Hacker News Reader

Comment by pjc50 | Hacker News Reader

Self-driving cars are not a silver bullet for congestion, because they don't reduce the physical space required by cars. For example, the Sao Paulo 180km traffic jam: http://www.bbc.co.uk/news/magazine-19660765

View on HN

Couldn't self driving cars even cause congestion to get worse, by making long commutes more convenient, and making driving more convenient than public transit?

Currently long commutes are a strong incentive to live closer to where you work. Reducing the inconvenience of commuting in a car means that more people will live further from the city center or public transit. Additionally, living further from the city center means housing or rent is cheaper. That means that more people will commute longer distances. This applies to other situations as well, like getting downtown on Friday night or to the beach on Saturday. Maybe self driving cars will be so convenient that people don't mind spending time in them, such that people spend far more time in cars than they do now.

Even with the extra congestion, it's a win because the congestion is no longer as stressful as when you have to deal with it as a driver. My commute to the Bay Area is about 2 hours each way. Let's say self-driving cars added 30 minutes due to extra (self driving) cars on the road. I'd be totally fine with that if it meant I wasn't white-knuckling it down I880 myself every day. I could get some work done while my cyber-driver stresses out.

When the commute increases to four hours, you can do a full eight-hour day in the car, arriving at work just in time to have lunch with colleagues before starting the commute home again!

ha, as if the commute time would come out of the work part of your day.

Instead you would get home and go right to bed so you can wake up in time to go to work.

In commute mode self-driving interconnected cars or shuttles could enter in some kind of virtual train mode that would enable them to drive at much lower safety distance or virtually bumper to bumper thereby vastly improving the road capacity while also reducing the drag each car has to deal with.

https://en.wikipedia.org/wiki/Safe_Road_Trains_for_the_Envir...

> some kind of virtual train mode that would enable them to drive at much lower safety distance or virtually bumper to bumper

You just described the Massachusetts Turnpike going into Boston on most mornings :-)

> Self-driving cars are not a silver bullet for congestion

If only we had self-driving cars. Everyone talks about them as if they are a done thing, but they're still under development and it seems there are still huge problems to overcome. That includes technical and legislative issues. Even if those problems can be overcome, will people actually want to buy them? Will they be able to afford them, or will these effectively be luxury/status-symbol vehicles? I think it's too easy to succumb to Nirvana Fallacy when we discuss them.

If they were commercially available as door-to-door vehicles without a human driver needed across broad areas of the US, I think it's fair to say that people would swarm to buy them. And I suspect enough of the secret sauce will be in software that I doubt if cost will be the main stumbling block even if techs appear first in luxury vehicles (as is the case with assistive driving features).

The bigger question is what realistic timeframe we're looking at. I know smart people who go on about exponential progress and all that think we're talking 10 years. I've also listened to researchers in this area who think it's 50+ years and IMO they're probably closer. The challenge is that accurately mapped streets only get you so far and there are a lot of real world situations a car would need to be able to handle that are well beyond the current level of tech.

The self driving car doesn't need to be able to handle things like the dirt road to my parents place to be a success for the vast majority of people.

I agree. But it does need to handle a significant portion of the paved road network, in a variety of weather conditions, and deal at least somewhat gracefully with construction zones and accidents. Even self-driving just on freeways would be useful but at that point it's primarily a convenience feature rather than something enabling significantly different uses of cars.

Self driving cars will create a virtuous cycle, ageing drivers who lose their licenses will be forced to buy them to stay mobile, this availability will decrease police/insurance/legislative tolerance of poor driving which will increase the number of drivers required to buy them. The boomers are going to make this happen.

People won't "buy" self driving cars, they will briefly borrow them from ZipCar, Uber or Ford.

The younger generation as a whole is much less interested in learning to drive or purchase a car. They will also be pushing to make this happen.

And thus we continue to sleepwalk into a state of total control.

Not sure how preferences for mass transit, longer independence of an ageing population, and not killing 33K people a year (USA+CA) in traffic accidents is sleepwalking into a state of total control.

Actually, the USA isn't where self driving cars will probably become popular first. It's the developing world, like China, that lacks the road infrastructure of the west. Self driving cars are a much cheaper way of optimizing fewer roads to do more. And an authoritarian government like China can mandate them (e.g, in a city like Beijing with serious congestion and pollution problems, and already a huge fleet of taxis).

That's an interesting proposition. Beijing traffic already consists of a high proportion of busses at peak times. Automating that will help, but maybe not transformatiional.

The greater advantage for China is how willing they are to push products into the market without knowing if they will succeed. Ready! Fire! Aim! is a powerful framework for finding the real answer when all else is speculation.

Self-driving cars will reduce physical space e.g. http://i.telegraph.co.uk/multimedia/archive/02748/Renault-Tw... Currently, people are using only 30% of the seats, on average. Secondly, carpooling is more feasible. Various optimizations depending on time, traffic light, reusing train tracks.

On this is just one vector. More remote work, VR meetings instead of face-2-face, delivery drones.

The Twizy is already available and not self-driving. People might be more accepting of smaller cars once accidents have become vanishingly rare, but that's a very long way in the future.

Remote work / non-face-to-face meetings are a social and managerial problem. Having your employees commute has no cost to managers and allows them to oversee them more easily. (See for example Yahoo had remote work and took it away; this kind of progress is not on a ratchet).

I wouldn't buy one of those. I wouldn't buy the google car either, it's ugly as hell. If I'm spending money I will get the car I want.

Why in the world would you think that you would buy a self-driving car? Your GooUberZipLyft has left the local depot and will be by to pick you up in three minutes (<link to realtime map here>) and your total cost for this trip will be $1.37. Thank you for using GooUberZipLyft!

They do, actually, optimize space if ALL cars on the road are self driving and cooperating, especially on a limited access freeway.

I expect developing world to take is up much faster, since they lack the roads of the west. People complain about traffic in the west, but man, living in Beijing for the last 8 years, we have some of the worst in the world. They will never build enough roads, there isn't space, but self driving cars could optimize the roads built.

Try Manila. ;)

I have. Honestly, the traffic didn't seem that bad for the largest city in the world, but I only stayed in the new part of the city and didn't see much of the old.

Rush hour and xmas season traffic gets horribad, that's going across the city though. Staying inside certain boroughs usually isn't as bad.

Self-driving cars are the necessary but not sufficient condition to solve traffic problems. Start with that and remove 1-person-1-car model [0], add ride-sharing software, ban human drivers [1] and done, no more traffic problems.

As for your example - this jam is a direct result of very high latency of human reaction. It causes stop-and-go in jams [2].

[0] Or it will rot and occupy parking space for 97% of time.

[1] This will likely to average speed and reduce time spent waiting for lights.

[2] https://www.youtube.com/watch?v=Suugn-p5C1M

> Start with that and remove 1-person-1-car model [0], add ride-sharing software, ban human drivers [1] and done

... you've got a train

(particularly on the Intermediate Capacity Transit System / Copenhagen-like models, automatically run small and light cars for very frequent service)

You've got a superior train that actually drops you off at work, rather than at a train station where you have to take a shuttle to a bus station, then a bus to a bus stop near work, then walk a few blocks.

That's just poor city planning, though.

Do you suppose that an organization that builds a bus station away from a train station and has to connect them with a shuttle will do much better at building reasonable, cost effective infrastructure for self-driving cars?

Self-driving cars are the necessary but not sufficient condition to solve traffic problems. Start with that and remove 1-person-1-car model [0]

There's no need to have self-driving cars to stop 1p1c. We've already got mass transit… Switch to semi-fullly seated buses and you've reduced the number of vehicles by a factor of 40.

Besides, since everyone starts work at the same time self-driving cars will have minimal effect on the number of cars on the roads at rush hour as long as there's only one occupant per car.

Much cheaper solutions than developing and constructing an entire new vehicle fleet would be to force multiple occupants per vehicle, and to have people start work over a wider time span.

IMO, even simpler than that would be to stop subsidising parking: raise parking fees, eliminate free road parking, force companies to charge for parking (this part may not be feasible, special land taxes on parking lots?). Whether or not you'd need special exemptions for businesses (deliveries, plumbers, etc.), I'm not sure.

People can feel free to drive around to their hearts' content, they just won't be able to stop and get out of their cars! ;)

Example of the problem with mass transit : http://www.uwgb.edu/dutchs/pseudosc/masstransit.htm

1) Money: generally speaking, depending on the number of passengers mass transit is a little cheaper (1 person), about the same (2 pers) or much more expensive (a family of 4-5) than public transport.

If you take into account that home -> work -> shopping -> home with public transport turns into home -> city center -> other city center -> work -> city center -> shopping -> city center -> other city center -> home ... The point is that there are many kinds of trips where the routes taken by public transport cause it to be more expensive than a car.

In the cities where I've worked, mass transit (2x bus + train) was only 20-30% cheaper than a car and only if you lived out in the suburbs and worked in the city. The other way around (living in city, working a bit out of the city), mass transit was more expensive than the car. Living in city, working in city, a fold-up bike was far more convenient.

2) Time: mass transit doesn't go where I live, and doesn't leave/arrive when I need to leave/arrive. Furthermore, it's regularly non-operative when I need it. This limits what I can do (I stayed late, we ordered in pizza, then hang out with a friend ... oops it's 22h30 ... no more buses to my place until 6 am. Yes it only happened to me maybe 5 times over a decade but fuck that shit)

3) Baggage limits: when using mass transit I can carry 25 kilograms. If I don't care about comfort of myself and other passengers, maybe 40-50 is possible. I don't imagine I will be able to do that anymore when I turn 60. If I have my kids with me, 20 kg is pushing it, and if I'm either slightly ill or ... I can barely take anything at all.

When I go shopping, I buy for the week at least (by far the least bothersome way to do shopping imho, a lot faster, and also 30% or so cheaper) I come home with 30-40 kilograms usually. Can't use public transport for that.

This is ignoring that a lot of things can't be reasonably/safely transported using public transport, like DIY building supplies (ladder, long bars, 2 by 4s, ...). This includes anything heavy or above a certain very restrictive size limit.

4) Comfort. I don't even get why they bother with 1st class on trains. Let's not pretend that even 1st class has 1/10th the comfort of a car. Commuter trains are torture.

Mass transit is usable as a commute solution. It cannot reasonably replace a car. We need to fix transport and mass transport as it currently exists is not the answer.

> The other way around (living in city, working a bit out of the city), mass transit was more expensive than the car.

Do you mean the monthly cost of transit pass/fares was more than monthly cost of parking + gas + insurance + car payments + maintenance?

Our monthly numbers are

Transit passes for a family of four: 250-450 USD

or one very cheap car used for commute: 300-600 USD + public transport

or bicycles + public tranport: 150-300 USD a cargo bike being the most expensive part of that.

I commute 2*50 min daily on bicycle since 2013.

To be fair it's actually 2*30 min nowdays, I just say 50 min because that's what how long the commute was with the car and also how long it took me to bike when I started.

Are those transit passes just covering commutes ? Or are they the closest thing public transport has to a car (global access to all routes) ?

Might want to include overt taxpayer subsidies in the former, plus any allowance for access to a car in those cases where mass transit doesn't cut it. It's kind of like including the capital/expense cost of the necessary standby fossil plants in the costs of "green energy" system.

Mass transit is usable as a commute solution. It cannot reasonably replace a car.

It can't replace all uses of a car. But what causes rush hour traffic isn't that you sometimes want to go shopping, or that you sometimes stay late at a friend - it's that too many people take the car all the time, and all the way.

By making mass transit the default and a car the exception, we'd be a long way towards a solution.

    "is a little cheaper (1 person), about the same (2 pers) ..."

Living in any city with a decent transit infrastructure (I know, that isn't by any means all the medium-large cities in north america) this calculation really isn't realistic.

It's more like owning a car is 4-8x as expensive as transit (only), so your break even is more like the 3-4 people. The calculation changes a little bit if you live in the suburbs, but it is still more expensive than most people realize to maintain a car (mostly people do this calculation wrong by ignoring depreciation and incidental costs, even financing)

For what it is worth, in the downtown area I live in, average monthly parking rates are about the same cost as unlimited transit usage....

This doesn't speak to your other points, and I totally get why people want to own cars, but they should be realistic about the choices.

In several cities I've lived in, it is entirely possible to a) not own a car b) have unlimited transit pass plus c) take a cab back from groceries every weekend plus d) rent a car or trunk whenever you need it on a weekend -- and still end up a couple thousand dollars ahead (or more) at the end of the year. And that's assuming a cheap car. Sure, it doesn't apply to everyone's situations but it's also not terribly unusual.

and increased my hatred of my commute by about the same amount. I don't want to be crammed into an uncomfortable seat feeling sick because the bus is lurching around while the stranger next to me tries to fall asleep on my shoulder. oh and it takes longer.

It's just like taxis now though. In Beijing, around 10-20% people go to work by taxi, and it can be a bit difficult to find one during the rush hours. I don't see how self driving cars change that at all. It's a luxury option for those of us who would otherwise buy private cars and don't want to be smashed in on the subway (for 50 Kuai vs. 5 Kuai).

remove 1-person-1-car model [0]

Westerners are going to be very reluctant to do this, especially if it means sharing with random strangers.

The developing world already has lots of "minibus" transport, and still has plenty of traffic jams.

> this jam is a direct result of very high latency of human reaction

I would say high latency is not the reason, but road greed, and defensive driving against unwanted and unannounced lane changes.

but proper design of the roadways coupled with self driving cars which don't do stupid things humans do (not let people in, merge well below speed, drive 10mph under the limit just because) should help.

A significant amount of congestion is bad exit and merging.

But you're right, another reason mass transit needs to be treated as a QOL issue and not just a tax issue.

Also, there's not only congestion but environmental issues.

Even electric cars don't fix those.

Reply on news.ycombinator.com