Author is missing a few important points that are highly relavent:
1) No insurance implies no liability. When people "buy" these mostly autonomous vehicles, who is assuming liability? Google?
2) "Death of car companies" - is Google going to start making cars? Are they going to compete against years of engineering? Just because they are building the brain doesn't mean they will be building the entire body.
3) The auto industry is heavily regulated, and moves very slow. 5-10 years is a very optimistic estimate.
4) Cars are moving mechanical parts. They require maintenance and repair. How often do you go to the car shop because of an accident? How often do you go for general repairs?
5) Sure, parking revenue might go down for the city. Traffic at those same spots where people did not go because it was hard to find parking will go up. How often have you considered going somewhere else because of parking?
6) Not sure what percentage of the legal system is bogged down by "driving" related crimes. Same for health industry, vehicle related health service.
The author is making the mistake of equating Google moving into this space as Google being able to do all. Google will not be able to manufacture cars for the near future. They will not be able to eliminate liability. They will also have to change driving culture and habits. Having discussed this with friends, a fair number of people just want their own car that they drive.
Having discussed this with friends, a fair number of people just want their own car that they drive.
I've encountered the same reaction, and I believe it is widespread. But here's the thing: even the people who insist on owning and driving their own cars, and refuse to let it drive them, can still benefit greatly from owning a self-driving car.
Imagine never having to park - you just get out of your car in front of your destination and it goes and finds a parking spot.
Imagine never having to get someone to drop you off at the airport. Your car can just go home afterwards.
Imagine getting your oil changed or other maintenance. The car can do this on its own when you don't need it.
I believe that we'll see a lot of marketing geared towards the people who would never let a car drive them around. This will be a great stepping-stone to help convince these people that it's safe. They'll get it for the convenience, but then eventually one day they'll use it when they're drunk or tired. Eventually, they'll learn to trust it.
Well, not everyone will see it this way. My example:
> Imagine never having to park
I like knowing where I parked. Further, we have something called valley parking, in most places that I would want to get off for a while.
But besides, in crowded cities, how do you actually imagine that? Most car owners in the NYC take subway to go downtown, because it would either take 2 hours to find a parking spot, or it would be very far from initial destination. With AC, I can imagine people just getting off at their destination (right in front of the door) and won't give a damn what their car will do in the meanwhile until they are back in 45 minute. So you will have jam packed streets with empty cars driving around waiting for their masters to come back -- total horror on the streets! Otherwise - what? Have the car to drive back home empty and come back in 45 minutes?? That putting even more traffic on the street and more pollution.
> Imagine never having to get someone to drop you off at the airport. Your car can just go home afterwards.
Sure, or I can take a cab which will be the same result - I will get to the airport after all. The only difference, I can have one or two to relax at home with wife, before the flight.
> Imagine getting your oil changed or other maintenance. The car can do this on its own when you don't need it.
Yes, so I am even more disconnected from the real world. Instead of making a Saturday trip to change an oil and chat with mechanics about good old days, about whether, girls, beer, or whatever it is, I am sure better staying home to chat on Facebook, while my car gets oil change on its own...
> I believe that we'll see a lot of marketing geared towards the people who would never let a car drive them around.
Probably not. I would guess all makers will jump ship quickly. There will be money to spend on advertising new technology, not the old one.
> This will be a great stepping-stone to help convince these people that it's safe.
If its going to be safe, noone will need to be convinced. I am sure, even reading this thread, you will find enough initial adopters for them to pass the word to the masses how safe AC really are.
Instead of making a Saturday trip to change an oil and chat with mechanics about good old days, about whether, girls, beer, or whatever it is, I am sure better staying home to chat on Facebook, while my car gets oil change on its own.
If that's how you prefer it, that's fine. But all my mechanic ever chats with me about is about how much I need his $49.99 radiator purge. It's not how I want to spend my time.
If its going to be safe, noone will need to be convinced.
Many people have an innate fear of things they can't control. Think about all the people who fear flying on commercial airliners -- quoting them a statistic about its safety doesn't make their fear go away. There will need to be a lot of convincing.
I like knowing where I parked. Further, we have something called valley parking, in most places that I would want to get off for a while.
You wouldn't need to know where you parked: a quick text message retrieves your car.
I will get to the airport after all. The only difference, I can have one or two to relax at home with wife, before the flight.
The end result for the self-driving car is identical to the cab case, you can still have your drinks because you will not be driving the car. However, removing the cabbie from the equation will cut costs dramatically.
Again -- where all those cars will be parked, knowing before I would either take a cab, or take a public transportation?
Will we get millions of parkings underground built the same time AC hit the streets? Dont think so.
Further, you saying that removing cabbie will cut the cost. Hmm.. my take is that it is the city that makes the most money out of cabbies. I can bet my left arm (Im lefty) that if the city will realize they losing large chunk of money due to AC driving everywhere, I am sure they will smack you with such an AC ride tax that taxis will become again a reasonable alternative!
I think it's an underrated step between self-driving cars and cars that drive unoccupied, as in your example of cars that go park themselves. Adverse weather conditions, mechanical failure, a simple flat tire -- if something goes wrong with a car, we expect the owner to be there to deal with the problem.
That's not to say that the problem can't be solved, but my guess is that it will take time and culture changes to adapt to these cars before people have the kind of comfort level to allow empty cars to roam around.
For mechanical failures, such as flat tires, the car could just pull to the side of the road and fire off a text message to you and AAA (maybe we require you to have something like a AAA membership if you want your car to drive itself). They come and fix the tire, tow you to an auto shop, or whatever needs to be done. You're inconvenienced, but certainly no more than you would be otherwise.
As for cultural changes, here's what I imagine will happen: At first, people will be uncomfortable with self-driving cars. As another commenter noted, people don't feel safe with things they don't control. But most users of the cars will need very little time to get used to it. They'll hop in their car, tell it where to go, and proceed to pay no attention at all to what the car is doing. Once enough people have had this experience, it won't be difficult to convince the public to allow the cars to strike out on their own.
My theory is that we'll see the rise of what the author calls robo-taxi companies e.g. Uber or Google can provide car-as-a-service, for a monthly fee or
1. robo-taxi companies will self-insure. Insurance is just spreading the risk. If you have a fleet of 10.000 cars, self-insurance will be cheaper
2. by "death" he means "significant decline", not "disappearance". A company that buys 10.000 cars has a purchasing power that a single customer doesn't. They can dictate the specs to a car company and they can pit several companies against each other to get the best deal. They can even co-design the cars for their own metrics (e.g. fuel efficiency, reliability, maintainability is super important for someone who has a large fleet).
Car companies will have no choice but to play ball, some of them will go bankrupt and the remaining will service a much smaller market.
3. You assume that self-driving cars need a lot of new regulations. They don't - they just need to be allowed on the road. The rules have already changed in Nevada in California. When the technology is proven, the states will have a hard time justifying not allowing technology that is proven to save a significant number of lives.
4. robo-taxi companies will bring repairs in-house or outsource to existing companies. Again, they'll have upper hand in negotiations due to providing a lot of business and the overall market for repairs will shrink dramatically. Google could bring all their statistics knowledge to predicting what and when will fail and use the past data to improve the parts that fail the most.
5. not sure what you're driving at
6. agreed, probably overstatement on the author's part
Google (or some other robo-taxi company) doesn't have to make their own cars, they'll have enough pricing power to pit multiple car manufacturers against each others to get the best deal, drive down the margins etc., just as Apple does with iPhone parts manufactuers (where Apple gets most of the profit from the phone).
They can self-insure - the liability never goes away but will be manageable from revenues.
As to people who insist on owning a car - most of them will change their mind because of price. If you leave college with $50k in student debt, are you going to spend $10k on a new car + monthly insurance + gas + cost of garage ($100/month in San Francisco if your apartment doesn't already come with it) + hassle of owning a car or will you pay $100/month for much more convenient car-as-a-service deal?
Give a 50% overhead and make the cab driver's cost $15 an hour. I used to pay about $50 for a half hour cab ride from LAX to Westwood, tip and surcharge subtracted, call it $40. I would still find a "no driver" taxi ride for $30 expensive.
The point I am trying to get at is that the economies of traditional taxi companies and future "robo-taxi" companies will not be all that different. Taxi companies will not be repairing cars. They already buy large fleets of cars. Also keep in mind that the taxi industry itself is regulated. Recall Udemey's issues in DC with the taxi lobby.
I do think Google will can be successful in the automotive industry, I just don't buy that they will be doing so by attempting to take down other existing industries. Google is entering a space that is much different than the web or electronic hardware. They have so much trouble breaking into the living room, not sure why you think they will make it to the driveway so quickly.
On (5), I meant a decrease in parking revenue is not necessarily a bad thing for a city. Easier parking also means less barriers to a purchase.
Good luck with this argument. I ran it on an earlier HN thread about the cost of a driver being only a small portion of the cost of a cab. But people don't want to listen. They think because they hand a driver a $50 note, the driver makes all the money.
Taxis cost money because of (a) running costs (b) finance/depreciation costs and (c) regulatory costs. (a) and (b) isn't going anywhere, and (c) certainly is not.
Drivers are only a small part of the picture. Removing them won't dramatically alter costs or availability.
Give a 50% overhead and make the cab driver's cost $15 an hour. I used to pay about $50 for a half hour cab ride from LAX to Westwood, tip and surcharge subtracted, call it $40. I would still find a "no driver" taxi ride for $30 expensive.
OK, now factor in lower insurance premiums...
Now cut the price of the car in half (no front seats, steering column, etc needed)...
Now cut out the pay for the time cabbies spend waiting for calls
Now cut the fleet size by 75% (better utilization due to intelligent routing, cabs never sleep, etc)
I have a feeling there are a bunch more ways this will get a lot cheaper...
>Now cut the price of the car in half (no front seats, steering column, etc needed)...
You're kidding, right? steering column, steering wheel + pedals cost a fraction of the cost of a car. That's assuming they aren't left in there for manual override (parking ,etc).
Front seats? Where are the passengers going to sit? Taking away 1 driver seat will just mean 1 more person capacity, which wouldn't affect revenue because very few cab calls are 1 seat over, resulting in two cabs.
Now that you've saved $500 on the cost of the cab by taking out the steering wheel (or making it removable) now add in the $x,000 worth of sensors.
>Now cut out the pay for the time cabbies spend waiting for calls
At $10/hr, the most you're going to pay for a cabbie sitting around all shift is $120. Which doesn't happen, because cab companies already effectively model utilisation patterns.
>Now cut the fleet size by 75%
Cab fleet size is dictated by peak usage patterns as a function of revenue, not by drivers.
Cabs can be autonomously profitable - many cities have privately owned cabs. The fixed costs of running a cab company are actually quite low in comparison to the marginal costs per cab.
Your 30 minute trip would cost €8.70 ($10.72). This rate includes everything, even insuring the driver, which I presume could get cheaper if these cars were all driverless. Driverless dispatching (instead of me having to find one on my phone and walk to it) could raise the utilization rate, and thus lower costs.
At the moment I believe these cars are given free/discounted parking in Amsterdam to encourage their usage. I can imagine that changing at some point, but I wouldn't factor that into my calculations as it's pretty easy to disrupt parking fees when all of your cars are driverless and highly utilized.
> robo-taxi companies will self-insure. Insurance is just spreading the risk. If you have a fleet of 10.000 cars, self-insurance will be cheaper
I can attest that Zipcar, in fact, already seems to do this. Insurance and gas are included with the hourly rental fee. In fact, just get rid of having to drive the damn thing and Zipcar already has the whole business model down....
1) You would still need some form of insurance. It's still possible for a human operator/owner to negligently cause accidents. (e.g. poor maintenance, chair strapped to the roof falling off at 65mph).
However, if premiums become low enough, car insurance policy might even be offered as an add-on to homeowner's or renter's insurance.
4) I think cars will evolve (or devolve) to require less maintenance, and more self-diagnosing.
5) If there are a lot of autonomous cars, owners can simply walk away from their cars, and the cars will coordinate amongst themselves to part in designated spots.
6) You know what bogs down city courts? People fighting traffic and parking tickets.
4) This has nothing to do with Google and more to do with auto manufacturers. I don't think the problem is that people can't drive their cars to the auto shop.
5) Agreed. Also better traffic flow if you can get the cars to work together.
6) You're probably right about city courts. I'd also love to see cops focusing on actual police work rather than waiting for people to run a stop sign. But we have traffic courts that handle these sort of things - we have assigned separate agencies to take care of it. It's hard to argue though that we have somehow failed to prosecute actual crime due to traffic and parking tickets - there are probably problems more severe that have caused that.
The author also is very naive when he says "no more bus drivers, taxi drivers, limo drivers". It may happen eventually, but not before a few decades. There are a lot of power struggles and political interests that will make this happen in a very very long time frame. Most probably, it will happen as it happens today with lots of aircraft's. While they can do lots of things autonomously, there must be a human to takes certain decisions and as a backup.
What I can see instead, is lots of truck drivers watching TV or just sleeping at the steering wheel, while their trucks do the job.
Bus drivers and those represented by unions and assciations will take a while to be eliminated from being necessary.
It hasn't been necessary to have train operartors/conductors for a few decades but they persist, despite trains being by their nature of prescribed, unchanging, predictable routes with preditable stops, prime candidates for automation. Monorails, for some reason are an exception.
One reason is that trains are hugely more efficient in terms of passengers carried per operator:
On a bus carrying 50 people, the driver is a much more significant portion of the cost than a driver and a conductor on a train carrying 2000 people.
So even though it's much easier to automate a train, there's less incentive to do so, and doing so still requires investment. It'll slowly happen anyway, of course, but there's no real rush.
The changes the author proposes will not occur for the vast majority of the population of the earth - not in China, or in India, or anywhere in Africa and not in many places in south east Asia or south/central America. He has the first world, US-centric view of most futurists. In places where people don't have reliable access to electricity, or even safe drinking water, absolutely no one would be interested in self-driving cars.
He also raises a disturbing point, but masks it as an advantage - "People said it would take years to get these things on the road. It took Google a few months of lobbying Nevada for it to happen. The US states and cities are broke, and Google has $43 billion in cash." In other words, he implies that this will happen quickly because corporations are able to strongarm broke government institutions into policies favouring them, a trend that is dangerous for the average citizen.
He also misses that self-driving cars becoming a norm leads to more car-centered infrastructure over public transportation, leading to a continuation of the vicious circle of over-reliance on cars.
But the developing world is developing fast. Some developing countries never bothered laying down landline telephone networks because it was cheaper and easier to just put up cell towers--in other words, they leapfrog old technologies. Why is it hard to suppose that developing countries might leapfrog regular cars and adopt autocars straightaway?
Because unlike wireless communication, autocars don't fundamentally change anything about the way cars use roads - they certainly don't make it cheaper or easier to implement the infrastructure.
You could certainly build smaller roads and smaller autocars, since you wouldn't have the margin of error involved with human drivers and larger vehicles.
> self-driving cars becoming a norm leads to more car-centered infrastructure over public transportation
Out of curiosity, why would this be? I can see that commuters who despise driving may switch to cars from public transport, but at least from personal experience, I don't see this as a major reason for people using public transport in the first place.
Is there another reason I'm missing? I'd imagine it would be more likely that self driving vehicles end up improving public transport (both in quality of service and popular usage).
I think there would be a feedback loop in place where self-driving cars demand better infrastructure to support them, and as this infrastructure improves, self-driving cars become more convenient to use, leading more people to want to use them, leading to more such infrastructure. Since this infrastructure gets funded, it's not much of a stretch to think that most transport money would go there.
Some of the technology might make public transport more efficient (self driving buses seem like a good idea), but cars would probably dominate.
It will eventually happen, but in order to do so, you first need a catalyst - a major metropolitan area such as Manhattan that converts to 100% robocars within a geographic area. The problem is that if you mix human and robocar drivers, any car company making a robocar is going to get sued into oblivion when they collide with a human driver, regardless of who is at fault. And, there are definitely ways that humans can cause robocars to have unavoidable accidents (swerving across the double yellow line at the last split second, etc).
I think that's a really good solution, honestly. Just deploy 100,000 robotaxis the size of smart cars in Midtown and ban all other vehicles.
On a side note, NYC pedestrians are already pretty aggressive at jaywalking and the fact that robocars stop on a dime when they sense a pedestrian in their way will create all kinds of new ways for pedestrians to harass robotaxis.
any car company making a robocar is going to get sued into oblivion when they collide with a human driver, regardless of who is at fault
I'd actually be somewhat surprised if this were the case. Self-driving cars will gather an enormous amount of sensor data and I bet they'll store that and decision tree logs in a "black box" of sorts. If this isn't being done now, I guarantee it'll be done after the first few lawsuits.
With that info in hand, it should be rather clear – at least in the majority of cases – who's at fault. The human driver can claim they swerved over the double line to avoid an obstacle, but the sensor information can probably confirm whether that's actually the case.
This isn't to say there won't be lawsuits, or that self-driving cars will never be at fault. I just expect there will be a rash of initial lawsuits. Then, as it becomes clear that the data recorded by self-driving cars makes them a difficult target to exploit (and assuming self-driving cars are in fact less error prone than human drivers), I expect lawsuits to taper off.
You are exactly right -- no company with half an ounce of brains is going to make a robocar that does anything except obey every traffic law on the books. Barring a software failure it's almost inconceivable to me that said cars even could cause an accident! But the flip side is much more damning: congratulations, you're now driving the speed limit everywhere you go. You're following at five car lengths. You stop for yellows and anybody within a quarter mile of a crosswalk. Basically, you drive like a Florida octogenarian, everyone else on the road hates you and it takes longer to go places. I can see a few die-hards going for that proposition, but not many.
Actually, a software-driven car could follow cars much closer than (mph / 10) car lengths. That particular rule of thumb is based off of human reaction times. A safety buffer is still a good thing, but software will react several orders of magnitude faster than a person.
Also, if software-driven cars become common enough, you will probably get places a lot faster, at least in heavy traffic. Traffic flow could be much better than it is with humans driving. There have been several studies on freeway gridlock traffic that show that small driving mistakes (e.g. slowing down more abruptly than necessary) can butterfly-effect out into massive bottlenecks. And of course if software-driven cars cause fewer accidents, there will be less gridlock due to those as well.
Flip side: with accurate and trusted self-driving cars in mass use, speed limits could be raised, car following distances can be reduced. There's no need to keep traffic controls and flow patterns that were designed for humans when instead there's a network of autonomous vehicles.
Wouldn't it be fantastic to get on a freeway where the speed limit's well over 80, because the vehicles have necessary reaction times and are guaranteed to be well maintained (why not have a car take itself in for routine and regular maintenance when it's not in use)?
And regardless, sign me up for Florida octogenarian driving patterns. I'd love to get in a vehicle that takes me point to point with no effort on my part. I can read, nap, work, or fully interact with other passengers – things I can't do when my attention must be focused on controlling the vehicle and monitoring traffic around my car.
You're right, but I think that most people who would like a robocar place the most value on doing something else while en route.
Even if it took longer (which, I agree, it would (unless the traffic was all robocars, which I can imagine actually being more efficient in many traffic scenarios, despite their slavish adherence to regulations)) I still think that most people would be happier doing something important or interesting for 35 minutes than they would be getting there in 27 minutes that they weren't able to use because they had to drive.
When cars started to take over from horse and cart, it still took 20+ years for them to become commonplace. It takes time for reliability to go up, and cost to come down. It takes time for the laws to catch up.
What seems like a blink in time to us now (1880s-1900) was 20 years for the people living it at the time. Cars were, at first, a freak sighting - then gradually, little by little, they caught on, and millions of horses were, ahem, retired.
A big step change isn't needed at all. Just gradual introduction and refinement until they are a compelling proposition.
As for the legal issue - the first generation of cars will require a human driver behind the wheel. They'll remain repsonsible for the foreseeable future. Likely the answer to long-term compensation for injured victims will be a shared insurance scheme - you pay the registration on your self-drive vehicle, and a portion of that goes into an insurance fund. If there is an accident and property/people are adversely affected, they get paid out from that fund according to a pre-determined formula ($x,000 for a limb on a 25 year old, $y,000,000 for a 45 paraplegic, etc). This isn't greatly different from schemes in place around the world today, where pinning fault on someone comes second to making sure those injured in traffic accidents are compensated + rehabilitated properly.
The biggest reason why it took a long time for cars to replace horses was high price (simply stated: most people couldn't afford a car even if they wanted one) and lack of infrastructure (it takes years to build roads).
We already have the infrastructure and the price will favor self-driving cars.
The laws heavily favor safety in cars (not everyone likes seat-belts and yet we have laws requiring people to use them; we have laws banning using a cell-phone while driving, driving while drunk; cars have to build in a way that withstand crashes etc.).
As long as it can be demonstrated that self-driving cars are safer (and they are), there will not only be pressure to allow them but to require them and eventually ban human drivers altogether.
Once the self-driving cars are allowed, the lower price and greater convenience will make it a blockbuster. I don't want to own a car, I don't want to operate it, I just want to get from Mission to SOMA fast and cheaply and self-driving cars will be able to offer that.
Roads existed long before cars. Even the Romans had paved roads. That is not a valid argument.
The reason people took a long time to take up cars was cost. It took time for cost to come down and reliability to go up. Once that happened, they were adopted widely. Even then, it took a long time for cars to disappear.
Any photo of San Francisco around 1915 or so will show plenty of horse-drawn vehicles on the roads, and cars will be present, but scarce. This is 7 years after the Model T entered production.
abruzzi has answered this well - not everyone can afford a self-drive car. They will not be cheap initially - far from it, in fact. I'll grant you that the cost-reduction time will be faster in this instance - nobody has to invent the modern production line - but the cost of development and warranting of all the parts will be expensive.
Replacing the vehicle fleet takes at least 10 years. Replacing it with self-driving vehicles could easily take twice that long, even if you started today, which we are patently not.
Price will favor self driving cars? What about all the lower income people that are limping along with $2000 fifteen year old cars? So if we give the industry 20 years to get to the point where you can but an old used first gen self driving car for $2k, how well do you think all those automation mechanisms will still function?
Doesn't even have to be a major metro area. Start small. Someplace without snow (Google cars still have problems with snow). Puerto Rico. The Bahamas. Hawaii. These are all small places where you can shake out the major issues before expanding.
Also, other places to start would be Nevada (as Google is doing), Texas, and other large states where there is a benefit to having something automated do the driving for 3-6 hours across large stretches of highway.
You're never going to have a major metro just ban human-driven cars overnight; you win by attrition, small bites very quickly.
Good point - better to start small. Vacation islands would be perfect as most people don't really want to drive on vacation anyway, and you don't have to worry about getting a shady taxi driver.
> the fact that robocars stop on a dime when they sense a pedestrian in their way will create all kinds of new ways for pedestrians to harass robotaxis.
Meh. I can solve this with one line of code. Every programming language has a random function =)
As a native, I can tell you that this is the solution employed by most human driven taxis already.
I would assume that any robocar would keep a record of its visual interface, like the airline blackbox. It would probably become much easier to judge who is at fault.
Hah. Unlikely. We still tax diesel cars several times more then petrol cars because we think they are dirtier. This is only just beginning to change. Tesla gave up here because of the frosty reception from our government. The authorities here tend to adopt a wait and see attitude rather than pushing the boundaries on things like this.
It makes sense on long haul trucking but it is going to have to get a LOT cheaper before it goes on passenger cars. 10+ yrs is my guess before you see it on luxury cars. Same as Tesla.
People who can afford a driver will replace them with Google Bots who won't sell their story to People.
I think this is missing one major thing: private car ownership will drastically decrease. It will be a luxury to own a car, almost everyone will be driven in available commodity cars. I personally hope this becomes public transit: available public cars just like some cities have available public bikes. Request it via your handheld (a la Uber), take it anywhere, hop out and it will either remain where it is or head to where it's needed. But it could just as easily by all private robotaxis.
Human drivers will be regulated and confined to certain areas. No amount of "but I like the feeling of driving!" can justify unnecessary death and destruction for too long, and anyway people will grow up not driving. It's more or less like paper books.
Eventually the infrastructure and car design will change, too. Our system is built for human drivers and cars. Obviously we won't abandon existing roads, at least not for a long time, but a lot can be changed cheaply that will have major effects once robocars are ubiquitous.
Without car accidents i bet the supply of organs will decrease dramatically, I wonder (in theory) if it might be enough to push them to sell on the free market.
Another thing it will due is drastically reduce the need for state police, I assume the majority of their work is traffic and traffic accident related.
The problem with organs is that for most of them, you need freshly-killed young people. Old people's organs are usually no good, as are the organs of those who die of natural causes (if their organs were any good they wouldn't have died), but healthy young people on motorcycles provide top-notch organs, assuming they weren't damaged in the accident itself.
YEah true, though I wonder in the new world of self-driving cars whether some incentives for motorcyle driving will go away.. that and I imagine a decent tranche of motorcyle accidents are caused by cars.
I have hoped/dreamed of this for 10 years. It will happen.
Another thing thats nice is we will be able to route efficiently. Tim Roughgarden came up with the "price of anarchy" of when we route selfishly (one of the most interesting papers I have ever read) check it out here: http://theory.stanford.edu/~tim/papers/optima.pdf
I understand the googly eyed take on "the way of the future" of cars, but self-driving cars are going to be too impractical to become ubiquitous for much longer than 5-10.
Maybe this all works great in metro areas with mild weather, but it's not going to work well in severe weather, poorly/wrongly marked roads, "offroad" areas, and construction zones, which is still a lot of the country at this point (and most of 3rd world). I've driven in some backwards places where there are no maps or GPS, and I've driven in near-whiteout conditions, and I'm sorry but no robotic system today can deal with all those variables.
It's also not going to handle human trade-offs well--e.g. sometimes it's better to crash your out-of-control car than to ram into a schoolbus full of kindergarteners.
You can argue these are all low probability situations or that eventually AI will solve them all and you're probably right, but I'd argue these situations are going to prevent full adoption for much longer than 5-10 years. It's going to take near-human intelligence to overcome all of them. Until then, humans will still drive cars, and probably a lot more often that you can imagine, even if everyone wanted the utopian vision to be true.
Sorry but I will always want to drive myself. I have a great car that I love and because of the way the world has changed since it was made (20 years or so) I don't think they'll ever make anything quite like this again. I intend to keep it forever.
Plus driving is way too fun. Driving != commuting... go find yourself a nice deserted mountain road, roll down the windows, and feel the breeze.
As long as your choices don't have any externalities, more power to you. People romanticize the days of riding horseback, too, but they don't romanticize the fact that a hundred years ago, every city street was covered in horse shit.
I imagine we'll still have recreational tracks and courses for driving, but "the joy of driving" doesn't justify the megadeaths caused by automobile accidents.
He addressed this point: you'll be able to drive yourself but it'll cost you much more than giving the control to the car (due to cost of the insurance) and much more than today (insurance spreads risks and in order to work, needs a lot of buyers, in this case drivers; if most people switch from current "person driver", expensive insurance to much cheaper "auto-driver" insurance, the "person driver" insurance will become more expensive because there will be no benefits of scale).
Ultimately, it doesn't matter what you will do, but what majority of people will do and his article is about those big-picture changes. Cars replaced horses as mode of transportation, with many consequences to society. Very rich people can still ride horses as a hobby, but that doesn't nullify the magnitude of changes brought by replacing horses by cars.
And your part of what will be known 50 years from now as an "old generation". When kids grow up with atomized cars (their parents will add the feature when their kids start driving out of fear, texting, friends, drinking, you fucking name it) they'll become accustomed to the experience and adopt it in droves.
Which leaves me with one question for you: do people still drive the Ford Model T on the road? My point should be obvious. Time changes things drastically. Either adopt the future or become obsolete. I can't stand people who can't stand change.
I don't follow... are the economies of scale that allow insurance companies to operate in the current market graduated such that if they only had to insure a fraction of the current driving population, they couldn't do it at the same cost per person?
Granted they will fight the shrinking of that market tooth and nail (and probably lobby to mandate insurance for all cars regardless of who or what is driving them), but it's not as though their cost structure stays the same once most of the population stops having accidents.
There is non-insignificant fixed cost to insurance business: people, advertising cost immediately come to mind.
But I see another scenario: traditional insurance business will not be willing to insure self-driving cars rationally (i.e. at much lower cost due to much lower incident rates) and robo-taxi companies will self-insure. That will deprive traditional insurers from a lot of revenue and since they are publicly traded companies, they won't just accept that and scale down to match smaller market. Instead, they'll try to prop up revenues in short term by rising prices which will make self-driving cars even more attractive to more people, fueling their own decline.
Insurance companies make money by investing the premium in investments. A correctly balanced insurance book will match premiums + claims over an expected time period. The idea is to invest the premium and make money off the investments. This is why Warren Buffet buys insurance companies, because he is a good investor.
If an insurance company is paying out more in claims than they are taking in in premium, that is when they will increase the premiums. If they are taking in more premium than they are paying in claims, they will generally identify areas where this is the case, and lower premiums to attract more customers.
There is a lot of ignorance about insurance companies on this thread. There is also a lot of ignorance that self-driving cars will require no insurance. Not all of the cost of insurance goes to insuring the driver hitting something. There are fire, theft and third party property + injury components. None of these will be removed, though some may be reduced over time in line with claims being lower.
Most of the extra cost of insurance in a large city has to do with the theft component (especially if not garaged). People who live in rural areas pay much less insurance than those in city areas, principally due to decreased theft rates. Autonomous cars will be stolen just like regular cars.
Are taxi companies currently allowed to self-insure?
I would bet current insurance companies are more likely to fight any legal distinction between traditional and self-driving cars as it would require a change to their (lucrative) business model.
Sure, but how hard would it be for Google to create a wholly-owned subsidiary that is an insurance company, which only happens to insure Google?
Patent troll Intellectual Ventures reportedly created hundreds of shell companies. Amazon creates tens of subsidiaries (e.g. Kindle is technically created by lab126, owned by Amazon but a separate company) etc. Whatever the legal roadblocks insurance companies might try to throw, I don't see how they can stop new insurance companies from being created and that's a simple loophole against any anti-self-insurance laws.
I've only been able to sympathize with what you're saying a very few times. Probably because I've always driven cars that I only view as utilitarian.
The interesting part to me is that the pervasiveness of this general sentiment in America is the reason I think self-driving cars will win out in the near-to-medium term over other options that are probably more efficient and cost effective. America is married to the car, and that's what makes the self-driving car the most likely to succeed new form of transport in my opinion.
Cars won't change 'Everything', self driving or not.
Thinking and building the infrastructures to handle smoothly different types of transportations for different purposes, managing the energy consumption and pollution problems, making efficient public transportation work in high population density areas, have smarter roads in the more rural areas and keep in mind communities' safety and well being while designing it all.
All these are large problem that cost a lot, require cooperation from big players from everywhere, and need to be done gradually and consistently over decades.
There was a very very interesting talk by Horace Dediu on this topic on his podcast [0]
Solving the low level infra problem changes everything, having self driving cars is the icing on the cake at best.
How will this technology work in say, New Zealand? There's a lot of backroads, gravel roads and roads that don't appear in Google Maps.
I guess it would be interesting to see a Google Car, flying around on some back gravel road in New Zealand at 100km/hr without crashing. :) (Google rally driving anyone?)
To me, there are still so many questions left unanswered...
First and foremost, the majority that believes AC will resolve heavy traffic in the cities is wrong. At the end, its about people, not cars. Think how you behave nowadays in the city crowded with cars -- you are perfectly aware not to f*ck with the driver, because if you step in his way you may get killed. We are humans, we make mistakes. Now, I would imagine those super cars must see everything around -- otherwise noone in the GOV in their sound mind would have approved AC to drive around humans. Therefore, I can imagine hoards of people taking advantage of this and just simply walking in front of a riding AC just because they know its not a human steering the wheel and that the car has no other choice than stop in front of the pedestrian, regardless if he stepped in and force the right of way or not. Heck, most likely the car will not even hunk or course at the pedestrian. Awesome! In the crowded cities, people will establish this behavior: before walking into the street, check the car - oh its AC, fine I am stepping in, no worries! And whos gonna enforce those violators? 100 more cameras at each corner??
Second, I still have no idea how on Earth would AC riding on a highway know that it is a black cat, not a black bag laying on the street. We humans see the difference faster than a blink of an eye. Are you telling me that Google invented perfect image recognition system that will be always right? What if someone paints a black cat on a bag that flies low on the highway?? -- will the car stop or not? I can see the news: google AC ran over Missis Jennings's cat -- noone is liable; noone will be responsible.
The bottom line: sooner or later AC will make a mistake. Hope it won't be in front of you.
Last, who is liable when the system makes a mistake? When you crash into other vehicle and there are victims. Are you liable? Can you go to jail? Will google go to jail? or car manufacturer? Are you telling me there are NO chances whatsoever that this will happen again?? [1] [2]
There are cameras on the autonomous cars, I'm sure they'd do the trick just fine.
> Second, I still have no idea how on Earth would AC riding on a highway know that it is a black cat, not a black bag laying on the street.
LIDAR. A black bag laying on the street is virtually invisible to it.
> The bottom line: sooner or later AC will make a mistake.
True, but they'll have to make a lot of mistakes to be worse than humans.
> Last, who is liable when the system makes a mistake?
That's a good question, and not one that can be answered categorically for all classes of motor vehicle accident. I'd imagine that the driver of an autonomous car would be liable for their actions in the same sense that a pilot is when they place their plane on autopilot (or a skipper when the use an autotiller to steer their ship). If it can be proven to be poor engineering on the manufacturer's part, then likely the manufacturer will be held responsible. Otherwise it will probably be the driver getting sent to jail. Or maybe a mixture of both if both parties can be proven to have acted negligently.
> ...but I can throw myself under riding AC.
Yet you're incapable of throwing yourself under a normal car? Doesn't this contradict your previous point about sabotage?
Though admittedly sabotage is a very common concern with automatic cars - what if somebody papers over a street sign or throws rocks at the car whilst it is on a sharp corner? In reality, malicious third parties are of concern, but not any more than if the vehicle was not autonomous. Human drivers can still be distracted by noisy passengers, intoxicated by alcohol and blinded by lasers; all you're doing by putting a robot in control is swapping the weaknesses of one system for the weaknesses of another. Ideally, there would be a mechanical backup system which could be controlled by the driver in the case of an emergency, which would effectively make autonomous cars more reliable than normal ones.
> LIDAR. A black bag laying on the street is virtually invisible to it.
so when the cats lays in front of it -- it will just run over it, not being able to see it?
> If it can be proven to be poor engineering
Usually after 10 years of deep investigation, given the parties have enough monetary resources to keep the momentum on the law wheel. Read the case I sent about the guy wrongly jailed. Had they not discovered, eventually that it was a car malfunction, he would have still been in jail.
> Yet you're incapable of throwing yourself under a normal car? Doesn't this contradict your previous point about sabotage?
I guess what I was trying to say is that I think getting killed by a car driven by machine feels easier on your conscious than being killed by a car who is driven by a human. I cant explain why.
> so when the cats lays in front of it -- it will just run over it, not being able to see it?
LIDAR stands for Light Detection And Ranging. Much like sonar (which is also used on an autonomous car) it is used to map 3D environments, producing models which can then be analysed by the computer. A plastic bag lying perfectly flat on a road would look like sensor noise, whereas a cat would be a noticeable obstacle (much like a fallen garbage bin or a basketball). This clip[0] from Udacity's CS373 class is great if you want to find out what an autonomous car "sees". In short, if you're a two dimensional cat, you're in trouble. Otherwise you should be fine :)
> Usually after 10 years of deep investigation,
What? I said "I'd imagine..." because that part of my post was pure speculation since AFAIK nobody has ever been in a serious autonomous car accident.
> I guess what I was trying to say is that I think getting killed by a car driven by machine feels easier on your conscious than being killed by a car who is driven by a human.
Unanswered or unaddressed by you? All of these questions have most certainly been answered or at least very seriously addressed with room for more conversation.
I was trying to say (pardon my English) that in the cities where the AC suppose to help, lots of people will be taking advantage of the "overprotective" system. When normally every time you are crossing the street you don't know what kind of driver is passing you, whether it is a good decent driver or a lunatic on acid, with AC all cars will drive and behave the same way. So people, knowing this will be likely to cross in front of the AC knowing it is smart enough to pull over safely enough not to hit the pedestrian. They will take advantage of knowing the computer's behavior.
>Long-haul truck driving will cease to exist. Think how much money trucking companies will save if they don't have to pay drivers or collision and liability insurance. That's about 3 million jobs in the States. Shipping of goods will be much cheaper.
Does this mean there won't be anyone inside the truck guarding valuable goods inside?
1. Find a truck carrying mobile devices
2. Stand in front of it since it's programmed to stop for pedestrians
3. Partner uses blow torch to get the goods
4. $$$$
I think there will always be someone at the wheel even with self-driving technology. Just think of other forms of commercial forms of transport aka trains. Why aren't their unmanned trains, couldn't we deploy the same tech and incorporate it with a flying drone that has sensors to see what's ahead on the track or even satellite imagery?
I've also thought about how self-driving cars will affect crime. If police don't have to enforce moving violations does that give them more time to patrol areas with higher crime OR do those departments shrink because of the lack of revenue from said moving violations?
This is the next big thing and it will happen. Just imagine a city full of robotic cars were you dont have to own a car anymore because you can always call one. Less traffic jams because they can drive at peak efficiency and can plan ahead because they all know where everyone wants to go and all of them will be electronic cars.
This has so many ups on society and our environment if i could buy/use one tomorrow i would do so.
One of the biggest downside of this a lot of people like cab/truck driver will lose their jobs due to this "revolution"
I love Google's Self Driving Cars. But do you think auto industry will allow self driving cars without any problems? For simple concept like Uber, we are already seeing so many hurdles. For any radical innovation to happen lots of naysayers and incumbents will try to kick innovators at every opportunity. Unfortunately, auto industry will use some sort of machination to lobby for evil regulations.
EDIT: To downvoters, I'm in favor of self driving cars but I'm worried about auto industry spoiling the party.
Self driving cars have the ability to increase the number of passenger miles driven.
The auto industry profitability depends on passenger miles - less miles, less cars + parts, less revenue.
Google isn't going to get into the car assembly business. One or more of the innovative car firms will just licence the self drive technology into an existing platform as a first step.
Car makers themselves have been experimenting with this stuff for decades. If you google around you'll find BMWs doing high speed laps of race circuits with no drivers.
A modern S-Class Benz can literally drive itself already - it has brake assist, lane assist, steering assist, parking assist and radar guided cruise control. It knows where it is going and how long it will take to get there. The driver really is only a small part of the driving equation in a car like this.
Why would Google license the tech when they could buy Tesla (TSLA market cap is $3 BB; CHEAP for the value they deliver) and turn out Google cars themselves? They then control the entire ecosystem, similar to Android, versus having to cede control to car companies who drag ass on innovation.
This is what I was thinking earlier today, and it makes a lot of sense. Google would have an opportunity to enter the automobile market very quickly and use its technology on all Tesla cars. The major problem to me is that a company like Tesla only manufactures more expensive cars, and as a result Google would be targeting a small percentage of the population initially with this move.
Actually Android is a counterexample--Google doesn't control the entire Android ecosystem, and they do have to cede control to phone manufacturers, who cede control to carriers, who drag ass on innovation, which is why you can buy a brand new phone today that has a years-old version of Android running on it and loads of crapware.
Google went from zero to making a large amount of computer hardware in a decade, partly through using the large existing contract manufacturing chain. The auto industry is not entirely different: Bosch, Rotax, ZF, etc would all design and build parts for Google...
I agree but Auto industry is not about car manufacturers. Players like auto insurance companies (who constitute billions of dollars in market cap) will not fade easily.
FYI, the article suggests self-driving cars have been approved in Nevada and California, which is true -- but in both states, a human passager (presumably with a driver's license) has to be in the vehicle. In Nevada, you actually have to have two passengers, which in some cases makes it more inconvenient than regular driving.
So ideas like sending your car home instead of parking it, or sending tractor trailers across the country without a driver, aren't going to be legal for the time being.
Very good points. Almost all of them. This, however...
> Think how devastating that would be to the car industry. People use their cars less than 10% of the time. Imagine if everyone in your city used a RoboTaxi instead, at say 60% utilization. That's 84% fewer cars required.
Somehow that does not ring true. 84% number assumes that no matter the utilization, car lifetime will be the same. As a driver, I know this is definitely not the case, especially in colder regions.
I doubt that car insurance is dominated by fixed costs - istr payout rates are over 70%. Even if the number of policy holders falls by half (which is unlikely to happen in a decade) premiums shouldn't skyrocket.
More robot cars on the road likely make it safer for the remaining humans, which will tend to lower premiums.
The only problem is if the most dangerous humans prefer to drive themselves. I can see factors in both directions.
The other day I was driving 2 of my kids (6 and 7) to one of their Summer rec programs. I told them that they would be driven by robotic/self-driving cars someday soon. The questions never stopped after that:
Never underestimate the Kronos Effect that Tim Wu talks about in The Master Switch. If what this article predicts starts to pan out, I'd expect a lot of lobbying dollars to scare people into keeping it illegal. Even if it is safer/better.
People still ride horses and enjoy doing it, but I don't see anyone advocating using horses for their daily driver (the Amish excluded). Driving for pleasure doesn't have to go away.
Exactly. And I can see a new sport/hobby of semi-automated driving where you can perform maneuvers that are difficult or impossible for the best humans today. http://www.youtube.com/watch?v=RY93kr8PaC4 is a simple example; imagine what you could do with a high-performance vehicle.
Speaking of dumb and short-sighted, how do you fail to see that robotic drivers can make vastly more efficient use of existing infrastructure than humans? How do you not see that reducing traffic jams lowers the amount of gasoline burnt idling, and carbon emitted to no effect? How do you not see that the same technology can be the basis for a very different transportation system? Can you not imagine a world where few people own cars (which are parked most of the the time) and, instead, rely on fleets of robotic taxis simply pick people up and drop them off where and when they're called? If crashes were a thing of the past, do you see how dramatically engineering would change? How much smaller and lighter cars could become? How much less fuel they'd use?
I mean, I'm all for envisioning better ways of living. I'm even more in favor of the guys who are not only envisioning it, they're actually building the means.
autophil probably goes too far, but he isn't entirely wrong--even perfectly efficient robotic cars powered by pixie dust are inefficient and not a panacea for the problems of traffic and pollution that car-centric infrastructure has caused.
True but assuming the future is that of mostly robo-taxi companies (and not individually owned self-driving cars), there will tremendous business pressure on robo-taxi companies to make their cars as efficient as possible, as the cost of gas (or other energy) will be part of their cost of providing the service.
Today car companies sell cars based on many criteria. Fuel efficiency is one of them but when gas prices aren't too high, individual customers don't care about that very much.
Someone who operates a fleet of 10.000 cars will make fuel efficiency their top-most priority because that will drive down the cost of their service and the cost will be a major reason why customers will pick one provider over another, not how stylish the car is or how fast it goes from 0 to 60.
E.g. if they realize that by putting solar panel on the roof of the car they can save money in the long term, they'll immediately put solar panel in every car. The progress in making cars more efficient will be much faster.
Anything like a car is always going to be space and weight inefficient. A Smart car has an order of magnitude more weight than a person, and occupies two orders of magnitude more space on the ground--you're not going to get significantly below that, so propelling those cars and fitting them all on the street already adds a bunch of overhead on top of the original problem of people-moving. That's before tackling engine efficiency or energy sources.
This isn't to say self-driving cars are useless. They might be just the kick needed to get people out of car ownership, which will enable the fundamental land use and infrastructure changes necessary to make more sustainable improvements. But they aren't the endgame for sustainability.
> Self driving cars is such a dumb, short-sighted idea. We don't need more cars on the road, and we don't need more roads either. It's not sustainable.
Doing heroin is a bad idea; it's harmful to the body and the community. But while fighting heroin is a good thing, it's also good to give addicts clean needles so they aren't harmed even more.
Cars aren't going anywhere, but if you can reduce the harm while promoting better alternatives, everyone wins.
Look, I'm an environmentalist (I fund projects that help pay for inholdings in state parks so they transition to what we'd consider the public domain), but I'm a realist as well. People need mobility. Its not going away, ever. Self driving cars + electric vehicles + intelligent energy management between a large electric grid and all those vehicles is the holy grail. It's going to happen much sooner than you think.
That's a really pathetic version of a "holy grail". Real progress would mean people living in compact communities where most trips are undertaken on foot or by bicycle. Your "holy grail" is just a new coat of paint on the same deteriorating sprawl.
Compact communities don't have to involve large populations. There are already small and medium-sized villages in Europe that have successfully converted to 100% car-free walkable spaces, and it isn't inconceivable for many existing US suburbs. You just have to move away from this design philosophy focused on strictly separating residential and commercial zones.
There is nothing intrinsically wrong about the idea of a car. It is what powers the car (currently gasoline) that is the problem. But the self-driving technology is applicable to electric cars, cars that run on fuel cells, what ever other modern and sustainable methods for powering a car exist.
> There is nothing intrinsically wrong about the idea of a car.
For a single commuter, it increases their weight footprint by an order of magnitude and their space footprint by over two orders of magnitude, which increases total energy consumption and traffic congestion compared to simply transporting the commuter themselves.
> We don't need more cars on the road, and we don't need more roads either. It's not sustainable.
If anything, autonomous cars is a great way to go to reducing the number. The car can drop you off at office and come back home by itself, so your significant other could use it in the meanwhile without really having to buy two cars.
I suspect after the technology becomes wide-spread, new laws will be made to minimise people buying (human-driven) vehicles. Cars will probably require some expensive licence to drive, older cars will not be allowed on the road unless an exemption is made etc. Sort of how Singapore does it.
If cars are self-driving, there'll be less of them. And they can better utilise existing road space (faster reaction times means they can tail-gate with comparative safety.)
What annoys me about posts like this in general is that they only see cars (and motorised transport in general) as a cost, but don't take into account any benefits.
It's a silly position to only talk about costs, completely disregard any benefits. It comes across like a petulant brainwashed teenager, and people should try and argue their point better.
1) No insurance implies no liability. When people "buy" these mostly autonomous vehicles, who is assuming liability? Google?
2) "Death of car companies" - is Google going to start making cars? Are they going to compete against years of engineering? Just because they are building the brain doesn't mean they will be building the entire body.
3) The auto industry is heavily regulated, and moves very slow. 5-10 years is a very optimistic estimate.
4) Cars are moving mechanical parts. They require maintenance and repair. How often do you go to the car shop because of an accident? How often do you go for general repairs?
5) Sure, parking revenue might go down for the city. Traffic at those same spots where people did not go because it was hard to find parking will go up. How often have you considered going somewhere else because of parking?
6) Not sure what percentage of the legal system is bogged down by "driving" related crimes. Same for health industry, vehicle related health service.
The author is making the mistake of equating Google moving into this space as Google being able to do all. Google will not be able to manufacture cars for the near future. They will not be able to eliminate liability. They will also have to change driving culture and habits. Having discussed this with friends, a fair number of people just want their own car that they drive.