From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!csd.unb.ca!morgan.ucs.mun.ca!nstn.ns.ca!bonnie.concordia.ca!thunder.mcrcim.mcgill.edu!snorkelwacker.mit.edu!spool.mu.edu!mips!think.com!zaphod.mps.ohio-state.edu!uwm.edu!lina Tue Apr  7 23:23:58 EDT 1992
Article 4900 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!csd.unb.ca!morgan.ucs.mun.ca!nstn.ns.ca!bonnie.concordia.ca!thunder.mcrcim.mcgill.edu!snorkelwacker.mit.edu!spool.mu.edu!mips!think.com!zaphod.mps.ohio-state.edu!uwm.edu!lina
c!uchinews!spssig.spss.com!markrose
>From: markrose@spss.com (Mark Rosenfelder)
Newsgroups: comp.ai.philosophy
Subject: Re: Rock/FSA -> Humongous LUT
Message-ID: <1992Apr03.164000.7009@spss.com>
Date: 3 Apr 92 16:40:00 GMT
References: <1992Apr1.132848.9773@cs.ucf.edu>
Organization: SPSS Inc.
Lines: 33
Nntp-Posting-Host: spssrs7.spss.com

In article <1992Apr1.132848.9773@cs.ucf.edu> clarke@acme.ucf.edu (Thomas Clarke) writes:
>Presumably the rock would pass a (bounded) Turing test, albeit with the  
>assistance of a lab full of equipment for determining the rock's internal  
>state. Placing the rock in front of a terminal CRT displaying the message "How 
>are you?"would produce a unique rock state due to optical radiation emitted by 
>the screen's phosphors etc. The equipment would then determine this state and  
>determine that the proper output would be the key presses "Fine. And  
>you?<return>".  The determination of the output must be done automatically,  
>free of human intelligence, however, to avoid cheating.  A sure way to  
>determine the output would be a look-up table (invoke quantum mechanics at  
>finite temperature to argue a finite number of rock states).  An algorithmic  
>approach could also be used, but I couldn't implement it, even in principle.  

Attempting to use the rock in so rigorous a fashion seems to miss the point
that (by hypothesis), the rock implements *any* FSA.  It's not a
matter of sneaking up on the rock and finding out what it's thinking.
It's a matter of a priori choosing an FSA, any FSA, and mapping its states 
onto the rock.  Actually attempting to read the "states" of the rock is
redundant.  If you find that the rock is in state X, it's because you
set the mapping up that way.

I can't say I understand why anyone would want to retain a definition of 
"implementation" which is this loose.  Wouldn't a definition in which
implementations of two different FSAs could be distinguished
be a bit more useful?

>[...]
>The rock in front of the CRT does nothing.  The rock plus lab plus humongous  
>look-up table passes the Turing test.  The intelligence is not solely in the  
>rock.  

That's an understatement, since in this case the lab plus humongous look-up
table *minus* the rock passes the Turing test, too.


