I think this is the wrong take, for the following reasons: - There is no reason to assume that the learnings from the 50 crashes weren't also applied to the ne

I think this is the wrong take, for the following reasons: - There is no reason to assume that the learnings from the 50 crashes weren't also applied to the newer model. In fact you'd expect that they all were. - Faults in a new design are likely to be front-loaded, meaning most of the crashes would have happened earlier than later. Therefore the new model seems to be a much safer design if it flew 10% of the miles without even 10% of the crashes (actually 0%).