Re: Novas
Robin Hanson (hanson@hss.caltech.edu)
Thu, 17 Oct 96 11:09:20 PDT
Anders Sandberg writes:
>On Wed, 16 Oct 1996 N.BOSTROM@lse.ac.uk wrote:
>> It is important that we use realistic assumptions about the
>> conditions of cosmic warfare in our simulations. Since there
>> are still many unsolved problems in fundamental physics, I
>> suggest we set up several simulations with different
>> parameters for weapon technology. There are several relevant
>> variables here, such as projectile speed, production cost,
>> identifiability of aggressor, technology level required for
>> its manufacture, discrimination capabilities ("hurt only the
>> bad guys"), etc. Maybe the time is ripe to begin to try to
>> really work this out.
>
>Good points. Personally I think this kind of hyperweapons we discuss
>above are very unlikely - they are too indiscriminate and require a lot
>of energy/work. "Perversion attacks" a la Vinge are more likely -
>infiltrate the enemy with nanites or AI viruses, and then destroy him
>subtly. One could imagine tiny berzerkers (gremlins?) that move around a
>la von Neumann probes, and when they find a civilization they study it,
>send out very subtle agents, and then plans its downfall using SI.
This perversion strategy requires more confidence that you are in fact
much more advanced than those you attack. Imagine you see signals
from what looks like a primitive civilization 100 light years away.
Whatever you send to destroy it will face a civilization 200 years
more advanced than what you saw. If you send a fast simple bomb (and
have it turn a corner), they may end up advancing enough to stop it or
survive it, but you may feel confident that they probably can't find
you from it. A perversion probe will be lots more complicated, and so
risk telling more about you, and will have be much more advanced than
they are if it is to succeed.
If your probe fails and reveals you, a probe from them to destroy you
may be coming at near light speed from them to you, so you really want
fast feedback about whether your attack succeeded or not. A
perversion probe will take longer, increasing your anxiety that what
looks like a slow progress of your attack really masks a return attack.
A bomb either works or it doesn't real fast.
Robin D. Hanson hanson@hss.caltech.edu http://hss.caltech.edu/~hanson/