3 ms·

If Toyota had anything to do with that analysis then it wasn't independent. My recollection of the Barr analysis is that a particular failure case was in fact

• kabouseng 11y ago
  As far as I followed the story, the BARR group only found it was possible, but couldn't replicate the condition. I would also be careful in stating who is independant. Mr Barr also have an incentive to claim the code is terrible since he sells static code analysis software and embedded programming bootcamps. That's great advertising for his business. Would your code stand up to the rigours of a Barr analysis? Would any automotive code from any other manufacturer stand up to Barr analysis? Would any other industry stand up to Barr analysis? We don't know the answers to these questions...It is for instance well known that if your code is squeeky clean according to one static analysis package, just running another static analyser will raise a host of other warnings.
• lambdaelite 11y ago
  I don't see why an independent analysis commissioned by Toyota would be any more or less credible because Toyota paid for it than an analysis by an expert witness paid for by a plaintiff in a civil suit. I should think they deserve equal scrutiny. Reading the transcript of the court proceedings, it sounded like Barr Group were able to induce a fault that could lead to an unintended acceleration, but the evidence that this fault actually happened was circumstantial. Also, I don't find the jury's decision against Toyota to be relevant, given that there are so many other elements that figure into the calculus of a jury decision. I can't fault Barr Group for trumpeting that they were a key expert witness in a successful trial of this magnitude—it's phenomenal advertising—but I don't see how it follows that their failure proposal is actually what happened (nor do I think that the Barr Group would say with certainty that it did happen).
  • tunesmith 11y ago
    How would it be anything other than circumstantial that it "actually happened"? Those cars crashed and people died. You could make the same case about any bug; "I duplicated the behavior, but that doesn't mean it actually happened in the original bug report". That kind of developer comment would raise eyebrows. Toyota's original defense was that it was user error because unintended acceleration couldn't happen; they blamed it entirely on driver error. They were making a "forall" argument, and all Barr had to do was show an "exists". He did. Toyota was completely wrong.
    • lambdaelite 11y ago
      > How would it be anything other than circumstantial that it "actually happened"? Because the fault could trigger a DTC. Barr's argument was that the right kind of bit flip event could cause a fault and it would not be logged as a DTC. The beauty of Barr's theory is that it leaves no evidence if it was the cause of a real-world accident. The plaintiffs successfully argued that Barr's theory was the most plausible explanation for the accident, not that it was the cause. > You could make the same case about any bug; "I duplicated the behavior, but that doesn't mean it actually happened in the original bug report". I think this analogy is misplaced, here, there would be no bug report other than "the system crashed" (no pun intended). In which case, your statement would be accurate and proper. > Toyota's original defense was that it was user error because unintended acceleration couldn't happen; they blamed it entirely on driver error. They were making a "forall" argument, and all Barr had to do was show an "exists". He did. Toyota was completely wrong. Toyota's defense was that user error was the most likely explanation; the plaintiff via Barr argued that a software defect was the most likely. The plaintiff prevailed, but that has no bearing on what actually (or likely) happened.