There’s a company called comma that has something close.

Designing Robo-vehicles and the infra to support them requires "real" engineering, not SV style fail-fast "engineering". The task should be left to the government, along the lines of the Eisenhower Interstate Highway System.

Somehow I get OpenSSL vibes from that description. Everyone would use it, nobody would pay for maintenance and the result would be billions of tons of steel controlled by a buggy vulnerable mess filled to the brim with a nightmarish labyrinth of #ifdefs and hacks ensuring compatibility with every embedded chipset that has ever existed.

I still think OpenSSL is a massive success story. It’s used everywhere and it’s good.

Not perfect, and has serious bugs but it’s been used by millions for decades. Even software that costs billions has zero days.

We could be in a world where everyone pays RSA and still have massive vulnerabilities.

We could be so lucky if all software ra exactly like OpenSSL.

A self-managed FOSS level 5 self-driving system, where you're not legally required to be actively monitoring it and ready to take over, seems very very hard.

The trouble with cars on roads is that you're putting others lives in danger, not just yours. I think we should require a huge and lengthy validation process for true autonomous level 5 self driving. So autonomous the car doesn't need a driver at all. It seems possible for an open source self driving system to go through this validation for one particular build of the system. (though not economically feasible). But the moment I start modifying my own system (say, increasing `DRIVING_AGGRESSION` by 10 percent), it's no longer the same system that was verified as safe.

But even if personal modification is out of the question, the safety verification process needs to be so thorough and costly it seems difficult for anyone but a major for-profit company to go through it.