you still have to get to bus stop
Edit: ca. 2008 the policy was 800–1,000 ft for bus stops and 1,000–1,200 ft for street car stops. On grades between 10–15% the standard is 500–600 ft for bus stops, and on grades above 15% the standard is 300–400 ft. Trunk routes have 24x7 service, and (I can't find it now) there are standards about the amount of service on the other routes but it's something along the lines of 80% of residences are supposed to have frequent service ~18 hours a day and reduced service in the wee hours. Currently the tightest headway I can find is 8 minutes for travel during peak hours along the busiest routes.
Muni is often inefficient but it can get you pretty much anywhere in the city (including places you can't walk to like Treasure Island) for most of the day.
Edit: For reference San Francisco is roughly 7 miles by 7 miles. So you're talking about transiting halfway across town.
Cars (with or without drivers) are competing for the same right of way as the bus, so more of them will simply slow everyone down. A 60ft bus can fit, what, about 100 people? A self-driving car can fit, what, 5? The current quantity of Lyft/Uber/Waymo/Cruise vehicles already has a negative impact on traffic. Put enough self driving cars on the street to make up for the capacity that buses provide and you're going to make traffic much, much worse.
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.
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 :()
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
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.
- 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
- Build a real time monitoring system of people travelling on these services to allow instant arrests by diverting the vehicles.
- Sell all the data to advertisers and unscrupulous individuals, who may or may not be stalking someone.
- Allow authorities to use the cars as weapons to kill dissidents and other threats to government.
Building something this complex and all-encompassing goes 90% of the way to building a dystopia.
As long as we're throwing out fantasies, I'd like us to spend $2B more, and have the Chinatown subway run all the way out to Fisherman's Wharf and the Marina district, and also expand capabilities west out to the avenues.
I'm amazed we got the Chinatown subway and I'm excited to see the city develop around it!
Like I just implicitly assumed that that is way too much money for the value it provides but admittedly I didn't think much about it. Can you quantify how the benefit might be worth that much?
I think Cruise Automation has roughly raised that amount over it's entire lifetime. Even if we had to fund 10 of those companies for a similar amount, that seems like a better investment than the dozen or so subway miles that would fund.
Subway lines can pay for themselves across hundreds of years. It’s the best kind of investment for cities because it operates on their timescales and use ultra cheap bonds responsibly and it doesn’t interfere with free markets.
just pick the low hanging fruit first and let driverless cars operate freely with data sharing. then build your pyramids.
Cities taking advantage of people with ultra low risk tolerance to make infrastructure investments at ultra low rates that only make sense with very long time horizons is dependent on markets.
Ahh, so by ‘we’ you didn’t actually mean a group you are part of but instead other people.
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.
Bad service is much worse on transit though.
People in NYC don’t just save money using the metro vs a Taxi they save time.
It can actually fit 150.
I am not in great shape but I know which I prefer.
You cut it in half because you can walk to the closest stop and cut it in half again before you’re starting at a random point not the worst possibility so 5280/8/2/2 = 165 feet. Except we don’t exit buildings or get on busses in the middle of the street so it’s not random.
1/8th of a mile is every 660 feet, but the farthest you can be from one is the midpoint so 330 feet. However for every trip starting at the midpoint there’s another at and the bus stop thus the average would be 1/2 that distance again or 165 feet.
Assuming a 1D world, you can beat this in practice.
As to unbearably slow, there’s this idea of express busses and buss only lanes etc. The core of cities really isn’t that big, it takes forever because traffic is slow or mass transit is infrequent.
Ideally you want a grid where bus lines are near intersections so going from North to East means crossing 1 street so you can go anywhere with 2 waits and a street crossing.