“ … problems with raw data with the FD's on can easily be taught and emphasized in training …” (#12)
One of the problems with assumptions (as above) is that they can be wrong considering the range of human behaviours and situations which might be encountered in operations, e.g. FD fixation.
IIRC there was a regional jet fatal accident in France due over focus on the FD during the approach.
Where a high integrity auto pilot is being used, switching off the FD does not downgrade the aircraft, it still has the same capability. Furthermore it could be argued that the less complicated monitoring scan pattern – focussing on the important parameters, adds safety capability to the operation as there is less opportunity for error.