[umvi]

Catastrophic as in 300 people will not die due to the flaw, yes.

But one-off car fatalities that kill 1-3 people happen regularly and they add up. The self-driving variety pop up with the highest visibility but if you go searching you'll find tons of accidents where brake failure at highway speeds cause a fatal crash.

I do concede that distracted driving and alcohol play a much bigger role in the large amount of car fatalities than software flaws. But I still stand by my original assertion that you are more likely to die due to the effects of a software flaw in your car than due to a software flaw in the 737 Max.

[CydeWeys]

How do software flaws in cars kill you exactly? The main example you bring up is brake failure at high speed, but that's not a software issue, that's a mechanical issue (and is oftentimes caused by neglecting maintenance on the part of the car owner).

Yes, there's the self-driving stuff, and there have been some egregious examples, but those systems also save lives by preventing accidents. Lane departure warnings, automatic braking, and electronic stability control all, on the balance of things, make driving much safer.

[kube-system]

> How do software flaws in cars kill you exactly?

Braking systems have been partially modulated by software for decades, i.e. ABS, TCS, ESC.

Additionally, other software controlled systems can induce mechanical issues. For example, in the case of the Toyota unintended acceleration debacle, an engine at WOT typically does not produce vacuum. However, power-assisted brakes almost universally are vacuum-powered. So, if the software-controlled throttle gets stuck wide open, you lose power-assist to the brakes.

https://users.ece.cmu.edu/~koopman/pubs/koopman14_toyota_ua_...

[zaarn]

Power assist not working in the breaks doesn't really equate to loosing the breaks entirely. You can still use them to slow down unless the break wire or hydraulics are literally cut.

[kube-system]

And successful control of the vehicle depends on the vehicle's specific characteristics and the physical ability and awareness of the driver. The point being: it has killed people.

[bumby]

Software is increasingly controlling safety critical systems in cars so I would expect software failures to take up an increasingly large number of fatal vehicle faults going forward. On safety systems that have been using software for decades, one can find examples of such potential failures [1]

[1]https://www.cnn.com/2010/WORLD/asiapcf/02/04/japan.prius.com...

[ulfw]

Only a heavy-software run car might (e.g. a Tesla autopilot or the Uber fatality). But I agree with you. I fail to see how software bug in a car would lead to a comparable outcome.

[bumby]

I don’t think most people realize how software dependent their “dumb” car is. From antilock braking to throttle response to steering response in some cases is largely controlled by software. It goes way beyond the infotainment systems we intuitively think of as software

[kube-system]

Electronically controlled throttles have been on the road for over 30 years now, and became extremely common in the mid 2000s.

[umvi]

I agree that cars are more safe now than ever before and that mechanical failure is more deadly than software failure.

The whole point of my comment was to put to bed the irrational fear of flying. You are still more safe travelling long distances in a faulty flight system such as the MAX than you are by car. There are just too variables to account for in cars, one of which includes increased software complexity.

[CydeWeys]

The 737 MAX crashed twice and killed 346 people. It's not an "irrational fear" to refuse to ever fly in one again.

Secondly, what are the exact figures you're using to show that the 737 MAX is safer than cars? And now compare it to other planes, the more realistic comparison? I'm not taking planes to places that are within driving distance. The 737 MAX was waaaay less safe than other planes.

[bumby]

I think we have to stop segmenting our thoughts into “software” and “hardware” and instead look at issues like the 737 Max as an integrated system failure. “Software” failures can easily manifest themselves into hardware failures; thinking of them as separate systems can lead to a complacency mindset of “its just software so we don’t have to be as rigorous in our design”

[Gibbon1]

> integrated system failure

I agree with you, that's exactly what it was. All of the components worked as designed. But it's pissing against the wind on HN to say that.