I watched the video. It's really a smoking gun that this is absolutely a deliberate attempt to detect the testing cycle and produce the desired output just for the test. This is exactly the kind of detailed analysis of the code that I wanted to see.

It's also interesting how, in the video, they talk about how other car companies were caught doing the same thing, and that this kind of cheating is widespread. But the companies who did it in the past are not well-known for it in the way the VW is today. In the video, I think they said it was BMW in, I think they said 2000, who were caught doing this and they got caught and then fixed the problem, and then were smug about it when VW was caught in the crosshairs this year.

I think that my initial impression that VW was being unnecessarily singled out by the media is still true. But my suspicion that this might be something more subtle than obvious test-cheating is not true. This is obvious cheating, and it's super clear-cut in the code, according to the demonstration. But I really needed that technical analysis to be convinced, because I know that the media twists things a lot.

Fascinating.