So teaching to the test would appear to have consequences.
So, no doubt, the little gnomes at VW working on the software (of course I'd like to see the actual, concrete evidence of this) figured, ah well, its just a test. We were taught that doing what's needed to pass the test is what we should focus on, so let's focus on it...
Next point - at least so far there is no actual evidence that software was deliberately written to cheat. We see a lot of statements like this (from here): "For years, millions of Volkswagen’s deisel cars contained software that turned their pollution controls on only when the cars were being tested by regulators."
So how do the cars know the "regulators" are testing them?
Deliberately writing software for cheat would have to require that there were a very specific set of real world circumstance that occurred only during a test - otherwise the performance would be uneven and customers would complain.
It would be nice to see emails or other confirming documents.
If we think all this through we would have to draw some conclusions:
Either the tests the EPA uses are so unrelated to actual driving that they are useless (hence the evil gnomes took advantage of that fact that the conditions were SO OBVIOUS they could build software to test for them)
Or its just happens that when VW cars are doing what they do they are more pollution conscientious when doing what the EPA tests for, i.e., the EPA tests are just stupid and measure nothing useful.
It will be fun to watch all this play out...