All of them optimize for those well defined regulatory test cases. It's all shades of gray as long as they are just tuning parameters to yield favorable results in the test scenarios without bothering with possible drawbacks in other situations. But things suddenly turn solid black when a method of optimization is employed that is explicitly disallowed in the rules.

It's not unlikely that this infraction was enabled by "motivated disbelief" questioning that software alone can be a "defeat device". And without court-actionable "defeat device" this thing would clearly be in the gray range, even if the "up to forty times" that are being talked about would probably make it one of the darker shades.

They might have even checked this with some lawyers, who can sometimes be very eager to look at the law in the way most favorable to the client: "Yes, it may hold in court" (and if not I will still get my money, also, the negative PR caused by the lawsuit even if you wirll will be your problem, not mine)

I've heard something similar for freezers or some other kitchen equipment but can't find it right now.

Since it can be done through software and those red to green sheets are all over cars and electrical equipment inventing new levels of A+++++...well, I'd guess: very likely.

This should be testable - I mean, it's not practical to review every car while it's running; but if we have a suspicion that such 'defeat devices' are installed by default, then it should be feasible to simply rent a few random cars of each manufacturer, ducttape some sensors on them, and take them for a loop on a normal highway - if the indications are suddenly significantly different from what you measured in normal testing, then the expected fines will pay for a detailed investigation.

Ford & Honda were find in 2008 for doing this: http://www2.epa.gov/sites/production/files/2014-06/documents...