And this is the challenge for all those involved in the development of flight models. It feels like more of a craft than a science now.
Some of the behaviours we want to calibrate typically don’t have any IRL data to support them - like how much a pilot would expect to deflect their rudder in X amount of crosswind. Pilots can rarely give clear answers and manufacturers don’t publish this sort of data either.
We can’t overcome that. But we could at least develop a standard set of flight model testing protocols and benchmarks and set some industry standards for FDE development.
This is going to become essential come MSFS2024 when I imagine CFD will become the norm.