One of my very favorite composers is Andrzej Panufnik (1914-1991). When Panufnik had to leave Poland because of the Warsaw Uprising in 1944, he left behind all of his manuscripts. When he returned, he found that they had all been destroyed.
Some of that work is lost forever, but Panufnik did something extraordinary. He reconstructed some of this work from memory. Notable among the pieces he reconstructed was his Tragic Overture (1942, lost 1944, reconstructed 1945, revised 1955). I own the score to this work and love it deeply. I can tell you it must have been a bear to rewrite, as there are heavy, complex harmonies and a lot of activity. We will never know for certain how accurate the reconstruction was, or even how accurate it was intended to be. But the thought of even making an attempt must have been daunting.
So now I want you composers to imagine one of your best, big orchestral works suddenly vanishing. All printed copies, computer files, and recordings are destroyed. Would you attempt to reconstruct it? How close to the original do you think you could make it, and would accuracy be important to you?
Personally, I don't think I would reconstruct anything, as tragic as the loss would feel to me. First of all, I don't think I could do it very well, even after hearing my own pieces hundreds of times. But I also think that the amount of effort it would take could be used to compose two or three new works in the same time. I'd probably use some of the more memorable ideas and motives if I could find a natural home for them in a new work, but accuracy would not be an issue in this case.
What's your take?