There is a huge difference here. Cars clock billions of hours of operation in an uncontrolled environment. These must deal with decades of use, abuse, neglect, sabotage, third party parts, etc. Anything that can happen WILL happen in a car.

Compared to this, rockets are tightly controlled environment and running for a very short time. There is also incredible amount of redundancy and supervision in a rocket but you will not be able to have double of every critical component in a car and then have separate controller to check if both are returning matching results and which might be faulty and you will not have a team of specialists obsessing every time something did not work exactly right.

Because rockets have tight tolerance requirements, but very low reliability requirements. Consider this case [1] where missile software leaked memory because it would blow up before it ran out of memory. It would hardly be reasonable to state: "If missiles can leak memory, then why shouldn't that be okay for a car." even though missiles are probably more advanced, complicated, and more difficult to make than a car. This is because the "features" are largely orthogonal from the "reliability" of the features which is really a measure of how consistently the features as promised are delivered.

In the rocket case, you need very tight tolerances, but it does not matter much if it fails 1 in 1000 times since the total-system-failure rate is on the order of 1 in 100. A single 1 in 1000 failure mode does not massively increase total-system-failure rate. However, in the car case the tolerances are worse, but the requirement to consistently deliver on those tolerances is vastly more important. Consider Ford Pinto which sold 3,000,000 cars and had 27 deaths attributed to the fuel tank design defect. A 1 in 100,000 failure mode is viewed as an unconscionable deathtrap, so adding a single 1 in 1000 failure mode would make your system 100x deadlier than the most egregious example of an unsafe system.

The prevailing rate of system errors tells you what sort of errors a quality assurance process can detect. If you have 1 in 1000 system errors, then it means you can probably detect things on that order otherwise your system error rate would be dominated by those higher frequency errors. However, you probably can not or do not bother to detect 1 in 1,000,000,000 errors since they would either happen so infrequently compared to other errors or not be worth solving since they have such a minor impact on overall reliability.

So, in summary, reliability is different from tolerances, techniques used on low reliability systems are not necessarily applicable to high reliability systems, rockets are on the order of 10,000-1,000,000 less reliable than cars, so technology used on rockets does not necessarily apply to high reliability industries.

[1] https://groups.google.com/forum/message/raw?msg=comp.lang.ad...

To the OP's point about infotainment being okay, I wonder if the core functional components of space flight (RCS, etc), are controlled through linux. It would surprise me if it wasn't used for high level things (displays), but not the "keep the rocket flying on course" stuff. I can't really find anything about what's controlled by that Linux box in the Falcon 9.

I'd love for somebody with knowledge to comment re: the SpaceX flight control platform. There's Linux in the UI to be sure, but I'd strongly suspect there's a hard-realtime OS that's doing the hardware-level communication to the flight hardware.

I only saw it used on the recent crew mission, but wasn't Linux basically used as an infotainment system there as well? The presenter even explicitly mentioned there was a backup system in case the fancy user interface (which was HTML+JS based) crapped out on them.

The space is a much more controlled environment and has low uptime compared to millions of cars every day