VW had one emissions equipment protocol mode for testing and another for on road.
Others had one mode.
Under the artificial lab conditions the other brand diesels met regulations. Using real world conditions with drivers using the accelerator pedal more aggressively and no tricks like taping seams the cars did not pass. But the engine had the same exact software code to regulate emissions equipment.
The other cars did not have one protocol for emission equipment when the vehicle was being tested and another when the car brain figured out it was on the road.
Yeah there's no comparison here. Lots of vehicles, both diesel and petrol, fall short in the real world to what they do under government testing, everything from fuel economy to various emission parameters to crash testing. It's called designing for the test. What VW did is very different, they cheated on the test.
Best analogy I can think of is other manufacturers studied just for the test, VW cheated on the test.
That will just shift the parameters, and end up with the same results as the manufacturers tune their vehicles to the new test procedure. The only real way to fix the problem is to evaluate a statistically significant number of cars in service on a somewhat regular basis and impose significant retroactive fines if the fleet of a particular model doesn't meet the standards.
I suspect that only VW will directly be penalized for 'cheating on the test'. I think that as some of this real world testing is happening, regulators are learning(?) just how big the disparity is between reality and studying for the test, and will change the parameters of the test to reduce the gap between reality and the test.
To the extent that the test changes to minimize that gap, all manufacturers (well - all gas/diesel engine car makers anyway) will feel that pain and have an incremental expense. And we'll all breathe better (eventually) for shrinking that gap.
The way I look at it, I'm comfortable if there's even a 2x gap between test performance and real world performance. 2x and down is at least within my own experience of how much I can vary the range on my Roadster based on how much lead I have in my boot that day, driving conditions, sprinting up hills, etc.. But 10x emissions, or passenger autos with higher emissions levels than trucks hauling stuff - that's not ok and it's not close.
So I don't expect the tests to be perfect - it's in the nature of a test to measure something specific in a controlled and repeatable fashion. But the test can be designed with a smaller delta between test and real life results. For eye.surgeon's analogy, it makes the test more useful. Who wants to hire somebody whose performance drops 10 fold over their test results?