Subj : Re: Incomputable
To   : comp.programming,comp.object
From : Chris Sonnack
Date : Wed Aug 17 2005 07:59 pm

Dmitry A. Kazakov writes:

>> The burning question is, what will happen when we turn it on?
>> Will we have a *mind* or just a big neural net.
> 
> A perceptron? (:-))

[grin]  You ever hear the one about how demos emit natural bogon
particles?  They cause equipment to fail.


>>> Purely fictitious, let you can compute random distributions, rather
>>> than their realizations (the only thing we can do now), then this
>>> class of computing will be incomputable for any Turing machine.
>> 
>> Couldn't you do that now by using a true random (hardware) source?
> 
> Not really. The simplest thing: you cannot take a random generator
> multiply it by another generator and get a third one - a product of.
> You can multiply realizations, but that won't give you a new
> independent generator.  The soul is gone... (:-))

Nice way to put it!

But can't you use two hardware sources?  (Maybe I just don't follow.)

I mean the thing about Q/C is that it wants to use the hardware in a
way that we work very hard to prevent it from working now!  I'm just
not seeing the difference between a Q/M-driven source of random input
(such as a semi-conductor junction) processed conventionally, and
doing it all with Q/C.

-- 
|_ CJSonnack <Chris@Sonnack.com> _____________| How's my programming? |
|_ http://www.Sonnack.com/ ___________________| Call: 1-800-DEV-NULL  |
|_____________________________________________|_______________________|

.