I keep thinking back to a 73 box which had a not at all convincing rudder response model at low speed ...
Then, as a couple of customers were FAA AOC airlines, the FAA dictated that the model be updated for the rudder hardover research .. I was the slowest getting away the night the boffins wanted to do some box testing so I was the bunny .. strange .. the box went from being totally unrealistic to very convincing ..
I am long convinced that a significant problem is computing power in the earlier boxes and where the tweaking efforts are concentrated .. but I stand willing to be corrected by those who have had far more experience in sim design etc., work ...