[net.space] Tipler's argument

REM@MIT-MC@sri-unix (10/20/82)

From: Robert Elton Maas <REM at MIT-MC>
I categoricaly reject your claim that if one builds self-replicating
robots and sets them loose on the Universe, all of them carefully
programed and checksummed etc., that you'll inhibit mutation/evolution
in these creatures. If anything can go wrong it will! Haven't you
learned that? There's a cost associated with error-correcting codes.
So if some machine manages to turn off its error-correcting code via
something breaking (a bit switched, a JSR instruction changed to a NOP
by accident, whatever) it'll have a slight advantage in surviving. So
the population of non-error-protected machines will grow slightly. One
of them will suffer a more serious mutation, that offers a true
advantage, and suddenly a popuolation explosion of mutants will take
over the galaxy. I don't believe the pressure of evolution can be
stopped by a one-time careful design of a replicating form. *any*
replicating form set loose in a big enough environment will eventually
evolve. Even if you have police robots going around checking each
robot to verify correct operatin of checksum codes, there are enough
places to hide in the galaxy that evolution can escape detection long
enough to take over.
<opinion of REM>