I seem to recall a proposal by Drexler for a "brute force" route to AI
via simulation. He had some estimate for the total number of compute
cycles it would take to simulate the entire history of life on earth.
You wouldn't try to actually replicate our history, but you would
be in effect simulating an alternate history where life might evolve
differently. The point was, supposedly using nanotech computers you could
in a reasonable time evolve AI's which would be as intelligent as humans.
This would imply that by running it a bit longer than the first appearance
of intelligence, you could simulate a whole history for that AI species:
its empires, its revolutions, its artists and engineers. They'd have
their own "library at alexandria", and many other tragedies and triumphs.
Each simulation would be as rich and significant as our own human history.
Once you've done this once, you could do it again. And again, and again.
Each time, with a few months or years of nanotech simulation (or a
few more cubic meters of nanotech computers), you have created as much
complexity and interesting information as the entire history of mankind.
Plus, you have access to it at a far more detailed level. Nothing will
be lost.
If this is really possible (I'm not sure the numbers work) then it
suggests to me that the loss of information caused by the destruction of
some portion of the human race is insignificant compared to the amount
which will be created and manipulated routinely by a nanotech culture.
>From that perspective, the loss of Alexandria or the loss of the United
States would both be no more important in practice than the loss of
structure when a bird plucks a worm from your lawn.
Hal