[0xDEFC0DE]

>I, for one, hope more engineers involved in fraud (eg VW), privacy breaches (eg Facebook) etc get tried and punished. It might help make programmers realize that yes, your work actually affects real humans and yes, you're partly responsible for getting it right.

You can't do this without a lot of ethical backing, supportive law resources, whistleblower protections, and better controls on letting people into the industry.

It will be an utter disaster if you don't have those protections.

[blackflame7000]

The halting problem shows that it is impossible to verify a program is bug free. Therefore anyone could say that some feature X that harms the user is actually just a bug and you would be hard pressed to beat reasonable doubt

Edit: I love downvotes without rebuttal.

[0xDEFC0DE]

Yeah how do we ever find fault in auto accidents since we don't understand all of physics

[blackflame7000]

Some guy named Newton solved the physics problem you’re referring to. He never got around to computational theory though.

[erik_seaberg]

By construction, there's at least one program whose behavior is undecidable. But there's a big subset of all possible programs that always halt (and others that never do), and we can choose to only ship decidable programs and avoid the weird edge cases as unreliable.

[pilsetnieks]

> anyone could say that some feature X that harms the user is actually just a bug and you would be hard pressed to beat reasonable doubt

That would fall under criminal negligence then.

[blackflame7000]

Ahh, so you see, additional regulation is not required.

[pilsetnieks]

The same way that harming people intentionally via software engineering is not regulated now, negligent harm isn't regulated either. I'm not expressing my support either way because this is a complex issue I haven't explored nearly enough - I'm just pointing out that current laws have stipulations for unintentional action or inaction as well.

Claiming ignorance, incompetence or other factors that might lead to unintentional harm does not get you off scot free.

[blackflame7000]

There isn't a single software engineer alive that hasn't made a negligible error. It's like penalizing doctors if their patient dies. The halting problem specifically prohibits a programmer from ever even knowing if they are without negligence or simply one iteration away from complete negligence. (ie integer overflow that occurred because the system was used longer than the original developer intended).

Also, I would wager that about 75% of bugs come from the seems of integrations where two peoples duties overlap. In that case who's fault is it if they both built half a bridge that doesn't line up?

[pilsetnieks]

> It's like penalizing doctors if their patient dies

Doctors should be punished if their patient dies for reasons caused by doctor's own fault or by some unintentional action or inaction. That's why malpractice insurance is a thriving industry.

> In that case who's fault is it if they both built half a bridge that doesn't line up?

The chief engineer or the architect, the ones who signed off on the project. And again, that's why those industries have professional liability insurance.