For debate, using the example of restrictions on raw data flying, this could create more problems than those supposed to be avoided.
The use of computed guidance (FD/AP) for an approach in high crosswinds reduces opportunity for pilots to gain knowledge of the drift angles involved and thus anticipate the manoeuvre for aligning the aircraft during a manual landing - the feel of the aircraft in higher crosswinds. Although pilots could make a mental note of the drift during a FD approach, but without consciously paying attention to this, thinking / flying, adding tacit knowledge - experience, pilots could be less skilled, less experienced.
Operators should be careful of introducing 'novel' safety interventions without considering side effects, which implies that the reasoning for raw data skills practice has to be understood by management.
This aspect runs throughout the thread; where is the need. Raw data flying can be fun, it can be encouraged, but is it really required; nice to have or must have - refer to bean counters.
In the example above, operators might encourage manual flight with computed guidance, but require a diversion for unaided approaches (failure case) in high crosswinds - safe without introducing problems.
But then the bean counters dislike diversions - but how much dislike in comparison with an accident. They lack experience in contemplating side effects, or an accident.