SAS all good points you make. I believe there is enough evidence to suggest Dave was startled by the first failure and probably did not react correctly to the second one which I recall was about 32 seconds after the first engine fails.
I take your point about being visual for an EOL and the appropriate use of the landing light. However, I am now convinced that if you can teach a guy to rely on the RADALT and manage the flare by numbers he has a ready ,made solution for Night, Over water and IMC EOLs. It works well if it is well taught in the FSTD.
In regards the Landing lights, I have long accepted that without extremely versatile thumbs on the Landing Light coolly hat, its use in an EOL is minimised by the changes in nose pitch angles during the flare and level off. Even with plenty of practice, trying to add the value at the critical stage proves very difficult. Incidentally, watching an experienced pilot complete a night CAT A rotation leads to some interesting attempt to manage the landing light.
Taken holistically, building a procedure on the RADALT and AI, especially when the RADALT information is on the AI as in the EC modern EFIS cockpits, provides a simple set of numbers for the pilot to remember and build upon.
What is everyone else doing in the FSTD in this respect?