This wouldn't be hard to fix.
Loan cars to a random selection of non-professional beta drivers. Graph the results, showing the median and sd.
It would be cheaper to do this than to get hit by a class action for misleading buyers.
Of course reality-based non-expert mpg would look very bad compared to the dishonest figures you get now. But that's because of manufacturing and design decision, and not really the fault of the drivers.
- We base regulations and taxes on absolute MpG, so overstating efficiency weakens the regulations and allows automakers to cheat on taxes.
- It misleads consumers about the cost to operate the vehicle.
- It is an issue when comparing efficiency of new cars to older cars produced before the systemic cheating.
Presumably not everyone cheated the same, as well.
This is going to sound backwards, but consider this: People buy cars based in part on MpG, however MpG is within itself an abstract thing (i.e. you don't actually get X miles per gallon, never did, never will too many other factors to include).
So if everyone removes 4% from an abstract number, the entire market has shifted down, but the relative difference between two models remains identical (e.g. if one car has 21 MpG and another has 25 MpG, even with 4% removed from both they retain their relative position).
If some were removing 4% while others weren't, that would legitimately be a problem. If they're all removing 4% then we're essentially where we would be at the start just with lower abstract numbers.