4 ms·

Two really popular locations, 3 miles apart, with a direct bus route servicing them and still 40 min travel time. This doesn't work: https://www.google.com/map

Two really popular locations, 3 miles apart, with a direct bus route servicing them and still 40 min travel time. This doesn't work: https://www.google.com/maps/dir/Dolores+Park,+Dolores+St+%26,+19th+St,+San+Francisco,+CA+94114/Marina+District,+San+Francisco,+CA/@37.7768012,-122.4338673,14.93z/data=!4m14!4m13!1m5!1m1!1s0x808f7e1779aa70a7:0xa618e4eff1228d60!2m2!1d-122.4271145!2d37.7597704!1m5!1m1!1s0x808580d79ae5a51b:0x3c50fe8f3930832!2m2!1d-122.4370011!2d37.8043849!3e3?hl=en&entry=ttu https://www.google.com/maps/dir/Dolores+Park,+Dolores+St+%26... I'd argue that we'll be able to make this much better with driverless cars than with public transit.

Panzer04 3y ago
It's a capacity issue. You just can't transport as many people with cars. In a completely empty city, the first car handily beats transit. Of course. In a traffic-laden city, separated PT tends to determine how long it takes to travel instead. 12AM it takes me 25min to get into the centre of the city - at peak hour it's more than an hour. Curiously, the train also happens to take approximately an hour to get into the city. If public transit exists, it becomes the upper-bound for traffic travel time as if it takes longer than that people start to take the constant-time PT trip instead of the variable car trip. In a software sense you can usually view decent PT (ie. not busses stuck in traffic) as having a high but constant time cost, whereas cars have a low but increasing cost corresponding to the traffic load (with non-separated transit incurring the same costs but being worse due to traffic stops :()
- alooPotato 3y ago
  Fair argument. Public transit caps downside case and upside case. Cars can have downside but have upside too. I want to have my cake and eat it too and I think we can with driverless cars. Lots of software tricks we can do with a network of driverless EVs + city traffic lights to dramatically increase capacity. My roadmap for SF would be: - allow waymo and others to scale driverless cars - stop spending ~$2B for a 1.7 mile subway to chinatown, spend that money on a traffic light control network (ideally just let google run it) - give the same per passenger subsidy to riders of driverless cars that current public transit passengers receive per trip - require that driverless cars in SF should send realtime destination info to the traffic light network - remove parking spots on high traffic streets effectively creating two more lanes - allow bus lanes to be exclusively used by driverless cars - tax driverless car rides based on occupancy+distance+time of day so you can traffic shape
  - alooPotato 3y ago
    All of these are possible to implement in a really short amount of time (because its mostly software) compared to public transit (10+ years to make a bus lane on van ness)
  - xnx 3y ago
    This is a great line of thinking. Public transit based on Waymo cars could be dramatically more efficient, cheaper, and more flexible.
  - inferiorhuman 3y ago
    Your premise is, to put it politely, completely wrong. Even considering how broad the scope was, Van Ness took well longer than it should've. The delays were almost entirely political, driven by one supervisor who was on a mission to preserve the street lights. Meanwhile the solution you've put forth is almost entirely political and even were there general acceptance of privatizing the roads that sort of project would be mired in just as much NIMBYism. Even then you're talking about, what, refunding federal monies that have already been spent? Levying a new tax (which will require a 2/3 voter approval), subsidizing private corporations, and kneecapping transit? And for what? A system that has a fraction of the capacity of the existing bus and tram infrastructure? Beyond the gall to suggest privatizing our roads, you're suggesting handing it all over to Google? Really? A corporation with a track record of zero customer or long-term product support is not in a position to run our roads.
    - alooPotato 3y ago
      Ok I agree I was being a little ambitious :) Let me tackle your points one at a time: - cities already pay private companies to help manage the traffic infrastructure so it doesn't seem too crazy. Seems hyperbolic to claim that's "privatizing roads" - subsidizing private corps - ok maybe we shouldn't do this. but we should stop subsidizing public transit. it should have to compete with private alternatives now that we have them. previously there were no alternatives to public transit so we kind of had to subsidize it for equity, but now I think private enterprise (given fair competition) could deliver the same outcome. Instead of subsidizing the method (public transit) we should subsidize the outcome we want (equitable access to transport - which might mean that you need to provide low income rides as a ride share provider) - on the money spent on the SF subway and bus lane, you're right, we're not getting that back. But we can just stop doing those going forward. - regarding van ness bus lane, I didn't know it was because of the street lamps. But that actually proves my point. Basically to make any improvement, we just have to steer clear of new physical infrastructure. I'm suggesting no new physical infra, just software. - regarding the tax on driverless rides, we kind of already have it. SF has a 3.25% tax on all ride sharing rides. We'd just need to update that to price the tax dynamically to help traffic shape. And if we don't get that piece, it's not critical. - on giving it to Google. Fine... not them then. Whoever can provide the best service for the lowest cost. Perfect for free markets. In fact, if we make the API to traffic system standardized, you could plugin and hot swap different providers. You could even back test new providers on historic data to see what their performance would have been like and swap to new providers as they demonstrate their improvement over the current system - cars have a fraction of the capacity of buses - yeah only if they are fully utilized and you're not counting getting to and from the bus station nor the loss of productivity waiting for a bus
Retric 3y ago
Bad public transit doesn’t mean public transit can’t be good. It just means your local transit happens to be bad. The crazy thing is bad transit needs a larger subsidy because people don’t want to use it. So good systems get better and bad systems get worse.
- alooPotato 3y ago
  i think its just easier to make uber style driverless EVs be great than it is to make public transit great
  - bluGill 3y ago
    Great for uber is still limited by traffic and is a lot more expensive than great transit which doesn't get stuck in traffic, and thus is much cheaper than uber for better service. Bad service is much worse on transit though.
  - Retric 3y ago
    The hard part of Uber style driverless EVs being great is having extra roads to handle the extra traffic not actually building the self driving bits. People in NYC don’t just save money using the metro vs a Taxi they save time.
    - xnx 3y ago
      NYC is the one US city where trains almost make sense, but there are a ton of downsides related to cost, safety, etc. I'd love to see all those dedicated right-of-ways put to more efficient use with self-driving vehicles.
      - bluGill 3y ago
        NYC has out of control costs. Look to Vancouver canada for a very different picture.

Panzer04 3y ago

It's a capacity issue. You just can't transport as many people with cars. In a completely empty city, the first car handily beats transit. Of course. In a traffic-laden city, separated PT tends to determine how long it takes to travel instead. 12AM it takes me 25min to get into the centre of the city - at peak hour it's more than an hour. Curiously, the train also happens to take approximately an hour to get into the city. If public transit exists, it becomes the upper-bound for traffic travel time as if it takes longer than that people start to take the constant-time PT trip instead of the variable car trip. In a software sense you can usually view decent PT (ie. not busses stuck in traffic) as having a high but constant time cost, whereas cars have a low but increasing cost corresponding to the traffic load (with non-separated transit incurring the same costs but being worse due to traffic stops :()

alooPotato 3y ago
Fair argument. Public transit caps downside case and upside case. Cars can have downside but have upside too. I want to have my cake and eat it too and I think we can with driverless cars. Lots of software tricks we can do with a network of driverless EVs + city traffic lights to dramatically increase capacity. My roadmap for SF would be: - allow waymo and others to scale driverless cars - stop spending ~$2B for a 1.7 mile subway to chinatown, spend that money on a traffic light control network (ideally just let google run it) - give the same per passenger subsidy to riders of driverless cars that current public transit passengers receive per trip - require that driverless cars in SF should send realtime destination info to the traffic light network - remove parking spots on high traffic streets effectively creating two more lanes - allow bus lanes to be exclusively used by driverless cars - tax driverless car rides based on occupancy+distance+time of day so you can traffic shape
- alooPotato 3y ago
  All of these are possible to implement in a really short amount of time (because its mostly software) compared to public transit (10+ years to make a bus lane on van ness)
- xnx 3y ago
  This is a great line of thinking. Public transit based on Waymo cars could be dramatically more efficient, cheaper, and more flexible.
- inferiorhuman 3y ago
  Your premise is, to put it politely, completely wrong. Even considering how broad the scope was, Van Ness took well longer than it should've. The delays were almost entirely political, driven by one supervisor who was on a mission to preserve the street lights. Meanwhile the solution you've put forth is almost entirely political and even were there general acceptance of privatizing the roads that sort of project would be mired in just as much NIMBYism. Even then you're talking about, what, refunding federal monies that have already been spent? Levying a new tax (which will require a 2/3 voter approval), subsidizing private corporations, and kneecapping transit? And for what? A system that has a fraction of the capacity of the existing bus and tram infrastructure? Beyond the gall to suggest privatizing our roads, you're suggesting handing it all over to Google? Really? A corporation with a track record of zero customer or long-term product support is not in a position to run our roads.
  - alooPotato 3y ago
    Ok I agree I was being a little ambitious :) Let me tackle your points one at a time: - cities already pay private companies to help manage the traffic infrastructure so it doesn't seem too crazy. Seems hyperbolic to claim that's "privatizing roads" - subsidizing private corps - ok maybe we shouldn't do this. but we should stop subsidizing public transit. it should have to compete with private alternatives now that we have them. previously there were no alternatives to public transit so we kind of had to subsidize it for equity, but now I think private enterprise (given fair competition) could deliver the same outcome. Instead of subsidizing the method (public transit) we should subsidize the outcome we want (equitable access to transport - which might mean that you need to provide low income rides as a ride share provider) - on the money spent on the SF subway and bus lane, you're right, we're not getting that back. But we can just stop doing those going forward. - regarding van ness bus lane, I didn't know it was because of the street lamps. But that actually proves my point. Basically to make any improvement, we just have to steer clear of new physical infrastructure. I'm suggesting no new physical infra, just software. - regarding the tax on driverless rides, we kind of already have it. SF has a 3.25% tax on all ride sharing rides. We'd just need to update that to price the tax dynamically to help traffic shape. And if we don't get that piece, it's not critical. - on giving it to Google. Fine... not them then. Whoever can provide the best service for the lowest cost. Perfect for free markets. In fact, if we make the API to traffic system standardized, you could plugin and hot swap different providers. You could even back test new providers on historic data to see what their performance would have been like and swap to new providers as they demonstrate their improvement over the current system - cars have a fraction of the capacity of buses - yeah only if they are fully utilized and you're not counting getting to and from the bus station nor the loss of productivity waiting for a bus

Retric 3y ago

Bad public transit doesn’t mean public transit can’t be good. It just means your local transit happens to be bad. The crazy thing is bad transit needs a larger subsidy because people don’t want to use it. So good systems get better and bad systems get worse.

alooPotato 3y ago
i think its just easier to make uber style driverless EVs be great than it is to make public transit great
- bluGill 3y ago
  Great for uber is still limited by traffic and is a lot more expensive than great transit which doesn't get stuck in traffic, and thus is much cheaper than uber for better service. Bad service is much worse on transit though.
- Retric 3y ago
  The hard part of Uber style driverless EVs being great is having extra roads to handle the extra traffic not actually building the self driving bits. People in NYC don’t just save money using the metro vs a Taxi they save time.
  - xnx 3y ago
    NYC is the one US city where trains almost make sense, but there are a ton of downsides related to cost, safety, etc. I'd love to see all those dedicated right-of-ways put to more efficient use with self-driving vehicles.
    - bluGill 3y ago
      NYC has out of control costs. Look to Vancouver canada for a very different picture.