Hacker News Reader

Comment by awjr | Hacker News Reader

A most telling point is the comment about how driverless cars would keep to the speed limits and how 'boring' this would be.

In the UK there is a reason why the police have stopped using the word 'accident' when reporting a road traffic collisions. There is almost always somebody at fault.

With around 2175 road deaths in the UK in 2012, I truly believe driverless cars are the only way we will ever see a massive reduction in this fatality rate.

I, for one, welcome our driverless car overlords, but expect this to be years in the making.

View on HN

A most telling point is the comment about how driverless cars would keep to the speed limits and how 'boring' this would be.

I think this comment is interesting because it contrasts the safety of driverless cars with the freedom and control of being the driver - and the responsibility that carries. As the article notes, "who would be responsible in the event of a collision" still needs to be considered. Ultimately, I see driverless cars as just another point on the question of "how much freedom/control/responsibility should people have over their own lives, as opposed to leaving it to someone/something".

With around 2175 road deaths in the UK in 2012

Out of 499,331:

http://www.ons.gov.uk/ons/rel/vsob1/death-reg-sum-tables/201...

I'm wondering how you would feel about 2175 deaths in the Air Industry. There would be uproar. Seriously these are mostly preventable and driverless cars will be 'normal' in 50 years time.

I can see a situation where the only people that know how to drive are commercial drivers.

I think people just need to be made aware of the potential positives.

You can have a drink and drive, you can watch a film, you can program or even sleep!

And these positives aren't even fundamental, addressing the safety gains etc..., but I think talking up these positives that will affect you the most is the best way to sell this technology. Most people don't think they are going to get hit by another car, but most people would love to be able to do something else in the boring morning commute.

Will you ever be able to drink and drive? What happens when the car needs manual control?

> What happens when the car needs manual control?

Then the notion of driverless cars has failed.

It's night and your driverless car is waiting in front of a red light. Suddenly you see a man with a shotgun walking up to your vehicle. How do you tell your driverless car to ignore the red light and get the hell out of there?

We already put ourselves in situations like that when we take mass transportation, sit in rooms with one exit, enter retail stores, and so on.

Also, consider that with enough driverless cars, the congestion at intersections should be reduced (many lights will no longer be needed, especially during low-traffic times like night.) The danger at congested stops would be equal to now, and it is nearly impossible to drive away from danger in a multi-lane column of stopped traffic.

That scenario is incredibly unlikely in the UK, but even then, you could have an emergency-call mechanism that could take control of the car remotely.

The entire point of driverless cars is that the software of the car is superior to the wetware of the human. Humans cannot be trusted to determine what the car does, directly, under any circumstances, even an emergency.

So the answer is, you don't. The sort of fine-grained, always on surveillance necessary for driverless cars to work would be applied outside the cars as well as inside. Pedestrians are dangerous, with or without a gun, and as soon as you walk outside and attempt to interact with the urban environment without the safety of a self-driving car, someone would be alerted. Your identity will be confirmed, you will be tracked and every self-driving car for miles will know about it.

So in your scenario, the proper authorities have been called before you were even aware of his presence, simply because there was a human near traffic.

Since this will never happen, while the thousands of deaths per year due to driver error is a reality, who cares?

considering in the uk the deaths caused by dangerous driving is 0.04% is it worth making everyone change to driverless cars? What happens if I want to drive my old reliable ford escort MK2? Do I not get to any more?

Yes, it is worth it. Initially we will provide tax rebates to driverless cars to encourage people to transition, then we add a "you are an anti-social luddite" tax to manual drive cars, then we make the latter illegal on public roads.

If you want to drive you old reliable fold escort you are free to do so on your own property, this has the added bonus that you can drive it as fast and as dangerously as you want (barring negligence regarding passengers or bystanders.) Once you enter the public thoroughfare we fine you and if the offense is egregious or leads to injury to others we put you in jail. Same response we give to someone who wants to drive 200kph.

Governments are doubtless enthusiastic about their introduction given that they will then have access to exquisite details for every single journey undertaken.

They pretty much have this already. I've been researching automatic number plate recognition systems recently and it is surprising to find that the police have a ANPR database, are scaling cameras numbers up to the high thousands and keep the data (not images) for up to 2 years and expect 50 million readings per day.

I think if you drive into London (from any direction) it is very safe to say you're tracked going in, and tracked going out.

However elsewhere in the UK it is much less common in general. There are entire counties where I'm yet to see a ANPR camera, just your regular old fashioned speed cameras.

At the level of a journey they already have a lot of coverage for that data with ANPR although this would increase granularity:

http://en.wikipedia.org/wiki/Police-enforced_ANPR_in_the_UK

Driverless cars will in the far future also be able to minimize the injuries and deaths when an accident cannot be avoided by deciding who should be hurt and who should not.

For example it might hit a motorcyclist with a helmet, rather than one wearing none. Or a single older pedestrian walking on the sidewalk rather than a young one suddenly crossing the street.

It's frightening.

Or by then, motorcyclists might not exist anymore since motorcycles have been deemed too dangerous and made illegal, and neither pedestrians since it has become too convenient to use automated transport and walking is considered a waste of time.

Frightening indeed.

I still go on walks to "waste time". Feels great. I don't see our legs being completely obsoleted yet.

Even if there are only driverless cars, you can imagine that some models will have a costly casualty limiting option that will ask cheap neighbouring cars to take the maximum damage when a collision occur.

By then, of course, the sewer people will have risen up and reclaimed control of the overworld. A slippery slope.

Frightening.

How is this frightening? Aren't these the decisions that humans should make if they had the ability?

I'd imagine most people would find the idea of an AI making real time life or death ethical decisions such as the Trolley Problem[1] a little troubling.

[1] http://en.wikipedia.org/wiki/Trolley_problem

There is almost always somebody at fault.

With people, yes.

With robots, perhaps not.

If you think of a collision as "someone doing something wrong" (as the police do) then a robot car can't really be responsible for a collision because all it can do is follow its programming. It's literally impossible for it to be 'careless' or 'dangerous' in the same way that a human driver can. Consequently, any accident between two robot cars will always have to share responsibility between the two owners (or possibly the manufacturers, but that's highly unlikely). It'd be exactly the same as if two human-driven cars hit each other in blameless circumstances today - the costs are just divided equally, or each car's insurer pays for the damage to that car, or something.

Robot cars will massively disrupt the insurance industry.

A human driver can certainly can be careless or dangerous, but in my experience this is not how blame is assigned in the case of a collision. It is assigned based a failure to conform to very specific rules of the road - operating left of center (or right, depending on the country), failure to yield, etc. Human factors are taken into consideration, but there is certainly an existing framework for assigning responsibility that appears to be quite compatible with autonomous operators. I would suggest that the shared responsibility case would be the same as it is today - when both operators are in violation.

But for the most part we are not held responsible for our actions on the road, the insurance company are. The obvious penalty for bad driving (banning) is rarely used and requires lots of evidence despite the numerous benefits of eliminating threats from the roads. For the most part collissions are treated more like a social problem than a safety or criminal one. People are "good" or "bad" rather than safe or unsafe. Until unsafe driving becomes socially unacceptable these ideas are just pie in the sky.

I expect it will disrupt the software car industry instead. Insurance companies could buy robot car manufacturers and dictate the logic to follow in case of collision. New standards could emerge, like assigning more blame if one car has an older firmware, which would put pressure on manufacturers to deliver "insurance-friendly" cars first and foremost.

> With around 2175 road deaths in the UK in 2012, I truly believe driverless cars are the only way we will ever see a massive reduction in this fatality rate.

Uhh what? Road fatalities have been decreasing year upon year since at least 1940(!). Want to see a "massive" decrease: Road fatalities have fallen by wopping 81% since then.

In the 2000s most years had a 12% (!) reduction in car deaths in the UK compared to the previous year (admittedly 2010 didn't trend that way, and it flattened in 2012-2013).

1,713 is the 2013 figure. By comparison the UK's suicide rate in the same year was 5,981. 2,955 from "drug related poisonings" (i.e. overdoses).

I think it is laughable to suggest driverless cars are about safety primarily. Cars are relatively safe and only getting safer (e.g. radar cruise control, auto-breaking, partial side impact crush zones, etc).

The primary benefit of driverless cars is convenience. It is pretty darn convenient to get an extra 2 hours of usable time per day because you don't have to drive to or from work (and don't have to pay a third party to drive you).

Well, there're people like me, who simply like to drive and treat it as a way of relaxation.. I put my favourite music or an audiobook, set the perfect temperature, sit in a comfy seat, cut out from the outside noise and let my thoughts calm down, focusing only on driving. I don't even want an automatic transmission, because there is something very satisfying in using stick. I'll be honest - I smile each morning when I get into my car and I wouldn't trade that for the convenience of a driverless car.

You clearly don't take the M25 to work.

I'm fairly doubtful that the present state of AI is able to deliver a driving program that equals the skill of a competent human driver. Moving around requires a lot more understanding than is generally believed.

Sadly, competent human drivers are rare. Humans are terrible at paying attention to boring things and accidents are too rare to keep people's attention. Worse it's the edge cases where someone is not paying attention that tends to lead to most accidents.

> I truly believe driverless cars are the only way we will ever see a massive reduction in this fatality rate.

What about reducing cars on the road through greater usage of public transportation?

I see driverless cars removing the need to own a personal car (within urban environments). Some people unfortunately want the private space that a car provides to them.

Driverless cars enable better car pooling. "Would you be happy sharing your journey and reducing your charge by 50%/66%/75%"

I see entrenched taxi firms having a massive issue with these.

Note I've assumed that it would be legal to have a car drive around without a person in the driving seat able to take control. However I think this is a critical legal point to get to to make driverless cars a persuasive transport solution.

> I see driverless cars removing the need to own a personal car (within urban environments). Some people unfortunately want the private space that a car provides to them.

It isn't that so much as... People will vandalise shared cars. Write on the walls, spill stuff and leave it, vomit, etc. This is less of a problem on manned public transport (e.g. buses, trains, etc) since there is an authority figure to limit it.

So best case scenario they install cameras to stop the vandalism, but that then means you're being watched all of the time which is a little uncomfortable, when in a legitimately driverless car you could have actual privacy in theory.

I think an employer provided driverless car system is believable. Essentially your employer scraps car parks, and then sends out its fleet of cars to pick up employees at home and brings them in. It is unlikely employees would vandalise the cars.

I look forward to their safe passes and patience, as a cyclist and pedestrian.

Reply on news.ycombinator.com