes happen outside the rush hour, and the US roads have more than enough capacity for that. The true problem is the congestion due to low occupancy during the rush hour.
And if we want to speculate a bit more, self-driving cars offer an option of more flexible mass transit for truly bottlenecked cities. They can the make transfer from a larger bus into 2-seater local vehicles completely seamless. This is a more distant possibility, as it will require new infrastructure.
> By increasing that to 3-4 people, you can easily double the effective road capacity
Isn't "sharing a vehicle with a stranger" like, the #1 reason why most people who could use transit for a given trip choose not to? Why, when it's a way less efficient—and probably way more expensive per-rider-mile to the govt paying for it—autonomous car going to change that?
The issue with cars, self driving or not, is that they fundamentally use up too much space in urban areas, and cannot efficiently navigate crossings. The problem with traffic isn't (only) that cars are not full, but that in dense, high demand areas where paths/modes cross (i.e. intersections), cars make it extremely difficult to efficiently navigate that shared space. People and bikes can navigate without stopping, bus and trams move such high volumes of people that it significantly reduces the number of vehicles on the road (factor of 20x-40x). Heavy rail (and some light rail) is usually grade separated, and thus doesn't need to worry about intersections at all.
Self driving cars solve almost none of these problems. Unless you're imagining a future where existing outside of a car is illegal (and driving a car manually is illegal) and cars just autonomously speeding through intersections by choreographing perfectly in sync without crashing, AVs will still have to negotiate with everyone else, and thus they still have to navigate the intersection in much the same way they do now. Increasing the diameter of the pipe means nothing if the spout (the intersection) is the same.
So basically, increasing road capacity doesn't matter that much. If you still have a massive amount of individual vehicles trying to navigate the slow capillary network in an urban core at the end of the pipe, you're still going to have traffic on the highway into the city.
Cars give you point to point access. In Seattle I have to decide today whether to spend $100+ on an Uber that will take 25 minutes to get from SeaTac to Ballard, or spend $3.25 to use light rail plus bus that will take at least one hour and 30 minutes. And to think that Seattle is just one of the better transit cities in the USA.
The sharing the bus with murder hobos argument is only relevant for daily commuters with relevantly decent transit routes that aren’t 3X longer than a car would be. Most people don’t get that far in the argument against transit.
> Isn't "sharing a vehicle with a stranger" like, the #1 reason why most people who could use transit for a given trip choose not to?
Partially. Shitty people with zero manners usually end up in low-paying jobs (or in finance), so they tend to concentrate in transit that provides subsidized commutes to them. But the majority of people in transit are normal.
The real reason for not using transit is its slowness. Drop two points on a map of a large city randomly and plot the route between them using cars and transit. Transit will end up on average requiring 3-4x the time by car, even during the rush hour. Of course, people tend to optimize their commutes by choosing the workplace accordingly (this results in a toxic density-misery spiral), so on average transit is not _that_ terrible.
But it totally breaks down when you add more constraints. Like having several kids. Try to plot a route using transit between your workplace, a grocery store, a daycare, and a school. You'll find that it's impossible unless you package everything non-work-related into one spot.
> Why, when it's a way less efficient—and probably way more expensive per-rider-mile to the govt paying for it—autonomous car going to change that?
Transit is WILDLY inefficient. It's amazing how inefficient and polluting it is.
No car platooning, flock routing, car trains, or anything similarly exotic is needed. Just mild carpooling. People are not doing that now because it's a logistics nightmare that removes most advantages of owning a car.