Hacker News Reader

Comment by astrocat | Hacker News Reader

>What if Waymo implements a critical safety feature and patents the software/algorithm?

Maybe I'm naive, but is this really different than today? Let's say Volvo implements some new kind of crumple-zone tech that drastically reduces fatalities in head on collisions, and patents the tech. Is this a big ethical problem?

I think the distinction here is that failure conditions in current vehicles are almost entirely human caused, so that new safety tech only matters after a human has already made a mistake. With autonomous cars, humans aren't making mistakes anymore, software is.

So essentially, we're looking at transitioning from protecting people from themselves, to protecting people from the machines. Which is something we FEEEL very differently about, even if the big picture isn't all that different. I mean, it's certainly something that changes the dynamics of liability, that just a matter of of who's paying the insurance bill.

View on HN

We have a historical record of this - a company invented the three point seat belt and felt it was too important to keep to themselves for the safety of car users around the world and freely licensed it so that every car company could incorporate it. That company was Volvo. https://priceonomics.com/volvo-gave-away-the-most-important-...

Not hating on your comment - it's actually rather interesting and informative. What is also interesting to me is how you wrote it:"That company's name? Xxxx" after a description. This reads like click bait. I've noticed an uptick in people writing like this. I wonder if this a trend others are seeing?

It reminds me of this: https://www.reddit.com/r/OutOfTheLoop/comments/1plnrg/why_do...

That's one of the oldest rhetorical devices in the book. It's a common technique of good storytelling. Clickbait is annoying because of its context, but the reason it works is because those techniques stimulate human curiosity, and increase their enjoyment of the story.

In fairness, inventing a three point seatbelt is not in the same league as designing and producing a working SDC. It's not even the same game.

That's true but the OP gave a mechanical example, too! :) On the other hand my out of loop observation is we're missing some tech revolution/discovery to make SDC possible because every single large car company and some big name tech companies are on the hunt and it feels like no one is close, though I have no concrete data on that, just watching the Waymo cars drive around my home.

The main issue is liability. Who is liable when software produces an accident? The owner? The company that made the car? What about insurance?

I doubt people will be able to own self driving cars. They are going to need the maintenance schedules of small aircraft to be safe. It's more likely they'll only be available as a service or taxi.

I also think this is a big reason for the apprehension of self driving cars. While some advocate safety, people are more than familiar with software bugs in our phones, computers and web apps.

Sure there is a difference. The software in mission critical applications like pacemakers, aircraft navigation, etc. has to be thoroughly tested and designed to be either failure proof or failing in a safe way or having a redundant system. But will the products Uber and Google develop have the same level of testing Boeing puts into aircraft or SpaceX puts into rockets?

Reply on news.ycombinator.com