4 ms·

A world of robotaxis requires a huge percentage increase in the number of cars on roads because you will need many empty cars on the roads at all times repositi

A world of robotaxis requires a huge percentage increase in the number of cars on roads because you will need many empty cars on the roads at all times repositioning or picking up passengers. I can’t see how that won’t negate and even be worse than any benefit from improved reaction times. Any attempt to solve for repositioning just takes you closer and closer to buses or trains.

1shooner 10mo ago
>you will need many empty cars on the roads at all times Not necessarily. FSD services would probably optimize for occupancy, which would mean limited and unpredictable availability for customers. Since the fleet is a capital investment rather than being floated by gig workers, supply would presumably be more constrained than Uber etc are now, and we see what happens at peak demand with them. I don't see that business logic changing between Uber drivers and FSD.
- idontwantthis 10mo ago
  Why not have the FSD services run on predetermined routes along high density corridors on a schedule? Then they would be filled optimally on average.
JasonADrury 10mo ago
Given reasonable lead times on bookings, you'd need less cars on the road per trip than now. Especially since demand tends to be predictable. The really big win here is parking and time spent looking for it. Self-driving trips can just be shorter
- idontwantthis 10mo ago
  If you have 100 people who want to get somewhere in their own car at the same time, that requires a minimum of 100 cars. The lead time doesn’t factor into it. Parking is a problem because of our dependence on cars. You also don’t need to park if you take a bus or a train and the things you want to get to are built near the bus and train stations.
  - JasonADrury 10mo ago
    Those cars have to spend less time on the road because they don't have to spend time on parking, this leads to less car-hours on the road for the same trips. Those cars will also be faster, safer and pick better routes than humans. > You also don’t need to park if you take a bus or a train and the things you want to get to are built near the bus and train stations. It is nonsensical to compare self-driving cars against buses or trains in a situation where we still have tons of private cars on the road. FSD can totally improve buses too, and well-designed multi-passenger vehicles can certainly easily end up making up most of the fleet. You could for example seat 4 people in completely isolated spaces without needing outrageously large vehicles. That'd be great for serving commutes where loads of people are going to the same areas.
    - idontwantthis 10mo ago
      We're not disagreeing with each other. If we can have robotaxis that primarily move more than 1 person at a time then that's a good thing. But if "loads of people are going to the same areas" then those robotaxis should resemble buses.
      - JasonADrury 10mo ago
        >But if "loads of people are going to the same areas" then those robotaxis should resemble buses. Should, but if you really want to sell it to the public, you want to keep the service as close as possible or better than what's currently available. That's more feasible with fewer passengers on a vehicle, too many stops will easily double your travel time.
sokoloff 10mo ago
Today’s cars that are driving around (slowly and less predictably) looking for parking are functionally empty even if they’re not literally so. Those foregone miles will partially offset the deadheading miles.

1shooner 10mo ago

>you will need many empty cars on the roads at all times Not necessarily. FSD services would probably optimize for occupancy, which would mean limited and unpredictable availability for customers. Since the fleet is a capital investment rather than being floated by gig workers, supply would presumably be more constrained than Uber etc are now, and we see what happens at peak demand with them. I don't see that business logic changing between Uber drivers and FSD.

idontwantthis 10mo ago
Why not have the FSD services run on predetermined routes along high density corridors on a schedule? Then they would be filled optimally on average.

JasonADrury 10mo ago

Given reasonable lead times on bookings, you'd need less cars on the road per trip than now. Especially since demand tends to be predictable. The really big win here is parking and time spent looking for it. Self-driving trips can just be shorter

idontwantthis 10mo ago
If you have 100 people who want to get somewhere in their own car at the same time, that requires a minimum of 100 cars. The lead time doesn’t factor into it. Parking is a problem because of our dependence on cars. You also don’t need to park if you take a bus or a train and the things you want to get to are built near the bus and train stations.
- JasonADrury 10mo ago
  Those cars have to spend less time on the road because they don't have to spend time on parking, this leads to less car-hours on the road for the same trips. Those cars will also be faster, safer and pick better routes than humans. > You also don’t need to park if you take a bus or a train and the things you want to get to are built near the bus and train stations. It is nonsensical to compare self-driving cars against buses or trains in a situation where we still have tons of private cars on the road. FSD can totally improve buses too, and well-designed multi-passenger vehicles can certainly easily end up making up most of the fleet. You could for example seat 4 people in completely isolated spaces without needing outrageously large vehicles. That'd be great for serving commutes where loads of people are going to the same areas.
  - idontwantthis 10mo ago
    We're not disagreeing with each other. If we can have robotaxis that primarily move more than 1 person at a time then that's a good thing. But if "loads of people are going to the same areas" then those robotaxis should resemble buses.
    - JasonADrury 10mo ago
      >But if "loads of people are going to the same areas" then those robotaxis should resemble buses. Should, but if you really want to sell it to the public, you want to keep the service as close as possible or better than what's currently available. That's more feasible with fewer passengers on a vehicle, too many stops will easily double your travel time.

sokoloff 10mo ago

Today’s cars that are driving around (slowly and less predictably) looking for parking are functionally empty even if they’re not literally so. Those foregone miles will partially offset the deadheading miles.