Alf,
Given that we agree that good NTS are a positive thing, and I think we agree that the best environment to examine both our own and other people's NTS are in the controlled environment of the simulator.
But - that is surely not to say that accident investigations and LOSA are of no value because they record outcomes rather than underlying reasons.
Clearly not in all, but in many cases there is sufficient evidence to infer underlying reasons (whether these are down to NTS or wider-CRM issues, such as management, supervision, regulation etc) If this can spur us to re-examine the way we do things, it can surely only help to improve the safety of our operations in the future?