How driverless cars will affect our cities | Hacker News Reader

How driverless cars will affect our cities

cityminded.org

36 pointsluu71 comments

I'm hopeful that driverless cars will be a boon to cities, particularly to the many second-tier American cities whose formerly vibrant cores have been decimated by the need to supply parking for suburban workers.

For the past sixty years, suburbanization has increased demand for parking and fed a feedback loop that's destroyed urban density. More people driving means greater demand for parking, which means buildings get torn down for lots and parking structures. Nodes of activity spread farther apart, so more people drive, and drive more often, further increasing demand for parking. The result is a country of paved-over cities with more parking spaces than people [0].

Driverless cars have the potential to break the feedback loop by decoupling destinations from nearby parking, shifting parking out of central cities and freeing land for all the things that make urban living great. Perhaps the feedback loop will start working in the other direction.

(I don't think driverless cars will lead to a redoubling of suburbanization, by the way, because I think big houses and wide open spaces are far down the list of reasons most suburbanites live in the suburbs.)

0. http://www.nytimes.com/2012/01/08/arts/design/taking-parking...

While driverless cars can help with parking w.r.t. city development, I don't think that will be too helpful as we already have issues with the capacity of roads. We should be aiming to reduce the importance of roads for cars in city planning, and I'm not sure how driverless cars will help that. Rather, I could see it being worse if cities make mistakes and try to build more road capacity, making congestion worse as more cars fill the new road space.

Driverless cars should be a huge boon to the efficiency of moving cars along existing roads.

There may be an awkward period during which both automated and human-driven cars are on the road at the same time, but by the time it's mostly robots out there, traffic will flow much better. Traffic congestion is caused mostly by human-led factors like route choice and poor cooperation (in merging, etc.) rather than merely the number of cars on the road.

Why are big houses far down the list? What's high on the list?

The perception and/or reality of safety.

Suburban flight in the US didn't occur until long after cars became ubiquitous. People had the means to move to suburbs for years but didn't do so en masse until the urban cores started to rot.

People moved out of cities when they became shitholes, and people stay out of cities out of a perception that they are shitholes. Said perception may be true or false depending on where you are.

There is also the issue that for people with families, many urban centers have in the past few decades reconstituted and cleaned up, but without the infrastructure that families are looking for: e.g., schools.

Suburban living is all about being safe from criminal hordes outside the gates - whether real or imagined.

| Suburban flight in the US didn't occur until long after cars became ubiquitous.

It seems you're not acquainted with the concept of "streetcar suburbs" [1]. The Wikipedia article goes into some detail; I recommend it for the curious who'd like to know how streetcar technology drove suburbanization before 1920.

FTA: "Because streetcar operators offered low fares and free transfers, commuting was finally affordable to nearly everyone."

This is not academic information about the distant past that no longer applies. Transit-oriented development [2] is impacting our cities positively today. It will be a long time before driverless cars have an impact.

1: http://en.wikipedia.org/wiki/Streetcar_suburb

2: http://en.wikipedia.org/wiki/Transit-oriented_development

I think we're running into a case of differing terminology. I'm well aware of streetcar suburbs - by our modern understanding of cities they are substantially more urban than they are suburban.

And that's part of the point exactly - before suburban flight the concept of "suburbia" was transit-aligned, grid-based, focused around walkable distances and small(ish) housing. People were fine living in these places, even long after cars became ubiquitous. The modern suburb was not born until the collapse of urban centers.

Today we understand all of the above as "urban". The bulk of the American population lives in the modern incarnation of suburbs, not the 1920s version of it. Still-extant streetcar suburbs are now sold more as urban neighborhoods than not.

The modern suburb calls for large houses, even larger lawns, and subdivision-style street layouts (with an unspoken design intent of keeping out undesirables, a core part of its history and continued appeal). Map[1] is a modern suburb, full of meandering streets, deadends, cul-de-sacs, and indirection - and that's what I was referring to, considering that modern suburbs such as this grossly outnumber streetcar-urban neighborhoods in the US today.

If you took someone from a typical suburban environment in the US and dropped them in a streetcar suburb they'd perceive it as distinctly urban.

[1] https://maps.google.com/maps?q=dallas+tx&hl=en&ll=33...

There is no coincidence that suburban flight coincided with desegregation.

Are you suggesting people like big houses with large lawns in quiet suburban neighborhoods with great schools because they are racist?

Racism was certainly one of the things that precipitated the shift in the equilibrium. Remember, at the time when people still lived in the cities, the suburbs weren't like they are today. There were good schools and decent housing in the cities, because the middle class white people hadn't moved away yet. As someone pointed out elsewhere, the availability of cars to make the suburban exodus happened long before the suburban exodus itself happened. And let's face it, race place a huge role in living choices today. Check out: http://www.city-data.com/forum/chicago. People are perfectly happy to be egalitarian right up to the point where it involves sending their kids to schools with heavy black populations. "Gentrification" is almost synonymous with moving blacks out of neighborhoods, and cities like to tout gentrification statistics, with their underlying tone of racism, without any of the bashfulness that is usually associated with even the hint of racism. People in Chicago considered it a great thing that the South Loop had gone from 35% white to 75% white in just one decade.

Look at it from a different perspective. Why is it that in Europe more of the population of the various metro areas lives in the city than in the surrounding suburbs, as compared to the U.S.? Do Americans just inherently like big houses and large lawns more than Europeans? And if they do, why didn't they move out of the city long before 1960, when it was eminently practical to do so?

That of course doesn't mean that suburbanites are racist today. After decades of suburban exodus the core cities are shells of their former selves, urban public school districts are almost exclusively for the poor, etc. There's lots of non-racist reasons, today, to not want to live in the city besides the inherent availability of lawns, etc. But given how racist society as a whole was in 1960, I think it's ridiculous to claim that the racial tension precipitated by the civil rights movement and desegregation didn't play a large role in tipping the balance between the cities and suburbs.

It's a coincidence by definition. Whether or not the two are linked is another matter, and someone has probably done research on it.

The idiomatic phrase "it's no coincidence" means "it's more than just a coincidence."

I know. I'm saying you should offer more than a one-line assertion.

People were really racist in the 1960's, to the point where they wanted blacks to not use the same restrooms. They didn't suddenly stop being racist when cities were legally desegregated. What happened consequently was what you would expect if a bunch of racists were stuck in cities with a bunch of black people who were no longer confined to their separate areas.

I know the popular assertions. Was hoping you had some actual research.

  | Suburban living is all about being safe from
  | criminal hordes outside the gates - whether real
  | or imagined.

Or being able to afford a house.

That really wasn't the initial problem, nor is it still in most places. You have highly anomalous cities like NYC or SF where living costs are ludicrous, but for the most part cities are affordable places to live.

Sure, you will always get bigger spaces for the same price - that's why they're called suburbs, but the marginal utility of space decreases pretty rapidly. There are many cities where owning a sufficiently large space is indeed affordable. In fact this describes most cities outside of SF and NYC where we're packed like sardines.

People stopped living in cities because drug and crime became rampant, and quality of life dropped precipitously. That was the instigator, and still a cultural cornerstone of the suburban lifestyle, despite the revitalization of urban America.

People were fine living in apartments rather than owning a whole building to themselves - until the environment went to shit. They had the ability to get big houses outside the city for decades but didn't take the option until their urban neighborhoods collapsed. I suspect they still are fine with not owning a large house.

My comment was more about the current situation, than the historical cause for sprawl or "white flight."

I get that - and my thought is that the whole concept of owning a big house is ex post facto rationalization, not a root issue. If cities were safer, transportation better, and family infrastructure competent, we'd see a dramatic move back into cities - a smaller space be damned.

And we do see this in some places, where buying a medium-sized flat in a nice neighborhood close to good schools is as desirable as a big house with a big lawn in a nice area.

If you look at suburban developments all around the country, you'll also notice an explosion of smaller houses and townhouses - places that are in many cases no larger than apartments in the city, just in upright standing form. Space is a detail to most of the population, not a core factor.

It's not always space. Just the idea of owning vs. renting.

For example, in Toronto the real estate prices are sky high in the city (e.g. I've heard of just a single floor of some of those large houses in The Annex going for upwards of $800k). There are a lot of condos being built (and already built), but they are all aimed at single people or childless couples, not necessarily at families. Two or three bedroom condos are few and far between, meaning that they aren't exactly cheap.

For purposes of this argument no one gives a damn about Toronto. It's another of the edge cases where attractiveness of downtown of a major city outweighs "normal" North American real estate factors.

Take a look where the most expensive houses in Hamilton or St. Catharines or Kitchener are and where the "bad" parts of town are.

My perception is that cost tops the list. That's partly about home size (why pay twice as much for a condo that's half the size?), but I think mostly about sheer high prices. [0]

Cost is also tied up with the issue of public versus private schools, and it's that issue more than any other that in my experience is the reason urbanites become suburbanites. Then generalized negative perceptions of American cities (as cesspools of crime, noise, high taxes, etc.) keep many suburbanites away.

Those perceptions are half-truths at best, but a perhaps even bigger problem is that for the great many people who haven't lived it, there's no consideration of the benefits of an urban lifestyle. Of course you're going to choose to live in the 'burbs if you only see the city's detriments and not also its commensurately higher (or better) advantages.

0. In cities like San Francisco, a lot of that cost is due to land-use regulation. But in a typical mid-sized Midwestern city like Cincinnati or Kansas City, where public transit is poor and parking is provided pretty much everywhere, only the true urban warriors try to get by without a car. So parking is a big driver of the cost of city living. Structured parking costs $40,000 or more per space, and median home prices are well under $200,000. Ten to twenty percent or more of the cost of a condo downtown is parking, and you still have to pay for a car. (And it should go without saying that the benefits of city living are greatly reduced if you need to drive a lot anyway.)

The problem is cities are full of poor people, for various lamentable and not so lamentable reasons. That fact drives all the other things that drive people out of cities: crime, bad schools, etc.

The sorry state of American cities is due in huge part to state distortion. The social services for the poor are heavily concentrated in the cities, so that's where they go (has the irony ever occurred to anyone that people complain about not being able to afford SF, NYC, etc, but at the same time those places are filled with poor people living in project housing?). Social services for middle class people (roads, good schools, etc) are concentrated in the suburbs, so that's where they go. Then the municipalities prevent all development, causing skyrocketing rents for people who want to try and live in the city anyway. City and state governments try as hard as they can to get people to leave their cities.

(has the irony ever occurred to anyone that people complain about not being able to afford SF, NYC, etc, but at the same time those places are filled with poor people living in project housing?)

I can't find it now, but I know Megan McArdle has written about how there is a bathtub-shaped curve (my word) for income distribution in many parts of NYC. You can be poor, and you will do okay. You can be really rich, and you will go well. But in the middle it becomes a desert.

Cost, I'm guessing. A studio apartment in Manhattan becomes a 5-bedroom house outside of the city. If you're willing to sit on NJ Transit or the LIRR for 3 hours a day, that is. (I don't see how self-driving cars are going to change commute times. There will still be traffic, and they'll still only go 60 miles an hour.)

I think human driver overcompensating for braking ahead, and human reaction time creating traffic "waves" contributes significantly to traffic. (At least that's what I've heard).

Also self driving cars wouldn't need as much distance between cars to be able to stop safely so you could physically fit more cars on a given stretch of road.

Traffic doesn't move at 60 mph during rush hour though.

Automated cars have the potential to reduce the lag in speedup after breaking, eliminate most accidents, and pick better routes, and all while letting you do other things with your time. Commute time is a huge reason to do this.

If I had to guess, I'd say better public schools, less crime, and affordable houses (which also makes bigger houses more affordable as a side effect).

This sounds like some perverse suburban fantasy. To an extent I'm okay with the idea of people following their preferences, but the fact is that the market is being undermined by municipalities in this case. All those land owners in San Francisco are halting further development of the city, driving prices artificially through the roof and driving people who might prefer to be in a more urban setting into the suburban wasteland. Where at least now they can tool around in their robotic cars. Fucking wonderful.

On a less bitter note: as someone who lives in the future (future circa 1930) and spends 0% of his day in unproductive commuting, you west coast folks are in for a treat in 20-25 years when (if) this happens. That extra hour or so a day of unstructured time to read, surf the net, or just veg is really wonderful.

Like virmundi brought up, landowners are not the bottleneck in San Francisco - if anything they'd love to knock over that 3-story Victorian and put a huge apartment building in its place. After all, those 5 apartments in his existing building cannot possibly be worth as much as 30 brand new ones, no matter how stratospheric the rent gets.

The root cause of almost every San Franciscan problem is rampant NIMBYism. Look at a bus map on the city and notice how closely the stops are spaced together - you have buses that stop on literally every block due to local lobbying, and any attempt to remove some redundant stops to make things run faster elicits an incredible wail from local constituents for whom the only thing that matters is themselves.

Development is the same story. People moved to SF for the pretty Victorians they get to see when walking down the street, and any attempts to change this landscape, no matter how mindful of history, art, and architecture, is struck down to preserve the status quo (wherein the status quo is a sick, perverse emulation of the early 1900s).

San Francisco is the city where everyone is trying to squeeze through a very small door, while the people who just squeezed through are busy trying to slam the door shut behind themselves.

I agree with you, I should have used "resident" instead of "land owner." The people that vote on the municipal boards, whoever they are.

I agree with the fantasy part. I do want to point out that the landowners are not necessarily halting building, but they are limited with what they can do. San Fran needs to build up. Unfortunately, they are not allowed (http://www.planetizen.com/node/39697).

I think a lot of these things also aren't thought through:

1) Driverless cars might possibly allow higher capacity/speeds on major highway thoroughfares, causing cities to expand. Or, they might totally solve the "last mile" problem of commuter rail. High-speed car travel dramatically increases wear and tear on roads, and building more highly-worn roads further out might not make as much sense as building out rail.[1]

2) Driverless cars may case semi-urban areas (like Palo Alto or Tyson's Corner in Virginia) to contract, because of the dramatic reduction in parking requirements.

It's not clear that the prevailing trends are towards bigger, sprawlier houses. Indeed, at least on the east coast, the development seems to be on moderate-density mixed commercial/residential in places like Northern Virginia. In the Chicago area (not east coast, I know), satellite cities like Aurora have been growing at 20-40% the last few decades. In the D.C. area, Arlington has been growing at double-digit rates over that period, and Alexandria has averaged about double digits. I can easily imagine driverless cars moving this trend along, replacing core cities with sprawling suburbs with core cities surrounded by dense satellite towns, connected by rail.

[1] One of the interesting things about say the Metro North railroad I take to work is how relatively undisruptive it is to the surrounding area. There are nice houses a stones-throw away from the rail line, while there would never be such houses that close to a highway. And even the 2x2 express arrangement fits the capacity of a major highway into the space of a very modest suburban road. The whole Metro North system moves 300,000 people into Manhattan each weekday and operates on a meager budget of $200 million/year. I don't think there is any way you could maintain a comparable highway network for that price, and that's before you factor-in gas/mileage on the vehicles (which, operating at higher speed, are going to use more energy and wear out faster--it's simple physics).

EDIT: I've seen 7-15 cents per vehicle mile quoted for maintenance and traffic services. Metro North probably represents a few billion vehicle miles equivalent.

Cars shouldn't be allowed inside cities at all, for obvious reasons (air polution, noise polution, overall well-being of all citizens).

And secondly internal combustion engine is an inventian appropriate for 20th century, certainly not for the present age (because of its 20-25% efficiency, and when you take into account that that's engine efficiency, and the power output is actually used to accelerate several tons of metal, actual cargo is insignificant in most cases, you end up with 2-3% efficiency). This is really one proof of the incredible human stupidity.

I would like to see most European cities in the following two decades, as having a traffic made up of bikes and few smaller electric cars.

On the topic of driverless cars, it is solving the wrong problem. The advantage most of you mention is more time available for people. That's not really true in my view, because: when I'm in a car, I feel physical distress, some kind of motion sickness, and a great portion of the population has similar symptomes. This greatly reduces the amount of activities one can perform in a car. And secondly, because we're talking about replacing cities (distances < 30km), the best solution is to replace cars with bikes, and you get the additional exercise that's missing from most people's lives, and you don't need to waste time in gyms, which are not as healthy, and don't result in as much endorphines/happiness (I don't have studies to cite for this).

And getting back to the topic of driverless cars. As a cyclist I would definitely feel safer around automated cars (but that's mostly because the place where I live is also inhabitated by many uncivilized drivers with no regard for other traffic participants, and no respect for the law, (Eastern Europe)). But then, the add efficiency of automated cars is quickly lost when you add cyclists and other vehicles to the equation.

The bottomline : wrong problem, very few benefits, except in places like Eastern Europe, where it can't/won't be implemented because of social and economical reasons.

Wouldn't a bicycle be a lousy transport during the winter or when it's raining? Or when you need to carry something like a suitcase to the airport that's 20 miles away? Your reasoning seems to be incompatible with a bus system too.

A reluctant upvote, but only because the author has put more thought into the matter than the typical futurist.

But for some reason, he still is victim to the same trap that befalls many futurists: A misunderstanding of the new problems that arise.

Cheap taxis that don't require parking and can drive closer together at higher speeds is great. But new problems arise:

1) Bottlenecks on streets due to the expansion of cars in dropoff zones. Imagine the typical airport dropoff zone on a busy day...and then think about that happening on nearly every street. Through-traffic is guaranteed to grind to a halt.

2) Smarter cars do not fix transportation networks. Braess's Paradox is not just a theoretical scenario; It has been observed in multiple cities and sub-city networks. Increasing road capacity can and will increase travel times...and it gets worse when everybody is independently optimizing their paths while driving in the same direction.

3) Supply and demand matching for taxis is an easy problem when you assume spatially random demand for travel in spatially random directions during temporally random times of the day. In the real world, taxis have a limit on the number of places they can be at the same time...and the non-randomness of times, origins, and destinations creates real problems with the availability of taxis...in addition to carshares, buses, trains, etc. Thus, the hypothetical futurist scenario relies upon an abundance of taxis, which may or may not occur due to the delicate balancing act between marginal utilization and capital/operating costs. A taxi that is used twice a day during rush hour is a taxi that does not exist.

4) Deadheading (driving to a demand-location while underutilized) has costs of its own: Now your taxi ride is paying twice as much for fuel for the same amount of of human transportation.

A more realistic future involves smaller net benefits which can be only be partially offset to the extent that modification of urban geography permits; We may not be able to turn a hilly isthmus into a flat plain with grid streets, but we can move away from our archaic zoning practices that ensure non-random travel vectors. Carpool lanes will continue to have benefit. If, and only if we allow it, transit systems will become more efficient and less costly on a vehicle mile basis, compounding the benefit of grade separation in the still-gridlocked future.

Driverless cars also means that taxis can be scaled a lot more according to real time demand. Unlike taxi drivers who have to work on a x hour shift, on demand driverless taxi can start and stop working at any time. Second, if people can call a taxi using a smart phone app then taxi's don't need to wondering around the city without passengers. These two factors combined will further dramatically lower the cost for taxi more, a lot more than the 2 thirds cost cut down this article predicted (http://www.theglobeandmail.com/report-on-business/economy/ec...). The cost (and effort required) of using such taxis as main means for transportation will be significantly lower than owning a car even without the concerns of parking expense (in rural areas).

Does that mean that cities will further spread out or the other way? Who knows. I think it means more freedom for people to select wherever they want to live. You can live in the middle of Manhattan without worrying how to get to the countryside with your dog. You can also live in the mountain and commute 5 hours a day to Manhattan to work (because that 5 hours won't be lost in driving any more)

Can anyone tell me if a driverless car hits and injures someone and it is the fault of the driverless car. Who is really at fault? Does the victim sue the driver, the software writer, both.

I can see data logging and pervasive monitoring will be needed in order to prove who was at fault, or what exactly happened.

In anyway I see somewhat of a legal obstacle more than a technical at this point.

EDIT: clarification, removed "hits me" reworded for clarity.

I think the legal situation is clearer than it looks. Hitting someone due to software failure isn't any different than if it happens due to a break or cruise control malfunction. As long as the equipment is approved by your local traffic authority and you did proper maintenance, you can prove it wasn't your fault. (not being your fault won't stop anyone who wants to sue you, but that's another matter entirely)

>> As long as the equipment is approved by your local traffic authority and you did proper maintenance, you can prove it wasn't your fault.

Which is exactly why I'm still 100% convinced this whole pipe dream will disappear the moment self-driving cars become available to consumers, and a few serious accidents happen. No car manufacturer on earth is going to continue selling cars with the promise they can 'drive themselves' if that makes them liable for accidents. Exhaustively testing a cruise control system is one thing, but ensuring at least 'nine nines' of reliability of all the hardware and software systems of a self-driving car, in every possible situation it could be used, is simply intractable. Anything less than nine nines reliability, and someone will die because of a problem with your self-driving car at least once every week or so.

> Hitting someone due to software failure isn't any different than if it happens due to a break or cruise control malfunction.

I kind of agree with that. That is why I inserted the bit about pervasive logging of events. This technology will increase that many times over. Any manufacturer getting into this will need that in order to prove beyond any doubt that their software did the right thing.

NOW bugs happens and in this case bugs will kill people at some point. So, can the manufacture be trusted with keeping logs themselves. The answer is no. Not if they have a strong incentive to make it look like it is not their fault. It it basically either their ass on the line or driver's ass. There will always be a question of "ok, so they do show that user pressed the pedal and tried to take manual control of the car, but how do we know they didn't insert that artificially in there...".

I think there will be a need for 3rd party or government mandated continuous logging. Cameras, following distance sensors, accelerators, local road conditions monitoring and so on, so during a law-suit this 3rd party's product (who supposedly should have not interest in falsifying the data) would be used in establishing fault.

Fault and liability aren't the same though. Even if an accident is caused by your brakes failing, it's still your car that caused the damage and it's you (your insurance) that will pay. Even if it doesn't result in a traffic citation on your driving records.

If I were judging this case, I'd call a car an inherently dangerous instrumentality and hold drivers strictly liable.

The question becomes who is the driver? A passenger is normally not liable in the event that a driver causes an accident currently. If your car is driving itself, then you're inherently not the driver, you're a passenger. Especially if driverless cars are allowed to be ridden in without needing to pay attention or assume manual control, which I believe is the end goal for this technology.

So if you want to hold drivers strictly liable for accidents, but the only person in the car is the passenger, then question still remains unanswered: who is liable?

Interesting perspective.

One can take various perspectives here: the person injured, the driver, the car manufacturer. It has ramifications for all of them.

The general questions is, "is the judicial system ready to deal with robots, sold to others, who end up injuring and killing people?". I don't know if it is.

Cruise control is there as others pointed. A car on a cruise control hitting someone will be the fault of the driver pretty much.

There are also law suits about failing breaks and other defects, those after extensive back and forth, end up eventually being a manufacturer's problem.

Are you a judge? do you have any background in traffic laws and/or driver liabilities? Or is this just your two cents...

Just my two cents (though as someone with a JD).

The idea that manufacturers would be liable for collisions is quite laughable. Is your computer manufacturer liable for the damage resulting from system crashes, bugs, etc? Of course not. Manufacturers are liable for design or manufacturing defects but a collision in a driverless car does not necessarily indicate a defect. They aren't magic--a certain number of errors are to be assumed. In some cases, it may not be physically possible to prevent a collision (computers react at the speed of electrons, but cars only at the speed of big heavy physical objects).

No, I think driverless cars will be handled by extending the concept of "inherently dangerous instrumentalities." If you're say demolishing a building with explosives, your standard of liability is not whether you were doing so carefully. It's strict liability: it's an inherently dangerous process so you better carry sufficient insurance because if it goes wrong you're strictly liable. Cars will, and should, be considered the same way, and will likely be handled through no-fault insurance systems paid for by drivers.

Interesting. Thanks for the response.

One reason computer manufactures aren't liable is because a computer is an "open" system (so to speak) where a user can introduce all sorts of external variables that may affect it's performance/stability (viruses, malware etc).

Ideally the software in the car is closed so malfunctions cannot be attributed to user intervention. And if one of the sole jobs of that software is preventing collisions thought a series or calculations, sensors etc, it seems like a failure would be more on the manufacture's end than the driver but thats just my opinion.

> One reason computer manufactures aren't liable is because a computer is an "open" system (so to speak) where a user can introduce all sorts of external variables that may affect it's performance/stability (viruses, malware etc).

That's not the legal reason (and manufacturers aren't any more liable on totally locked down platforms than on open ones). The legal reason is that we don't hold manufacturers liable for the ordinary dangerous of using their product (as opposed to the extra-ordinary dangers of using a defective product).

I'd imagine most collisions in driverless cars won't be from real design defects in the software, but rather from the simple fact that the real world is very complicated. If a driverless car is going around a blind turn and has spotty cell reception, or a deer jumps out in the middle of the road, or the road is wet, the algorithm could be doing everything right and still get into an accident. Again, driverless cars aren't magic.

I don't think there is any situation in where manufacturers are liable for something other than a major, reproducible, foreseeable defect in the software.

The question is, if the manufacturer isn't liable, who is? Sometimes life sucks and nobody is liable, but the law tends to avoid that. Hence my prediction that the liability will fall to drivers as it does today, on the theory that "you should expect that your 3,000 lb hunk of metal barreling along at 60 mph could hurt somebody."

liable for collisions is quite laughable.

Without a root cause analysis of the accident I am not sure I would call it laughable. How would driverless unintended acceleration be any different from Toyota's recent unintended acceleration recall and liability? If root cause is a defect then they will certainly be liable. Now that isn't to say any and every crash involving a self driving car automatically becomes their problem, it will be on a case by case basis just like now.

Right, that's why I said manufacturers would certainly be liable for design or manufacturing defects, just as they are now. But I can't imagine they'll be liable for routine fender-benders that aren't the result of some systematic defect.

I really don't see what's so hard about it. I think people are over-thinking the situation because they're anthropomorphizing the software. If a driverless car collides, the software isn't "at fault" instead of the "driver". Fault is a legal classification assigned to people. Rather, the car failed, just like cars can fail today, for all sorts of reasons. If that failure can be traced to some systematic problem the manufacturer could have foreseen, then the manufacturer might be liable. But if it failed for the various random reasons anything fails, because nothing is perfect and very complex things tend to be very much not perfect, then such failures will just be deemed an inherent risk of the vehicle, and whichever person owns and operates it will be liable, just as they are liable today for the inherent risks of the things they do.

Wow, what a breathtakingly misinterpreted vision of the future of cars. Off the top of my head:

1) Parking lots will not dwindle in usage. What a silly notion. In fact, the demand for cheap driverless taxis will probably make public transit all but obsolete, thus causing more vehicles to occupy the road. Just because the cars will be in high demand doesn't mean they won't need a place to stop and wait on occasion. I could definitely see more efficient and sophisticated parking structures becoming more commonplace, ones where human usage isn't a consideration. Where do you think all those cars will go at 3am? They're not going to just keep driving around...

2) the author seems to think at some point, driverless vehicles will become so ubiquitous that policy would ever pass that stripped the usage of freeways away from drivers. As if that was even enforceable, there will always be a significant enough subset of the population that won't own or choose to use a driverless car. That's what makes driverless cars rather than driverless car infrastructure such an important development. It doesn't ask the world to change around it, or ask people to change.

> Cities will greatly expand, again: Faster and more efficient transportation will convert locations that are currently too remote for most users into feasible alternatives, abundant with space.

Here's my alternative idea: cities will contract because you won't need to ever have your own car anymore (with all the required space that entails), city centers will keep making driving your own car more of a pain (as is already happening in european cities) and as gas prices keep increasing the incentive of owning and using your own car will drop, just as the cost of using self-driving electric cars on short in-city commuting will do.

Either way, in many places (like San Francisco) the constraint isn't places that are too far away, it's places which have too much traffic between here and there. If you could replace a freeway full of human-directed cars with a freeway full of robot-directed cars, this would let you put more people not just further into rural areas, but around existing neighbor cities too. (One of the reasons the SF peninsula is interested in new "transit-oriented developments" near Caltrain instead of other developments: highway capacity on US-101 and 280.)

True, but the constraints to construction and therefore to densification - particularly in the Bay Area - are immense.

When it's hard to build in the city and it's the urban fringe is a miserably long commute away - housing prices go up. When it's hard to build in the city and the urban fringe suddenly looks feasible, because traffic moves faster - you see city expansion.

Agreed. Owning a car, let alone two or more for a family, is currently very difficult or prohibitively expensive in an urban setting. If you can have the same access as owning, but without the hassle, this removes a barrier for many people. In fact, car-sharing would likely have higher availability and therefore lower wait time in densely populated areas, in some ways increasing your "travel freedom" compared to an exurb.

"a technology that condenses the time needed for commuting along the same route – and allows doing so in the back seat – will make [affordable, spacious dream] homes more agreeable"

As a New Yorker who spends a lot of time in back seats, I think the author is exaggerating the effect driverless cars will have on the palatability of commute times. I am more sensitive to commute times than my friends in the Bay Area, gladly moving to cut a 20 minute commute to 10. Even though I have a car taking me door-to-door, and even though I'm reading my email and screwing around on Hacker News the whole time.

In other words, no matter how great, easy, convenient or pleasant things get, people will find a way to complain.

>"a technology that condenses the time needed for commuting along the same route – and allows doing so in the back seat – will make [affordable, spacious dream] homes more agreeable"

Ugh, more sprawl.

Hummmm... Too optimistic. I think the driverless car thing is going to be hampered by rural living as well as the dependence on fossil fuel. Refueling is still a major limitation to this. I don't want my vehicle parking 5 miles away from my work using up more fuel at my expense. It sounds like a recipe for me hating cars in general.

This kind of thing would also give techno terrorists a very easy way to cause havoc on the masses... I think we would need to take cyber security quite a bit more serious before this could take place.

People want mass transit ... as long as it looks like a car.

People want point to point mass transit.

I'd love to use Mass Transit for my daily commute. But at 1 hour for taking the bus vs 10 minutes to drive, it just doesn't make sense.

People want a transit system that takes them where they want to go when they want to go, that is clean, that they don't have to share with strangers, that they won't have someone asking them for "spare" change [1], that has their personal accouterments available for them, that allows them to do other things with their time, that isn't terribly expensive, that doesn't smell like urine.

[1] http://www.davebarry.com/gg/newyork.html

[1] "I once watched three German tourists -- this is a true anecdote -- attempt to get off the northbound No. 5 Lexington Avenue IRT train at Grand Central Station during rush hour."

Yeah, taking the Lexington Avenue Line is usually a mistake if there's any alternative (e.g. walking, crawling, just staying home, or getting kidnappend and carted off to southeast asia for random organs to be harvested...) Rush hour just makes it worse.

If automated cars stop just one drunk driver from killing someone, I think the whole invention will be worth it if only for just that.

Because I've come to realize people are just going to drive drunk no matter what the penalties may be and judges are going to keep letting them back on the road. So maybe robots will cause better behavior despite some horrible people in society.

  | If automated cars stop just one drunk driver
  | from killing someone, I think the whole
  | invention will be worth it if only for just that.

Well, not if they cause more deaths from malfunction.

That's a valid concern but sensors and computing hardware is becoming so inexpensive it's possible to have two or more redudant systems that check each other's results before causing an action (or if one fails).

That is how the space shuttle's computers worked based on 1980s technology, I think there were six of them checking each other's results.

One of my biggest questions is what does this do for Children and young teens? Can they now ride in their families car by themselves? Do rich 12 year olds get their own car?

Driverless cars could open up entirely new age ranges to "driving" (10-16 and the elderly who can't physically drive anymore). And if these kids are riding around by themselves, what impact does this have on their social development. I enjoyed getting picked up by my mom or dad and the discussions we had in the car; and I look forward to picking up my children and talking with them about their days. But this may not be something future parents feel is important and those lost interactions would be a shame IMO...

Some of the comments here point out how driverless cars would allow people to just call for a car when they need it, a sort of on demand taxi.

I'd like to take it one step further and point out that it may make electric cars even more viable. If a car is going to be waiting until someone calls for one, rather than driving around the city looking for people - we could have a certain factor of "extra" cars charging while they wait.

When you then request a car, the system would send you one that is sufficiently charged. When you arrive the car will reenter the system and recharge.

I welcome driverless cars!

Tens of thousands of people die annually because of car accidents. Driverless cars should make sure that number decreases _dramatically_.

Second, driverless cars should use fuel more efficiently and try not to cause too many jams. Optimise traffic and make it less polluting.

Driverless cars don't need to be stored in/near your home; they can be parked in big neighbourhood parking facilities near-by and come pick you up when needed.

Driverless cars should put an end to the auto insurance mafia.

There will be far less car accidents.

Reply on news.ycombinator.com