I listened to that episode, and sort of regret it. I've heard these arguments, and they still aren't convincing. Have you read the article on their spaghetti code?

To be clear, I believe that most cases of unintended acceleration are caused by pedal confusion/floormats. Just because that's true of most accidents, that does not mean it's true of all of them.

Maybe it's because I work in the embedded space, but I've seen code written as indicated and terrible code like that is not reliable. It might work 99.999% of the time, but given 100,000 units, and a problem might occur every couple weeks. There is nothing magical about automotive software.