While I am for software transparency when it matters, I do have to say I'm much less excited about the software being gone over by lawyers (or programmers in the pay of lawyers).
Give me a page of C code, and I can find ten faults. I can complain about variable naming schemes, I can complain about indentation I don't like. I can complain even if it's actually my preferred indentation, because you won't even know better. I can always complain about architecture because there are always pros and cons, which means I can play up the cons and ignore the pros, along with ignoring the fact that we have to use some architecture and "perfect" was never on the table. I can say this other architecture should have been used instead, and get you involved in a battle of pro and con analysis that is a perfectly good engineering discussion but will sound like dissembling on the witness stand. I can complain about the bug fix for an issue that I can't imagine how it comes up, but came up in testing. I can, basically, complain all day long, even if it is literally the best C code ever written.
Now, I grant that history suggests I'm unlikely to encounter the best C code ever written in one of these contexts, but my point is that since nothing can survive lawyer scrutiny, lawyer scrutiny is actually information-free.
The solution to "Not having any software standards" can't be "Requiring absolute, unattainable perfection", because the only sane response to that is to stop writing software.
The real question is not "Is this software perfect?", but "Did it function correctly?" And I'd be a lot more comfortable if somebody took the source code and the hardware and actually showed a case where it is wrong by physically producing that case, and not just theorizing about how the software might go wrong.
All that said, for all I know this thing's a pile of crap. Certainly if I had to lay money, that's the way I'd bet. My point is more that this writeup doesn't meet my standard for determining that it's a pile of crap, and as much fun as it may be to pile on law enforcement, if the price is letting the "lawyerly-perfection" standard pass without comment, that price is not worth paying!
"In all criminal prosecutions, the accused shall enjoy the right to a speedy and public trial...and to be informed of the nature and cause of the accusation; to be confronted with the witnesses against him..."
The real question IMHO is neither "Is this software perfect" nor "Did it function correctly?", it is "is the software quality sufficient for the risks associated with failure?"
What are the bad events resulting from failure? I see mainly people being wrongly punished (and suing) and people wrongly not being removed from of their cars when actually drunk.
On the other hand, the source reports "incomplete verification of design, and incomplete “white box” and “black box” testing". That sounds like the "well it mostly works I guess" level. To me, that's insufficient.
Give me a page of C code, and I can find ten faults. I can complain about variable naming schemes, I can complain about indentation I don't like. I can complain even if it's actually my preferred indentation, because you won't even know better. I can always complain about architecture because there are always pros and cons, which means I can play up the cons and ignore the pros, along with ignoring the fact that we have to use some architecture and "perfect" was never on the table. I can say this other architecture should have been used instead, and get you involved in a battle of pro and con analysis that is a perfectly good engineering discussion but will sound like dissembling on the witness stand. I can complain about the bug fix for an issue that I can't imagine how it comes up, but came up in testing. I can, basically, complain all day long, even if it is literally the best C code ever written.
Now, I grant that history suggests I'm unlikely to encounter the best C code ever written in one of these contexts, but my point is that since nothing can survive lawyer scrutiny, lawyer scrutiny is actually information-free.
The solution to "Not having any software standards" can't be "Requiring absolute, unattainable perfection", because the only sane response to that is to stop writing software.
The real question is not "Is this software perfect?", but "Did it function correctly?" And I'd be a lot more comfortable if somebody took the source code and the hardware and actually showed a case where it is wrong by physically producing that case, and not just theorizing about how the software might go wrong.
All that said, for all I know this thing's a pile of crap. Certainly if I had to lay money, that's the way I'd bet. My point is more that this writeup doesn't meet my standard for determining that it's a pile of crap, and as much fun as it may be to pile on law enforcement, if the price is letting the "lawyerly-perfection" standard pass without comment, that price is not worth paying!