From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!news-server.csri.toronto.edu!neat.cs.toronto.edu!cbo Tue Mar 24 09:58:15 EST 1992
Article 4685 of comp.ai.philosophy:
Xref: newshub.ccs.yorku.ca sci.philosophy.tech:2410 comp.ai.philosophy:4685
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!news-server.csri.toronto.edu!neat.cs.toronto.edu!cbo
Newsgroups: sci.philosophy.tech,comp.ai.philosophy
>From: cbo@cs.toronto.edu (Calvin Bruce Ostrum)
Subject: Re: A rock implements every FSA
Message-ID: <92Mar24.055613est.14362@neat.cs.toronto.edu>
Organization: Department of Computer Science, University of Toronto
References: <92Mar18.182726est.14357@neat.cs.toronto.edu> <1992Mar19.000544.22634@bronze.ucs.indiana.edu> <92Mar23.003224est.14362@neat.cs.toronto.edu> <1992Mar24.025128.9379@bronze.ucs.indiana.edu> <1992Mar24.042009.12510@organpipe.uug.arizona.edu>
Date: 24 Mar 92 10:56:42 GMT
Lines: 82

Bill Skaggs writes:
bs|   This is not really an original point -- Putnam says the same
bs| thing, I think -- but bringing counterfactuals into the
bs| picture only muddies it.  The problem, as Hofstadter very
bs| adroitly shows in "G\"odel, Escher, Bach" and "Metamagical
bs| Themas", is that making the condition (the "If not X" part)
bs| of a counterfactual true requires changing some aspect of
bs| the world, but it is often not obvious what aspect or aspects
bs| to change.  As Hofstadter puts it, it is not obvious which
bs| things are "slippable" and which are not.

This problem has been addressed by a very large number of philosophers
in a very large number of places. It is closely related to the problem
I mentioned in my last post, which is that it is not clear *which* 
counterfactuals have to be true in order to make some statement about,
for example, intelligence, true as well.

The simplest criterion generally advanced for deciding whether a
counterfactual is true is the Ramsey test, rendered roughly as: "if A were
true, then B would be true" is true in world W just in case B is true
in all possible worlds which are as similar as possible to W given that
A is true in each of these worlds.  The problem with this test is that
no-one has ever specified an adequately general similarity metric for
all possible worlds.

In fact, Putnam does complain very strongly about this problem.  He
claims that it renders physicalism a completely hopeless doctrine, and
as things currently stand, I agree with him.  A paper where he makes
this point quite vividly is "Why there isn't a ready made world", in
which he complains:

hp| 'It's all physics, except that there's this similarity metric', just
hp|  doesnt make *sense*.

However, as I pointed out in my last post, for the case of a reasonably
isolated automata, there is a pretty obvious class of counterfactuals
to consider, and it is also pretty easy to see what would make these
counterfactuals true.  For the purposes at hand, it seems quite easy to
imagine the closest possible world to ours in which the automaton was
in a different state, and received a different input symbol, because 
these are close enough to separable, isolated facts which can be changed
pointwise without disturbing any other facts. 

bs|   Consider the aphorism, "If wishes were horses, beggars
bs| would ride."  A clever aphorism, no doubt, but is it
bs| true?  Is is *objectively* true? 

What it is, is out of date!  Let me provide you with the up-to-date
version courtesty of Jerry Fodor's grandmother:

jf| "If wishes weren't causally isolated from horses, then beggars would
jf| ride ceteris paribus" as Granny is also always saying.

Silliness aside, it's hardly fair to cite a figurative aphorism as
being an impediment to a theory of counterfactuals.  More importantly,
it is hardly fair to insist that we have a complete theory of 
counterfactuals before we attempt to use them to provide a theory
of something else.  Still more importantly, if we have decided, for
whatever reason, that an account in terms of counterfactuals is required,
then the fact that counterfactuals "muddy the picture" is no excuse for
remaining with the clear one.  This is exactly the case of the drunkard
searching for his keys under the streetlight.

bs| If some well-formed counterfactuals
bs| are capable of being objectively true and others are not,
bs| how do we tell the difference?  

If we have the similarity metric, the answer is easy.  In lieu of that, 
we have some intuitions about the similarity metric which seem to be 
pretty reasonable in lots of cases.  I would include the cases needed  
in our current discussion in this class.

What makes these intuitions reasonable and/or reliable?  What exactly is
a similarity metric and where does it come from?  Good questions all.

---------------------------------------------------------------------------
Calvin Ostrum                                            cbo@cs.toronto.edu
---------------------------------------------------------------------------
One further suggestion: if you undertake the task of philosophical
detection, drop the dangerous little catch phrase which advises you to
keep an "open mind".   -- Ayn Rand
---------------------------------------------------------------------------


