Localiser Green,
You say that:
It is pretty fair to say that the chance of both engines on a twin failing at the same time for mechanical reasons is effecitvely zero. If the chance of even a single engine failure is 1 in a million, then if I remember correctly from doing probability at school, the chance that both engines will fail at the same time (for purely mechanical reasons) is 1 in a million multiplied by 1 in a million, which is 1 in 100,000 billion.
I have to say that I'm less than convinced by your 'O' level probability theories, what if the engine fails due to a faulty component? I would suggest that the probability that the other engine has the same part fitted is somewhat better than one in one hundred thousand billion.
So yes, I do think your argument is flawed.