> From: [EMAIL PROTECTED]
> 
> > Nope.  You are comparing apples and orangutangs.  See my other post
for
> > why there would be no 'selection' taking place.  I am sure there is a
> > system that could be created the way you explain here, but that
system
> > would not be as good as the system I describe.  In some ways it could
be
> > better, by orders of magnitude, but my system would be error free.  
> 

> I see your point about apples and orangutangs and I agree. But I think
that your system is less likely to be developed precisely because it will
have no techniques for innovation. Selection is the best way to search
for new ideas since it is an iterative process that explores "idea space"
(to paraphrase Dennett). 

Not really.  Is _Natural selection_ more powerful than _Artificial
selection_?  No.  Is _Artificial selection_ more powerful than _Sexual
Selection_?  No.

Is a random search better than a directed search?  Occasionally you might
get lucky, but for the rest of 99.999999999999999999999999999999999% of
the time, No.
 
We aren't talking about a stupid machine, we are talking about an
Artificial _Intelligence_.  Not only that, it is given all of the
knowledge of mankind up to that point, all of the Axioms, and their
proofs, and the capability to do research (a 'directed search' if you
will).  

(How do you think the machine societies in the Upliftiverse and the
Trekiverse got started?)

> As to the specifics of your specs for the system: My only question (and
one which I am inadequate to answer) has to do with Goedel's theorem.
Would not your system run into questions for which there are no logical
answers and would that not throw it into one of those Star Trek "Does not
compute" loops with eventual self-destruction associated with tacky smoke
special effects? That is there are some problems for which there are no
mathematical or logical solutions. So the machine could not achieve the
sort of perfection that you describe. 

Their are known examples like the derivative of sin x^2.  But most of
these problems are apparent, and of only little value.  Any problems that
it would come across that it could not solve, it would deal with in other
ways.  I doubt it would into those kinds of problems for the four most
important aspects of the system: The compiler, the 'program writer', the
operating system, and the BIOS/firmware.  Less functional (important)
parts of the system would be more likely to run into those kinds of
issues.


 
> > FYI all this discussion is not really off topic.  The uplift universe
has
> > machine intelligence's (computer civilizations and just ordinary
> > computers).  And the book _Earth_ had a lot to do with stuff too.

Reply via email to