I'd agree with you that that tends to happen in practice. I think there are two effects at work here:

1) The driver is still driving the car like it's balanced. With a balanced car, it is going it's fastest when you push it right up to the limit. With an unbalanced car, by the time you get to the limit, it's too late. For instance, in our rear wing example, a driver who is used to having the whole car start to slide at once will wonder why the damn thing just won't turn in. In reality, the car is turning in as well or even better than before, but there is excess rear grip so it feels off. As a result, the driver tries thinks there should be more front grip, tries to turn more, and scrubs off speed.

2) The alignment is not considered. Adding a rear wing means the load on the rear end will now be dynamically changing with speed, while the front may not. For most vehicles, this means the camber on the rear wheels will also be changing with speed. If the car is still running the same rear camber it used to, then it will be too far positive relative to the previous setup. If the previous setup was optimized, this one will obviously not be. Additionally, if tires, springs, sway bars, or another modification has been made to add grip to the front and the front camber and caster are not changed, you'll run into a similiar problem there with the static settings.


As you say, these effects can make the car actually slower when there is only a small increase in grip. In cases where the increase is large, these effects may be masked. In either case however, these issues are costing time and should be addressed.

 

sorry to say this was a 40 years old idea that was banned right away in all top professional racing league, due to high maintenance and error failure of the system, many teams can't figure out how to tune their suspension to work with this design, at the end it was just an old concept that did not work with any race car. Glad to see someone from MIT to bring this back after 40 years ! I hope it will work this time.