Xref: utzoo comp.ai:5540 talk.philosophy.misc:3439 sci.philosophy.tech:1920
Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!wuarchive!brutus.cs.uiuc.edu!apple!sun-barr!newstop!texsun!convex!muse!cash
From: cash@muse.uucp (Peter Cash)
Newsgroups: comp.ai,talk.philosophy.misc,sci.philosophy.tech
Subject: Re: Algorithms, Turing, Semantics
Message-ID: <4593@convex.UUCP>
Date: 15 Jan 90 03:48:43 GMT
References: <12883@phoenix.Princeton.EDU> <91Eq02wy7eX=01@amdahl.uts.amdahl.com>
Sender: news@convex.UUCP
Reply-To: cash@convex.COM (Peter Cash)
Organization: Convex Computer Corporation, Richardson, Tx.
Lines: 62

In article <91Eq02wy7eX=01@amdahl.uts.amdahl.com> kp@amdahl.uts.amdahl.com (Ken Presting) writes:
>In article <12883@phoenix.Princeton.EDU> kadickey@phoenix.Princeton.EDU (Kent Andrew Dickey) writes:
>>
>>I clearly must be missing something very important to all of these
>>discussions.  Therefore, before I spew out my version of the world, I
>>have three questions:
>>
>>1) Could someone name one process which cannot be broken down into an
>>algorithm?  That is, many people have said the mind may not be simulated
>>through an algorith, but personally, no other examples of
>>non-alogorithmical processes occur to me.  Please, someone, tell me what
>>obvious example I'm missing.
>
>There has been a discussion here lately as to whether all processes are
>algorithmic, and some controversy perhaps remains.  Algorithms are finite
>sets of rules determining transitions among a finite set of states.
>Physical systems usually have an infinite number of states, so any
>algorithm can at best approximate the behavior of a physical system. In
>a physical system with unstable components (where a small perturbation
>can produce large changes) the approximation may never be good enough.

You are, of course, operating with a very narrow definition of what an
"algorithm" is (Turing's definition).  But to me, it sounds very odd to say
that an algorithm can "approximate" a physical system; I might concede that
a *model* can--in some strict sense--"approximate" a physical system, but an
algorithm, never.  An algorithm is a formal statement of rules (e.g., the
rules used to solve a problem, or construct a model, etc.), it is not the
thing itself.

For example, a topographical map can be said to correspond to the terrain
(I can't bring myself to say that it "approximates" the terrain).  But by
no stretch of the imagination do the cartographers "algorithms" (his
procedures and methods), correspond to the terrain.  

It seems to me that this is a confusion that arises out of conflating the
"states" of a Turing machine with the "states" of a physical system.  They
have something in common only in the most abstract way.  To speak of them
in the same breath begs the question of artificial intelligence.  If we
think that people are in some sense "systems" that have "states" like
Turing machines, then the outcome has already been decided:  people can be
"approximated" by algorithms.  Thus, the really important point that must
be decided first is in what sense (if any) people are like Turing machines.

>...[lots of formal stuff about semantics deleted]...

>I've ignored issues of consciousness, intentions, and lots of others,
>because with even this simple type of semantics, lots of really
>horrible problems can be productively studied.

But you imply that there *is* a connection between the concerns of the
formal semanticist and "consciousness, intentions, and lots of others".
This connection is what is really interesting and important here, and this
connection is what needs to be demonstrated.  After all, much of the
interest in semantics would wane if people didn't assume that there is a
connection.  But is this more than an unfounded assumption?  If you think
that this assumption has a foundation, I would like to know what it is.


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
             |      Die Welt ist alles, was Zerfall ist.     |
Peter Cash   |       (apologies to Ludwig Wittgenstein)      |    cash@convex
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~