Comment by stackcollision | Hacker News Reader

Comment by stackcollision | Hacker News Reader

I don't think robotic cars will ever happen, and I'll explain why.

Let's look at the recent MetroNorth railroad accident. Immediately afterwards, people were inevitably throwing around the idea of "why do we need train drivers at all?". The reason we need drivers is to satisfy a fundamental human misconception:

A while back, the DC Metro installed an automated train control system, with the intent of making the system independent of human operators. They flipped the trains onto automatic, and for a while everything ran fine. Then, one day, a sensor failed and one train plowed straight into the back of another at full speed while it was sitting in the station. That was the end of automatic train control in DC.[1]

Keep in mind that railroad traffic is a much more controlled environment than the open road.

Everyone knows that people make mistakes. We're (mostly) all engineers here, and we understand that technology is also flawed, like its creators. However, the public at large has this opinion that technology is magic; computers _don't_ make mistakes. The second that belief is proven wrong, people are immediately afraid of it.

Maybe one day a few robot cars will make it onto the road. But then someone is going to get killed by one. And despite these robotic cars having a stellar record up until their first accident, especially when compared to their human-operated counterparts, they will be taken off the road so fast your head will spin.

[1] http://en.wikipedia.org/wiki/June_2009_Washington_Metro_trai...

And if it should happen that a human operator would have likely been able to avoid that fatality? Is it a mistake in all cases to doubt the validity of the premise that humans shouldn't be allowed to drive their own cars?

It is this loss of control which creates skepticism, not magical thinking about the nature of technology. Most people are surrounded by examples of technology not working, they know it's fallible, they would be afraid of autonomous cars for the right reasons - because the intent of autonomous cars is to deny them the ability not to trust technology blindly.

I think the word "never" should never be used for any problem that could be solved algorithmically.

There is no doubt about that we'll have robotic cars. But when, thats a million dollar question.

Wait. Are you saying 100,000 years from now, we will not have had robotic cars?

I agree that people think computers are magic and as soon as one fails they get out their pitchforks, but I disagree that automated cars will never happen. It will happen just more slowly.

The changes will be gradual and they have already happened to a degree. Automatic brakes have already been implemented, more expansive GPS is happening. Next I could see an expansion on cruise control.

Its a combo between marketing new features and iterating on current technology.

I think the invention of the automated car will be a gradient, not a sudden occurrence.

There are already cars today with adaptive cruise control and lane correction. On a highway, the car essentially drives itself. I imagine that with 10 or 20 years of incremental improvements we'll see something that's very close to a completely self driving car.

The thing is everyone says this. But I have yet to hear one person actually argue that "self-driving cars make mistakes and therefore should be banned". I think it's a strawman.

When the time comes to debate it, everyone is going to make the argument that they are still safer than human drivers. That wasn't true for the train operators.

You should probably phrase that as "Why people will resist the idea of robotic cars for some time" rather than "will never happen". Because they're fair points, but surmountable.

Are you offering 1000:1 odds, for example, on your "never" claim?

Reply on news.ycombinator.com