[gurps] Failing autopilots and RVO

[Knapp]

> If your Trade Federation is building 100,000 killer droids, is each one
> going to be trained individually, or do you think it's a heck of a lot more
> likely each one will have THE SAME initial database of "pattern libraries,
> shape, group status, move history, killer moves, etc."

I agree and it is the same with training humans. I also suspect that
future killer droids will have DNA based "personalities". You would in
a very complex simulation make the best you can and then have huge
Alife wars. The top 10,000 programs that come out would be your new
fighters. Of course to get to this point would be a big jump from what
is done or needed in the present. This is speculation on my part.

> That database is the program.

No it is not. The database is the data, the program is the method to
use and interpolate the data. Also out in the real world the input
will never be exactly the same. This is not like a game world.

> And even more likely, that the selection critera will tend to cause "the
> program" to converge on similar tactics over time, or at least wander
> between common "attractors", no matter how much training it gets.

Same as humans.

> Perhaps I did phrase my rating of an LAI inappropriately.  The software in
> your example is still limited to modifying numbers within its framework, it
> is not rewriting the framework itself.

Again, same as humans. Only the researcher rethink and change things
but for very rare instances. A US GI grunt does not change the average
tactics of the whole army. They also vary between "attractors". On a
whole you can predict what the US Army will do but that does not mean
you will win, if they are doing the right thing most of the time.

> --
> Zan Lynx

-- 
Douglas E Knapp

Why do we live?