Comment by sologoub | Hacker News Reader

Comment by sologoub | Hacker News Reader

Parent Full thread

Looking at the actual facts, there appear to be little justification for this violence, especially since it affects the human safety drivers the most and they are definitely not any kind of aggressor here.

Some recent quotes from the USA Today article covering the Waymo self-driving car hailing service:

“"But mostly it just makes me feel safe. One time, the Waymo (vehicle) paused before turning, and I wondered why. Then a car ran the red light and crashed into the median. It saw that car way before I did."”

“But the fact remains that after millions of miles of city driving, another 7 billion miles of virtual testing and countless more tests undertaken at a private faux-city facility in California, Waymo vehicles have yet to cause a major accident. When fender-benders do happen, often it's because human drivers bump into the robot cars.”

Source: https://www.usatoday.com/story/money/2018/10/10/waymo-self-d...

The real issue is that we need more miles driven to really be certain that this safety record will hold. Thwarting this testing will probably endanger more people because the tech will become mainstream whether anyone likes it or not. It would be a lot better if the tech is over-tested than under-tested.

Yes. How many seconds per trip does Waymo's conservative driving cost? Probably not very many. Left turn delays, the main complaint, are a problem only at intersections with heavy traffic but no left turn signal. That can be dealt with partly by routing and partly by bugging municipalities for left turn signals at key points.

UPS drivers almost never turn left.[1]. It is both safer and more efficient to plan your route using almost only right turns.

I am sure Waymo knows this but I'd guess they don't want passengers to feel like the car is taking "crazy" routes. People could act in unpredictable ways if they felt the car was out of control and turning away from their destination.

In the long term, when people are more comfortable with the technology, I suspect they'll change this.

[1] https://theconversation.com/why-ups-drivers-dont-turn-left-a...

As your parent says: "Context suggests people are justifiably upset, not that these specific actions where justified. "

As in then being upset is justified, not their retaliation.

We are all entitled to our opinions. My opinion is that these vehicles are a lot safer than most human drivers and I do not see any justification for the vitriol displayed.

Think about it - running a car off the road so you can yell at them to display your displeasure? That’s violence and no one should be subjected to that.

Despite what people like to say, our institutions work and work a lot better than many around the world. If you want change, you have to work for it and convince others of your point. Democracy is not about acting out, it’s about civil discourse. All sides deserve to be heard, but in a civil manner.

> My opinion is that these vehicles are a lot safer than most human drivers

Eventually, that’s likely to be true, but currently these are sill in the experimental stage and regularly have issues. The collective benifit is probably a net win long term. But, they are not replacing google street view cars or something so as to spread that risk out. Instead they are concentrating that risk into a small subset of residential streets in some poor neighborhoods.

Even if it was already slightly safer than human drivers the extra traffic really risks these people’s lives. But because they are low income they don’t expect to be able to afford self driving cars in the next 10+ years. So, that’s meaningful harm without the exception to benifit.

Now how a tiny subset of a community responds is predictable but not justifiable.

> Instead they are concentrating that risk into a small subset of residential streets in some poor neighborhoods.

These cars already safer than human drivers and they are common on the streets around Google. So unless you mean "risk" as "risk of being slightly inconvenienced", neither of your points is accurate.

> But because they are low income they don’t expect to be able to afford self driving cars in the next 10+ years.

It is pretty clear that Waymo primarily intends to sell a taxi service, not cars directly. This taxi service would be cheaper and safer than any currently available so people of all incomes would benefit both directly and indirectly (fewer DUI drivers).

Taxi are expensive luxuries, poor people use busses.

Also, the last study I found had self driving car tests result in 4x the rate of injury’s vs normal drivers. This should be improving over time and the sample size was 11 injuries over 1.1 million miles, but even still systems in R&D are inherently risky.

Ride-Hailing Apps May Benefit Poor and Minority Communities The Most, Study Suggests.[1]

Waymo has caused a single crash in 5 million miles of driving and no one was hurt.[2] They have been involved in slightly more incidents than would be predicted for a human driver (30 vs. 21) in those 5 million miles but it was not at fault for the others.

[1] http://fortune.com/2018/06/30/uber-lyft-poor-minority-commun...

[2] https://www.wired.com/story/waymo-crash-self-driving-google-...

First injury is more likely in residential areas. Tapping a car is less of an issue than tapping a pedestrian which are harder to detect and predict.

Second while the number is still a low enough number not to be particularly relevant other than to say the gross order of magnitude. However...

Google's autonomous vehicles have crashed 36 times in the Mountain View area

https://www.businessinsider.com/cruise-waymo-apple-which-sel...

Waymo spokesperson called the research "deeply flawed" and said it "reflects a poor understanding of the data as it fails to distinguish between a vehicle that drives hundreds of miles per day on public roads, with one that seldom leaves its garage."

In other words they don’t disagree with the number of crashes, even when excluding other areas.

> First injury is more likely in residential areas.

I'm not sure what new argument you are attempting here. Mountain View has lots of heavily residential areas if you are implying otherwise.

The Waymo spokesman is completely right, this is a very stupid way of ranking safety. This article is more recent than the one I cited and includes more than twice the number of testing miles so the raw number of incidents is unsurprisingly higher.

> These cars already safer than human drivers

I have seen no such proof, source?

Reply on news.ycombinator.com