As a example, just look at the 737 MAX, a airplane viewed as a literal deathtrap and a complete indictment of airplane software development, which had 2 crashes in 400,000 flights. At a average flight distance of ~500 miles, the 737 MAX, a airplane literally multiple orders of magnitude worse than any other and a complete failure of the approval process, only resulted in 1 fatality per 100 million person-miles. That is safer than the driving average of 1.1 fatalities per 100 million person-miles. What is viewed as by far the most atrocious outcome in the airplane sector, a system literally hundreds or thousands of times more dangerous than the average, is above average in the car sector; that is how far the gap is.
And this is ignoring driverless car software which is being developed to even lower standards than regular car software and which has no actual independent quality requirements. In contrast, far more lives are at risk with respect to driverless cars, so they should be developed to standards probably around 1000x better than airplane software which puts them around 0.0001% of the way there to a acceptable level of quality even if the software was already as reliable as regular car software.
You mean the standards set by the FAA, which regularly defers to the private industry to regulate itself, due to complexity that it can't afford to keep up with? Those standards?
As you note, unfortunately the standards as employed in practice have recently regressed dramatically to allow a system as unsafe as the 737 MAX to fly. So it seems only prudent to go back to the historical, far more rigorous, process which did result in the amazing historical safety regime. And, since self driving cars are a even harder and more dangerous technology with far more downside risk, they should use a standard even more rigorous than the historical airplane standards which are still by far better, from a safety perspective, than any other consistently applied software process, such as cars, by multiple orders of magnitude.
So these were all miles where people were unwilling testers as well. The difference being airplane auto-pilot is primarily focused on keeping the plane in the air, not dodging pedestrians. I think it does a lot of warnings for the pilots as well, but I'm not in the industry.
It's really comparing apples to oranges, isn't it?
> since self driving cars are a even harder and more dangerous technology with far more downside risk, they should use a standard even more rigorous than the historical airplane standards
So this is on The United States Department of Transportation to be far more involved in regulating the software that governs self driving. But the question is, was the FAA as relaxed in governing the aerospace industry 60 some odd years ago as USDT is now, and did they just get lucky that flying planes is actually easier than driving cars? Or is there actually some negligence involved?
The FDA just wants there to be standards, and the companies have a vested interest in making sure those standards ensure safe medical devices.
It seems that you're trying to suggest either 1) that the standards as mentioned necessarily aren't True Standards - as a blanket statement - because they're set by the industry or 2) that these standards aren't True Standards - specific to this particular discussion - because there is no reason to think they're necessarily more rigorous than the standards for driverless car software. If you're trying to suggest something else, please understand that it is unclear.
I'm all for a dissenting opinion, but when it is not presented in an understandable way the discussion becomes hard to follow.
(I don't mean to suggest that the argument is wrong when I call out the True Scotsman vibes; just that it is the argument I see.)
From the person you're responding to:
> Yes, those standards that produced a safety regime of 20 years of flights without software being even implicated in a single fatality. ...
> As you note, unfortunately the standards as employed in practice have recently regressed dramatically to allow a system as unsafe as the 737 MAX to fly. ...
Your point, if I'm not mistaken, being that the FAA has clearly allowed for less stringent regulations in recent years. (Please correct me if I'm wrong. You can ask questions but it does not seem that they lead so immediately to the answers you expect.)
That being said, please consider that I am genuinely trying to understand you:
> I'm all for a dissenting opinion, but when it is not presented in an understandable way the discussion becomes hard to follow.
Now they are. This regulations are a result of several tragedies culminating in a big disaster:
Imagine a thousand employees, half your HQ staff, that leave. There are several possible reasons:
- they're older -> more corporate knowledge and knew the "magic recipes"
- skill based mobility -> they're so good they can go anywhere, and Chicago isn't Seattle
- etc
https://hbr.org/2001/10/inside-boeings-big-move
This was widely thought on ... what was I browsing back then? Reddit? as a fore-warner of bad stuff to come
The fore-warning came to fruition when the issues at various Boeing plants started developing over the next ten years (insider videos talking with an employee about being on drugs during crucial assembly work and shift mates being similarly high, "get it through your station" attitudes) and the Dreamliner issues that were hand-waved away (no training, 1-hour tablet self directed training, certification delays, and eventually crashes).
I remember it being discussed on SlashDot back then.
(Of course this had some substantial downsides.)
It's estimated 31,720 people died in car crashes, in the US, in 2021[2].
[1] https://en.wikipedia.org/wiki/Category:Aviation_accidents_an...
[2] https://www.transportation.gov/briefing-room/nhtsa-estimates...
Before an airplane goes into circulation, it has gone through extremely rigorous testing. An airplane also always has a pilot.
The same can't be said for driverless cars. They don't go through extreme rigorous testing, they don't have a driver, and they're controlled by software that is most likely a far cry from aviation standards.
Apparently there is ZERO learning or forethought about risks that can be anticipated or suggested from other technologies!
As others have pointed out, aircraft certification (while it has its problems) is very rigorous. Driverless car certification, up to date, is basically nothing.
But the other aspect is that aircraft autopilots are solving a far easier problem. Developing a car autopilot for busy streets is many orders of magnitude more difficult than an airplane (or boat) autopilot.
It's a qualitatively different change than any of the above (driven by a messianic view that AI is here and now and about to change everything), plus an instance of the
https://en.wikipedia.org/wiki/Boiling_frog
apologue.
You put your faith in myriad bureaucracies and trust networks that you have no hope of understanding. Driverless cars are just one more added to the list.